0% ont trouvé ce document utile (0 vote)

288 vues653 pages

Cramsg 2015

Q: What role do unsaturated soil mechanics play in the development of housing on expansive soils in Africa, and how have they evolved?

Unsaturated soil mechanics play a crucial role in developing housing on expansive soils by providing frameworks to understand and predict the behavior of these challenging soil conditions, which are prevalent in Africa . They have evolved with the development of theories such as those by Fredlund et al., which use net normal stress and matric suction to predict shear strength and understand moisture interactions . Despite advancements, practical application is hindered by resource limitations for thorough parameter testing, driving the use of empirical methods adapted to the local context and resource constraints .

Q: How have recent developments in geotechnical practices affected the assessment of unsaturated soils?

Recent developments in geotechnical practices, such as the application of methods like micro-porous membrane technology, have brought improvements in the measurement of soil-water characteristic curves, aiding in the assessment of water retention and shear strength in unsaturated soils . These advancements have enabled more accurate predictions of soil behavior under varying moisture conditions, providing insights into the matric suction effects and making it feasible to apply more sophisticated models. Such progress enhances the ability to predict and manage soil-related risks in engineering projects, improving the design and stability of structures on unsaturated soils .

Q: Identify the challenges faced in implementing advanced soil mechanics models for expansive clay across Africa.

Implementing advanced soil mechanics models for expansive clay in Africa is challenged by the high costs and extensive time needed for parameter testing, which limits their practicality for economic housing projects . Although theoretically robust methods exist, the scarcity of resources and financial limitations impede their use. The empirical methods tailored for road-building are often inappropriate for housing foundations, necessitating a need for more tailored procedures that are economical and effective for expansive soils . Additionally, a general lack of cooperation among professionals and an over-reliance on empirical methods hamper innovations in soil mechanics, which are essential for addressing the heaving problems caused by expansive clays .

Q: Analyze the interaction between socio-political factors and geotechnical engineering practices in African countries.

The interaction between socio-political factors and geotechnical engineering practices in African countries is complex; political influences, such as budget allocation and policy decisions, greatly affect the execution of geotechnical projects . The inefficiency in road maintenance policies and inadequate financing mean that engineers may face constraints in applying optimal designs or new materials. Political decisions shape priorities that often overlook long-term infrastructure needs in favor of immediate economic gains, and entrenched practices resist the adoption of new methodologies and technologies critical for advancement in construction practices .

Q: What challenges are faced by geotechnical engineering professionals in African countries regarding the use of local materials for road construction?

Geotechnical engineers in African countries face challenges in using local materials for road construction due to insufficient geotechnical characteristics of these materials, such as low California Bearing Ratio (CBR) and inappropriate plasticity indices which hinder their use as base layers in roadways . Additionally, the use of innovative materials is limited by a lack of characterization and an ingrained reliance on imported materials, which are often chosen for their premium quality despite higher costs. The financial and practical constraints support the call for creative use of known local materials to reduce construction costs while maintaining safety standards . The ongoing economic challenges and lack of government support further exacerbate the situation .

Q: Evaluate the effectiveness of current testing methods for the identification of expansive soils and propose improvements based on recent advancements.

Current testing methods for identifying expansive soils, heavily reliant on Atterberg limits and empirical data, often fail to provide reliable predictions for heave potential . These methods are more suited for road construction, lacking specificity for housing foundation issues . Improvements can be made by incorporating recent advancements in soil mechanics, such as more precise soil-water retention curve measurements and matric suction assessments using newer technologies like micro-porous membrane applications . Such improvements could result in more accurate evaluations of expansive soils, enabling better preventive engineering designs to mitigate related risks .

Q: Discuss the significance of collapse potential in soils and how it is determined in geotechnical practice.

Collapse potential in soils is significant because it affects the stability and safety of structures built on such foundations. It refers to the sudden collapse of soil structure when it becomes saturated, leading to settlement issues . Determination of collapse potential involves identifying soil characteristics, such as silt and clay content, and applying tests like the plate load test for assessing collapsibility . Several criteria exist for predicting collapse potential, and these tests help in evaluating soil behavior under specific moisture conditions crucial for infrastructure safety .

Q: What role does effective stress play in the assessment of shear strength in unsaturated soils, and how is it calculated?

Effective stress plays a critical role in assessing the shear strength of unsaturated soils by accounting for the pore air pressure and matric suction components. It is calculated using the equation: σ' = σn-ua +χ(ua-uw) where σ' is the effective stress, σn-ua is the net normal stress, ua is the pore air pressure, uw is the pore water pressure, and χ is a parameter related to soil saturation . The shear strength of unsaturated soil further involves parameters such as soil cohesion, internal friction angle, and matric suction, which are used to derive the shear strength using the equation: τf = c'+ (σn-ua)f tan ϕ’ +(ua-uw)f tan ϕb .

Q: What are the socio-economic impacts of inadequate geotechnical education on infrastructure development in Africa?

Inadequate geotechnical education impacts socio-economic development in Africa by limiting the availability of skilled professionals to develop necessary infrastructure, exacerbating challenges in projects that require specialized knowledge, such as those involving complex soil conditions . The shortage of trained geotechnical engineers leads to reliance on outdated methods and prevents the adoption of innovative solutions, affecting the quality and longevity of infrastructure projects. This educational gap contributes to inefficiencies in road and building construction, impacting economic growth negatively by increasing both construction costs and failure rates .

Q: Explain the implications of the economic crises of the late 1980s on geotechnical engineering education in Africa.

The economic crises of the late 1980s had significant implications on geotechnical engineering education in Africa, particularly due to structural adjustment policies imposed by the International Monetary Fund (FMI) that affected investments in infrastructure and education . These policies led to reduced financial support for engineering programs and hindered the renewal of teaching staff, which is now aging, thereby increasing the necessity for educational reform and training to maintain current levels of expertise. These changes have resulted in a decrease in specialized knowledge being passed to new students and a shortage of adequately trained geotechnical engineers, as evidenced by the low ratio of one geotechnical engineer per 20 civil engineers trained .

Transféré par

AhmedJardak

Nous prenons très au sérieux les droits relatifs au contenu. Si vous pensez qu’il s’agit de votre contenu, signalez une atteinte au droit d’auteur ici.

Formats disponibles

Téléchargez aux formats PDF, TXT ou lisez en ligne sur Scribd

0% ont trouvé ce document utile (0 vote)

288 vues653 pages

Cramsg 2015

Transféré par

AhmedJardak

Nous prenons très au sérieux les droits relatifs au contenu. Si vous pensez qu’il s’agit de votre contenu, signalez une atteinte au droit d’auteur ici.

Formats disponibles

Téléchargez aux formats PDF, TXT ou lisez en ligne sur Scribd

Comptes Rendus du 16ème Congrès Régional Africain de la Mécanique des

Sols et de la Géotechnique

LA GÉOTECHNIQUE INNOVANTE POUR L’AFRIQUE

Proceedings of the 16thAfrican Regional Conference on Soil Mechanics and

Geotechnical Engineering

INNOVATIVE GEOTECHNICS FOR AFRICA

TUNISIE/TUNISIA 27-30 AVRIL/APRIL 2015

EDITEURS / EDITORS:
Mounir BOUASSIDA
Mehrez KHEMAKHEM
Slah Eddine HAFFOUDHI
ii
Président du Congrès/ Chairman of the Conference: Slah Eddine HAFFOUDHI

Comitéd’organisation/Organizing committee

Président/ Chairman: Mehrez KHEMAKHEM

Membres/ Members
Samia BOUSSETTA
Imen SAID
Faten SAIHI
Wissem FRIKHA
Kamel ZAGHOUANI
Essaieb HAMDI
Moncef MAKNI
Zouheir BOUARADA

Comité Scientifique/Scientific Committee

Président/ Chairman : Mounir BOUASSIDA

Membres/ Members
A. ABBO (Australie/Australia) B. JELLALI (Tunisie/Tunisia)
H. ABDELMOHSEN (Egypte/Egypt) A. LOULIZI (Tunisie/Tunisia)
S. AMPADU (Ghana) J. NEJI (Tunisie/Tunisia)
R. BAHAR (Algérie/Algeria) S. LEPPLA (Allemagne/Germany)
M. BOUAZZA (Australie/Australia) M. LISYUK (Russie/Russia)
M. DAFALLAH (Arabie Saoudite/SaudiArabia) B. MCCABE (Irlande/ Irland)
N. DENIES (Belgique/Belgium) M. A. SAKR (Egypte/Egypt)
E. HAMDI (Tunisie/Tunisia) T.G. SITHARAM (Inde/India)
S.W. JACOBSZ (Afrique du Sud/South Africa) A. STUEDLEIN (USA)
D. KALUMBA (Afrique du Sud/South Africa) H. TUMI (Libye/Libya)
M. KANA (Cameroun) E. VORSTER (Afrique du Sud/South Africa)
M. KARRAY (Canada) M. ZERHOUNI (France)

Listedes Sponsors/List of Sponsors:

Société Pays Site Web Société Pays Site Web

APAGEO Tunisie Tunisie [Link] HUESKER Germany [Link]

HYDROSOL FONDATIONS Tunisie [Link] SEDIDRILL France [Link]
AFRICSOIL Tunisie - JEAN-LUTZ France [Link]
SIMPRO Tunisie [Link] TENCATE Netherlands [Link]
TERRASOL TUNISIE Tunisie [Link] TENSAR France [Link]
SOLS-MESURES France [Link] AFITEX-ALGERIE Algérie [Link]
GEOKON USA [Link]/ NAUE Germany [Link]
PAJOT France [Link]/ GLOTZL France [Link]
TECHNOLABO Tunisie [Link] TEREDO Italy [Link]
GEOCONSEIL Tunisie [Link]/ TEXINOV France [Link]
GEOSENSE LTD England [Link] CBAO France [Link]
ITECH France [Link] MACCAFERRI Italy [Link]
TEC SYSTEM France [Link] SISGEO Italy [Link]
VECTRA France [Link] CONTROLAB France [Link]
GEOTOMOGRAPHIE Germany [Link] FONDASOL France [Link]
CETA Tunisie [Link] IN SITU England [Link]
LIM France [Link] LABOSOL Sénégal [Link]
PLATIPUS-ANCHORS England [Link] GGS Tunisie [Link]

iii
Table de matières

Table of contents

Préface/ Preface

CONFERENCES PLENIERES
KEYNOTE LECTURES
1. Pratique de la géotechnique routière en Afrique intertropicale 3
Etienne Marcelin KANA (Cameroun)
2. Development in deep excavation and deep foundation technologies 17
R. Katzenbach, S. Leppla, M. Seip (Germany)
3. Landslides in saturated and unsaturated soils 33
E. E. Alonso (Spain)
4. Normalisation en Géotechnique 61
M.I. Zerhouni (France)
5. Controlled Low Strength Materials (CLSM) with Native Clays for Geotechnical Civil Infrastructure: 69
Design, Challenges, and Sustainability Assessments
A. J. Puppala, T. V. Bheemasetti, D. H. Marshall (USA)
SESSION 1 : COMPORTEMENT DES SOLS 79
SESSION 1 : SOIL BEHAVIOUR
1. Analysis of the time-dependent deformation behavior of soft soils 81
R. Katzenbach, S. Leppla (Germany)
2. Determination of the modulus of deformation using static and dynamic plate load tests 89
R. Katzenbach, S. Leppla(Germany)
3. Effects of Matric Suction on the Shear Strength of Highly Plastic Compacted Clay 97
A.M. Elsharief, O.A. Abdulaziz (Sudan)
4. Comparison of shear strength characteristics of cohesive soils evaluated by the triaxial Mohr’s circle and 105
stress path methods
A.K.M. Zein (Sudan)
5. Saturation and Sample reconstitution Effects on the Mechanical Behavior of Chlef Sand in Northern 113
Algeria-An Experimental study
N. Della, M. Belkhatir, J. Canou, J.C. Dupla (Algeria, France)
6. Influence de l’initialisation des contraintes sur les résultats des calculs par la méthode des éléments finis. 121
M.D. Ngom, S. Burlon, N. Droniuc, J.P. Magnan (France)
7. Comportement sur chemin œdométrique à succion contrôlée de l’argile du barrage de Boughrara 129
(Tlemcen).
F.E.M. Derfouf, S. Taïbi, N. Abou-Bekr, M.A. Allal (Algérie)
8. Analyse de la réponse hydromécanique des sols compressibles saturés 137
A. Zadjaoui (Algérie)
9. Des Lois de Comportement des Géomatériaux Pertinentes pour les Calculs aux Eléments Finis 147
E. Bourgeois, D. Remaud (France)
10. Hydraulic Properties of South African clayey soils as Landfill barriers 155
F.N. Okonta, L.J. Moatlhodi, O.O. Ojuri, P. Meruti (South Africa, Nigeria)
11. Effect of gravel fractions on direct shear strength of sand 161
F.N. Okonta, L.J. Moatlhodi (South Africa)

iv
12. Assessment of clays by small-scale suction tests 167
P. Stott, E. Theron (South Africa)
13. Shortcomings in current methods of identification and assessment of expansive clays 173
P. Stott, E. Theron (South Africa)
SESSION 2 : CARACTERISATION DES SOLS 179
SESSION 2 : SOIL CHARACTERIZATION
1. Utilisation du CPT'u et du pressiomètre pour la qualification de la classe de sol et pour l'estimation de 181
Gmax.
J.C. Gress (France)
2. Undrained Shear Strength of a Libyan Marine Soil 189
S. Khatrush, G. El-Gehani (Libya)
3. Application of the nchrp-1-28 resilient modulus model for characterizing granular material in sudan 197
Omer G. Omer, A.M. Elsharief (Sudan)
4. Geotechnical properties of polymer stabilised backfill soils 205
F.N. Okonta; L.J. Moatlhodi; T. Makharamedzha, M.T. Nhlapho (South Africa)
5. Efficient Geophysical Survey for Geotechnical Site Investigation 211
F.S. Elburas, Bahia M. Ben Ghawar, M.T. Shakshuki (Libya)
6. Determination of Jet Grout Column Diameter Using Seismic Crosshole Methods 219
T. Fechner, S. Mackens, J.C.G. Guerreros, E. Niederleithinger (Germany)
7. An Experimental Study on the Identification and Classification of Collapse Potential of Some Soils in 225
Sudan
M.A. Osman, E.M. Ali Ahmed
SESSION 3 :FONDATIONS PROFONDES ET SUPERFICIELLES 231
SESSION 3 : DEEP & SHALLOW FOUNDATIONS
1. Optimisation and safety of urban geotechnical construction projects 233
R. Katzenbach, S. Leppla (Germany)
2. Intérêt des bases de données géotechniques dans l’étude des fondations 241
F. Debiche, R. MiticheKettab, F.A. Nechnech (Algérie)
3. Comportement d’un pieu isolé sous sollicitation axiale cyclique–Modélisation par éléments finis et 249
construction des boucles hystérétiques de frottement latéral
M. Khouaouci, A. Bouafia (Algérie)
4. Numerical Analysis of Shallow Foundations on Purely Cohesive Soil under Eccentric Loading 257
N.E.H. Khitas, A. Mabrouki, D. Benmeddour, M. Mellas (Algeria)
5. Comparaison entre la capacité portante in situ et théorique de différents pieux battus dans les terrains 265
alluvionnaires.
E. Kenmogne, E. Wemeyi, J.R. Martin (Cameroun)
6. Développement d’une loi t-z cyclique à partir d’essais en laboratoire et d’expérimentations en vraie 273
grandeur
Z. Abchir, S. Burlon, R. Frank, A. Le Kouby (France)
7. Deep foundations and underground structures – Tailor-made solutions for soft soil 281
M. Schoepf (Germany)
8. Investigation of the influence of vertical loads on the lateral response of pile foundations in clayey soils 289
L. Hazzar, M.N. Hussien, M. Karray (Canada)
9. Particle image velocimetry analysis on the sinking of shallow foundation in 2D 297
P. Pizette, N.E. Abriak (France)

v
10. Numerical Modeling of Piles in Expansive Soil 301
N. ElSakhawy, T. Nageeb, R. El Badawy (Egypt)
11. Consequences of Geotechnical Uncertainties in Predicting Settlement of Shallow Foundation on Sands 307
M. I. Lingwanda, S. Larsson, D.L. Nyaoro (Tanzania, Sweden)
12. Rehabilitation and underpinning of a building classified as a national patrimony 315
S. Boubaker (Tunisia)
13. Retaining Structures and Special Foundations at the Platinum Tower, in Maputo 323
A. Pinto, R. Tomásio, R. Coelho and R. Nicolas (Portugal)
SESSION 4: AMELIORATION DES SOLS 331
SESSION 4: GROUND IMPROVEMENT
1. Evaluation of Index and Strength Characteristics of Clay Soil Stabilized with Lime as Subgrade Material 333
in Jonglei, South Sudan
S. Jjuuko, P.B. Matio, D. Kalumba (South Africa, Uganda)
2. Reinforcement of Base Course Consist of Black Cotton Soil using Do-nou Technology 339
Y. Daito, Y. Fukubayashi, M. Kimura (Japan)
3. Analyse statique par éléments finis des fondations superficielles sur sol renforcé par inclusions rigides 345
N. Djeraf, A. Nechnech (Algérie)
4. Etude de la résistance au cisaillement d’une argile surconsolidée expansive traitée aux liants hydrauliques 351
A. Mahamedi, M. Khemissa (Algérie)
5. On the behaviour of columnar-reinforced foundations 359
M. Bouassida, M. Klai, S. Tabchouche, M. Mellas (Tunisie, Algérie)
6. Reduction of clay soil collapsibility using cement-polymeric fiber admixture 365
M.M. El-Emam, M.F. Attom, Z.H. Khan (United Arab Emirates)
7. Effect of Salinity on Hydrated Lime Stabilized Swelling Clays from Sudan 375
A.M. Elhassan, A.M. Elsharief (Sudan)
[Link]-encased columns (GEC) and local available mineral aggregate (e.g. sand) 379
An innovative and economic foundation system for embankment on soft soil
L. Sakou, J. Bruhier, R. Arab, R. Medjo, P. Assinder (Germany, France, Cameroun)
9. Use of Stone Columns for liquefaction mitigation 387
Z. Bensalem, W. Frikha, M. Bouassida (Tunisia)
10. Study on effect of Sisal fiber-reinforcement on Tensile strength of clay soil using newly developed 393
tensile strength measuring apparatus.
M. Chebbi, M. Jamei, H. Guiras (Tunisia)
11. Experimental study of Tunis soft soil improved by deep mixing column 399
W. Frikha, H. Zargayouna, S. Boussetta, M. Bouassida (Tunisia)
12. Experimental study for the mechanical characterization of Tunis soft soil reinforced by a group of sand 407
columns
W. Frikha, W. Kaffel, M. Bouassida (Tunisia)
13. Numerical analysis of embankments on soft clay improved by PVDs 415
M.A. Mansour, A.M. Samieh, M.H. Mahmoud (Egypt, Qatar)
14. Analysis of performance of preloading vertical drain system for improvement of soft clay soil 423
comparative study between 2D And 3D numerical techniques (Egypt)
M. H. Rabie, M. A. El Hamed
15. High strength earth blocks reinforced with natural and polymeric fibers 433
M.M. El-Emam, A.K. Al-Tamimi, A. Aljallad, M. Al-Moukdad, A. Al-Marei, and M. Noaman (United Arab
Emirates)

vi
SESSION 5: GLISSEMENT DES TERRAINS ET STABILITE DES PENTES 443
SESSION 5: LANDSLIDES AND SLOPE STABILITY
1. Analyse de stabilité et stabilisation du versant urbanisé de Tizi-N’Béchar (wilaya de Sétif, Algérie) 445
A. Seddiki, M. Khemissa (Algérie)
2. The present-day landslide of the Moulares phosphates mines (west-central Tunisia): Origin and evolution 451
F. Dhahri, R. Benassi, K. Zeyeni, N. Boukadi (Tunisia)
3. Etude à la rupture des murs de soutènements ancrés par la méthode cinématique des solides rigides 457
F.Z. Benamara, L. Belabed (Algérie)
4. Contribution to study and mapping of the risks induced by Tigzirt landslides (Algeria) 463
L. Djerbal, I. Khoudi, N. Alimrina, R. Bahar& B. Melbouci (Algérie)
5. Comparative study of analytical and numerical methods in slope stability. A case study of an open slope 471
mining in DRC
J. ShimatuMbuyi, H. TshimangaNzokona and V. FarialaMuchanga (RDC)
SESSION 6: GEOTECHNIQUE ENVIRONNEMENTALE 477
SESSION 6 : ENVIRONMENTAL GEOTECHNICS
1. Faisabilité de la récupération d’énergie à partir du biogaz dans la décharge de Sousse, Tunisie 479
M. Zairi, A. Kammoun, N. Hcine, Y. Frikha, S. Abid (Tunisia)
2. Valorisation de sable de dune pour la conception de barrières ouvragées dans les régions arides 485
M.K. Gueddouda, I. Goual, M.S. Goual, S. Taibi, N. Abou-bekr, M. Lamara (Algérie, France)
3. Geogrid reinforced structures for important infrastructure projects 493
O. Naciri, L. Vollmert (Germany)
4. Case history report: Geotextile encased columns (GEC’s) - Highway A2 in Poland 501
L. Sakou, J. Bruhier, R. Arab, P. Assinder, J. Sobolewski (Germany, France)
5. Expérience rwandaise dans les ouvrages de soutènement en sol renforcé 509
L. Sakou, J. Bruhier, R. Arab, R. Medjo, P. Assinder(Germany, France)
6. Traitement de talusmarneux par ungeosynthétiquealvéolaire :cas de l’autorouteestouest en Algérie. 517
S. Tabti, F. Cherifi (Algérie)
7. Formulationof a newroad material based on treateddredged marine sediments 525
W. Maherzi, F. Ben Abdelghani(France, Tunisia)
8. Chemical andphysicalcharacterizationofGabes marine sediments 529
A. Missaoui, I. Said, E. Hamdi, Z. Lafhaj (Tunisia, France)
9. MSE wall foundationcomposedof granular soilmattressesreinforcedwithgeogrids: the Herb Grey Parkway 537
(Windsor, Canada) casestudy
P. Rimoldi, A. Grimod (Italy, France)
10. Influenceof Earth Pressure Cells Installation Methods on Measurement Accuracy 545
A. A. Mirghasemiand D. Safaifard (Iran)
SESSION 7: SOLS LATERITIQUES 551
SESSION 7: LATERITIC SOILS
1. The Effect of Pretreatment on Index and Compaction properties of a Lateritic Gravel 553
I.K Ampadu, P. Ackah, F. Owusu-Nimo, F. J. F. Ayeh (Ghana)
2. Basic engineering properties of laterites in Muglad basin, Western Kordofan, Sudan 561
T. Elkhawad, H. Elarabi, S. Bashier (Sudan)
3. A laboratory investigation of the effect of used oil contamination on the direct shear behavior of a 567
lateritic soil
F.J.F. Ayeh, S.I.K. Ampadu, F. Owusu-Nimo, F.K. Boadu and A. Lawer (Ghana)

vii
4. Quasi-static numerical modeling of an ore carrier hold 573
S. Daoud, S. Ennour, I. Said, M. Bouassida (Tunisia)
5. Correlations between CBR index, modified proctor’s characteristics, and identification parameters of 579
lateritic soils in Cameroon
E.M. Kana, M. Mamba, T.R.F. Fangleu (Cameroon)
6. Interaction sols latéritiques-rejets des concentrateurs: quel(s) impact(s)? 587
G. MukokoKalenda, J.F. Thimus; D. NgoieNsenga (Belgique)
7. Prediction of Compaction Density of Lateritic Soil Base Pavement Layers from Dynamic Cone 595
Penetration Values in Uganda
S. Jjuuko, D. Kalumba, H.K. Mutaasa, J. Lukanda (South Africa)
SESSION 8: ARGILES GONFLANTES 603
SESSION 8: EXPANSIVE CLAYS
1. Duplicate testing conducted on the input parameters for the estimation of potential expansiveness of clay. 605
W. Badenhorst, E. Theron, P. Stott (South Africa)
2. Effect of cyclic loading on the behavior of swelling clays 611
A. Medjnoun, R. Bahar (Algeria)
3. L’évaluation de la susceptibilité au retrait et gonflement des sols argileux du site d’un projet urbain à 619
Diamniadio (Dakar, Sénégal)
A. Mbengue,M.M. Ndiaye, I.K. Cisse (Sénégal)
4. Problematic soil mechanics in the Algerian arid and semi-arid regions (case of M’sila expansive clays) 625
M. Khemissa, A. Mahamedi, L. Mekki (Algeria)
5. Influence de l’ajout du système chaux ciment sur le potentiel de gonflement et de la microstructure d’une 631
argile naturelle d’Algérie
S.A BourokbaMrabent, A. Hachichi, L. Bengra, M. Boudia(Algérie)

viii
Préface

L'Association Tunisienne de Mécanique des Sols (ATMS), en collaboration avec la Société Internationale de
Mécanique des Sols et de la Géotechnique (ISSMGE), a le grand plaisir d'accueillir le 16ème Congrès Régional
Africain de la Mécanique des Sols et de la Géotechnique.
Le congrès est organisé à Hammamet (Tunisie) du 27 au 30 avril 2015, avec la participation de chercheurs,
d'enseignants, d’étudiants et d'ingénieurs praticiens de plusieurs pays, notamment d’Afrique.
Le thème du congrès '' la géotechnique innovante pour l'Afrique'' reflète l'importance pour les géotechniciens
Africains de saisir l'opportunité que nous offrent les innovations apportées ces dernières années à la
géotechnique. Cela devra contribuer au développement de notre continent qui dispose du meilleur potentiel de
croissance dans le domaine des infrastructures dans les années à venir.
Les innovations en géotechnique concernent les trois axes suivants:
 La caractérisation : basée sur des instruments pour effectuer des essais in situ et au laboratoire plus fiables et
plus précis, bénéficiant des progrès enregistrés dans les systèmes de mesure et d'acquisition des données.
 Le calcul : basé sur des outils plus puissants et des travaux de recherche boostés par les nouvelles
technologies de communication facilitant les échanges et la coopération entre les académiciens des
différents pays.
 Les travaux de fondations et d'amélioration des sols basés sur des innovations technologiques et des
techniques d'exécution en constante évolution permettant aux entreprises d'affronter des terrains de plus en
plus difficiles.
Plus de soixante-dix communications sont présentées dans ce recueil et ont traité les principaux thèmes suivants :
- Le comportement des sols.
- La caractérisation des sols.
- Les fondations profondes et superficielles.
- L’amélioration des sols
- Les glissements des terrains et la stabilisation des talus.
- La géotechnique environnementale.
- Les sols latéritiques.
- Les argiles gonflantes.
Au nom de toute l'équipe de l'Association Tunisienne de Mécanique des Sols (ATMS), nous tenons à remercier
tous ceux qui ont contribué à l'élaboration du présent document, particulièrement à Mme S. Boussetta et Mr E.
Hamdi, et à l'organisation du congrès: membres du comité scientifique, membres du comité d'organisation,
auteurs, conférenciers et sponsors.

Le Président du congrès et de l'ATMS Le Président du comité d’organisation

Slah Eddine HAFFOUDHI Mehrez KHEMAKHEM

ix
Preface

The Tunisian Society of Soil Mechanics, in collaboration with the International Society for Soil Mechanics and
Geotechnical Engineering, hosts with great pleasure the 16th African Regional Conference of Soil Mechanics
and Geotechnical Engineering.
The conference held at Hammamet (Tunisia) from 27-30 April 2015 is attended by researches, faculties, students
and engineers from several countries, especially African.
The conference theme « Innovative Geotechnics for Africa”, reflects the great opportunity for African
geotechnical engineers to adopt the recent innovations in Geotechnical Engineering. This will contribute for
development of Africa which owns the best increasing-potential in infrastructures.
Geotechnical innovations embrace three main subjects:
 Characterization: in using reliable and high precision tools for laboratory and in situ testing benefited from
the progress made in measurement devices and acquisition data.
 Based computational powerful tools and boosted research investigations using novel communication
technologies which facilitate the exchange and collaboration between academicians from several countries.
 Ground improvement and foundation works aided by non-stop innovative techniques and technologies
enabling entrepreneurs to tackle challenging terrains.
The 16th ARCSMGE proceedings comprise more than seventy papers accepted for publication. Those papers
deal with the following main themes:
- Soilbehavior
- Soilcharacterization
- Deep and shallowfoundations
- Groundimprovement
- Landslides and slopestability
- EnvironmentalGeotechnics
- Lateriticsoils
- Expansive clays
On behalf of the ATMS team we greatly thank contributors for the edition of this document, especially to Mrs S.
Boussetta and Mr E. Hamdi, and the organization of the conference: members of the scientific committee,
members of the organizing committee, authors, speakers and sponsors.

The Chairman of 16th ARCSMGE 2015

The President of ATMS The Chairman of Organizing Committee
Slah Eddine HAFFOUDHI Mehrez KHEMAKHEM

x
Dear Colleagues and Friends,

On behalf of the International Society of Soil Mechanics and Geotechnical Engineering it is for me a great
honour and pleasure to welcome you to the 16th African Regional Conference on Soil Mechanics and
Geotechnical Engineering that will take place in Hammamet, Tunisia, from 27th to 30th of April 2015. This
important event will be followed by the symposium ISP'7-Pressio2015 to mark the 60th anniversary of the
MénardPressuremeter.

Choosing the topic "INNOVATIVE GEOTECHNICS FOR AFRICA" was to emphasize the role played by
geotechnics in the development in Africa. Papers deal with the following themes, as they relate to our topic:

Foundations, Soil improvement, Landslides and slope stability, Reinforced soil, Soil behavior, Soil
characterization, Lateritic soils, Swelling clays, Pavement design, Seismic hazards, Soil hydraulics,
Environment geotechnics, Geophysics and Case histories.

The reader will find something of use and interest under each of the themes.

The first Regional Conference for Africa was held in Pretoria in 1955, and was a good start for the series of the
conferences that followed. The proceedings of these conferences present a valuable reference on geotechnical
problems particular to Africa and engineering solutions to local problems. They are also a valuable source of
data on the properties of African soils. The proceedings of the 16th Regional Conference follow in the mold of its
forerunners and add to the store of knowledge represented in the conference series.

On behalf of the ISSMGE, I would like to take this opportunity to express my sincere appreciation and deep
gratitude to the Tunisian Association of Soil Mechanics (ATMS) and the Organizing Committee of the 16th
ARCMSG for their devotion, time and effort, without which this conference would not have become true. The
valuable contributions from all of you, from the invited lecturers and from the chairpersons are also greatly
recognized and appreciated.

Last but not least, I would like to wish all delegates a beneficial 16th ARCSMG and a pleasant stay in
Hammamet.

Fatma El ZahraaAly Baligh

Vice-President Africa

xi
CONFERENCES PLENIERES

KEYNOTE LECTURES

1
2
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Pratique de la géotechnique routière en Afrique intertropicale

Geotechnical engineering practice for roads works in tropics regions ofAfrica

Dr Etienne Marcelin KANA

Director in charge of research, education and cooperation, LABOGENIE, Cameroun

Résumé - De plus en plus, le transport routier se confirmecomme l’un des facteurs primordiaux de la croissance
économique des pays en voie d’émergence du continent africain. La genèse et la revue historique de ce mode de
transport indiquent qu’il n’est introduit en Afrique intertropicalequ’au début des années 1950,à l’orée des
indépendances des pays de cette région. Cette période est en effet marquée par la construction des premières
routes et la naissance concomitante des laboratoires nationaux de génie civil qui y constituent le socle de
l’expertise locale pourl’exercice de la géotechnique routière. Ces laboratoiresy effectuent,hormis les
missionsclassiques de géotechnique (caractérisation des matériaux sols et définition des contraintes de stabilité
des ouvrages géotechniques, etc..), mais également celles d’expertise sur le dimensionnement des chaussées, les
matériaux qui rentrent dans les travaux routiers (béton, fers à béton, géotextiles, bitumes, enrobés bitumineux,
etc.) et sur celle de leur mise en œuvre.
Du quotidien de ces géotechniciens, ressort le côté artistique de la pratique de la science
gé[Link], au regard des avancées scientifiques et technologiques actuelles, une caractérisation
plus idoine des sols et l’adaptation des constructions routières à l’environnement intertropical africain, s’y avère
indispensable. Plusieurs travaux de recherches ont dans ce sensdéjà été réalisés, et certains se poursuivent
avec,ou sans, une collaboration étrangère (CEBTP1, LCPC2, TRL3, CRR4). Pour un développement durable de
cette partie du globe, certaines d’entre elles gagneraient néanmoins à être réactualisées, et leurs résultats mieux
diffuséset pris en compte dans les différentes spécifications et normes locales.

Mots clés :géotechnique, routes, Afrique intertropicale, développement durable, recherches

Abstract - The road transport is more and more becoming, one of the keys factors of the economic growth of
most of theemergingAfrican’s countries. The early stages and the historicalreview of this mode of transport
indicate that it was only introduced in African’s tropics region during the early fifties, when countries there, were
about to gain their independences. During that period,construction of the first roads was noted, as well as the
concomitant creation of national civil engineering laboratories. Those laboratories became and still constituted
there, the bedrock of the local expertise in the geotechnical engineering activities. They carried out not only the
classical geotechnical duties (soils characterization, soil stability design for constructions, etc), but they are also
in charge of pavementsdesign and the quality control of road works, particularly the quality of building materials
such as: concrete, steel reinforcement bar, geotextile, asphalt concrete, etc. Their day to day activities brings out
the artistic face of the geotechnical engineering science.
To upgrade the soil characterization of African’s tropics soils, as well as the adaptation of the road works to
thatenvironment, studiesand research have been, and are continuing to be carried out, with or without
collaboration of non-African experts. Due to the non-stopscientific and technological evolution, it seemed
necessary for some of them to be renewed, and the dissemination of their results largely extended.

Keywords: geotechnical engineering, roads, African’s tropics, sustainable development, research

1- INTRODUCTION

La géotechnique routière estdéfinie ici,comme l'application de la géotechnique au domaine routier. Dans un

projet routier,elleconsiste dans le cadre d’une mission classique à définir les contraintes géotechniques du site
qui s’exercent sur le projet d'infrastructure (contraintes de stabilité) et à caractériser les matériaux (sols, roches)
en proposant leur valorisation dans l'ouvrage routier (remblais, couche de forme).

1
Centre Expérimental de Recherches et d'Études du Bâtiment et des Travaux Publics (C E B T P), France
2
Laboratoire Central des Ponts et Chaussées, (Transformé en IFSTTAR), France
3
Transport Road Laboratory, United Kingdom
4
Centre de Recherche Routière, Belgique

3
Pour ce faire,elle procède au préalable à la réalisation d’un état des lieux de la nature des solsau droit de
l’emprise du projet, pour en préciser les caractéristiques et les difficultéssusceptibles d'être rencontréespar leur
interaction avec l’ouvrage. Son objectif est d’indiquer aux autres intervenants dans la réalisation du projet, les
principales techniques compatibles avec les caractéristiques géotechniques et de stabilité des matériaux, les
contraintes hydrogéologiques, mais aussi avec la géométrie du projet (profil en long et en travers).Elle
contribueensuite à l'étude du projet de terrassement proprement dit, en proposant «des destinations privilégiées»
au sein de l'ouvrage routier (remblais, partie supérieure des terrassements, couche de forme, blocs techniques des
ouvrages d’assainissement tels que les buses, les aménagements paysagers, etc...).
Si en général le mouvement des terres est un document essentiel pour l'élaboration du projet de terrassement,
et constituel'aboutissement d’une étude géotechnique classique, en Afrique intertropicale, cette étude
concerne certes :
 les travaux de terrassement (utilisation du sol comme matériaux de construction en déblai/remblai) ;
 les soutènements et stabilisation de talus ;
 les fondations et remblais d’accès ou de couverture des ouvrages d'artet d’assainissement ;
Mais aussi :
 les ouvrages d’art et d’assainissement (qualité des matériaux et de leur mise en œuvre) ;
 la structure de la chaussée (dimensionnement, choix des matériaux et qualité de leur mise en œuvre) ;
 l’auscultation, l’entretien et la réhabilitation des chaussées (auscultation, choix des matériaux et qualité
de leur mise en œuvre).
“The ground is one of the most highly variable of engineering material. It is therefore not surprising that
geotechnical design depends on local site conditions and local engineering experience… Engineering practices,
relating to investigation and design methods…will therefore vary between different regions.” (Patrick et al,
2013). Il est donc«au sens de la géotechnique»fort hasardeux, de rentrer dans les dédalescertes très captivantes
de l’exécution de ces activités dans chacun des dizaines de pays que compte cette partie du globe. Dans la
présente communication,il ne sera question que deleurs aspects communs les plus généraux. Il s’agit en
l’occurrence de leur consistance, de leur acteurs, des difficultés et perspectives de développement de cette
activité. Il semble toutefois opportun de commencer au préalable parsituer cette région et d’en rappeler leclimat
et la géologie.

2- SITUATION ET CLIMAT DE LA REGION INTERTROPICALE AFRICAINE

L’Afrique intertropicale, comprend les 47 pays dont la majeure partie du territoire est géographiquement située
entre le Tropique du Cancer(23°27’, lat. nord) et le Tropique du Capricorne (23°27’, lat. sud).Les valeurs
moyennes des précipitations y varient annuellement entre les isohyètes de 250mm à plus de 2000mm.
Leurrépartition permet d’y distinguer globalement trois zones climatiques (figure 1) :
 la zone à climat aride, intertropical sahélien au nord et kalaharien au sud, marquée par une tendance
pluvio-orageuse annuelle monomodale de courte durée (1 à 3mois) qui concentre de 300 à 800mm de
hauteur d’eau. Le nord du Mali, le Botswana, la Namibie, le Niger, le Sud-Soudan, le nord Tchad, la
Somalie, le Djibouti, l’Éthiopie, une partie de l’Afrique australe, le Sénégal, le Burkina-Faso et le nord
du Kenya sont concernés ;
 la zone à climat équatorial, située en Afrique centrale avec généralementplus de 2000 mm d’eau
enregistrés à travers un régime bimodal de 6 à 8 mois (L’Hote et al., 1996). C’est la zone la plus humide
de toute l’Afrique, avec un maximum pluviométrique pouvant atteindre 9000mm (IDENAU, ville du
Sud-Ouest Cameroun) ;
 la zone à climat tropical ou soudano-guinéen, caractérisée par une moyenne comprise entre 800mm et
2000mm avec des disparités locales de type monomodale (2,5 à 4 mois de pluies) ou bimodale (2 et 3
mois et 4 et 3 mois). C’est la zone de la majorité des pays ouest africain, situés sur le littoral côtier de
l’Atlantique (Golfe de Guinée).
La température des pays de l’Afrique intertropicale suit la tendance pluviométrique ainsi que le relief. Les
températures moyennes annuelles sont comprises entre 16°C et 35°C. Les écarts moyens thermiques saisonniers
entre les zones sont de 4°C à 6°C. Tandis que la variation journalière de température (jour/nuit et saison
pluvieuse/saison sèche) enregistrée varie entre 8°C (climat tropical humide et équatorial) et 15°C (climat aride et
soudano-guinéen).

4
Tropique du Cancer

Equateur

Tropique du Capricorne

Figure 1:Zones climatiques de l’Afrique

3- APERCU DE LA GEOLOGIE DE L’AFRIQUE

Les traces de l’histoire de la terre se retrouvent en Afrique (figure 2). En effet les traces d’avant le paléozoïque, à
l’instar des orogenèses ouzzalienne (Archéen) et éburnéenne (protérozoïque), sont rencontrés en Afrique de
l’Ouest, Australe et équatoriale.L’orogenèse kibarienne quant à elle, est visible en Afrique du Sud, en Tanzanie
et en RD Congo. Audébut du Paléozoïque, le bloc continental Pannotia se fragmente en Laurussia au nord et
Gondwana au sud (auquel appartient le futur continent africain). Une nouvelle convergence des deux blocs
conduit à la formation de la Pangée à la fin de la pé[Link] un premier temps, le bloc continental est soumis à
l'érosion des chaînes panafricaines. Celle-ci alimente quelques grands bassins en sédiments détritiques
(notamment des grès riches en quartz). Quelques transgressions océaniques, par exemple dans le nord du
continent, permettent la mise en place de dépôts marins.

Figure 2: Carte géologique simplifiée de l'Afrique, par J, SOUGY, CCGM

5
La période Méso-cénozoïque y est marquée sur le plan tectonique par trois évènements majeurs qui
correspondent à l'ouverture de l'Atlantique nord (rifting à partir du Trias en Afrique du nord), de l'océan Indien
(Jurassique), puis de l'Atlantique sud (rifting crétacé dans le nord et la partie équatoriale du continent). Cette
ouverture océanique engendre la formation de marges et isole définitivement l'Afrique des autres continents. Des
zones en surrection (Afrique australe) alimentent des zones de subsidence(zones centrales et sahariennes) en
sédiments détritiques.
Ce bref aperçu du climat et de la géologie de l’Afriquemontre que,même si les roches d’Afrique ont la même
nature géologique que celles des autres régions, elles ont eu des transformations particulièresliées aux climats
spécifiques d’Afrique, qui sontdifférentes de celles des roches des pays du bloc continental européen.C’est une
donne importante pour la pratique dans ce continent de la géotechnique en général, et de celle de la géotechnique
routièreen particulier.
L’exercice de ce pan de la géotechniquedans la partie intertropicale de ce continent, est intimement lié à la
genèse et l’histoire de l’introduction de la route en Afrique. Elles sont ci-après brièvement résumées.

4- GENESE ET HISTOIRE DE LA ROUTE EN AFRIQUE

Dans le monde, les premières routes sont liées à l'invention de la roue, du char et des chariots (Rival, 2005). En
Europe, jusqu’aux années- 400avant J.C., les Romains utilisaient des pistes pour se déplacer de Rome vers les
cités environnantes. La nécessité d'une meilleure défense alliée à une volonté d'expansion et d'hégémonie sur
l'Italie a conduit ces derniers à mettre en place un réseau de solides routes empierrées, mieux adapté à leurs
besoins militaires.
Les Chinois quant à eux, auraient déjà disposé d'un vaste réseau très ancien de routes commerciales entre
l'Asie et la méditerranée qui, passant par l’Afrique (Alexandrie en Egypte), reliait la ville de Chang'an (actuelle
Xi’an) en Chine à la ville d’Antioche, située en Syriemédiévale (aujourd'hui Turquie). Il s’agit du réseau de la
route de la soie, dont les plus anciennes traces connues remontent au milieu du premier millénaire avant J.C.
(voir Figure 3B).
En Afrique par contre, du moyen âge jusqu'au XIXe siècle, le commerce transsaharien, situé dans le
prolongement de la route de la soie (figure 3A) (Nantet,2008), était caravanier et reposait essentiellement sur
l'utilisation du dromadaire (Hugon, 1994). Il relie le « pays des noirs » (Bilad-al-Soudan en arabe) à la
Méditerranée et à la mer Rouge. Ce n’est que dès les années 60, que ce commerce commence à être effectué par
camion bien que le dromadaire y soit encore utilisé (Grégoire, 2010 ; Lombard et al, 2002). Comme autrefois,
« le Sahara est aujourd’hui traversé par quelques grands axes nord-sud » qui mettent en relation des villes parfois
très éloignées (Ninot, 2005).

B
A

Figure 3:(A)Carte du commerce transsaharien du VIIIe au XVIe siècles,

(B) Principales routes de la soie entre 500 av. J.-C. et 500 après. J.-C. (en latin)

Excepté les avancées très mal connues des constructions routières inhérentes à l’érection des pyramides
d’Egypte, et du Nigéria où l’on note l’existence d’un important réseau de routes (1782 km) construitesà partir
des années 1925, le début du développement des réseaux routiers carrossablesen Afrique intertropicale est
estimé à partir des années 1950.

6
« Les routes y surpassent progressivement le rail, émergent et contribuent…à la formation progressive, mais
quelque peu anarchique, d’entités supranationales. » (Lombard et Nino, 2010).En effet, pour les gouvernements
postindépendances des années 1960, la route a une double vocation.

Routes
Routes revêtues revêtues en 1968
en 1968
Carte routes revêtues
en Afrique de l’Ouest
et du Centre

Routes revêtues en 2008

Routes revêtues en 2009

Source : Carte Michelin,
1970 et 2009

Figure 4:Évolution de la construction des routes revêtues entre 1968 et 2009

La première d’une part, est la substitution au transport ferroviaire pour desservir les zones agricoles et
urbaines majeures. A titre d’exemple, l’on peut citer : les routes du « bassin arachidier » au Sénégal, celles de la
« zone coton » au Mali,de la « zone café cacao » en Côte d’Ivoire, de l’Ouest, du sud-ouest et du Centre au
Cameroun, de la région de Koumassi au Ghana et du sud-ouest et du nord du Nigeria.
La seconde vocation d’autre part, est le rattachement à la capitale des parties du territoire national jusque-là
oubliées. L’intégration des périphéries y reste une œuvre en cours (c’est seulement au tournant de 2000 que les
régions de Kayes au Mali, de Tambacounda au Sénégal, etde l’Est du Cameroun, ont été connectées aux
capitales nationales par des axes bitumés. Certaines de ces régions à l’instar de celles des zones forestières de
l’Afrique centrale et de l’Ouest à climat équatorial, n’étaient desservies que par des routes en terre appropriées
pour l’évacuation des grumes de bois vers les ports maritimes ou fluviaux.
A partir des années 50, jusqu’à la fin des années 90, la construction routière connait un véritable essor qui
n’est freiné que par la crise économique des années 90. Les pays africains comptent environ 2,06 millions de
kilomètres de routes en 2001, soit une densité routière de 6,84 km aux 100 km2. En 2005, seuls 580 066 km de
routes, soit 22,7% du réseau routier africain, étaient bitumés (CEA, 2009).
Après la fin de la guerre mondiale et partir du début des années 1950, les européens (Belges, Français,
Portugais et Britanniques)lancentdans leurs colonies d’Afrique, des travaux de construction des routes destinées
surtout à favoriser l’exportation de la production agricole et minière de ces territoires. Concomitamment, ils y
créent pour une mutualisation de leur capacité, des laboratoires de génie civil pour leur permettre une meilleure
insertion de leurs projets dans ces nouveaux environnements. C’est dès lors à travers ces laboratoires que
commence à s’effectuer la géotechnique en général, et la géotechnique routière en particulier en Afrique
intertropicale. Ceux-cirestent jusqu’à présent, dans cette partie du globe, le socle de cette activité.
Le récent réseau routier, à cause du coût relativement élevé des investissements y relatifs (400 à 750 milles
US$ /Km), est fortement lié au niveau économique de chacun des pays concernés, l’on y relève une
multiplication des routes et pistes praticables qui reste en dehors du Nigéria (figure 4), étroitement associée aux
grands et petits projets de développement agricole, forestier ou minier (Ninot, 2005).

5- CONSISTANCE ET OUTILS DE LA GEOTECHNIQUE ROUTIERE EN AFRIQUE INTERTROPICALE

Comme indiqué en introduction de cette communication, dans cette partie du globe, la pratique de la
géotechnique routière comprendcertes les études et le suivi géotechnique traditionnel des projets routiers, mais
elle inclutaussi l’ingénierie des matériaux de construction, d’entretien ou de réhabilitation des structures et

7
revêtements routiers. En principe, un projet de géotechnique routière qui doit s’intégrer dans l’environnement où
il est réalisé, comprenden Afrique intertropicale chronologiquement:
 la reconnaissance géotechnique ;
 le choix et la formulation éventuelledes matériaux de la structure de chaussée ;
 le dimensionnement des ouvrages géotechniques et de la chaussée ;
 le suivi des travaux de terrassements et de la mise en œuvre idoine des matériaux ;
 l’auscultation et l’entretien post-construction.
Les spécificités relatives de la pratique de la géotechnique routière seront illustrées ci-après, à travers la
description de deux de ses principales activités à savoir : lareconnaissance géotechnique et le dimensionnement
des chaussées.

5.1. La reconnaissance géotechnique

Nonobstant le niveau technologique des équipements qui est tributaire du PIB du pays concerné, le mode de
conduite et le matériel technique de la reconnaissancegéotechnique des itinérairessont identiques à ceux des
autres régions du monde. Leur choix dépend des désidératas des maîtres d’ouvrages qui sont généralement des
administrations publiques en charge des travaux routiers. Les objectifs poursuivisrestent eux aussi identiques
dans le principe à ceux des autres régions à savoir : la réalisation d’un état des lieux de la nature des solsmeubles
et autres roches au droit de l’emprise du projet, afin de préciser aux autres intervenants dans la réalisation de
l’ouvrage, leurs caractéristiques ainsi que les contraintes et difficultés susceptibles d'être rencontrées. Cette
reconnaissance permet ainsi de proposer des palliatifs aux différents risques et aléas géotechniques, en leur
indiquant les principes techniques compatibles avec les caractéristiques géotechniques et de stabilité des
matériaux (pente des talus,...), les contraintes hydrogéologiques, mais aussi avec la géométrie du projet (profil en
long et en travers).
L’on note néanmoinsd’un pays à l’autre,et parfois dans le même pays, une certaine disparité dans la densité
et les types d’essais réalisés, ainsi que sur celles des spécifications à respecter (i.e. : normes de classification des
sols), non liée à l’expertise géotechnique ou technologique. Cette disparité est plutôt relative à l’origine du
financement des projets suivis, dont le pays ou groupe de pays, dans la plupart des cas, est similaire à celui du
Bureau d’études techniques responsable de l’exécution globale du projet.
La catégorisation ou la classification des sols et roches est l’un des résultats attendus de la reconnaissance
ainsi menée. Paradoxalement, il s’avère généralement qu’à cause de la particularité géomorphologique de cette
région, certains de ses sols à l’instar des latérites, des coraux, des scories volcaniques, des matériaux coquillers,
rentrent difficilement dans les classes des normes usitées. En effet, ces normes ont été confectionnées pour les
sols des régions tempérées à climat continental d’une part, et certains des matériaux qui y sont régulièrement
utilisés le sont sans codification particulière d’autre part.
Parmi les classifications les plus couramment utilisées l’on note celles :
 du « UCS » (Unified Classification System) ;
 du Guide de Terrassement Routier (GTR) ;
 del’American Association of States Highway and Transportation Officials” (AASHTO), USA;
 du Transport Research Laboratory (TRL), Overseas Development Administration, Grande Bretagne;
 du Centre de Recherches Routière (CRR), Belgique.
Des recherches pour une classification des sols africainsont été menées de la fin du 19ème siècle jusqu’au
ème
20 siècle, notamment par les français, les britanniques et les portugais. Mais celles-ci avaientsurtout été
orientées jusqu’aux années 1950, pour des besoins agricoles ou d’exploration minière (Pedro Sanchez, 2010).
Depuis lors, plusieurs travaux axés sur le génie civil ont été réalisésen vue de remédier à la situation sus
évoquée. Parmi ces derniers,l’on peutnoter ceux récapitulés dans les documents suivants :
o Tanzania Pavement and Materials Design Manual , 1999 ;
o South African Mechanistic Design Model/SAMDM1996 ;
o Recommandationspour l’utilisation en corps de chaussées des graveleux latéritiques naturels,
LBTP, 1977
Il est par ailleurs signalé les travaux de Gidigasu et al (1991), Ampadu et al (2004), Idrissa Tockol (2011)
sur les laté[Link] ailleurs, pour favoriser les échanges et discussions sur ce sujet, il est observé que le 14ème
CRA MSG en 2007 à Yaoundéa été intitulé : « Les Sols d’Afrique ». Ces efforts devraient être poursuivies et
surtout mieux coordonnées au niveau de cette région. L’on peut en effetregretter que parmi les différents travaux
effectués au cours dudit congrès, fort peu (1.5%) se sont penchées vers une caractérisation physico-mécanique
orientée sur le comportementdans le temps, de ces sols vis-à-vis des sollicitations routières. Encore moins, vers
une comparaison économico-technique desdits sols par rapport aux autres matériaux utilisé[Link]’en est-il du
dimensionnement des chaussées ?

8
5.2 Ledimensionnement des chaussées

Pour l’Afrique francophone de l’Ouest (THIAW, 2006 ; COMBERE, 2008) et du Centre (NGAGNON et al.
2008), Le document de référence du dimensionnement des chaussées utilisé est le « Guide pratique de
dimensionnement des chaussées pour les pays tropicaux et désertiques » du CEBTP,édité en 1972 et révisé en
1984. Il est l’adaptation aux pays tropicaux d’Afrique et des Antilles, de la méthode analytique rationnelle
utilisée en France par le LCPC. Les structures de chaussées y sont dimensionnées pour des conditions de trafic
de 30% de poids lourds, pour un essieu standard de 130kN (contrainte de l’empreinte du pneu 662kPa).
C’est une méthode semi-empirique dont le dimensionnement,fait à partir des abaques, est complété par
l’évaluation des contraintes et déformations dont les principaux paramètres et le protocole de mise en œuvre sont
décrits dans les documents techniques et les travaux de Yoder-Witzcak (1975) et Kerkhoven et Dormon.
Cette méthode CEBTP,adaptée à des structures monocouches et bicouches,est plus utilisée pour des
chaussées flexibles à faible trafic T1 à T3 (Tableau 2),qui représentent pour l’heure,environ 85% du linéaire
revêtu des routes en Afrique intertropicale. Pour les chaussées à fort trafic (T4 et plus) ainsi que les structures à
trois couches et plus, il y est recommandé d’utiliser la méthode analytique rationnelle (les abaques, les
catalogues LCPC 1994 et 1998 et le logiciel ALIZÉ). Pour des trafics inférieurs à T1, ce guide recommande
l’utilisation des approches semi-empiriques basées sur la variation du nombre d’essieux équivalents standard
(NE) de 130 KN.
Pour le calcul de la contrainte admissible, l’équation semi-empirique 1 ci-dessous est proposée.

(1)

Où σz est en Kg/m2 ; N est le trafic en essieu de 130kN ; CBR du sol à 95% de l’OPM

La vérification de la loi de fatigue des matériaux obéit à l’équation 2 :

(2)

Où α=5 pour les enrobés et 1 à pour les matériaux traités aux liantshydrauliques ; σ et σo sont
respectivement les contraintes admissible et calculée ; N et No sont respectivement les nombres de
passagescalculé et celui du trafic.
Les modules des matériaux sont déterminés à partir des équations 3 et 4
- pour les matériaux granulaires non liés et sol (module statique);

Estatique =50xCBR, pour les matériaux moyen à grossier (3a)

Estatique =30xCBR, pour les matériaux fins à très fins (3b)

- pour les matériaux traités aux liants hydrauliques (ciment).

Estatique=1000 à 2000* Rc7 (4)

Rc7 : la résistance des matériaux à 7 jours [1000 pour les matériaux plastiques, 2000 pour les matériaux
granulaires).

Les classes de trafic exprimées en nombre cumulé de passages d’un essieu équivalent pour la méthode
CEBTP sont données par le tableau 2 suivant :

9
Tableau 1: Classification géotechnique des sols, CEBTP (1980)

Classe de la plateforme en fonction du Description des sols

CBR (%) à 95% de l’OPM
S1 < 5 Sol silteux, argileux ou sol tourbeux
5 <S2<10 Argile, sable argileux, limons,
argilelatéritique
10 <S3< 15 Sable et graves
15 <S4 <30 Graviers et arènes à grains moyens
S5 >30 Arènes à grains grossiers

Tableau 2: Classes de Trafic suivant le CEBTP (1980)

Essieu équivalent de 130KN Essieu équivalent de 82KN

T1< 5.105 T1< 3.106
5.105< T2< 1,5.106 3.106< T2< 107
1,5.106< T3< 4.106 107< T3< 2,5.107
4.106< T4< 107 2,5.107< T4< 6.107
107< T5< 2.107 6.107< T5< 108

Pour les pays anglophones, le document de base utilisé pour le dimensionnement est « A Guide to structural
design of bitumen-surfaced roads in tropical and sub-tropical countries, Overseas Road Note 31, TRL (1962,
1966, 1977, 1993) ».Il estdéveloppé à partir de la méthode semi-empirique britannique de dimensionnement
routier, et contient des similarités avec celui du CEBTP (1984), notamment pour les chaussées flexibles et semi-
rigides. Cependant le mode de conception est fondamentalement différent. Les valeurs de l’essieu standard de
base y est de 80 KN (contrainte de l’empreinte du pneu : 750 kPa). Six classes de plateforme sont usitées en
fonction du CBR caractéristique à l’OPN du sol en place (S1pour CBR<2 à S6 pour CBR>30). Tandis que 8
classes permettent de distinguer les différents trafics des routes.
Contrairement à la méthode CEBTP, les révisions régulières de ce guide ont intégré certaines préoccupations
relatives à la spécificité climatique de la région. Il s’agit notamment de la prise en compte pour chaque structure
proposée, de l’exploitation des résultats d’auscultations antérieures des chaussées existantes dans les pays
concernés. Le dimensionnement avec cette méthode est ainsi basé sur l’évaluation de la valeur résiduelle des
structures des chaussées à la fin de l’horizon de service (TRL, Note 5, (1988). Elle est faite sur la base de la
comparaison des mesures de déflexions et de CBR préconisées par le calcul, à celle d’une chaussée à structure
similaire existante. Ces données conduisent à la sélection des matériaux qui ont fait la preuve de leur
performance dans le contexte local de référence. La conception de nouvelles structures se fait en deux étapes qui
sont :
 la sélection initiale des matériaux à partir des résultats des essais normatifs pour le même type de
trafic conformément à la TRL Note 40 (1978) et des conditions environnementales locales (TRL
Note 31);
 l’analyse des données statistiques des détériorations (structurale et fonctionnelle) observées qui
permet de retenir une structure de chaussée apte à supporter un type de trafic équivalent à une
charge précise et dont le contexte local (climat, sol, drainage, etc.) du tronçon est connu sur une
période d’années très étendue pouvant aller à 30 ans (Haas et al. 1994).

Les équations semi-empiriques de base y ont été développées par Powell et al. (1984) pour :
les matériaux granulaires :

MR (MPa) = 17.62CBR (5)

Où MR est module résilient qui traduit le rapport entre la contrainte déviatorique (σ1-σ3) et la déformation
axiale résiduelle (ε) ;
la fatigue structurale de l’enrobé ;

10
Ncrack = 1.66x10-6 x(εt) -4.32 (6)

Ncrackest le nombre de passages pour la fissuration par fatigue; εt la déformation horizontale à la base de
l’enrobé (traction) ;
la déformation permanente des matériaux granulaires (orniérage) du sol
Nrut = 6.18x10-8 x(εv) 3.95 (7)

Nrut est le nombre de passages pour atteindre l’orniérage de 10,2mm; εv: la déformation verticale au sommet
du sol (compression)
La méthode du TRL est en définitive une méthode basée sur la performance résiduelle et comporte des
avantages liés à des possibilités de réaliser des analyses très fines lors de l’auscultation des chaussées pendant le
cycle de vie de chaque axe routier (Paterson, 1987 ; Assaf ,2011)

5.3 Limites des méthodes de dimensionnement existantes

Le coût élevédes travaux d’entretien routier,qui atteint près de 5,2 milliards USD par an en Afrique Sub-
saharienne (SITRASS, 2003), et les données d’auscultations des chaussées réalisés à partir de 2004, ont montré
que la recherche de solutions à la problématique du dimensionnement structural et de la durabilité des routes en
Afrique Sub-saharienne demeure à l’ordre du jour. » (KOUBIKANA PAMBOU, 2013; KANA, 2011). Diverses
causes ont été identifiées au nombre desquelles figurent : les erreurs de dimensionnement structural, la gestion
inadéquate des ressources mobilisées ainsi que le non-respect des échéanciers pour l’entretien et/ou la
réhabilitation des axes routiers.
En effet, s’agissant du dimensionnement structural, des analyses comparatives réalisées sur des structures de
chaussées construites en utilisant différents outils (catalogue SATCC, logiciel ALIZÉ et les lois de fatigue de
l’Asphalt Institute et de Shell) ont révélé l’existence d’erreurs de dimensionnement structural notamment des très
grands écarts constatés pour les valeurs de CBR supérieures à 30% (Combéré, 2008). Les travaux de Tockol
(1993) ont abouti à des résultats similaires pour les détériorations observées sur les routes du Mali et du Niger.
Du côté de l’Afrique australe et de l’Est, la perte de portance et la dégradation des chaussées semi-rigides ont été
observées également de façon préoccupante dès les années 1980 (Paige-Green, 2009). Le Rwanda qui a opté
pour l’AASHTO 2002 a enregistré également des détériorations sévères après seulement 4 ans de service de son
réseau routier (Ndéka, 2006).
Les travaux de Patrick et Dongal (2001) ont permis de comprendre les impacts des erreurs sur les valeurs de
CBR > 30% sur les structures de réseaux routiers de la Nouvelle Zélande ayant des conditions similaires
environnementale et de trafic) à l’Afrique subsaharienne.
Chacune des deux méthodes ci-dessus résumées, qui sont les plus utilisées par les praticiens de la
géotechnique routière en Afrique intertropicale, présente des limites dont les plus pertinentes sont récapitulées
ci-après

5.3.1 Limites de la méthode CEBTP

Les limites pour cette méthode, révéléesà travers les résultats de plusieurs travaux de recherche (Thiaw 2006 ;
Ba 2008 ; Combéré 2008 ; Mbodji, 2011 ; ALBTP, 2014), portent sur:

 les valeurs de la porosité/drainage des matériaux qui sont obtenues par déduction au lieu d’une
détermination expérimentale ;
 la remise en cause de la formulation des caractéristiques mécaniques des matériaux notamment
cellesdes graves traitées aux liantshydrauliques ;
 la détermination du module résilient des matériaux concassés à partir du CBR ;
 l’usage des valeurs pessimistes de performance mécanique des graves naturelles non traitées (GNT)
déterminées en utilisant la norme française basée uniquement sur les mesures de dureté (Los Angeles et
du Micro-Deval) des éléments granulaires.
 L’absence de modules fiables des matériaux au regard des différents climats, permettant une évaluation
non subjective des contraintes et déformations à partir des logiciels de dimensionnement actuels.

11
Figure 5:Route Lilongwe ABC au Malawi, en bon état et âgée de 20 ans avec un trafic moyen

Au-delà de ces travaux, il est constaté que les caractéristiques présentées des matériaux locaux à
l’instar des graves latéritiques, scories volcaniques ou coraux, y sont soient absentes soient
insuffisantes. Il est y présenté par exemple que « Les valeurs indicatives de déflexions caractéristiques
admissibles sur la couche de fondation sous essieu de 13 tonnes … peuvent être modulées entre 100/100 et
300/100 mm selon les trafics (T3 à T1) ». Ces valeurs n’ont pas un fondement pratique suffisant, et
peuvent exclure plusieurs matériaux naturels locaux en couche de fondation pour ces trafics. Les
matériaux latéritiques, sur la seule base de leur valeur du CBR et d’indice de plasticité, y sont
déconseillés en couche de base des chaussées.…pour la couche de base, …leurs caractéristiques
géotechniques sont généralement insuffisantes pour qu’ils puissent être utilisés (CEBTP, 1980). Des études et
planches d’essais réalisées dans certains pays, notamment au Cameroun (LABOGENIE, 1994) au
Burkina Faso (LNBTP, 1993) et au Malawi (AFCAP, 2011) ne confirment pas cette assertion. Les
résultats de celles-ci devraient être diffusées, poursuivies et intégrées dans les pratiques.

5.3.2 Limites de la méthode TRL

Les principales limites sont liées aux aspects suivants (Koubikana, 2010):
 la mise en œuvre appropriée de cette méthode exige une actualisation régulière des bases de données
qui reste une action encore à y faire ancrer dans les esprits et surtout avoir les moyens financiers de
cette politique ;
 l’absence de corrélations fiables locales entre les résultats de la TRL de Grande-Bretagne et ceux des
laboratoires de pays concernés ;
 L’inefficience actuelle des politiques d’entretien routier de ces pays.
 L’utilisation de techniques et matériaux nouveaux peut être difficilement effective, si ces derniers n’y
sont pas au préalable, suffisamment caractérisés.

« À l’heure où l’énergie devient de plus en plus onéreuse, les laboratoires des travaux publics doivent faire
preuve d’imagination et concevoir l’utilisation dans les meilleures conditions des matériaux locaux afin de
minimiser le coût de construction des chaussées tout en restant dans les limites de sécurité acceptables. Les
solutions luxueuses, fortement consommatrices d’énergie, doivent être, dans la mesure du possible, évitées mais
l’utilisation des matériaux locaux suppose que les ressources disponibles soient connues, aussi faut-il insister à la
fois sur l’importance des inventaires à réaliser au moins de part et d’autre des axes de développement du pays et
sur la nécessité des études régionales sur le terrain » (CEBTP, 1980). Il s’agit ici, d’une exhortation des experts
d’Afrique intertropicale de ce métier,à la mise d’un accent particulier sur le volet artistico-technique de la
géotechnique autour duquel s’exercent ou devraient s’exercer leurs activités quotidiennes.

6- ACTEURS DE LA GEOTECHNIQUE ROUTIERE EN AFRIQUE INTERTROPICALE

La géotechnique routière en Afrique intertropicale,est un domaine où le géotechnicien africain, pionnier dans son
domaine, exprime indubitablement par son activité quotidienne, le côté artistique de la science géotechnique. En
effet,afin d’apprécier leur interaction sur le fonctionnement de la route dans le temps, et sans expériences
préalables, il doitavoir une parfaite connaissance du comportement des différents matériaux pouvant constituer le

12
corps des chaussées,des caractéristiques physico-mécaniques et hydrogéologiques des sols supportset enfin, des
contraintes du reste de l’environnement sur cet ouvrage. Il est par son statut, contraint de proposer des solutions
constructives et s’assurer de leur mise en œuvre adéquate, afin que le fonctionnement global de l’ouvrage projeté
soit homogène, et permette le développement durable de son [Link] principaux acteurs de l’exercice de cette
activité comprennent : le personnel technique des laboratoires de génie civil, des géotechniciens des Bureaux
d’études techniques (BET) et le personnel des Administrations publiques en charge des travaux routiers.

6.1 Personnel des laboratoires de génie civil

Jusqu’au début des années 80, à quelques exceptions près, le personnel d’encadrement technique de ces
laboratoires d’Afrique intertropicale,est essentiellement constitué par des experts europé[Link] dehors du
personnel interne aux entreprises de BTP, ces experts sont en majorité issus,pour les états francophones, du
Centre Expérimental de Recherches et d'Études du Bâtiment et des Travaux Publics (C E B T P),et du Transport
Research Laboratory (TRL) pour les pays anglophones.L’encadrement techniquelocal composé de géologues et
d’ingénieurs géotechniciens ou de génie civil, n’intègre progressivement ces laboratoires qu’à partir de la fin des
années 70, après leur formation dans des universités ou écoles d’ingénieurs d’Europe, des Etats unis ou du
Canada, ou dans lesnouvelles universités qui voient le jour en Afrique en cette période.
Cet encadrement, utilise pour la conception, la construction ou la réhabilitation des chaussées, les
recommandations des guides compilés par les experts européens dans des guides utilisables exclusivement dans
pays, et dont les principaux sont les suivants :
Pour les pays francophones du centre et de l’ouest,
- le Guide de dimensionnement des chaussées dans les pays tropicaux du CEBTP (1972, 1984),
- le Guide des terrassements routiers, réalisation des remblais et des couches de formes (LCPC, SETRA,
1992),
- le Guide pour le renforcement des chaussées dans les pays tropicaux ;
Pour les pays Anglophones:
- A Guide to structural design of bitumen-surfaced roads in tropical and sub-tropical countries,
Overseas Road Note 31, TRL (1962, 1966, 1977, 1993)
- A Guide to the pavement evaluation and maintenance of bitumen-surfaced roads in tropical and sub-
tropical
countries, TRL (1999)
- Hydrological design manual for slope stability in the tropics, M G Anderson, D M Lloyd, M J Kemp,
1997, pp64

Dans la pratique, ce personnel est souvent arrivé à proposer des ajustements aux recommandations de ces
documents qui ne sont malheureusement pas formalisés (AFCAP 5 , 2011)ni incorporés dans des règles qui
concernent leur environnement propre (ALBTP 6 , 2014), à l’instar de l’inexactitude de la fonction Ip=
f(CBR).Certaines tentatives dans ce sens, notamment celles du LBTP7 de Côte d’ivoire (1986) et de la D.E.N. du
MINTP8au Cameroun (1987), parce que non régulièrement actualisées, ouinsuffisamment vulgariséessont restées
sans impact véritable.
Le personnel d’exécution de ces laboratoires africains, formé pour certains, grâce à des stages dans les
laboratoires d’Europe, maîtrise les modes opératoires des principaux essais courants de laboratoire et in-situ,
suivant les normes AASTHO, AFNOR, DIN et BS. La crise économique de la fin des années 80, et son
corollaire de politique d’ajustement structurel imposé dans ces pays par l’entremise du FMI 9 , a atténué le
processus de renouvèlement de ce personnel, dont l’âge est actuellement devenu relativement avancé. Un besoin
de formation pour cette relève,est l’une des préoccupations de l’heure de ces laboratoires.

6-2 Autres acteurs

Une partie non négligeable des acteurs dans la pratique de la géotechnique en Afrique intertropicale est
constituée par le personnel des administrations publiques en charge des travaux routiers, ainsi que de celui des
Bureaux d’études techniques privés. Il en assure le financement pour le premier, et la charge officielle du
contrôle technique des prestations pour le second. Leur impact sur la qualité finale de l’ouvragene saurait donc
être négligé. Pour les acteurs publics de certains de ces pays où progressivement la géotechnique renaît de ses
cendres, il pourrait encore y avoir une perception assez étriquéedes enjeuxdu financement des travaux

5
African Community Access Program
6
Association des laboratoires du Bâtiment et des Travaux Publics d’Afrique
7
Laboratoire du Bâtiment et des Travaux Publics
8
Division des Etudes et de la Normalisation du Ministère des Travaux Publics
9
Fonds Monétaire International

13
derecherche en géotechnique. Lesproblèmes techniques peuvent être noyés dans les difficultés d’affectation
budgétaire. En effet, des évaluations économiques des projets ont montré que 21 à 40 % seulement des besoins
annuels en fonds d’entretien routier ont été consentis par les gouvernements d’Afrique subsaharienne (Gwilliams
et al., 2008).
Les enseignants des écoles et universités en charge de la formation en géotechnique sont également, un
maillon essentiel de cette activité. Ils sont néanmoins en nombrerelativement insuffisant. La moyenne au
Cameroun était, dans les années 70 à 90, d’un enseignant de géotechnique pour 19 ingénieurs de génie civil
formés/an. Avec un ratio d’un ingénieur spécialisé en géotechnique/20 ingénieurs de génie civil formés. Alors
que la moyenne d’ingénieurs formés par pays en Europe se situe vers 30 000ingénieurs/an, contre 700 000
ingénieurs/anen Inde, le Sénégal dépasse rarement les 100 ingénieurs formés par an (Meissa Fall, 2010).
« L’enseignement a pour objectif de transmettre les connaissances de base et l’expérience que donne la
pratique…. Beaucoup des formateurs sont des ingénieurs dont l’enseignement est une activité annexe. Les
enseignants permanents des universités sont souvent entraînés vers des spécialisations qui les détournent de la
maîtrise globale des problèmes dont est responsable la géotechnique »(Magnan, 2014). Lestravaux de recherche
de ces derniers, jusqu’à une période très récente, qui précède le lancement des journées africaines de la
géotechnique (JAG) à Yamoussoukro en Côte d’Ivoire en 2010, ont donc parfois été en décalage par rapport aux
préoccupations des professionnels. Entre ces acteurs, il n’y avait que peu de coopération.

7- CONCLUSION ET PERSPECTIVES

Au regard des nombreux projets routiers engagés en vue de son émergence, et de la nécessité de leur conception
et réalisation adéquate pour l’atteinte de cet objectif, il est loisible de croire à une croissance de l’activité de la
géotechnique routière en Afrique intertropicale. Il y a toutefois lieu d’y procéder, tant au niveau universitaire
qu’à celui des laboratoires, à la poursuite des recherches prospectives et opérationnelles en vue de l’adaptation
des spécifications utilisées avec les différents environnements de cette région. Une large diffusion régionale des
résultats existants et ou nouvellement obtenus de ces recherches est à promouvoir. Celles-ci devraient en outre
s’appesantir davantage sur la caractérisation et la classification des sols en place, et sur l’utilisation optimale de
ces derniers dans la construction routière.

QUELQUES RÉFÉRENCES BIBLIOGRAPHIQUES

AFCAP, (2011), Performance Review of Design Standards and Technical Specifications for Low Volume Sealed
Roads in Malawi, p 8
Anne Hugon, (1994), Vers Tombouctou, Gallimard, p. 23
Bernard Lugan, (2001), Atlas historique de l'Afrique, p 18
Bernard Nantet, (2008) Histoire du Sahara et des Sahariens, 124-148
BOUASSIDA et al, (2007), Les sols d’Afrique, Proceedings of the XIV th ARC SMGE, Yaoundé, 150 pages
CEA, (2009), Rapport d’examen africain sur les transports, Nations Unis, Commission Economique pour
l’Afrique, p2
Emmanuel G. (2010), Migrations et circulations transsahariennes, Université Ouverte de Besançon
Fall M. et al, (2011), Evolution of Lateritic Soils Geotechnical Parameters during a Multi-Cyclic OPM
compaction and correlation with Road Traffic, Geomaterials, 59-69
GIDIGASU M. D. (1991), Geological and geotechnical engineering, Vol 9, Issue3
GIDIGASU M. D., (1967), Lateritic Soil Engineering-Pedogenesis and Engineering Principles, Elsevier,
Amsterdam
GIDIGASU, M.D. (1984). Variability of Geotechnical Properties of Subgrade Soils in a Residual Profile Over
Phyllite. Proceedings of the 7th ARC SMFE, Accra, Ghana, V1: 95-104.
GODARD X., (1996), Transport local, transport international : quelle articulation, Actes du SITRASS IV,
Lyon/Arcueil,LET/INRETS, p. 339-343
KANA et al. (2014), Projet de révision des règles techniques de génie civil en Afrique, 6ème JAG, Ouagadougou,
p4
LOMBARD J., NINOT O., (2002), Impasses et défis dans le transport routier sénégalais, M. C. Diop (éd.),
Karthala, p.109-162.
Magnan J. P. (2014), Enseigner la géotechnique : pourquoi, pour qui, quoi et comment ?, 6ème JAG,
Ouagadougou
Michel Rival, (2005), Grandes Inventions de L'humanité, Larousse
NINOT O., (2005), Pour une lecture critique des circulations entre villes et campagnes dans les pays du sud :
l’exemple du Sénégal, Presses Universitaires de Rennes. p. 455-466
Patrick A. et al, (2013), Modern Geotechnical Design Codes of Practice, Implementation, Application and
Development, IOS Press, Millpress, Vol1

14
Pedro S., (2010) Development of soil classification in Africa, The soil of Africa, p48
Remillon A., (1967) Les recherches Routières Entreprises en Afrique D’expression Française, Application à la
Conception et au Renforcement des Chaussées économiques, ITBTP
TRRL. (1990), Tropical Soils: Formation, Classification and Engineering Properties, Berkshire, UK.
Norme NF P 98-086: Dimensionnement des structures de chaussées routières

15
16
Innovative Geotechnichs for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Development in deep excavation and deep foundation technologies

Développement en excavations profondes et des technologies des fondations profondes
R. Katzenbach*1, S. Leppla1, M. Seip2
1
Technische Universität Darmstadt, Institute and Laboratory of Geotechnics, Germany
2
Ingenieursozietät Professor Dr.-Ing. Katzenbach GmbH, Frankfurt am Main, Germany
*
Corresponding Author

ABSTRACT -Regarding economic and safety aspects of any construction like high-rise buildings, bridges,
infrastructures and underground structures, the main objective is sustainability. To achieve this main objective a
high-level design based on an adequate soil investigation and an independent peer review for the planning,
design and the construction stages are necessary. Using hybrid foundation technologies like the Combined Pile-
Raft Foundation and in-situ load test of deep foundation elements using pressure cells an optimization of deep
foundation systems is possible including an increasing level of safety. For the successful construction of deep
foundation systems main regulations and recommendations have to be considered. This keynote paper explains
the basics of the qualified design and construction of deep excavations and deep foundations and shows the
application in engineering practice.

RÉSUMÉ-En ce qui concerneles aspectséconomiques et sécuritaires detoute constructioncommeles gratte-ciel,

les ponts, les infrastructures et les structures souterraines,l'objectif essentiel est le développement durable. Pour
atteindre cet objectif, il est nécessaire d’accomplir une conception de haut niveau sur la base d’une
reconnaissance adéquate du sol et une évaluation indépendante par les pairs pour la planification, la conception
et les étapes de la construction. En utilisant des technologies de base hybrides commeles fondations mixtes
semelle-pieux et les tests de charge in-situ d’éléments de fondations profondes en utilisant des cellules de
pression, une optimisation des systèmes de fondation profonde est possible, y compris une augmentation du
niveau de sécurité. Pour une construction réussie des systèmes de fondation profonde, des règlements et des
recommandations principaux doivent être prises en considération. Ce document principal explique les bases de la
conception qualifié et la construction des excavations et des fondations profondes et montre l’application dans la
pratique de l’ingénierie.

1 INTRODUCTION

The construction of deep excavations and deep foundations in inner cities are special challenges due to the
limited space for the construction works and the effects on existing buildings and underground structures
(Katzenbach & Kurze 2013, Katzenbach et al. 2014). Figure 1 gives an impression from engineering practice.

Figure [Link] works of piles close to a underground metro station in Frankfurt am Main, Germany (left), and close
to sensitive structures of a World Heritage Property in Barcelona, Spain (right).

17
An economic and safe design of deep excavations and deep foundations in difficult soil conditions in urban
areas is based on the following main aspects:
 qualified experts for planning, design and construction
 interaction between architects, structural engineers and geotechnical engineers
 adequate soil investigation
 design of deep foundation systems using the Finite-Element-Method (FEM) in combination with enhanced
in-situ load tests for calibrating the soil parameters used in the numerical simulations
 quality assurance by an independent peer review process (4-eye-principle) combined with the observational
method if necessary

2 QUALITY CONTROL MANAGEMENT BY 4-EYE-PRINCIPLE

The quality control management is defined by the 4-eye-principle. This principle is a process of an independent
peer review as shown in Figure 2. It consists of 3 levels. The investor, the experts for planning and design and
the construction company belong to the first level. Planning and design are done according to the requirements
of the investor and all relevant documents to obtain the building permission are prepared. The building
authorities are the second level and are responsible for the building permission which is given to the investor.
The third level consists of the publicly certified experts. They are appointed by the building authorities but work
as independent experts. They are responsible for the technical supervision of the planning, design and the
construction. The independent peer review by publicly certified experts for geotechnical engineering checks that
all information including the results of the soil investigation, the laboratory and field tests and the boundary
conditions defined for the geotechnical design are complete and correct.

Figure 2. Independent peer review process (4-eye-principle).

3 OBSERVATIONAL METHOD

For projects with difficult boundary conditions it is necessary to apply the observational method to review the
design during the construction time and, if necessary, during service time. The observational method is always a
combination of the common geotechnical investigations before and during the construction phase and the
geodetic survey together with the theoretical modeling and a plan of contingency actions (CEN 2008). Figure 3
shows the principle of the observational method.
Only monitoring to ensure the stability and the serviceability of the structure is not sufficient and according
to the standardization not permitted for this purpose. Overall the observational method is an institutionalized
controlling instrument to verify the soil and rock mechanical modeling.
The identification of all potential failure mechanisms is essential for defining the monitoring program. The
program has to be designed in that way that all these mechanisms can be observed. The measurements need to be
of an adequate accuracy to allow the identification of critical tendencies. The required accuracy as well as the
boundary values need to be identified within the design phase and before the development of the monitoring
program. Contingency actions need to be planned before the construction works start depending on the ductility
of the bearing structure. The observational method must not be seen as a potential alternative for a

18
comprehensive soil investigation campaign. A comprehensive soil investigation campaign is in any way of
essential importance. Additionally the observational method is a tool of quality assurance and allows the
verification of the parameters and calculations applied in the design phase. The observational method helps to
achieve an economic and save construction.

Figure 3. Principle of the observational method.

4 IN-SITU LOAD TESTS FOR DEEP FOUNDATION ELEMENTS

Soil parameters are determined by project- and site-related soil investigations with core drillings and laboratory
tests. Those tests are important and essential for the initial definition of soil mechanical parameters of the soil
layers, but usually not sufficient for an entire and realistic capture of the complex conditions, caused by the
interaction of the subsoil and the construction (Katzenbach 2005).
In order to reliably determine the ultimate bearing capacity of deep foundation elements, load tests need to be
carried out. For those tests often very high counter weights or strong anchor systems are necessary. By using the
Osterberg method high loads can be reached without installing anchors or counter weights. Hydraulic jacks (O-
Cells) induce the load in the deep foundation elements using the pile itself partly as abutment. Such pile load
tests allow the testing of different soil layers and the determination of the skin friction and the base resistance
separately with only one pile by a multi-level test pile with e.g. two jack-levels. The results of the field tests
allow a calibration of the numerical simulations. Figure 4 shows the principle scheme of load tests on deep
foundation elements.

Figure 4. Principle scheme of pile load tests.

5 COMBINED PILE-RAFT FOUNDATION (CPRF)

The CPRF is a hybrid foundation system that combines the effects of a foundation raft and deep foundation
elements like piles and barrettes (Poulos 2001, Randolph 1994). The bearing capacity and the deformation
behavior are affected by the interactions between the deep foundation elements, the foundation raft and the

19
subsoil. For an optimized and safe design of a CPRF the calculation method has to consider these interactions
(Hanisch et al. 2002, ISSMGE 2013).
Due to the stiffness of the foundation raft the total load of the building Ftot,k is transferred into the subsoil via
contact pressure under the raft σ(x,y) and via the deep foundation elements like piles. The total resistance of the
CPRF Rtot,k(s) consists of the resistance of the raft Rraft,k(s) and of the resistance of the piles Σ Rpile,k,j(s) as
explained in Equation (1). The resistance Rpile,k,j(s) of a single pile “j“consists of theskin friction qs,k,j(s,z) and the
base resistance qb,k,j(s), as show in Equation (2) to (4).
m
Rtot ,k  s    R pile,k , j  s   Rraft ,k  s  (1)
j 1

R pile ,k , j  s   Rb ,k , j  s   Rs ,k , j  s  (2)

  D2
Rb ,k , j  s   qb ,k , j  s   (3)
4
Rs ,k , j  s    qs ,k , j  s, z     D  dz (4)

Figure 5 shows the soil-structure interaction of a CPRF. The bearing capacity and the load-settlement
behavior are affected by the interactions between the different elements of the hybrid foundation system CPRF.
The distribution of the total building load between the different bearing structures of a CPRF is described by the
CPRF coefficient αCPRF which defines the ratio between the amount of load carried by the piles Σ Rpile,k,j (s) and
the total load of the building Ftot,k as explained in Equation (5).

 CPRF 
R pile,k, j s
(5)
R tot,k
A CPRF coefficient of zero describes a raft foundation without any deep foundation elements, a CPRF
coefficient of one represents a classic pile group, neglecting the existence of a raft. For the design and the
construction of CPRFs the standard regulations for deep foundation elements have to be considered.
Additionally the international CPRF-Guideline, published by the International Society of Soil Mechanics and
Geotechnical Engineering (ISSMGE), has to be respected.

Figure 5. Soil-structure interaction of a CPRF.

20
6 CONSTRUCTION OF DEEP FOUNDATION ELEMENTS

For the transfer of high loads into the subsoil different types of deep foundation elements are available. Table
1 gives an overview including the advantages and disadvantages of the different types.

Table1. Types of deep foundations.

For the successful construction of deep foundations special equipment and large experiences are necessary.
The main European regulations for the construction of deep foundations are:

 EN 1536 and SPEC 18140 for bored piles

 EN 12794 for pre-cast driven piles
 EN 12699 and SPEC 18538 for displacement piles
 EN 14199 and SPEC 18539 for micropiles
 EN 12063 for sheet-pile foundations
 EN 1538 in combination with EN 1536 and SPEC 18140 for barrettes (elements of diaphragm walls used as
deep foundation elements)

21
The drilling of the borehole should always be carried out in combination with a steel casing or a
stabilization fluid. During drilling and befor reaching the total depth the base of the casing must be
deeper than the excavation level. For pile drillings under the groundwater table at any time inside the
casing a counter weight against the groundwater has to be guaranteed to avoid a hydraulic failure. This
is realized by water or another fluid inside the casing. It is recommended that the water level inside the
casing should be 1.5 m above the groundwater level. Figure 6 shows the production process of a bored
pile. When the casing is pulled out, the concrete level should always be higher than the base of the
casing.

Figure [Link] process of a bored pile.

7 EXAMPLE FROM ENGINEERING PRACTICE: CLASSIC PILE FOUNDATION

7.1 Project overview

In the centre of Frankfurt am Main, Germany, on a construction site of 17,400 m2 the high-rise
building complex “Palais Quartier” has been realized (Janke et al. 2010, Katzenbach & Leppla 2014).
Figure 7 shows the completed high-rise building complex from the south.

Figure 7. View from the south of the high-rise building complex “Palais Quartier”, Frankfurt am Main, Germany.

22
The complex consists of several structures with a total of 180,00 m2 floor space, thereof 60,000 m2
in the sublevels (Figure 8). The project includes the historic building “Thurn und Taxis-Palais” whose
façade has been preserved (Unit A). The office building (Unit B), which is the highest building of the
project with a height of 136 m, has 34 floors each with a floor space of 1,340 m2. The hotel building
(Unit C) has a height of 99 m with 24 upper floors. The retail area (Unit D) run along the total length
of the eastern part of the site and consists of eight upper floors with a total height of 43 m. The
underground parking with five floors spans across the complete project area. With an 8 m high first
sublevel, partially with mezzanine floor, and four more sublevels the foundation depth results to 22 m
below the surface. Thereby excavation bottom is at 80 m above sea level (msl). A total of 302
foundation piles (diameter up to 1.86 m, length up to 27 m) reach down to a depths of 53.2 m to 70.1
m above sea level depending on the structural requirements. As shown in the sectional view of Figure
8 the upper part of the piles is in the Frankfurt Clay and the base of the piles is set in the rocky
Frankfurt Limestone. Due to the different load areas and the difficult deformation behavior of the
different structures the project is classified into the Geotechnical Category GC 3 according to EC 7.
The Geotechnical Category GC 3 it the category with the highest factor of complexity.

Figure 8. High-rise building complex PalaisQuartier: overview (left) and cross section A-A (right).

7.2 Optimized pile foundation

Regarding the large number of piles and the high pile loads, a pile load test has been carried out for
optimization of the classic pile foundation. Osterberg-Cells (O-Cells) have been installed in two levels
in order to assess the influence of pile shaft grouting on the limit of skin friction of the piles in the
Frankfurt Limestone (Figure 9). The test pile with a total length of
12.9 m and a diameter of 1.68 m consists of three segments and has been installed in the Frankfurt
Limestone layer 31.7 m below the surface. The upper pile segment above the upper cell level and the
middle pile segment between the two O-Cells can be tested independently. Pile shaft grouting has been
carried out for the middle pile segment.

23
Figure 9. Setup of pile load test.

In the first phase of the test, the upper part was loaded by using the middle and the lower part as
abutment. A limit of 24 MN could be reached (Figure 10 on the left). The upper segment was lifted
about 2 cm, the settlement of the middle and lower part was 1 cm. The mobilized skin friction was
about 830 kN/m2.

Subsequently the upper pile segment was uncoupled by discharging the upper O-Cell. In the second
test phase the middle pile segment was loaded by using the lower segment as abutment. The limit load
of the middle segment with shaft grouting was 27.5 MN (Figure 10 on the right). The skin friction was
1,040 kN/m2, this means 24 % higher than without shaft grouting. Based on the results of the pile load
test using the Osterberg method, the majority of the 290 foundation piles were made by applying shaft
grouting. Due to the pile load test the total length of piles was reduced significantly. The reduction is
about
15 % of the total pile length.

Figure 10. Load displacement curves of test phase 1 (left) and test phase 2 (right).

24
8 EXAMPLE FROM ENGINEERING PRACTICE: COMBINED PILE-RAFT FOUNDATION (CPRF)

8.1 Project overview

South-west of the centre of Frankfurt am Main, Germany, a new building complex is realized on a
construction site with 21,000 m2. An overview of the project area is given in Figure 11 on the left. The
new building complex is situated nearby the historic monastery in the east, the river Main in the south,
a metro tunnel in the west and a street and metro tunnel in the north. The building complex consist of
6 parts, 4 of them are high-rise buildings. Regarding the complex construction process in several steps
and the soil-structure interaction between existing buildings, new buildings, the metro and street
tunnels and an existing sewer line the project is classified into the Geotechnical Category GC 3
according to EC 7 (CEN 2008).

Figure 11. Project overview with building Panorama in the south-west (left) and cross section 1-1 (right).

8.2 Soil and groundwater conditions

The soil conditions can be summarized as follows:

 fillings with a thickness of 2 m to 10 m

 quaternary sand and gravel down to 11 m under surface with a thickness of 1 m to 9 m
 tertiary Frankfurt Clay, consisting of alternating layers of clay, limestone and sand, down to 35 m
under surface
 Frankfurt Limestone
The upper groundwater level is in the quaternary sand and gravel layer between 4 m and 6.5 m
below the surface. The lower groundwater level is in the tertiary limestone and sand layers within the
Frankfurt Clay as well as in the Frankfurt Limestone. The lower groundwater level is normally
confined.

8.3 Optimization of the foundation system

Depending on the loads of the building complex the foundation systems vary in relation to the loads:

 spread foundations
 classic pile foundations

25
 micropiles
 Combined Pile-Raft Foundation (CPRF)

For the optimization of the deep foundation elements, two pile load tests were carried out on the
construction site. In addition to the determination of the characteristic skin friction and base resistance
the results of the pile load tests were used to re-calculate the distribution of the stiffness of the soil
model was calibrated by numerical back-analysis using the Finite-Element-Method (FEM). For
example the test setup of test piles TP 1 is shown in Figure 12 on the left. TP 1 ends in the Frankfurt
Clay. TP 2 ends in the Frankfurt Limestone. The load on the piles was given by O-Cells. The test piles
consist of the upper, the middle and the lower pile segments, divided by the upper and the lower O-
Cells. The test pile was equipped with measurement devices:

 strain gauges in the upper, the middle and the lower pile segment
 stack extensometer from the surface to the pile top, to the upper O-Cell, to the lower O-Cell and to
the pile toe
 extensometer between the O-Cells
 displacement transducer inside the O-Cell

According to EC 7 the results of the pile load tests were reduced by the factors of variation (Seip et
al. 2014). The characteristic skin friction and ase resistance under consideration of the factors of
variation are as follows:

 skin friction of Frankfurt Clay: qs,k = 0.16 MN/m2

 base resistance of Frankfurt Clay: qb,k = 1.80 MN/m2
 skin friction of Frankfurt Limestone:qs,k = 1.30 MN/m2

Using the back-analysis of the pile load tests the developed numerical soil model was calibrated.
For example Figure 12 shows on the right the measured displacements of the pile load test in-situ and
the calculated displacements of the back-analysis in one of the test phases. The soil mechanical
parameters detected during the soil investigation were updated according to the results of the back-
analysis. These analysis were the basis for the following 2-dimensional and 3-dimensional numerical
simulations.

Figure 12. Test setup of pile load test TP 1 (left) and measurement and back-analysis of pile load test TP 1 (right).

26
8.4 Numerical analysis of the foundation system, the retaining system and the soil-structure
interaction

For the determination of the complex soil-structure interaction between existing buildings, new
buildings, the metro and street tunnels and the sewer line, several 2- and 3-dimensional, non-linear
numerical analysis using FEM have been carried out. For example the numerical model of the south-
west of the project area with the Panorama building is shown in Figure 13.
The foundation system of this part of the project area combines a CPRF and a classic pile
foundation. It has to be guaranteed that the existing sewer line does not get any load from the new
buildings. For the reduction of the deformation of the sewer line the retaining structure does not get
any vertical loads from the superstructure. For the settlement relevant loads the predicted settlements
of the building is up to a maximum of 3 cm. The predicted settlements of the sewer line are smaller
than 1.5 cm. The piles of the CPRF have a diameter of 1.5 m and a length between 14 m and 20 m.
The CPRF-coefficient is αCPRF = 0.6 according to the CPRF-Guideline (ISSMGE 2013). The piles of
the classic pile foundation have a diameter of 1.5 m and a length between 20 m and 24 m.
In the area of the metro tunnel in addition a 2-dimensional numerical analysis has been carried out
in order to get more detailed information about the deformations, the variation of stresses during the
different construction phases and the influence on the sealing system of the metro tunnel caused by the
excavation and the construction of the retaining system.
The soil mechanical parameters have been varied for the study of the sensitivity of the metro
tunnel. According to the results of the numerical simulations and the experiences made in the region of
the project area, horizontal displacements of the metro tunnel of less than 1 cm have been expected.
Differential displacements at the joints of the tunnel blocks have been expected in the range of only a
few millimeters. Based on the 2-dimensional and 3-dimensional numerical analysis no limitation of
the serviceability and of course no damages have been expected. The measured displacements of the
tunnel blocks were less than 0.5 cm and the maximum differential displacement at the joints of the
tunnel blocks was 0.1 cm.

Figure 13. Numerical model of the area of the building Panorama, view from the north-east.

8.5 Retaining system for building Panorama

For the retaining system of the excavation in most cases bored pile walls have been used. The bored
pile walls have been stabilized by struts and anchors. The maximum depth of the excavation in the
south-west of Panorama building is 14.3 m for up to 4 sublevels.

In order to adapt the retaining structure to the underground structures (sewer line, sublevels and
retaining structures of the existing buildings) the piles with a diameter of 88 cm were combined with

27
jet grouting columns to ensure a watertight pit and a maximum of flexibility regarding the difficult
geometry.

The planning and design have been significantly influenced by the following aspects:

 sewer line directly beside the retaining wall

 existing underground parking with up to 3 sublevels north of the sewer line
 excavation for the building Palazzi in the east
 several cables and sewer lines in the south
 metro tunnel in the west with a distance of about 20 m
 several remaining sublevels and former retainings systems
In the areas with existing sublevels behind the retaining system the upper level of anchors have
been carried out with a back anchoring into the existing, back-filled basement.

8.6 Construction phase

For the construction of the excavation in the south-west of the project area the following special
geotechnical constructions have been carried out:
 3,700 m piles for retaining wall, d = 88 cm
 10,500 m anchors for retaining wall
 290 m special steel bars for anchoring into the existing sublevels
 36 t of steel for struts and girder beams
 790 m of foundation piles, d = 150 cm
 415 m jet grout columns
Due to the tight time schedule up to 4 big drilling rigs have been used by the contractor, i.e.
Bilfinger Constrution, for the construction of the retaining walls (Figure 14 on the left). After reaching
the depth of the existing foundation raft, this reinforced concrete element with a thickness of up to 1.5
m was deconstructed. After tensioning the anchor in the upper and the middle level the foundation
piles have been constructed. Due to the number of large construction vehicles the organization of the
construction works has been a great challenge for Bilfinger Construction (Figure 14 on the right).
To keep the tight schedule the installation of the anchors was carried out with up to 4 anchor
drilling rigs parallel to the excavaton works by Bilfinger Construction (Figure 15 on the left).
After the construction of the blinding concrete layer for preparation of the reinforcement of the
foundation raft, the excavation including the retaining system and the foundation piles have been
finished (Figure 15 on the right).

Figure 14. Construction of the bored pile wall (left) and construction of the pile foundation (right).

28
Figure 15. Anchor installing during excavaton (left) and construction of the blinding concrete layer (right).

9 EXAPLE FROM ENGINEERING PRACTICE: IMPORTANCE OF INDEPENDENT PEER REVIEW

9.1 Project overview

During the independent peer review process for a new highl-rise building in Frankfurt am Main,
Germany, the publicly certified expert for structural engineering informed the building authorities to
involve a publicly certified expert for geotechnics. The high-rise building has 38 floors above the
surface and 3 sublevels. The total floor space is 57,450 m2. The total height of the building is 136 m.
In an early planning stage it was estimated that the soil and groundwater conditions are common for
the Frankfurt subsoil:
 fill to a depth of 10 m followed by quaternary sand and gravel to variable depth
 beneath the quaternary sand and gravel is the Frankfurt Clay consisting of stiff to semi hard clay
with thin limestone layers and sand layers
 the basis is the Frankfurt Limestone

Due to this estimation the foundation system was planned as a CPRF reaching the Frankfurt Clay.
According to the complexity the construction project was classified into the Geotechnical Category
GC 3 (EC 7). For the construction of a CPRF the international CPRF-Guideline (ISSMGE 2013)
requires a publicly certified expert for geotechnics for the independent review process.

9.2 Safe and optimized design of the CPRF

For the foundation a CPRF was plannend as shown in Figure 16 a). During the review process
additional soil investigations were carried out. The results indicated that there is a discontinuity in the
intersection between the settlement active Frankfurt Clay and the rocky Frankfurt Limestone of
several meters. The top of the Frankfurt Limestone is not as deep as estimated so the pile of the CPRF
would have been partly in the limestone (Figure 16 b)).
In this case the piles would have had a much higher resistance due to the very stiff Frankfurt
Limestone. Damages would occur in the foundation raft and the superstructure because the piles
founded in the rock would get a higher load than the other piles found in the Frankfurt Clay.
Differential settlements would also occur.
On basis of the report of the publicly certified expert for geotechnics the design of the foundation
system was adapted and the piles of the CPRF were shortened (Figure 16 c)). The different stages of
the foundation design are summarized in Figure 16:
 left: first planning stage
 middle: after additional soil investigations
 right: realized foundation system after the report of the publicly certified expert for geotechnics

29
The whole construction was realized without any damage. Figure 17 shows the finalized high-rise building.

Figure 16. Development of the CPRF in different planning stages.

Figure 17. View on the finalized high-rise building.

10 CONCLUSIONS

For a safe and sustainable construction of deep excavations and deep foundation systems in urban
areas it is necessary to consider the soil-structure interaction. Normally those construction projects are
classified into the Geotechnical Category GC 3, which is the category with the highest factor of
complexity, according to EC 7. Due to the factor of complexity the following aspects are essential for
an optimized and a successful design and construction:
 adequate soil investigation with core drillings, laboratory and field tests
 in-situ load test
 numerical simulations considering the non-linear material behavior of soil
 independent peer review for geotechnical and structural engineering during planning, design and
construction
 independent supervision on the construction site
 observational method including a monitoring program
 contingency action plan

30
For the optimization of deep foundation systems in-situ load tests on the construction site are very
utile. The results of the in-situ load test have two main effects. On one hand the detected skin friction
and the base resistance may help to reduce the total pile length significantly. On the other hand the
results give the possibility to check the estimated bearing capacity of the deep foundation elements for
risk mitigation. In addition the pile load tests give the possibility for a calibration of numerical
simulations by a back analysis. Only by these numerical simulations it is possible to investigate the
load-deformation behaviour of complex foundation systems.
The main aspect of a safe and sustainable design of deep excavations and deep foundation systems
is the adequate consideration of the soil-structure interaction during all planning and design stage
because of the continously changing load-deformation behavior (Wörner & Pfeiffer 1998, Russo &
Viggiani 1998).

REFERENCES

CEN European Committee of Standardisation (1999) EN 12063 Execution of special geotechnical works - Sheet-pile walls.
CEN European Committee of Standardisation (2001) EN 12699 Execution of special geotechnical works - Displacement
piles.
CEN European Committee of Standardisation (2007) EN 12794Pre-cast concrete products -Foundation piles.
CEN European Committee of Standardization (2009) Eurocode 7: Geotechnical design - Part 1: General Rules.
CEN European Committee of Standardisation (2010) EN 1536 Execution of special geotechnical works - Bored piles.
CEN European Committee of Stnadardisation (2010) EN 1538 Execution of spezial geotechnical works - Diaphragm walls.
CEN European Committee of Standardisation (2010) SPEC 18140 Execution of special geotechnical works - Bored piles.
CEN European Committee of Standardisation (2012) EN 14199 Execution of special geotechnical works - Micropiles.
CEN European Committee of Standardisation (2012) SPEC 18538 Execution of special geotechnical works - Displacement
piles.
CEN European Committee of Standardisation (2012) SPEC 18539 Execution of special geotechnical works - Micropiles.
Hanisch, J., Katzenbach, R., König, G. (2002) Kombinierte Pfahl-Plattengründungen. Ernst & Sohn Verlag, Berlin,
Germany.
ISSMGE International Society of Soil Mechanics and Geotechnical Engineering (2013) Combined Pile-Raft Foundation
Guideline (Ed. Katzenbach, R., Choudhury, D.), Darmstadt, Germany.
Janke, O., Zoll, V., Sommer, F., Waberseck, T. (2010) PalaisQuartier (FrankfurtHochVier) - Herausfordernde
Deckelbauweise im Herzen der City. Mitteilungen des Institutes und der Versuchsanstalt für Geotechnik der Technischen
Universität Darmstadt, Germany, Heft 86, 113-124.
Katzenbach, R. (2005) Optimized design of high-rise building foundations in settlement-sensitive soils. International
Geotechnical Conference of Soil-Structure Interaction, 26.-28. May, St. Petersburg, Russia, 39-46.
Katzenbach, R., Leppla, S. (2014) Deep foundation systems for high-rise buildings in difficult soil conditions. Geotechnical
Engineering Journal of SEAGS & AGSSEA, Vol. 45, No. 2, June, 115-123.
Katzenbach, R., Leppla, S., Krajewski, W. (2014) Numerical analysis and verification of the soil-structure interaction in the
course of large construction projects in inner cities. International Conference on Soil-Structure Interaction: Underground
structures and retaining walls, 16.-18. June, St. Petersburg, Russia, 28-34.
Katzenbach, R., Kurze, S. (2013) Enhanced inner urban tunneling for the protection of World Heritage Properties. 12th
International Conference Underground Construction Prague 2013, 22.-24. April, Prague, Czech Republic, 8 pages.
Poulos, H.G. (2001) Piled-raft foundation: design and applications. Géotechnique 51, No. 2, 95-113.
Randolph, M.F. (1994) Design method for pile groups and piled rafts. 13th International Conference on Soil Mechancis and
Foundation Engineering, 5.-10. January,New Delhi, India, Vol. 5, 61-82.
Russo, G., Viggiane, C. (1998) Factors controlling soil-structure interaction for piled-rafts. Darmstadt Geotechnics No. 4,
Vol. 2, Technische Universität Darmstadt, Institute and Laboratory of Geotechnics, Germany, 297-321.
Seip, M., Vogler, M., Kurze, S., Kuttig, H. (2014) Großprojekt MainTor in Frankfurt am Main - Probebelastungen an
Großbohrpfählen zur Optimierung des Gründungsdesigns. Mitteilungen des Institutes für Grundbau und
Bodenmechnanik, TU Braunschweig, Germany, Heft 98, 23-41.
Wörner, J.-D., Pfeiffer, R. (1998) Importance of the development of stiffness and load during the construction period for the
soil-structure interaction. Darmstadt Geotechnics No. 4, Vol. 2, Technische Universität Darmstadt, Institute and
Laboratory of Geotechnics, Germany, 383-396.

31
32
Innovative Geotechnichs for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Landslides in saturated and unsaturated soils

Glissements de terrain dans les sols saturés et non saturés

E. E. Alonso*1
1
Department of Geotechnical Engineering and Geosciences, UPC, Barcelona, Spain
*
Corresponding Author

ABSTRACT – The paper reviews a few aspects of landslides in saturated and unsaturated soils. The relevance of
sedimentation and shearing planes in clayey soils is first discussed with reference to a number of case histories.
Residual strength is a fundamental parameter to discuss observations, which is probably a strong indication of
the widespread occurrence of progressive failure. The position and geometry of the failure surface is shown to be
a consequence of deformation processes. Landslide run-out is illustrated through a theoretical exercise which
shows the direct correlation between run-out and brittleness index. In a final chapter some aspects of the
behavior of partially saturated soils subjected to rain infiltration is discussed. The common observation of
shallow slides is explained, as well as the role of permeability, both intrinsic and induced by the deformation
process. Analysis of a real case highlights the behavior of slopes in weathered soil profiles.
RÉSUMÉ-Le document passe en revue quelques aspects de glissements de terrain dans les sols saturés et non
saturés. L’importance des plans de sédimentation et de cisaillement dans les sols argileux est d'abord discutée
selon un certain nombre de cas d’étude. La résistance résiduelle est un paramètre fondamental pour discuter des
observations, ce qui est probablement une bonne indication de la fréquence des cas de rupture progressive. La
position et la géométrie de la surface de rupture est signalée comme étant une conséquence des phénomènes de
déformation. La distance de propagation du glissement de terrain est illustrée moyennant un exercice théorique
qui montre la corrélation directe entre le distance de propagation et l'indice de fragilité. Dans un dernier chapitre
certains aspects du comportement des sols non saturés soumis à l'infiltration de pluie sont discutés. L'observation
typique de glissements superficiels est expliquée, ainsi que le rôle de la perméabilité, à la fois intrinsèque et
induite par le processus de déformation. L’analyse d'un cas réel met en évidence le comportement des pentes
dans les profils de sols altérés.

1. INTRODUCTION

Landslides are a vast subject which receives attention and contributions from Geology, Engineering Geology,
Geotechnical Engineering, Soil and Rock Mechanics and Geophysics. Therefore, some choices were
unavoidable when deciding the specific topics covered by this Report. The guiding idea was to present aspects
directly related with the experience and research activities of the author. Current classifications of mass
movements (Hungr et al, 2013) provide a synthetic picture of the variety of natural phenomena leading to the
instability of geological materials exposed to environmental and man-made actions. In order to limit the scope of
the presentation, the paper covers those landslides whose evolving geometry is characterized by soil volumes
which essentially maintain its original bulk stiffness and strength. The motion is explained by localized shearing
surfaces, where strains accumulate. These motions often involve clay materials especially in the vicinity of the
shearing surfaces. Soft rocks and hard clayey soils are often found. These landslides are relatively frequent in
practice and may be classified as translational, rotational or mixed-type landslides.
In an introductory chapter the relevance of sedimentation planes and pre-sheared surfaces is discussed. Then,
the complex phenomena associated to the concept of progressive failure in brittle soils will be discussed in
connection with some real cases and the development of computational tools. Landslide run-out and velocity of
the moving mass is a relevant subject when discussing the risk associated with instabilities. Recent progress in
this important subject will be presented. A chapter will deal with slopes in unsaturated materials, a topic directly
related with weather action. The degradation of clayey rocks is often at the origin of shallow movements in these
materials. This topic will also be mentioned.

2. THE RELEVANCE OF SEDIMENTATION PLANES AND PRE SHEARED SURFACES

Sedimentary planes are often a preferred place for shearing in the presence of tectonic activity, which has
been always present in many geological environments. This is a consequence of the large natural extent of

33
sedimentary layers and possibly on the microscopic nature of the planes (preferred orientation of particles).
Figure 1 shows a striated surface in a sedimentary plane, horizontally layered, of Guadalquivir, Tertiary, marine,
overconsolidated, brittle and high plasticity blue clay. Aznalcóllar rockfill dam failed catastrophically by sliding
on a Guadalquivir clay sedimentation plane (Alonso and Gens, 2006a, b; Gens and Alonso, 2006). A recent
analysis of the development of the progressive failure during the forward construction of the dam indicated that a
horizontal shearing band developed under the dam and provoked its forward sliding. The fact is that the sliding
surface was actually located in a sedimentary plane, not exactly horizontal, but dipping a few degrees (2-3º) in
the downstream direction. It is a clear indication of some reduction of strength of the sedimentary layer, even if it
is small, despite the highly homogeneous nature of this clay.
In a General Report on “Cuttings and Natural Slopes (Alonso, 1998) the back-analyzed strength of a few well
documented landslides was compared with laboratory determinations of strength. The case histories were
grouped into slides along sedimentation surfaces or else, slides crossing the stratification
Four cases belonged to the first category. The measured friction angles for the four slides, together with the
backanalyzed (BA) effective friction angle, were represented in a single plot (Fig. 2). Direct shear tests include
peak and residual conditions for the bulk or intact material (represented in the figure as DSTP(int), and
DSTR(int) respectively) and results for the shear tests performed on joints or the actual sliding surface
(represented as DSTDP or DSTDR for peak and residual conditions). When an effective cohesion is measured, it
is indicated as an added strength value in the proximity of the symbol. Otherwise the effective cohesion is zero.
In one case (La Riba slide) two results are given for the φ’ values associated with DSTP(int).They correspond to
two values of the void ratio as shown in the figure. Residual friction angle as measured in the ring shear device is
represented as RS.

Figure [Link] sedimentary plane of Guadalquivir blue

34
Figure [Link] along bedding planes. Synthesis of results in terms of effective friction angle

The striking result is that the back analyzed friction angle provides consistently the lowest friction, below the
values measured in ring shear tests. Interestingly, when the actual failure surfaces were tested in the shear box,
the measured friction corresponds to the backanalyzed field value. Several reasons for this result were advanced:
The effect of normal stress on residual friction angle, in the sense of reducing its value, and chemical changes
experienced by the shearing surface in nature. Pre-shearing by tectonic or geomorphological actions seems to be
very effective.
In the second group of landslides, a similar exercise led to the results shown in Figure 3.

Two of the cases reported in Figure 3 (Leizarán, Sort) correspond to first time failures. In both cases a finite
but small average effective cohesion is necessary to explain the actual slide by backanalysis. Therefore, ring
shear test results provide strengths lower than the actual strengths mobilized “in situ”. In both cases, however,
“residual” conditions as measured in direct shear tests after a complete reversal cycle of relative displacements
are close to backcalculated strength values. This is an indication of the relevance of progressive failure “in situ”.
In the three remaining cases in Figure 3 the residual strength, determined in direct shear tests (imposing
reversals) were close to the back-analyzed friction values.
The set of cases mentioned indicate that residual strength is a very relevant parameter to explain instability in
clay slopes, most of the time overconsolidated clays. Again, these results points towards the widespread
occurrence of progressive failure. It is then natural to plot the back-analyzed friction angles of the case records
mentioned in plots relating friction angle with clay plasticity. This is done in Figure 4. The correlations given in
the figure between residual friction angle and Plasticity Index were reported by Lupini et al. (1981). The figure
stresses the relevance of residual conditions in natural slides. It also points out the significance of slides along
sedimentary planes, because the friction explaining the field instability is, in the cases analyzed, smaller than the
laboratory determined residual friction.
It is concluded that further discussion on progressive failure mechanisms and the reasons leading to
unexpected and low friction angles are relevant issues.

35
Figure [Link] across bedding planes. Synthesis of results in terms of effective friction angle

3. PROGRESSIVE FAILURE IN OVERCONSOLIDATED CLAYS

Overconsolidated clays, especially when they exhibit a high plasticity, are brittle materials. The reduction of
peak strength drained envelope parameters to the residual state is dramatic. This is, in part, a consequence of the
low residual strength. Figure 5 shows one example: the evolution of drained strength of Guadalquivir marine
Tertiary clay with shear displacements, as measured in direct shear and ring shear tests. One can expect very
significant progressive failure mechanisms in this material if the peak strength is mobilized in points of the
failure surface.

Figure [Link] friction angles of slides across bedding planes or through clay matrix as a function of Plasticity Index

36
However, there are almost no field records showing the development of progressive failure in landslides. One
pioneering case is the Selborne failure experiment (Cooper, 1996; Cooper et al., 1998). The slope in this case
was excavated in OC high plasticity Gault clay and it was forced to fail by injecting water in an array of
boreholes. Figure 6 shows the position of inclinometers installed and the interpretation of measurements. Shear
stress is plotted along the shearing surface in terms of time. In this case, peak conditions were reached almost
simultaneously on the toe and crest of the slope. Failure progressed towards the center part of the sliding surface.

Figure [Link] envelopes of Guadalquivir blue clay

Figure 6. Shear strength states around slip surface. Selborne failure experiment (Cooper, 1996)

Another interesting case is the failure of Aznalcóllar dam, founded on Guadalquivir clays. This failure has
been analyzed in a number of papers (Alonso and Gens, 2006a,b; Gens and Alonso, 2006; Zabala and Alonso.
2011). The dam failure was actually a sliding failure. This is clear when observing the dam displacement which
led to the spill of acid tailings (Fig. 7).

Figure 7. Aznalcóllar dam failure. Cross-section of slide. Geometry after the slide, as interpreted from borehole data and
surface topography (Alonso and Gens, 2006a)

37
A simplified analysis developed by Gens and Alonso (2006) based on elasticity theory and a one-dimensional
dissipation of excess pore pressures suggested that the foundation clay in the vicinity of the observed position of
the failure surface experienced a progressive degradation of strength. This process, which was understood as a
progressive failure mechanism , was required to explain the case because no failure could be achieved by
conventional limit equilibrium or elastoplastic finite element analysis if the clay maintains its peak shear strength
(given in Fig. 5). One of the shortcomings of the analysis reported in Gens and Alonso (2006) is that the
numerical model used at the time (an elastoplastic Mohr-Coulomb material for the foundation clay) was unable
to reproduce the actual failure surface. In fact, a circular failure surface was predicted, far from the actual planar
surface. A possible explanation for the planar failure surface and its actual location was the existence of a weaker
layer at that position, a hypothesis never confirmed. In fact, borehole field records showed always a very
homogeneous and uniform clay stratum under the dam.
The inability of some standard FE methods to properly predict the actual position and geometry of the failure
surface was resolved in Zabala and Alonso (2011), who modeled the dam construction by the Material Point
Method. The foundation clay was characterized by a brittle non-associated Mohr-Coulomb model which could
reproduce the failure envelopes given in Figure 5. The consolidation problem was also solved in a coupled
manner with the mechanical problem. This analysis was able to reproduce in a surprisingly accurate way the
geometry of the failure surface (Fig.8). Figure 8 shows that the concentration of plastic shearing strains started
underneath the downstream slope of the rockfill dam. The shearing band extended first downstream and then in
the upstream direction. The shearing band defines a horizontal failure surface that eventually bends upwards
below the upstream toe of the dam slope and crosses the tailings’ deposit. Then, a forward motion of the dam
started.
The development of progressive failure and the actual shape and position of the final failure surface depends
on virtually any aspect which contributes to the instability: Geometry of the slope, details of constitutive
behavior, evolution of pore water pressures and the sequence of external actions eventually leading to failure.
Aznalcóllar was a case of a “downstream construction” of the rockfill dam (Fig. 8 shows the sequence of dam
construction). This advancing construction was especially dangerous because it facilitated a progressive failure
in the foundation soils. In order to illustrate these comments, consider some alternatives.
The first one is to assume that the entire dam is built in a single step in its final position. The weight was applied
slowly in this case (7300 days, to simulate the total construction time of the dam). This is a case which could
approximate the typical dam construction in horizontal layers starting at the base of the dam. Let’s assume that
the strength parameters are identical to the values adopted for the Guadalquivir clay, namely, ’peak = 24º; c’peak =
65 kPa and ’res = 11º; c’res = 0 kPa
Figure 9 shows this case. Contours of plastic shearing strains (1% and 1%) are plotted at the end of
construction. The figure indicates that no shearing strain is developed in the foundation. Failure is not predicted.
This result shows the extreme effect of the construction sequence to explain the dam failure.
Consider another case: The forward sequence of dam construction is maintained but the foundation soil is
represented by a classical elasto-plastic Mohr-Coulomb model characterized by c' = 0 and ' = 17º. These
strength parameters explain the failure of Aznalcóllar in a classical limit equilibrium analysis. Figure 10 shows
the failure surface obtained at the end of the forward construction of the dam. The failure surface bifurcates in
the downstream part of the slide. The overall shape is circular and extends to deeper clay levels, if compared
with the actual observations (Fig. 7). The actual shape and position of the failure surface can only be predicted if
the actual construction sequence and the brittle nature of the soil are conveniently modelled.
Summarizing the findings reported, the planar failure surface observed in the Aznalcóllar dam failure is a
consequence of (a) the particular construction sequence (a forward construction method of the rockfill retaining
dam) and (b) the softening of available strength (a rapid reduction of cohesion immediately after peak strength
and a progressive decrease of the friction angle). This is the conclusion for the specific geometry of the dam and
the specific dissipation of pore pressures of the real case.

38
Figure 8. Aznalcóllar dam failure. Sequence of contours of equal equivalent plastic strain, 1% and 5%. cv = 0.001 cm2/s, K0
=1

Figure 9. Contours of plastic shearing deformation of 0.1 % and 1% at the end of single step construction

39
Loading the dam in a single stage at a slow rate (applying the entire weight in a period of 7300 days) does
not induce any significant shear straining of the dam, even if the full degradation model for shear strength is
maintained.
If strength degradation is not considered and a constant average friction (and zero cohesion) explaining the
failure under limiting equilibrium conditions is adopted, the calculation shows a deep circular failure surface
even if a staged construction is imposed.

Figure 10. Contours of plastic shearing deformation of 1 % and 5% at the end construction (c’ = 0; ’ = 17º).

The set of results presented underline the significant effects of a number of practical aspects in the development
of failure. The forward advancing method of dam construction seems to be particularly critical to induce a
progressive failure. Strength softening of cohesion and friction in brittle materials results in a flat surface area.
This is an unexpected geometry that may lead to an erroneous conclusion, namely that a sedimentary feature (a
weaker sedimentation layer) was the reason for the observed failure geometry.

4. RUN OUT

Consider the following case history. We will call it the Leizarán slide. On May 2, 1992 a small rotational slide
(85 m long by 40 m wide) destroyed a house located in the close vicinity of a 22 m high embankment toe, killing
two people occupying the house (Fig. 11). The highway embankment was built by compacting, in horizontal
layers, heterogeneous coarse rock and rock-soil mixtures coming from quarry excavations. The natural ground
had an average slope of 22º and the embankment was built once a thin cover of colluvial soil was removed.
Figure 12 shows a plan view of the failure and Figure 13 is a representative profile along the central axis of the
slide. The rock substratum (unweathered ophites) exhibited a progressive weathering as the surface was
approached. Small reconnaissance trenches were excavated as part of the project design. They showed that the
thickness of soil-like highly weathered ophite could range from 3 to 5 m. Below, the intensity of weathering
decreased until the parent rock was found.
Heavy rains were recorded at the site during the two months preceding the failure. Springs are a frequent
occurrence on both sides of the highway axis. Immediately after the slide, water seeping out at the contact of the
embankment and the natural ground was observed.

40
Figure 11. General view of the embankment and the induced slide.

Figure 12. Plan view of Leizarán slide. The displaced gravel mantle at the embankment-natural soil contact provides an
indication of the slide displacement.

Figure 13. Leizarán. The natural soil profile before the construction of the highway embankment.

The position of these springs is marked by arrows in Figure 12. They suggest that the piezometric surface at
the time of the failure was close to the slope surface. A longitudinal trench filled with gravel was built along the

41
embankment toe. Its capability to reduce the phreatic level inside the relatively impervious weathered ophite is
probably very limited.
Figure 14 shows a central profile of the slide after failure. The profile shows also a likely position of the sliding
surface located within the weathered ophite. The strength of this material was the key soil parameter to analyze
the failure. The slide moved forward about 20 m and destroyed the house. The swimming pool (Fig. 12), a few
meters apart from the house, was not damaged.

Figure 14. Cross-section of Leizarán slide after failure.

A picture of the left flank of the slide is shown in Figure 15. The scarp left in the ground, shown in the figure,
had a height of 45 m. Block samples were taken at this location, in the highly weathered ophite, in order to
perform laboratory tests. The weathered ophite turned out to be a sand-clay mixture with a significant proportion
of gravel in it (Fig. 16). It classifies as SM-ML. Significant aspects are its relatively high liquid limit (wL = 41%)
low plasticity index (IP = 6%) and high void ratio (e  1.1).
Drained direct shear tests, as well as ring shear tests, were carried out. Figures 17 and 18 summarize the
strength envelopes obtained. Peak conditions were characterized by the strength parameters c’ = 70 kPa, ’ =
26.5º. “Residual” conditions as measured after 4 cm of accumulated relative displacement on the direct shear box
reduced c’ to 1530 kPa and maintained the drained friction angle (’ = 26.5º). Consistently, the ring shear tests
led to the same residual friction angle, ’ = 26.5º.
Back-analysis calculations were carried out in an effort to match as closely as possible all the
relevant geometrical data known from field observations after the failure. It was concluded that the
observed failure was fully consistent with a mobilized friction angle ’ = 26.5º, a small effective
cohesion c’ = 15 kPa along the failure surface and a phreatic level coincident with the surface of the
natural ground. A zero effective cohesion leads also to an acceptable simulation of the actual failure. It
only requires minor modifications in the position of the piezometric surface.

Figure 15. Leizarán slide. Lateral scarp in a residual ophitic soil.

42
Figure 16. Close-up of the weathered ophite. Ophite gravels are scattered in a red coloured clay matrix.

The failure was certainly induced by the weight imposed by the embankment. For this reason it can be
classified as a man-induced first time failure. The back-analysis required also the specification of strength
parameters for the less weathered ophite (c’ = 100 kPa; ’ = 26.5º) and for the embankment fill (c’ = 10 kPa; ’
= 28º). Moderate changes in these parameters do not modify conclusions, however, given the short lengths of the
failure surface inside these materials.

Figure 17. Strength envelopes determined in direct shear tests on undisturbed residual ophitic soil. Leizarán slide.

43
Figure 18. Residual strength envelope determined by ring shear tests on residual ophitic soil. Leizarán slide.

A conventional analysis of the slide would stop here. A satisfactory explanation for the reasons leading to the
failure has been given. However, the main question which explains the dangerousness of the slide, namely the
displacement and velocity of the slide, remains unanswered. This question is fundamental because it is at the
origin of the damage induced by the slide. In fact, if the slide had travelled 10 m instead of 20 m, the people
living in the house would have survived.
The stress-relative displacement records measured in the direct shear tests performed adds relevant
information (Fig. 19). The weathered ophite exhibits a brittle behavior.
The Bishop (1967) brittleness index:

 p  res
IB 
 res
(1)

where p is the peak strength and res is the residual strength, is IB = 0.6, when ߪ௡ᇱ = 50 kPa and IB = 0.46 for
ߪ௡ᇱ = 200 kPa. Being a first time failure in a brittle material, a progressive failure during the embankment loading
had most likely developed. The final failure mechanism involved a further decrease in the available strength
along the failure surface, once the unstable mass was in an incipient motion. Therefore, it accelerated. It came to
rest when the change in geometry equilibrated again (statically) the acting and resisting forces. This process
required, in this case, a forward slide displacement of 20 m. The challenge is to predict the motion once the slide
is unstable.
Such a challenge finds an answer if the stability of the slope is formulated in dynamic terms. This
is an easy exercise in simple geometries (one dimensional slides). In a general situation some advanced
numerical methods are being developed, such as the smoothed particle hydrodynamics (SPH) (Bui and
Fukagawa, 2008; Bui et al., 2011) and the material point method (MPM), among others.

The MPM formulation for a mechanical problem was presented in Sulsky and Schreyer (1996) where the
equation of dynamic momentum balance is discretized. Different authors have also extended the MPM to solve
coupled hydro-mechanical problems such as Zabala and Alonso (2011), Jassim et al. (2012) and Abe et al.
(2014).

44
Figure 19. Leizarán slide. Direct shear tests on undisturbed block samples taken in the vicinity of the sliding surface.
Effective normal stress: (a)  n = 50 kPa; (b)  n = 200 kPa.

Consider an academic exercise with the purpose of showing the capability of the MPM. The objective of the
analysis performed is to determine the relationship between clay brittleness and run-out. Of course, the problem
cannot be formulated in such a simple manner because the run-out will depend on additional factors. Therefore, a
particular case has been selected.

The instability of a slope, 6 m high and 37º steep, was analysed. The slope failure was triggered by increasing
the pore water pressure at the lower boundary simulating a phreatic level rise. The Selborne cut failure
experiment relied also in an increase in pore pressure to induce instability. This is a plane strain simulation. The
lower boundary was fixed and horizontal displacements along vertical contours were prevented. The water
pressure is zero along the slope surface, the lateral contours are impervious and saturated conditions are
considered during the calculation. The mesh was refined in the region where the failure is expected in order to
get more accurate results and optimise the computational cost.
A Mohr-Coulomb strain softening constitutive model from peak (’peak = 35º; c’peak = 5 kPa) to residual
conditions (’res = 25º; c’res = 0.5 kPa) describes the brittle behaviour of the soil. This is a “base” case. Changes
in brittleness were induced by changing residual parameters at constant peak strength values.
The calculation starts with the application of an increase in pore pressure along the lower boundary (during 1
second, up to 40 kPa). Afterwards, the water pressure on the boundary is maintained constant during all the
simulation.
The increase of pore pressure reduces the effective stresses in the slope. This fact leads to some points to
reach the peak conditions and the strain softening effect, implicit in the constitutive model, decreases the
strength parameters of the plastic zones down to the residual yield surface. As a result, the gravitational stresses
are sufficient to induce a progressive failure.
Figure 20 illustrates the progressive failure phenomenon. The evolution of the mobilized shear strength along
the shear band shows that the degradation of the material initiates at the foot of the slope and propagates
upwards.
A parametric study of 24 simulations was carried out with the aim of studying the slope stability and the
post-failure behaviour as a function of the soil brittleness. The initial geometry and boundary conditions are the
same for all simulations and identical to the one described in previously.

45
Figure 20. Distribution of the mobilized shear strength (normalized) along the initial shear band at different times.

Figure 21. Relationship between run-out

run (m) and the corresponding IB for all the simulations

Figure 21 shows the correlation between run-outs

run and IB. Run-out
out is defined here as the distance between the
toe of the initial slope and the toe of the slope after failure once equilibrium has been re re-established. Note that
this definition is not necessarily equivalent
equivalent to the maximum displacement achieved by any point of the slope.
The results indicate that a threshold (IB*) can be defined. Slopes with higher values of IB* become unstable. In
this study, the threshold IB* is around 0.5. Just above this value, the run-outs
r outs increase suddenly to 7 meters. Run Run-
out increases nonlinearly with IB in the manner indicated in the [Link]. To fully understand Figure 21 21, it is
important to highlight that when IB exceeds 0.83, the toe of the mobilized mass reach the boundary of tthe
domain, with is located at 26 m of the initial toe. Therefore, the maximum run-out
run out does not exceed 26 m.
Three stable geometries after failure
fail are illustrated in Figure 22 for three different values of IB at a constant
c’r. Therefore IB is controlled by the residual friction angle. The same pattern of displacements is observed in all
the cases. However, the run-outout increases and the slope becomes flatter with increasing values of brittleness.
Figure 23 shows the evolution of the velocity of a material point
point initially located just above the initial shear
band. This plot illustrates the different phases of an instability process. Initially, and for a given period of time,
the slope is stable and displacements and velocities are zero. At this stage, the progressive
progressive failure takes place.
When the initial failure mechanism develops, the velocity increases rapidly and the mobilized mass moves
forward. After a peak value of the velocity (in the cases analyzed it varies from 1 to 2 m/s), the velocity
decreases untill the slope reaches a new stable profile.
The maximum slide acceleration is calculated at the early stages of the motion. Maximum velocities increase
with decreasing residual friction and therefore with increasing IB. The higher the IB, the longer the time interval
to reach a final stable geometry.
The final stable geometry after failure is controlled by the residual strength. Lower residual frictional angles
leads to flatter final slopes.

46
Figure 22. Final geometry for three simulations in soils with different brittleness indices. cr’=0.5 kPa in all cases. The
displacements of the material points are indicated in the indicated colour scale.

Figure 23. Velocity of the material point P for the simulations 14, 15, 16 and 17

5. SLOPES IN PARTIALLY SATURATED SOILS

Slopes in soils of semi-arid

arid and arid regions of the world often exist in a state of partial saturation. Likewise,
slopes in compacted and residual soils arc also partially saturated and, in general, the geometry of a slope
favouring
ring the drainage towards the toe, is a mechanism to reduce the saturation of shallow levels. Soils covering
bedrock formations are frequently the result of weathering which lead to an increase in void ratio, an increase in
water content and a very marked reduction of strength. When affected by atmosphere-induced
atmosphere induced infiltration and
evaporation, suction will change in the soil. Under extreme events, slides may develop. Experience indicates that
very often rain-induced
induced failures are shallow translational and
a rotational
ational slides. Figures 24 and 25 show two
examples.

47
Consider an apparently simple problem: the stability of an infinite unsaturated slope subjected to rain
infiltration (Fig. 26). The shear strength will be defined in terms of suction as follows:

 f  c ' (  pa ) tan  ' ( pa  p ) tan b

(2)

where pa and pw are the air and water pressures and  and  are the shear and normal stress on a critical
sliding plane. The strength law is defined in terms of a cohesion and two friction angles, the latter one
characterizing the contribution of suction (pa–pw). A measure of safety is given by the Margin of Safety, defined
as

M   f 
(3)

Figure 24. Shallow slides induced by rainfall in Eocene marls. Bagà, Barcelona, Spain

Figure 25. Upstream natural slopes in Castrovido reservoir, Burgos, Spain. Weald siltstones and claystones. Felling the trees
in the lower part of the valley exposed the bare soil to intense atmospheric action, which resulted in numerous shallow slides

The slope will be subjected to a time history of infiltration and evaporation which is characterized in Figure
27 by a measured record in the city of Barcelona for a period of 6 years. The random character of this record is
obvious. Therefore, the problem may be formulated as follows: Find the solution of the infiltration equation:

48

( S r n )  div ( v  )  0
t (4)

Subjected to an inflow-outflow flux at the upper boundary, described by a random process in time:

Figure 26. Formulation of the one dimensional slope stability under unsaturated soil conditions and atmospheric action

Figure [Link] record measured in a meteorological station in Barcelona area during a period of 6 yrs. Plotted values:
averages of 10 days. Evaporation was measured in an evaporation tank 1.25 m in diameter, corrected to transform free water
surface to soil conditions

N
I (t )   An cos n2t  Bn sin n2t (5)
n 1

where An and Bn are random variables and n describes a set of frequencies. In the infiltration equation Sr and
n describe the degree of saturation and the porosity and vw is the water flow rate. Solution of this problem
requires information on the soil permeability under saturated-unsaturated conditions and the water retention
curve. In addition, once the suction distribution at every time is known, the Margin of Safety may be calculated
as indicated above. Details of the solution process are given in Rahardjo et al. (2008).
Consider first the solution of the infiltration problem for a simple sinusoidal variation of inflow/outflow at
the soil surface. It is plotted in Figures 28 and 29 for two frequencies, slow and high. The first one represents
seasonal changes. The second one describes short term weather conditions. Under low frequency infiltration the
soil reacts with a sizeable delay in its response. In addition, changes in suction are noticed at relatively large
depths (a few meters in Fig. 28). The high frequency infiltration is characterized by a rapid response of suction
close to the surface, by a negligible delay and by a short penetration of suction changes (Fig. 29).
A real infiltration record includes a wide range of frequencies and the solution to the problem will involve a
combination of long term and short term responses. The stochastic process given in Figure 26 was modeled as
outlined above and a closed form solution could be found for the random function describing the Safety Margin
(SM). A realization of SM is sketched in Figure 30. The slope will fail whenever SM reaches a zero value for the
first time. This is a “first passage” problem in random function analysis. The problem could also be solved in
analytic terms and expressions for the probability of failure were derived. They are plotted in Figure 31 in terms
of the slope angle for different depths of the failure plane for a reference period of one year. Other required soil
properties are, in the example solved: K = 0.7.10-7m/s; ’ = 30º; b = 25º; c’ = 1000z N/m2; as = 10-5 m2/N
(defines the retention curve). These parameters describe a partially saturated silty soil.
The theoretical solution in Figure 31 shows that the most likely failure for a significant range of slope angles
is a shallow failure. If the slope inclination increases the “mechanical” conditions dominate and a deep failure is
more likely. Figure 32 illustrates the effect of ground water level. Shallow water levels result in a higher

49
probability of failure. The reason is that shallow groundwater leads to wet soils, which imply low suctions and
therefore reduced strengths. It is also interesting to check that under deep groundwater conditions shallow
failures are more likely for an extended range of slope angles (weather effects dominate).

Figure 28. Low frequency sinusoidal infiltration

Figure 29. High frequency sinusoidal infiltration

Figure 30. A "first passage" problem

50
Figure 31. Probability of failure of an infinite slope for different slope angles and depths of failure plane

Figure 32. Effect of depth of groundwater level and slope angle

The analysis described stresses the effect of weather but the slope geometry and soil conditions
were a first approximation to actual cases. Soil instability, even if it is moderate in terms of run-out,
results in soil cracking (Fig. 33). Dilatant soils will experience a volume increase in the vicinity of the
shearing bands. In addition, since actual motions are rarely kinematically admissible, they imply
deformations of the soil mass which translate into fissuring at several scales.

51
Figure 33. Slope instability in a Miocene formation, Sabadell, Barcelona, Spain

The consequence is a dramatic change in infiltration conditions of the slope if compared with the
academic case of a homogeneous soil.
Changes in permeability induced by the fissuring induced by the deformation process may be
addressed by the technique described in Olivella and Alonso (2008). The idea was to embed a "joint
element" into a regular joint element for a FE analysis. The joint opens when tensile strains are
calculated. Then, the permeability of the developed joint is calculated by a Poiseuille-type of
expression and the overall element permeability is modified accordingly. This procedure may be
applied to any constitutive model describing soil behavior.
Consider now another academic example of increased complexity (Fig. 34). A slope is generated
by a cut in a partially saturated soil described by an elastoplastic constitutive model (BBM: Alonso et
al., 1990). The slope was created by a 15 m high cut in a period of 40 days. A water table is assumed
to exist at the bottom of the model. Initial suction is in equilibrium with this position of the water
level. A constant infiltration rate simulates rainfall conditions. Two cases are compared: a "cracked"
model which will react to tensile strains by a rapid increase in permeability and a classical
homogeneous model where permeability remains essentially constant. As the water infiltrates suction
reduces (in both cases) and a localization of shear strains begins at the slope toe. The deformation
process continues and eventually a fully developed shearing band defines a failure surface. The
shearing band in a dilatant material leads to tensile straining and to a substantial increase in
permeability in the "cracked" model. The implications of this increase in permeability are shown in
Figure 35. The calculated pore pressure in the cracked model reaches maximum values along the
failure surface. In fact the contours of constant water pressure define nicely the position and shape of
the failure surface.

52
Wetting Layers to excavate

P = -0.1 MPa
w

Figure 34. FE model of cut slope

Figure 35. Calculated pore water pressures. a) "Cracked" model; b) Uniform soil model

An additional consequence is the overall reaction of the slope In terms of the calculated evolution of suction
of a central point of the failure surface. This is shown in Figure 36. Three cases are represented: no rain, constant
permeability and a "cracked" model. The first case is trivial: suction is constant in time. Suction reduces
progressively and smoothly in the second case beyond the time required for the water to infiltrate to the position

53
of the point represented. In the third case the suction reacts as in the constant permeability case but it reduces
dramatically when the dilatant shearing
shearin arrives at the reference point.

The consequences of this change in hydraulic regime are illustrated in Figure 37, which compares the
forward slope displacements of a central point in the slope surface. The constant permeability case shows the
slope instability, but in a progressive manner:
manner: it is difficult to select a failure time. This goes against
observations. Failure is often observed as a very definite motion at a given time. This is the case of the "cracked"
slope. This is explained by the rapid decrease of suction (or even the development
development of positive pore pressures)
along the failure surface, because of the rapid increase in its permeability. Certainly, other phenomena may
explain the sudden failure of the slope, in particular a strain softening behavior. The analysis made highlights
that sudden failures may be also a consequence of internal changes in permeability associated with the
deformation of the slope.

Figure 36. Evolution of suction in a central point of the failure surface

Figure 37. Evolution of forward displacements in a central point of the slope face

54
5.1. A case history: Ancona slide, Italy

The case was initially described in Alonso et al. (2003). The case has been reviewed and a more
comprehensive computational analysis is presented here. A brief background of this case is included
here. In December 1982 a large flowslide in OC clays destroyed some suburbs of Ancona City. In
1991 a European EPOCH Project was funded to analyze the stability of slopes in clay in the context of
Mediterranean climate. A particular slope (“Villa Blasi” slope) was selected to perform the study.
Borings were drilled, specimens for laboratory testing were taken and inclinometers and piezometers
were located in the slope. Rainfall data was recorded and field observations during a period of one
year (October 1992October 1993) were available. The main emphasis of the analysis performed was
to relate slope safety with rainfall and hydraulic and strength properties of the soil profile.
The case illustrates the relevance of changes in permeability of the soil profile. This is a common
case in practice. In particular, weathered profiles of a particular rock substratum exhibit a “layered”
sequence of permeability and other soil properties. This situation was observed in the case described
here. Figure 38 shows a longitudinal profile and the position of boreholes, inclinometers and
piezometer sensors. Three layers could be identified: Layer : Remoulded brown Ancona clay, rich in
organic matter (CH-MH); Layer : Stiff brown Ancona silty clay with sandy inclusions (CH-MH) and
Layer : Pliocene substratum of blue Ancona clay (stiff blue silty clay with sandy inclusions).
Piezometer data indicated that two hydraulic regimes coexisted in the slope: A deep flow regime
associated with a phreatic level (4-14 m deep) nearly constant in both wet and dry seasons and a
superficial flow regime, directly controlled by rainfall.
One of the major uncertainties was a correct identification of the saturated values of permeability
of the three soil layers. It was decided to approach the selection of permeability by comparing
piezometric observations, under the recorded rainfall regime, with model predictions. A number of
calculation cases were defined. They are collected in Table 1.

Table 1. Rainfall infiltration analysis. Cases analysed. Saturated water permeability (m/s)

Case Layer  Layer  Layer  Nr of different materials

a 10-9 10-9 10-9 1

b 10-8 10-9 10-9 2
c 10-7 10-9 10-9 2
d 10-6 10-9 10-9 2
f 10-8 10-8 10-9 2
g 10-7 10-8 10-9 3
h 10-6 10-8 10-9 3
j 10-7 10-7 10-9 2
k 10-6 10-7 10-9 3

55
Figure 38. Longitudinal profile of Ancona slope

The rainfall recorded in the reference period is shown in Figure 39. Modelling infiltration required attention
to the actual capability of the soil to infiltrate water compared with the actual rainfall intensity. Two cases were
distinguished:
Controlled infiltration: intensity of rainfall, I, is smaller than the maximum infiltration, Qi. Inflow is
controlled by the rain intensity. Prescribed flow boundary : Qi = I.
Flooded slope: If rainfall intensity I > Qi, only part of the rainfall infiltrates. The infiltration depends on the
computed permeability and degree of saturation of the slope surface. Boundary condition for a flooded slope
(zero suction: s = papw = 0) applied at the surface of the slope.
One example of calculations is given in Figure 40. Comparison of the reaction of two piezometers are
compared with calculations for different combinations of profile permeability

Figure 39. Rainfall regime in Villa Blasi slope

Figure 40. Comparison of measured and calculated piezometer readings. a) Open tube piezometer C in substratum: Depth of
measuring chamber: 9.80 m; b) Electrical piezometer P5698 in  layer: Depth of sensor: 3.15 m

56
It became clear that some combinations of permeability reproduce measurements in a better and more
consistent way than others. In particular, cases h and i seem to be the winners. They indicate a progressive
reduction of permeability with depth. A consequence of this permeability layering is that rainfall leads to a flow
parallel to the slope in the upper more pervious layer. Vertical infiltration towards inner layers takes place at a
much reduced rate (Fig. 41).

Figure 41. Computed water velocity field at mid slope (case h: K = 106 m/s; K = 108 m/s and K = 109 m/s) at t = 288
days (July 15th, 1993)

Inclinometer data served to identify the nature of the field motion. The upper alpha layer was unstable. The
instability can be described as a surface translational slide. The shearing surface was located at the transition
between  and  layers at a depth of around 2.503.00 m (Fig. 42).
The calculated displacements at mid slope follow a similar pattern (Fig. 43). In fact, A BBM model,
calibrated by means of laboratory test results, describes the three soil layers. The main reason which explains
these results is hydraulic. In fact, the permeability sequence favors the development of positive pore water
pressures in the upper part of the slope, even if suctions are maintained at depth. This is shown in Figure 44.
Calculation shows that pore pressures in  layer reach a hydrostatic regime after a given rainfall period. Pore
water pressure becomes positive and reaches a maximum value for a prolonged period of time. Eventually, as
water infiltration accumulates, this process continues downwards.
Therefore, two contributing factors explain shallow instabilities observed in typical weathered profiles: the
reduced strength of the upper levels and the sequence of permeability changes.

Figure 42. Measured inclinometer profiles of horizontal displacements at borehole B

57
Figure 43. Calculated displacements at mid slope

Figure 44.
4 Calculated profiles of pore pressures at borehole B

6. CONCLUDING REMARKS

This paper starts by stressing the relevant role of sedimentation surfaces and shearing bands in explaining the
development of landslides in clay formations. It has been argued that the residual strength is fundamental
information because of the progressive
rogressive failure mechanisms operating in practice.
Progressive failure is a widespread phenomenon in brittle overconsolidated clayey formations. Field
observations as well as modelling results explain that the shape and position of the final failure surface depends
on the process of deformation. No simple rules can be given.
Risk associated with landsliding depends on the motion dynamics after instability. Some modern
computational tools are capable of dealing in an integrated manner with the pre- pre and post
post-rupture evolution of
the slope. The results of a theoretical
heoretical exercise showing the relationship between run-out
run out and brittleness index are
reported. The final geometry of the slope after run out is controlled by residual strength.

58
The final section discusses some aspects of the behavior of unsaturated slopes. Probably the most relevant
feature to explain the slope behaviour subjected to rain infiltration is the distribution of permeability (and water
retention). Weathering, in the case of a clayey substratum, results in well-defined horizons and marked
transitions in properties (void ratio, permeability, strength). It appears that failure surfaces are located at internal
boundaries defining sharp changes in permeability. Positive pore pressures may develop at these interfaces.
Open discontinuities, induced during the deformation process, imply significant chanes in permeability which
have a profound effect on the slope response against rainfall action. Other soil environments, not mentioned in
the report (loess, volcanic and tropical soils) require specific analysis

ACKNOWLEDGEMENTS

The author expresses his gratitude to his colleagues Prof. Olivella, Dr. N. Pinyol, Dr. F. Zabala and
Eng. A. Yerro for the help received during the preparation of this paper.

REFERENCES

Abe K., Soga K., Bandara S. (2014) Material Point Method for Coupled Hydromechanical Problems. Journal of
Geotechnical and Geoenvironmental Engineering. 1–16
Alonso E.E., Gens A., Josa A. (1990) A constitutive model for partially saturated soils. Géotechnique 40(3), 405430
Alonso, E. (1998) Cuttings and natural slopes. General Report. The Geotechnics of Hard Soils-Soft Rocks 2, 9911000
Alonso E.E., Gens A., Delahaye C.H. (2003) Influence of rainfall on the deformation and stability of a slope in
overconsolidated clays: a case study. Hydrogeology Journal 11, 174192
Alonso E.E., Gens A. (2006a) Aznalcóllar dam failure. Part 1: Field observations and material properties. Géotechnique
56(3), 165-183
Alonso E.E., Gens A. (2006b) Aznalcóllar dam failure. Part 3: Dynamics of the motion. Géotechnique 56(3), 203-210
Bishop A.W. (1967) Progressive failure with special reference to the mechanism causing it. Proc. Geotechnical Conference
142–150.
Bui H., Fukagawa R. (2008) Lagrangian meshfree particles method (SPH) for large deformation and failure flows of
geomaterial using elastic–plastic soil constitutive model. International Journal for Numerical and Analytical Methods in
Geomechanics 32, 1537–1570
Bui H.H., Fukagawa R., Sako K., Wells J.C. (2011) Slope stability analysis and discontinuous slope failure simulation by
elasto-plastic smoothed particle hydrodynamics (SPH). Géotechnique 61(7), 565–574
Cooper M. (1996) The progressive development of a failure slip surface in over-consolidated clay at Selborne, UK. Proc. 7th
Int. Symp. on Landslides 2, 683–688
Cooper M., Bromhead E., Petley D., Grant D. (1998) The Selborne cutting stability experiment. Géotechnique 48(1), 83–101
Gens A., Alonso E.E. (2006) Aznalcóllar dam failure. Part 2: Stability conditions and failure mechanism. Géotechnique
56(3), 185201
Hungr O., Leroueil S., Picarelli L. (2013) The Varnes classification of landslides types, an update. Landslides 11, 167194
Jassim I., Stolle D., Vermeer P. (2012). Two-phase dynamic analysis by material point method. International Journal for
Numerical and Analytical Methods in Geomechanics. 37(15), 2502–2522
Lupini J.F., Skinner A.E., Vaughan P.R. (1981)The drained residual strength of cohesivesoils. Géotechnique 31(2), 181213
Olivella S., Alonso E.E. (2008) Gas flow through clay barriers. Géotechnique 58(3), 157176
Rahardjo H., Rezaur R.B., Leong E.C., Alonso E.E., Lloret A., Gens A. (2008) Monitoring and modeling of slope response to
climate changes. Proc. 10th International Symposium on Landslides and Engineered Slopes 12, 6784.
Sulsky D., Schreyer H.L. (1996) Axisymmetric form of the material point method with applications to upsetting and Taylor
impact problems. Computer Methods in Applied Mechanics and Engineering 139(1–4), 409–429
Zabala F., Alonso E.E. (2011) Progressive failure of Aznalcóllar dam using the material point method. Géotechnique 61(9),
795–808

59
60
Innovative Geotechnichs for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Normalisation en Géotechnique
Standardization in Geotechnics
M.I. ZERHOUNI*1
1
Président de la Commission de Normalisation Reconnaissances et Essais Géotechniques – FONDASOL
Argenteuil – France

RÉSUMÉ -Une norme est un document de référence approuvé par un institut de normalisation reconnu tel
qu’AFNOR, en France. Elle définit des caractéristiques et des règles volontaires applicables aux activités. Elle
est le consensus entre l’ensemble des parties prenantes d’un marché ou d’un secteur d’activité. Ces dernières
décennies la normalisation dans le domaine de la géotechnique s’est largement intensifiée sous l’impulsion des
instances de normalisation européenne (CEN : Comité Européen de Normalisation) et internationale (ISO :
International Organization for Standardization), ainsi qu’au niveau des instituts de normalisation nationaux des
pays membres de ces instances, comme par exemple AFNOR, DIN, BSI, etc. au travers des groupes et comités
de normalisation miroirs. Nous présentons dans les paragraphes qui suivent un aperçu de l’organisation des
structures normatives œuvrant pour la normalisation en géotechnique et une synthèse des normes publiées ou en
cours d’élaboration.

ABSTRACT– A standard is a reference document approved by a recognized standardization institute such as

AFNOR, in France. It defines the characteristics and voluntary rules applicable to activities. It is the consensus
of all stakeholders in a market or industry. Recently the standardization in the field of geotechnics has largely
intensified in the European (CEN: European Committee for Standardization) and international (ISO:
International Organization for Standardization) standardization committees and at the level of national standards
institutes of the member countries of these bodies, such as AFNOR, DIN, BSI, etc. through the standardization
groups and mirrors committees. We present in the following paragraphs an overview of the organization of
normative structures working for geotechnical standardization and synthesis of published or in development
standards.

1 INTRODUCTION

Une norme est un document de référence approuvé par un institut de normalisation reconnu tel qu’AFNOR, en
France. Elle définit des caractéristiques et des règles volontaires applicables aux activité[Link] est le consensus
entre l’ensemble des parties prenantes d’un marché ou d’un secteur d’activité. Elle permet de définir un langage
commun entre les acteurs économiques -producteurs, utilisateurs et consommateurs-, de clarifier, d’harmoniser
les pratiques et de définir le niveau de qualité, de sécurité, de compatibilité et de moindre impact
environnemental des produits, services et pratiques.
Les normes facilitent les échanges commerciaux, tant nationaux qu’internationaux, et contribuentà mieux
structurer l’économie et à faciliter la vie quotidienne de chacun.
Les champs couverts par les normes sont aussi variés que les activités économiques et répondent aux
questions de société.
C’est ainsi que les recommandations d’une norme peuvent porter aussi bien sur des produits, des procédés,
des bonnes pratiques, des méthodes de mesure et d’essais, des systèmes d’organisation, des méthodes de calcul,
etc.
Ces dernières décennies la normalisation dans le domaine de la géotechnique s’est largement intensifiée sous
l’impulsion des instances de normalisation européenne (CEN : Comité Européen de Normalisation) et
internationale (ISO : International Organization for Standardization), ainsi qu’au niveau des instituts de
normalisation nationaux des pays membres de ces instances, comme par exemple AFNOR, DIN, BSI, etc. au
travers des groupes et comités de normalisation miroirs.
Nous présentons dans les paragraphes qui suivent un aperçu de l’organisation des structures normatives
œuvrant pour la normalisation en géotechnique et une synthèse des normes publiées ou en cours d’élaboration.

61
2 LES ORGANISMES DE NORMALISATION

Dans un souci d’harmonisation et pour faciliter les échanges, les normes sont majoritairement élaborées au
niveau international. En effet, désormais les normes peuvent être élaborées non seulement au niveau du pays
sous l’égide de l’institut ou du bureau de normalisation national correspondant, comme l’AFNOR en France,
mais également au niveau européen sous l’égide du CEN et de ses commissions techniques, ou encore au niveau
international ISO qui dispose également de commissions techniques correspondantes selon le domaine couvert
par la norme considérée.
Des accords internationaux permettent d’harmoniser et de transposer les normes entre ces niveaux CEN et
ISO et les pays qui s’y rattachent.
L’un des principaux accords de coopération technique entre CEN et ISO est connu sous le nom d’agrément
de Vienne(CEN et ISO, 2001).
Cet accord qui régit aujourd’hui le fonctionnement de la normalisation CEN et ISO, prévoit deux modes
essentiels pour le développement collaboratif de normes : le mode oùl’ISO est pilote (ISO lead) et le mode où le
CEN est pilote (CEN lead). Les projets de normes et les documents sont élaborés par l’un des modes et sont
soumisà l'approbation simultanée de l'autre mode.
Les avantages attendus de l'utilisation de cet accord conformément aux guides et aux lignes directrices de
mise en œuvre, comprennent :
- Accroître la transparence des travaux en cours au CEN aux membres de l'ISO et réciproquement, et de leur
possibilité d'influencer le contenu des normes ;
- éviter la duplication des travaux et des instances normatives, permettant ainsi de mobiliser plus
efficacement l'expertise des membres d’instances au profit de la normalisation internationale ;
- réduire les durées pour l'élaboration, la disponibilité et la maintenance des normes en établissant un
consensus commun qu'une seule fois.

Ainsi, il est possible de représenter la structure de l’organisation de la normalisation, comme indiqué sur le
schéma de la figure 1.

Figure [Link] de la normalisation – Niveaux et organismes de normalisation (Daubilly, 2014)

Dans ce contexte, aujourd’hui plus de 80 % des normes du domaine de la construction, du bâtiment et des
travaux publics, dont fait partie la géotechnique, sont désormais élaborées soit au niveau européen, avec un
pilotage du CEN, soit au niveau international, avec un pilotage de l’ISO. Les normes élaborées au niveau du
CEN et celles élaborées par l’ISO dans le cadre de l’accord de Vienne conduisent obligatoirement les pays
européens à reprendre et à transposer les normes homologuées européennes dans leur collection nationale de
normes (ex. en France NF EN… ou NF EN/ISO… ou NF ISO…) et à supprimer les normes du pays (ex. NF)
qui sont en contradiction avec ces normes européennes.

62
3 LES INSTANCES DE NORMALISATION EN GEOTECHNIQUE

En France, différentes commissions de normalisation (CN), sont en charge des travaux normatifs dans le
domaine de la géotechnique et des domaines connexes et apparentés. La plupart de ces commissions dépendent
du Bureau de Normalisation des Transports, des Routes et de leurs Aménagements (BNTRA).
Au niveau Européen (CEN) et au niveau International (ISO), ce sont les comités techniques (TC : Technical
Committee) dédiés aux domaines géotechnique et aux domaines connexes qui ont en charge l’élaboration des
normes de ces domaines. A ces comités techniques sont rattachés des groupes de travail (WG : Working Group)
affectés aux travaux d’élaboration d’une collection identifiée de normes.
Les commissions de normalisation françaises (CN) ont pour missions de gérer les normes nationales et de
servir de groupe miroir pour les comités techniques du CEN et de l’ISO et notamment préparer les positions
françaises pour les discussions européennes et internationales relatives à l’élaboration, la maintenance et à
l’évolution des normes.
Le tableau 2 ci-après donne la correspondance entre les commissions de normalisation françaises et les
différents comités techniques CEN et ISO.

4 LES DIFFERENTS TYPES DE NORME

Une norme est un document de référence qui apporte des réponses à des questions techniques et commerciales
que se posent de façon répétée les acteurs, sur des produits, des biens d’équipements ou des services. Elle est
élaborée en consensus par l’ensemble des acteurs d’un marché (producteurs, utilisateurs, laboratoires, pouvoirs
publics, consommateurs, etc.). Elle a pour vocation de présenter l’état de l’art, reconnu par l’ensemble des
parties concernées par une technique ou une pratique répétitive.

Tableau 1. Correspondance des commissions françaises, européennes et internationales (Magnan, 2014)

Domaine Nombre de France CEN ISO
normes(1)
Justification des ouvrages 9 CN JOG TC250/SC7 TC182
géotechniques
Reconnaissances et essais 114 CN REG TC341 TC182
géotechniques
Exécution des travaux 15 CN ETG TC288 -

Géosynthétiques 93 CN GSY TC189 TC122

Terrassements 42 CN ET TC396 -

Granulats 67 CN GRA TC154 -

Paravalanches 10 CN PRN - -

Missions géotechniques 1 CN MG - -
(1) Situation en 2014.

Etant le fruit d’une réflexion approfondie des experts qui participent à sa mise au point, une norme est établie
par consensus et approuvée par un organisme reconnu. Elle fournit des règles, des lignes directrices ou des
caractéristiques pour des activités, ou leurs résultats, garantissant un niveau d’ordre optimal dans un contexte
donné.
Une norme doit être univoque et non ambiguë. Son respect doit garantir un résultat et donc des décisions
basées sur ce résultat de manière reproductible et avec une marge d’erreur connue à l’avance. C’est pourquoi elle
est utilisée : dans les marchés comme spécification technique, dans les contrats commerciaux pour clarifier les
relations client-fournisseur, dans l’évaluation de la compétence et de la capacité à atteindre la qualité en vue
d’accréditation ou de certification, et dans bien d’autres applications.
De par son mode d’élaboration, elle constitue un référentiel accepté par tous.
Une norme est d’application volontaire et contractuelle. Mais de manière réglementaire, les pouvoirs publics
d’un pays peuvent rendre certaines normes d’application obligatoire. C’est notamment le cas pour des raisons
d’ordre public, de sécurité publique, de protection de l’environnement, de protection du patrimoine national, de
loyauté des transactions commerciales, de défense du consommateur, etc.
A titre d’exemple, il y a en France près de 150 normes homologuées d’application obligatoire, ce qui
représente environ 1% de l’ensemble des normes homologuées. Parmi ces normes d’application obligatoire, on
peut citer par exemple, certaines parties de l’Eurocode 8 (NF EN 1998), relatif à la sécurité des structures sous

63
séisme, certaines normes sur la sécurité au travail et également des normes sur la prévention en cas de travaux à
réaliser à proximité de réseaux (NF S70-003).
Les principaux types de normes ou de documents normatifs qui peuvent être publiées par les instituts de
normalisation sont résumés dans le tableau 1. En France, ces normes peuvent être d’origine française,
européenne (EN), ou internationale (ISO).

La référence d’une norme comporte des sigles qui permettent d’identifier son statut. Par exemple en France,
la codification est la suivante :
- « NF », « NF EN », « NF ISO » et « NF EN ISO » : il s’agit là de norme homologuée, d’origine,
respectivement, française, européenne, internationale ou européenne et internationale en cas d’accord.
L’homologation est une officialisation publique d’une norme en raison de sa destination (référence dans
la réglementation, les contrats ou les marchés publics, codification des règles de l'art, …) ;
- « XP », « CEN/TS », « ISO/TS » : lorsqu’il s’agit de norme expérimentale, d’origine, respectivement,
française, européenne ou internationale. Les normes expérimentales doivent faire l'objet, dans un délai
n'excédant pas cinq ans après leur publication, d'un nouvel examen par la commission de normalisation
compétente, en vue soit de les soumettre à l'homologation, soit de les remettre à l'étude, soit de les
supprimer ;
- « FD », « CEN/TR », « ISO/TR » : lorsqu’il s’agit de fascicule de documentation(technical report),
d’origine, respectivement, française, européenne ou internationale. C’est un document de normalisation à
caractèreessentiellement informatif ;
- « AC », « CWA », « IWA » : lorsqu’il s’agit d’un accord technique(workshop agreement), d’origine,
respectivement, française, européenne ou internationale.

Tableau2. Types de normes et documents de normalisation – Codification et tableau de correspondance

64
5 LE PANEL NORMATIF EN GEOTECHNIQUE

Dans le domaine de la géotechnique et ceux connexes ou apparentés, le catalogue de normes comporte plusieurs
catégories de normes qui peuvent être listées comme suit, avec :
- Des normes de conception, justification et calcul, comme par exemple les Eurocodes, leurs annexes
nationales et les normes d’application nationale associées ;
- Des normes d’essais, comme les normes de prélèvements et d’essais géotechniques in situ ou en
laboratoire et les essais sur ouvrages géotechniques et leur instrumentation ;
- Des normes de classification et de spécification de produits, comme par exemple les géotextiles,
géomembranes et produits apparentés, les granulats, les bétons, les liants hydrauliques y compris ceux
destinés au traitement des sols, les graves traitées, les classifications de sols, les filets et dispositifs de
protection contre la chute de blocs, etc. ;
- Des normes d’exécution de travaux géotechniques spéciaux, comme celles relatives à la réalisation de
pieux, de parois, de sol renforcé, d’injections, d’amélioration des sols, etc. ;
- Des normes d’organisation et de certification, comme celles relatives à la qualification des personnels et
des entreprises, ou encore celle relative aux Missions d’ingénierie géotechnique.

5.1 Normes de conception

Dans cette catégorie, on trouve aujourd’hui principalement les normes de l’Eurocode 7 (NF EN1997-1 et NF
EN1997-2), ainsi que l’annexe nationale correspondante.
Les Eurocodes sont des codes européens (CEN) de conception et de calcul des ouvrages de bâtiment et de
génie [Link] ont le statut de normes volontaires, et ont été transposés en normes nationales dans les états
membres du Comité Européen de Normalisation (CEN).
En France, depuis 2010 (fin de la période de transition), ils remplacent officiellement les normes ou
règlements nationaux équivalents existants.
Les Eurocodes servent de documents de référence pour les usages suivants :

- Comme moyen de prouver la conformité des bâtiments et des ouvrages de génie civil aux exigences
essentielles de la Directive Produits de la Construction (Directive DPC, 89/106/CEE et règlement UE,
305/2011), en particulier :
o l’Exigence Essentielle N°1 –« Stabilité et résistance mécanique » ;
o l’Exigence Essentielle N° 2 –« Sécurité en cas d’incendie » ;
o l’Exigence Essentielle N° 7 –« Utilisation durable des ressources naturelles ».

- Comme base de spécification des contrats pour les travaux de construction et les services techniques
associés (ex. Ingénierie, CCTP, DCE) ;

- Comme cadre d’établissement de spécifications techniques harmonisées pour les produits de construction
(marquage CE et agréments ATE).

La transposition de l’Eurocode dans chaque pays européen, comme en France a conduit à la publication des
normes homologuées suivantes :
- La norme NF EN 1997-1 : Calcul géotechnique - Règles générales(publiée en juin 2005 par l’AFNOR
(P94-251-1) ;
- l’Annexe Nationale française NF- EN 1997-1/AN, publiée en septembre 2006 par l’AFNOR ;
- l’Amendement A1 publié en avril 2014. Ce document apporte des corrections au texte initial de et
complète la section 8 de la norme EN1997-1.
- La norme NF EN 1997-2 : Calcul géotechnique – Reconnaissance des terrains et essais
géotechniques(publiée en septembre 2007 par l’AFNOR (P94-252 – 2ème tirage).
- (corrections de forme + section 8)
La transposition est complétée par les normes d’application nationale suivantes :
Les normes françaises d’accompagnement
- Fondations superficielles (NF P94-261 : juin 2013)
- Fondations profondes (NF P94-262 : 2ème tirage de 01/2013)
- Murs de soutènement (NF P94-281 : avril 2014)
- Ecrans de soutènement (NF P94-282 : mars 2009)
- Remblais renforcés et clouage (NF P94-270 : juillet 2009 – 2ème tirage)

65
Ces normes annulent et remplacent certains textes de référence comme le fascicule 62 – titre V ou les normes
NF P94-220 et NF P94-240.

5.2 Normes d’essais

Dans cette catégorie, on trouve aujourd’hui principalement les normes d’essais aussi bien in-situ qu’en
laboratoire référencés par l’Eurocode 7 ou par les autres normes associées à l’Eurocode 7, comme les normes
d’exécution de travaux.
Comme on l’a vu précedemment, l’élaboration de ces normes d’essais est prises en charge et pilotée (lead)
soit par le comité technique du CEN/TC341 soit celui de l’ISO/TC182).
Les collections de normes existantes à ce jour sont :
- Série des normes NF EN/ISO 17892-xx : Celles-ci, initialement publiées sous la forme de normes
expérimentales (XP/TS) couvrent les principaux essais géotechniques d’identification et mécaniques
réalisés en laboratoire. Ils sont actuellement en cours de révision par le CEN/TC341 et plus
particulièrement par le groupe de travail WG6 dédié ;
- Série des normes NF EN/ISO 22282-xx : Celles-ci sont relatives aux essais géohydrauliques réalisés in-
situ, comme les essais de pompage et les essais de perméabilité en forage. Elaborées par le groupe de
travail WG1 du CEN/TC341, elles ont été homologuées et publiées en 2013.
- Série des normes NF EN/ISO 22475-xx qui couvrent les Méthodes de prélèvement et les mesurages
piézométriques (exécution des forages, des prélèvements, piézométrie…). La norme homologuée a été
publiée en 2007. Une consultation au niveau CEN et ISO est en cours pour sa révision prochaine.
- Série des normes NF EN/ISO 22476-xx relatives aux essais géotechniques réalisés en place. Les normes
couvrant les principaux essais comme les essais pénétrométriques statique et dynamique, l’essai au
carottier SPT, l’essai pressiométrique Ménard ont été élaborées par le CEN/TC341 et ont été publiées en
tant que normes homologuées.
- Série des normes NF EN/ISO 22477-xx : relatives aux essais géotechniques réaliséssur des structures
géotechniques. Parmi celles-ci, on peut citer les essais de pieux, les essais de clous, les essais de tirants,
etc.

5.3 Normes de classification et de spécification de produits

Certaines normes définissent des spécifications et des caractéristiques de produits en fonction notamment de la
destination et du domaine d’application des produits concernés. Ces normes sont utilisées entre autres pour aider
au choix des produits adaptés et également comme référence pour le marquage et la certification de produits,
notamment dans le cadre des règlements couvrant les produits de la construction.
Parmi les produits pouvant être utilisés dans le domaine géotechnique, on peut citer :
- Les produits de fondations :
o NF EN 12794 : Produits préfabriqués en béton - Pieux de fondation
o NF EN 10248-xx : Palplanches laminées à chaud en aciers non alliés
o NF EN 10249-xx : Palplanches profilées à froid en aciers non alliés
o NF EN 10305-xx : Tubes de précision en acier
- Les produits géosynthétiques et produits apparentés : Le tableau 3 ci-après fournit un exemple de
normes produits rattachées à cette catégories.

Tableau3. Exemple de normes de spécifications de produits géosynthétiques

66
- Les Granulats et enrochements (CEN/TC154) : Le tableau 4 ci-après fournit un exemple de normes
produits rattachées à cette catégorie.

Tableau4. Exemple de normes de spécifications de produits géosynthétiques

5.4 Normes d’exécution de travaux géotechniques spéciaux

Ces normes définissent les processus et règles d’exécution de travaux géotechnique spéciaux. Elles sont
élaborées au niveau du TC288 du CEN. Aujourd’hui, la majeure partie des techniques pratiqués d’exécution de
fondations, d’ancrages, de renforcement et d’amélioration des sols et des travaux d’injection, etc. font l’objet de
normes homologuées.
Le tableau 4 fournit en exemple, un extrait des normes publiées entrant dans cette catégorie.

5.5 Normes d’organisation

Il est possible de regrouper dans cette catégorie :

- Les normes (expérimentales) de qualification des entreprises et du personnel
o TS EN ISO22475-2 : Reconnaissances et essais géotechniques - Critères de qualification des
entreprises et du personnel
o TS EN ISO22475-3 : Reconnaissances et essais géotechniques – Evaluation de la conformité
des entreprises et du personnel
- La norme française homologuée :NF P94-500 : Missions d’ingénierie géotechnique –publiée par
l’AFNOR en 2013.

67
Tableau5. Exemple de normes de travaux géotechniques spéciaux

6 CONCLUSION

L’élaboration des normes se fait désormais à l’échelle internationale EN ou ISO. Ceci va bien entendu en faveur
d’une harmonisation des pratiques et des produits.
En revanche, il devient parfois plus compliqué de suivre et de participer à cette élaboration des normes qui
exigent une présence et une implication dans les comités et groupes de travail en charge de la rédaction des
normes. Cette participation bien que facilitée aujourd’hui par les outils collaboratifs informatiques nécessite
malgré tout une assiduité régulière et une veille normative permanente.
Le panel normatif en géotechnique est vaste, avec des centaines de normes répertoriées (calcul, essais, produits,
exécution,…)
La veille technique et normative à assurer doit être rigoureuse et continue, en particulier pour les laboratoires et
les concepteurs (Ingénieristes, Maitres d’Œuvre…) qui participent à l’établissement des cahiers des charges et
des contrats, ainsi qu’à ceux en charge du suivi d’exécution des travaux.

REMERCIEMENTS

L’auteur remercie Jean-Pierre Magnan, les secrétaires des Commissions de normalisation du domaine
géotechnique et connexes et les différents intervenants à la journée sur la normalisation organisée par le CFMS
le 08 octobre 2015.

REFERENCES

AFNOR Normalisation (2006) Eventail des documents normatifs, Fascicule AFNOR, téléchargeable sur [Link]
AFNOR Normalisation (2013) Règles pour la normalisation française, Partie 1 - Instances et procédures de travail, Fascicule
AFNOR, 5eme édition, déc. 2013
Site internet de l’ISO – [Link]
Site internet du CEN – [Link]
Daubilly B. (2014) Organisation générale de la normalisation – Présentation à la journée CFMS du 08 octobre 2014 - Paris
Magnan J-P. (2014) Organisation de la normalisation en France – Présentation à la journée CFMS du 08 octobre 2014 -
Paris
ISO et CEN, (2001) Agreement on technical co-operation between ISO and CEN- VA codified – Version 3.3
Zerhouni M.I., Bigot G. (2002) La normalisation en géotechnique.Géologues132, 65-71.

68
Innovative Geotechnichs for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Controlled Low Strength Materials (CLSM) with Native Clays for Geotechnical
Civil Infrastructure: Design, Challenges, and Sustainability Assessments

Matériaux à faible résistance contrôlée (MFRC) à base de sols d’origine argileuse pour les
ouvrages géotechniques et infrastructures civiles: Conception, Défis, et vérifications de
Durabilité
Anand J. Puppala and Tejo V. Bheemasetti
University of Texas at Arlington, Arlington, Texas USA
*Corresponding Author: Anand J. Puppala; Email: anand@[Link]
David H. Marshall
Tarrant Regional Water District, Fort Worth, Texas

ABSTRACT – Native clayey soils possess high plasticity characteristics, which provide inevitable challenges
and problems in civil infrastructure. Typically, these soils are excavated and transported to landfills during
pipeline construction projects. Excavation, transportation and maintenance of these soils increase the project
costs and present potentially challenging sustainability issues. Effective use of these excavated soils can present
a sustainable engineering solution. An attempt was made in this research study to substitute native high plasticity
soil for aggregate bedding material in the preparation of controlled low strength material (CLSM) to be used as a
bedding material to support large diameter pipelines. Laboratory and field studies were conducted to evaluate the
properties of CLSM in controlled and field conditions. This paper highlights the use of CLSM with native high
plasticity soil, the challenges therein, and the sustainable aspects of adopting this innovative material for civil
infrastructure projects.

RESUME – Les sols d’origine argileusesont caractérisés par une forte plasticité qui conduit inévitablement à des
défis et des problèmes pour les infrastructures civiles. Ces sols sont typiquement excavés et transportés vers les
dépôts au cours de la construction de projets de pipelines. L’excavation, le transport et le dépôt de ces sols
augmentent le coût du projet et posent des problèmes sérieux de durabilité. Un usage efficace de ces sols excavés
peut constituer une solution durable en génie civil. En guise de solution, le but de la présente étude est d’utiliser,
pour les pipelines à grand diamètre, un matériau de pose à base d’agrégats où le sol à forte plasticité soit
substitué en vue de formuler le matériau à faible résistance contrôlée (MFRC). Des études en laboratoire et en
place ont été conduites pour déterminer les propriétés des MFRC en conditions contrôlées et de terrain. Ce
papier donne un éclairage quant à l’usage des MFRC avec un sol d’origine à forte plasticité, les défis qui en
découlent, et les aspects de durabilité en vue d’adopter ce nouveau matériau dans les projets d’infrastructures
civiles.

1 INTRODUCTION AND BACKGROUND

Sustainability is becoming a key aspect in major construction projects, as developing sustainable materials or
using native or reusable materials can substantially reduce project costs and minimize negative impacts on the
environment (Abreu et al., 2008). Societal benefits can be accomplished in minimizing the stress to the
communities surrounding the construction projects. The research team at The University of Texas at Arlington
(UTA) is attempting to develop a sustainable Controlled Low Strength Material (CLSM) for a major pipeline
construction project by using in situ excavated high plasticity material prevalent along the pipeline alignment
(Puppala et al. 2007, Raavi 2012, and Chittoori et al. 2014).
The present study, which is an on-going research conducted at UTA, has focused on the development of CLSM
mix designs, using native high plastic clay materials present along the pipeline alignment. If successful, the clay-
based CLSMs will enhance the sustainability aspects of major construction projects by reusing large amounts of
excavated fine clayey soils. Reutilization of excavated materials will positively impact the economy and the
environment, as landfilling these materials is expensive. Also, landfill operations at a project site often increase
vehicle-related emissions due to the transportation of material from the site to the landfill. Reutilizing the
excavated material can also minimize other issues such as damage to local city pavements around the
construction project site due to constant use by the dump trucks. Hence, the development of native clay-based
CLSMs is regarded an innovative and an appropriate sustainable solution for infrastructure projects.
The mix designs in detail, along with the literature available on various natural and manmade materials that can
be utilized in the CLSM mix designs, are extensively covered by Kumar et al. (2007) and Puppala et al. (2013).

69
Details of the research investigations were also summarized in Puppala et al. (2013), indicating that an attempt
was made in the laboratory to develop CLSMs, using native clays as fine aggregate materials.
As a part of the research study, flow, density, strength, and excavatability tests were conducted on CLSMs mixed
with fat clays (CH alone) and a mixture of fat and lean clays (CL and CH material mixture). The main focus of
the analysis was centered on whether the CLSMs designed have characteristics that are expected of traditional
CLSMs to be used in pipe bedding and pipe zone materials. Properties under consideration included flowability,
density, unconfined compressive strength (UCS), setting time, and excavatability. Both flowability and density
test results on several mix designs attempted were reported in Puppala et al. (2013). This present paper covers
undrained shear strength test results obtained by performing unconfined compression strength (UCS) tests, along
with the interpretation of stiffness properties and excavatability assessments of the same CLSM mixes.
Controlled low strength material (CLSM) is widely known as flowable fill and has become a popular material for
projects involving void fill, foundation support, bridge approaches, and conduit bedding (Folliard et al., 2008;
Bertola et al., 2013). CLSM with cement and fly ash additives has been proven an effective bedding material for
pipelines due to the material’s self-compacting behavior and strength performance (Rajah et al., 2012; Boschert
J. and Butler J., 2013). Research conducted by Achtemichuk et al. (2009) proved the effective performance of
CLSM material made from recycled concrete aggregates. This highlighted the versatile utility and high
performance of CLSM mixture as a bedding material for pipelines. It was also evident from previous studies that
the effective utilization of the native soil can substantially reduce the project costs and minimize the negative
impacts on the environment (Puppala and Hanchanloet, 1999; Abreu et al., 2008; Chittoori et al., 2012; Puppala
et al. 2012a).
The American Concrete Institute recognized CLSM as an effective material in the construction industry in the
1990’s (ACI 1994). Since then, CLSM has been successfully applied as a backfill material in the trenches,
minimizing the settlements of embankments (Hitch, 1998; Snethen and Benson, 1998). The flowability and self-
compaction characteristics of CLSM were successfully applied on a weak pavement base/subgrade for
immediate restoration by Lin et al. (2007). In a pipeline construction project, Farrag (2011) studied different
CLSM mixtures with fly ash additive for strength properties such as bearing resistance, compressive strength,
and resilient modulus. Investigations revealed that CLSM material with large amounts of fly ash is highly
susceptible to freeze/thaw damage. Field trench studies revealed that CLSM behavior is inconsistent in different
places due to sedimentation of CLSM material during transportation. Bertola et al. (2013) conducted
investigations on CLSM material manufactured, utilizing rapid hardening cement for pavement applications.
During this study, the short-term strength parameters, workability, and volumetric properties of the mixtures
were addressed. It was suggested that CLSM is an effective material for paving applications, increases the speed
of construction, and provides necessary long-term strength to pavements.
Raavi (2012) and Chittoori et al. (2014) demonstrated the effective usage of high plasticity clays in the mix
design for a pipeline project which was designed to supply water for the next 100 years. These studies addressed
the effectiveness of the CLSM-cast specimens when subjected to durability cycles. The sustainability aspects of
CLSM as an effective bedding material are highlighted in this research.

2 CONVENTIONAL MIX DESIGN OF CLSM

The components of controlled low strength material vary, depending on the availability of component materials,
as well as on the project-specific requirements and the application area of CLSM mixtures. The main
components of CLSM are aggregates (fine and coarse), cement, fly ash, lime, and water. Admixtures are also
used to enhance or retard some of the desirable properties such as setting time and density, depending on the
requirement. CLSM has mostly evolved using only conventional materials as aggregates. However, use of native
soils in place of conventional aggregates is the cheapest viable alternative for large projects, as it greatly reduces
the dumping of excavated material, saving land space and transportation costs. However, the CLSM mixed with
the soil has to pass the workability and constructability-related properties, which are discussed in the following:
1. When first mixed, it should be flowable (> 250 mm or 10 in. slump).
2. When cured, it should have a minimum strength greater than the soil it is replacing (>100 kPa [15 psi] for
most soils).
3. Its final strength should be low enough to allow it to be excavated (<2,000 kPa [300 psi] [ACI, 2005]).
These specifications change with the requirement and area of application of CLSM. A reasonable strength range
for CLSM is a 28-day strength minimum of 350 kPa (50 psi) and maximum of 1,000 kPa (150 psi) to allow for
some variation in actual reported test results (Finney et al. 2008). The lower-bound unconfined strength of 350
kPa (50 psi) is similar to the undrained shear strength of very stiff clay. The typical upper limit of 1,000 kPa
(150 psi) is similar to the unconfined strength of hard clay (Crouch et al.1998).
When CLSM is manufactured from the native soil, one major concern is the consistent strength of the mixture
produced. The grain size distribution of native soil varies more than processed sand, which necessitates varying
the mix design by accounting for the size variation in a given soil type. Duritsch (1993) conducted experiments

70
using clay-bonded sand, which resulted in higher water content to achieve required fluidity. However, greater
water content may cause aggregate segregation, increased bleeding, and reduced strength (Wu et al. 2005).
Therefore, the selection of suitable water content, making the material exhibit the best engineering performance,
is the first priority for the mix design. It should be noted that CLSM preparation using native clays has been
attempted mostly in the field. However, literature explaining the mix design is lacking. For example, the use of
native soil has been successfully implemented for the West Sacramento Force Main (WSFM) project, as a part of
Lower North West Interceptor program, which included construction of 6.4 km (4 mi.) through urban portions of
West Sacramento, California. The advantage of using trench-side soil outweighs the disadvantages, as long as
the variability of native soil materials is understood and strict attention to the quality control is maintained
(Finney et al. 2008).

2.1 Chemical Admixtures

Chemical admixtures are the materials that are added either before or during mixing to enhance or retard its
properties, such as flowability, setting time, strength, water content, and density of the mix (ACI 229R-99). Air-
entraining agents (AEAs) are the most commonly used chemical admixtures in CLSM. Some of the other
admixtures used are water-reducing admixtures, set retarders or accelerators, and super plasticizers (Pons et al.
1998; ACI 229R35 94). However, when used in high quantities, these admixtures may cause segregation
problems (ACI 229R-99). AEAs have been used to improve flowability and to enhance insulating characteristics.
Air-modified CLSMs typically contain 15 to 35% air by volume. Higher air content imparts greater workability
of the mixture. This condition leads to reduction in water content, which in turn minimizes segregation and
bleeding problems (Hoopes, 1998). Air volumes help in lowering the cost of CLSM by 10 to 30% (Riggs and
Keck, 1998). ACI 229R-99 mentions that air content in excess of 6% may increase segregation in the mixtures.
Water-reducing admixtures can be used in CLSM mixtures withlow fine content, which increase the overall
compressive strength, accelerate hardening, and decrease subsidence or settlement properties (ACI 229R-99).

2.2 CLSM - Properties

This section provides information on the properties of CLSM that most often affect its performance in key
applications. According to Folliard et al. (2008), fresh properties of CLSM are flowability, segregation, bleeding,
hardening time, and subsidence. Hardened CLSM properties include compressive strength, excavatability,
permeability, shear strength, resilient modulus, consolidation, shrinkage, and thermal conductivity.
In the case of pipeline applications, properties such as flowability, setting time, compressive strength, and
density are of major importance. Hence, this literature search mainly focused on gathering information on these
four fundamental properties of CLSM and how they are determined. This is described briefly in the following
sections.

2.2.1 Flowability

One of the important properties of CLSM is its ability to flow easily into confined areas, without the need for
conventional placing and compacting equipment. The CLSM’s self-leveling property greatly reduces the labor
required and increases the speed of construction. ASTM D 6103, “Flow Consistency of Controlled Low Strength
Material” is the most commonly accepted test procedure for addressing the flowability. A plastic cylinder with
dimensions of 150 mm (6 in.) and 76 mm (3 in.) inside diameter was lifted, allowing the CLSM to slump and
increase in diameter. The final diameter is typically used to differentiate between various degrees of flowability.
A final diameter of 203 mm (8 in.) or higher is the typical value of a flowable mixture.

2.2.2 Setting time

Setting time is the approximate period of time required for CLSM to gain sufficient strength to support the
weight of a person (ACI 229R-99). The setting time of CLSM material is affected by several parameters,
including mixture proportions, climatic conditions and drainage conditions. ASTM C 403, “Time of Setting of
Concrete Mixtures by penetration Resistance”) is the commonly used test in procedure in the laboratory. Other
techniques such as dynamic cone penetrometer and Kelly ball are also used for CLSM.

2.2.3 Unconfined Compressive Strength

Unconfined compressive strength is the significant property measured and is most commonly found in state
DOT specifications. CLSM compressive strength values are often used as an index for excavatability, when
future excavation may be required. Materials and mixture proportions must be selected to ensure that these

71
strength values are not exceeded in the long term. The development of CSLM compressive strength is different
from conventional concrete in that it is thought to have two components of strength: particulate and non-
particulate (Bhat and Lovell, 1996). The non-particulate component of strength results from the cementitious
reaction of cement and fly ash with water, whereas the particulate component of strength is similar in nature to
that of granular soil.
Water-cement ratio plays an important role in the development of unconfined compressive strength (Bhat and
Lovell, 1996). ASTM D 4832, “Preparation and Testing of Controlled Low-Strength Material (CLSM) Test
Cylinders” is the most common method used by state DOTs for evaluating CLSM strength. The characteristic
low strength of CLSM creates problems in testing cylinders because the many load frames used by researchers
are in the range of 1,300 to 2,200 KN (Folliard et al. 1999). But, for these cylinders with compressive strength of
1.0 MPa, the maximum load at failure is only about 18 KN, approximately 1 percent of the load frame capacity,
which makes the precision of the load frame doubtful (Folliard et al. 2008). This problem can be eliminated
when smaller-diameter cylinders are used (Folliard et al. 2008).

2.2.4 Density and Excavatability

Density of the CLSM mix primarily depends on the type of aggregate used. When used as backfill, high-density
CLSMs are hard to excavate at the time of repair of pipes; low-density CLSMs are preferred to facilitate ease of
excavation. Typical density of flowable fills reported in the literature ranged between 115 and 145 pcf (Pierce et
al. 2002). Hamilton County developed an empirical approach to predicting the long term strength or
excavatability of CLSMs (Du et al. 2002), which is both the function of 28-day uncompressive strength and
density of CLSM in the field. If the calculated value of removability modulus is less than 1.0, the specific CLSM
is considered to be removable.

3 CLSM MIX DESIGN USING NATIVE HIGH PLASTICITY CLAY

The key objectives of this research study were to develop the CLSM mix design using native soils selected from
Eagle ford geological formation. This material was to be used as a bedding material for a 150-mile water
pipeline. Utilizing the native high plasticity soil for a project of this large scale reduces the project cost by
cutting down the excavating, transportation, and maintenance of land fill costs. By reducing the carbon
emissions, it also provides a more sustainable solution than traditional concrete bedding materials. The present
CLSM mix with high plasticity clay was designed to satisfy the four target properties including flowability,
unconfined compressive strength, density, and setting time, and also to understand the impact of materialand
mixture proportions on CLSM characteristics. In order to achieve this, various CLSM mix proportions had to be
designed and evaluated for their properties in the laboratory. In the later sections, the laboratory and field studies
conducted on CLSM with native soil are discussed.

3.1 Laboratory Sample Preparation

As there was no standardized method established for preparing the soil samples, the test procedure adopted by
Folliard et al. (2008) was closely followed for preparing samples for flowability, density, compressive strength,
and setting time. This procedure is particularly applicable for medium stiff to stiff clayey soils. In order to obtain
a uniform soil-binder mixture, the bulk and undisturbed soil samples were first oven dried (at 60oC) and
pulverized to obtain fraction passing through a US Sieve 40 (0.425 mm). The procedures followed for soil-
binder mixing and specimen preparation is explained in following steps.
The approximate quantity of representative pulverized soil required for preparing the desired number of soil
samples was obtained. The appropriate amount of binders (cement, lime and fly ash) was determined and
weighed. The total water content, which included the in situ natural water content and the amount of water from
the designed mix proportion, was determined. The quantities of cement and/or fly ash or lime were mixed in dry
conditions separately, and then this mix was added and mixed with the soil in dry state, prior to the addition of
water, in a commercially available dough mixer. The mixing rate of the outer spindle was at 60 rpm, and the
inner spindle was rotated at 152 rpm. These rates were arrived at by trial and error to facilitate sufficient mixing
time without forming soil binder lumps. The established water content was slowly introduced into the soil-binder
mix. A flexible spatula was used to prevent the soil from sticking to the sides and bottom of the mixing bowl to
ensure that there were no soil-binder lumps in the mix developed.
Trial mixings were made to establish the particular mixing time that yields uniform soil - binder mixture. The
total mixing time in this study was 8 to 10 minutes. Finally, the soil binder mixture was transferred to the bowl
for flow, strength, and setting time tests. UCS samples were then casted for 7-day and 28-day strength tests. The
samples were left on the countertop for the first week, and then the 28-day samples were moved to the 100%
relative humidity room. They were removed from the humidity room at the time of testing, while 7-day samples

72
were left on the countertop until the time of testing. Humidity-controlled curing was not performed on the 7-day
samples, as they were still in wet condition. Hence, countertop curing was considered, and this method is similar
to the one outlined in the NCHRP report 597 by Folliard et al. in 2007.

TABLE 1 Summary of Basic Soil Characterization

TABLE 2 Summary of CLSM properties

3.3 Field Application of CLSM as a bedding material to support large diameter pipeline

The field testing in this research was primarily focused on determining the stiffness of CLSM prepared by using
native clayey soil. The Spectral Analysis of Surface Wave (SASW) method, (a nondestructive technique), was
employed. The test site was a part of the Integrated Pipeline (IPL) project running along the J1 line, located in
Mansfield, Texas. The SASW tests were performed in a prove-out section of the pipeline, which measured a
length of 500 feet. Figures showing the test site section with the pipe and the pouring of CLSM are presented
below.

Figure 1 500 ft. long pipeline section with CLSM around the pipe zone section

73
3.3.1 Stiffness monitoring of CLSM in field

The spectral analysis of the surface wave method (SASW) was performed inside a pipeline. Seventeen (17) test
sections were selected, along the prove-out section of the pipeline. The prove-out section extended to a length of
500 ft. Five (5) points, (South 2, South 1, Center, North 1, and North 2), were selected at each section to
determine the variation in stiffness of CLSM across the pipe of each section. Figure 2 illustrates the 17 test
sections selected, and Figure 3 depicts the cross-section of the pipe defining the alignment of test points at each
section and the procedure followed to conduct the field investigations. Sections 1070-50 and 1071-50 represent
joints in the pipeline. In addition to pipe testing, SASW tests were performed at joints to determine the variations
of stiffness of CLSM in comparison to the stiffness of CLSM at the other sections. As mentioned in the
experimental program, SASW tests were performed with a geophone spacing of 2 ft. A geophone spacing of 2 ft.
yields shear wave velocity measurements up to a depth of 4 ft.

4 CHALLENGES IN CLSM PREPARED WITH NATIVE HIGH PLASTICITY SOIL

In the earlier sections, the laboratory and field investigations of CLSM prepared by using native high plasticity
soil depicted a sustainable alternative to traditional CLSM mix design. However, there are certain challenges
involved in adopting this CLSM mix design. Reducing the plasticity characteristics of clay, setting time, and
flowability of high plasticity clay requires high water content. Chemical additives, other than set accelerator,
such as air entraining agents, which lower the required water content, should to be tried to meet the target setting
time, durability, and uniform stiffness increase of CLSM in field conditions.

Figure 2 Seventeen test sections tested inside the pipe

Figure 3 Cross section of the pipe

74
Figure 4 Stiffness of CLSM at section 1075-25

5 GEOSTATISTICAL MODELING

The field stiffness of the bedding material depends upon several factors such as environmental conditions,
construction process, volume of CLSM occupied, moisture content, and time. The seismic studies conducted
revealed stiffness values at regular intervals. However, in order to ascertain the field stiffness of CLSM along the
pipeline, geostatistical modeling was performed on the field data obtained from seismic studies. Geostatistics is
separate branch of statistics, which is mainly used in mining engineering to predict ore locations by describing
the probability distribution of existing ore (Krige, 1951).
The stiffness measurements of bedding material were utilized to determine the correlation between individual
measurements and their impact on unsampled locations. Using the correlation, the stiffness of the bedding
material after the 28-day curing period was mapped over the entire pipeline, using geostatistics, as shown in
Figure 5. From the figure, it can be inferred that the stiffness value of the CLSM was more than 380 MPa. Also,
there was uniform stiffness development along the pipeline at the end of the 28-day curing period. This indicates
that the CLSM prepared, using native clayey soil, performed effectively in real field conditions.

Figure 5 Contour map of CLSM stiffness after 28-day curing period

6 SUSTAINABILITY ASSESSMENTS

The present reuse applications have significant environmentally-related sustainability benefits since the
excavated material has proven to have the potential to be reused for pipe zone material applications. Usually, the
excavated clayey soil has to be landfilled; therefore, this application as a reuse fill or zone material application
will reduce landfilling costs. This will also reduce the use of aggregate base materials to support pipelines, which
means less quarrying-related blasting-induced emissions and less quarry fine-waste production. Other important
environmental and economical benefits include major reductions in transportation-related carbon emissions
arising from trucks hauling fill and bedding materials to pipeline construction and quarry locations. Overall, a
comprehensive reusability of excavated material from this project will result in major sustainability in
construction projects. Figure 6 shows the three sustainability-related social, economic, and environmental
impacts of reutilization of excavated material in a pipeline project. Chittoori et al. (2012) also noted that the
higher the overlap is among the social, economic, and environmental impacts, the more sustainable the project is.

75
Figure 6 Sustainability impacts from excavated soil reuse in pipeline construction project (Chittoori et al. 2012)

In order to quantify the sustainability benefits of using in situ excavated material versus imported material, a
hypothetical section of a pipeline was assumed and analyzed with respect to cost benefits and potential carbon
footprint analyses. A typical section of pipeline, highlighting the bedding, backfill, and haunch layers along with
the pipe section was selected. Two scenarios were considered in the analyses. The first one assumed that the
excavated soil would be reused as bedding and haunch layers, while the second scenario assumed that imported
soil from a nearby quarry would be used to provide material as bedding and haunch layers. It should be noted
here that both the scenarios assumed that the excavated material would be reused as a backfill layer application.
Comparisons of CO2 emissions and costs were made for both scenarios (Chittoori et al. 2012). Probable costs
and emissions due to various steps, including transportation of excavated material from site to landfill and
importing of material from quarry to the site, were estimated. In the second case, extra costs and emissions due
to purchasing and transporting of quarry material were included in the analysis.
The assumed section had a length of 30.5 m (100 ft.) and a bedding and haunch thicknesses of 180 mm (7 in.)
and 1070 mm (42 in.), respectively. The depth and width of the trench were assumed to be 4.6 m (15 ft.) and 3.7
m (12 ft.), respectively. The diameter of the pipe was 2.1 m (7 ft.). The nearest landfill to dump the excavated
material and the quarry from which to import bedding and haunch materials was assumed to be at a distance of
96.5 km (60 miles) and 40 km (25 miles), respectively. The volume of the each of the layers was calculated
based on these sections, and the amounts of soil volumes were calculated. Assumptions were also made
regarding the landfill costs and the price per gallon of fuel for transportation of materials. Mileage of a standard
dump truck was used in this analysis to calculate the number of trips required for landfilling and importing
materials. The following provides final values based on these calculations.
For the first scenario, the total cost of the excavated material being used as a pipeline zone material, the total
material costs for a 30 m short section was approximately $12.6 per m, and the carbon footprint was
approximately 0.015 metric ton per m length of the pipeline. For the second case, of using material from
quarries, the cost was $81.7 per m, and the carbon footprint was around 0.08 metric ton per m. Though CLSM
preparation can increase both costs and carbon numbers, the solution is still sustainable, as material reuse,
landfilling costs, and transportation costs can be reduced. More cost and carbon analyses will be attempted in the
future based on the actual costs incurred in the real field construction works.
Currently, the research team is looking at the use of CLSM for preparing treated foundation clayey subgrades to
support pavements. This could provide considerable benefits, as CLSMs could deliver a more resilient
foundation to support pavement infrastructures (Bheemasetti et al. 2015). Overall, CLSMs with native clays can
provide sustainable benefits to infrastructure by alleviating distresses to structures and provide cost effective
solutions by reducing landfilling and environmental issues posed by construction projects.

7 SUMMARY AND CONCLUSIONS

The use of existing native soils, such as high plasticity clays, presents challenges in design and construction of
civil infrastructures. The high plasticity characteristics and pore water pressures induced cause excessive
swelling pressures, resulting in damage to various civil structures. Consequently, practitioners often excavate
these soils in an attempt to provide a more reliable foundation for infrastructure. However, excavation of these

76
soils not only increases the budget costs, but also impacts sustainability. In this study, the UTA geotechnical
team attempted to use native clayey soil as an ingredient for controlled low strength materials. Hence, laboratory
and field investigations were performed to evaluate the strength characteristics of CLSM prepared with native
clays.
• Several laboratory mix designs were conducted using native clayey soils. Stabilizers such as cement and lime
were added to achieve the strength; 5% cement and 10% lime resulted in desirable results.
• The flowablity, density, and unconfined compressive strength properties obtained for CLSM, prepared using
native clayey soils, met the traditional CLSM specifications. However, high water content is required to meet the
flowability criteria due to high liquid limit. Field trail sections were built with cement-based CLSM only.
• The field investigations were performed using the non-destructive seismic method, SASW, to evaluate the
stiffness of CLSM as a bedding material for the water pipeline. The seismic analysis concluded that the stiffness
of CLSM is enhanced with time in field conditions at all the sections.
• The geostatistical modeling was performed to evaluate the stiffness of bedding material along the pipeline.
Uniform stiffness development was observed along the pipeline at the end of the 28-day curing period, with
stiffness values more than 380 MPa.
• The sustainable studies conducted on the CLSM revealed cost savings of 84.5% per m when using the
excavated material, and decreased the carbon emissions by 81.2%.

ACKNOWLEDGEMENTS

The authors would like to thank Shelly Hattan of TRWD, and Matt Gaughan of the IPL team for their assistance
with various research activities related to soil sampling and coordination among various groups and research
teams. We would also like to thank Bhaskar Chittoori from Boise State University for his support. We also thank
Anil Raavi and Rathna Mothkuri, former UTA Graduate Research Assistants, for their assistance with the
project. The researchers also thank the Fugro group for their help in transferring soil samples to UTA
Geotechnical laboratories.

REFERENCES

Abreu, D. G., I. Jefferson, P. A. Braithwaite, and D. N. Chapman, (2008). “Why is sustainability important in geotechnical
engineering?” Proc., GeoCongress 2008, ASCE, Reston, VA, 821-828.
Achtemichuk S., M. H. Shehata, (2009). “The Utilization of recycled concrete aggregate to produce controlled low-strength
materials without using Portland cement” Cement and Concrete composites, V 31, pp. 564-569.
ACI Committee 229. (1994). “Controlled low-strength materials (CLSM) (ACI 229R-94).” Concr. Int., 16(7), 58-64.
American Concrete Institute (ACI) Committee 229. (1994). “Controlled Low-Strength Materials,” Report 229.94, Vol.16,
No.7, American Concrete Institute, Framington Hills, 55-64.
Bhat, S.T. and C. W Lovell, (1995), “Flowable Fill Using Waste Foundry Sand,” Research Report, Purdue University, W.
Lafayette, Ind.
Bheemasetti, T. V., A. Pedarla, A. J. Puppala, R. Acharya, (2015). “Design of Sustainable High Volume Roads using
Controlled Low Strength Material from Native High Plasticity Soil” Transportation Research Record, Washington, Dc
(In Press).
Bertola, F., M. Bassani, F. Canonico, and M. Bianchi, (2013). “Use of Rapid-Hardening Cement for Controlled Low-Strength
Materials for Pavement Applications” Transportation Research Record No. 2363, pp.77-87.
Boschert J., and J. Butler, (2013). “CLSM as a Pipe Bedding: Computing Predicted Load using the Modified Marston
Equation” ASCE Pipelines 2013, pp. 1202-1213.
Chittoori, B., A. J. Puppala, R, Reddy, and D. Marshall, (2012). “Sustainable reutilization of excavated trench material.”
GeoCongress 2012, ASCE, Reston, VA, 4280-4289.
Chittoori, B., A. J. Puppala, A. Raavi, (2014). “Strength and Stiffness Characterization of Controlled Low-Strength Material
Using Native High –Plasticity Clay.” J. Mater. Civ. Eng., 10.1061/ (ASCE)MT.1943-5533.0000965, 04014007, 1-8.
Crouch, L. K., R. Gamble, J. F. Brogdon, and C. J. Tucker, (1998) “Use of High-Fines Limestone Screenings as Aggregate
for Controlled Low-Strength Material (CLSM),” The Design and Application of Controlled Low- Strength Materials
(Flowable Fill), ASTM STP 1331, A.K. Howard and J.L. Hitch, Eds., American Society for Testing and Materials.
Du, L., K. Miqiel, K. J. Folliard, S. Nazarian, and D. Trejo, (2006) “Rapid-Setting CLSM for Bridge Approach Repair,” ACI
Materials Journal, Vol. 103, No. 5, 312-318.
Duritsch. D. W., (1993). “The Use of Non-Toxic Spent Foundry Sand into Controlled Low-Strength Materials in Ohio,” The
Institute of Advanced Manufacturing Sciences, Cincinnati, 6 pp.
Farrag, K., (2011). “Controlled Low strength Material used Around Buried Pipelines,” Transportation Research Record
No.2251, pp. 157-164.
Folliard, K. J., David Trejo, and S. A. Sabol, (1999). “Controlled Low-Strength Material for Backfill, Utility Bedding, Void
Fill, and Bridge Approaches,” NCHRP 24-12 Phase I Interim Report, Delaware Transportation Institute, University of
Delaware, Newark.
Folliard, K. J., David Trejo, S. A. Sabol, and D, Leshchinsky, (2008). “Development of a Recommended Practice for Use of
Controlled Low-Strength Material in Highway Construction”, NCHRP 597 Report, 5-39.

77
Finney, A. J., E. F. Shorey, J. Anderson, (2008). “Use of Native Soil in Place of Aggregate in Controlled Low Strength
Material (CLSM)” Pipeline Congress, ASCE Journal of Materials in Civil Engineering.
Hitch, J. L., (1998). “Test methods for controlled low-strength material (CLSM): Past, present, and future.” ASTM Spec.
Tech. Publ., No. 1331, 3-10.
Hoopes, R.J., (1998). “Engineering Properties of Air-Modified Controlled Low-Strength Material,” The Design and
Application of Controlled Low- Strength Materials (Flowable Fill), ASTM STP 1331, A. K. Howard and J. L. Hitch,
Eds., American Society for Testing Materials.
Krige, Danie G. (1951). “A Statistical Approach to Some Basic Mine Valuation Problems on the Witwatersrand.” J. of the
Chem., Metal. And Mining Soc. of South Africa 52 (6): 119-139.
Kumar, A., B. S. Walia, and A. Abjaj, (2007). “Influence of fly ash, lime, and polyester fibers on compaction and strength
properties of expansive soil.” J. Mater. Civ. Eng., 19(3), 242–248.
Lin, D. F., H. L. Luo, H. Y, Wang, and M. J. Hung, (2007). “Successful Application of CLSM on a weak pavement
Base/Subgrade for Heavy Truck Traffic.” J. Performance of Constructed Facilities, 10. 1061/ (ASCE) 0887-3828, 70-77.
Pons, F., J. S. Landwermeyer, L. Kerns, (1998). “Development of Engineering Properties for Regular and Quick-Set
Flowable Fill.” The Design and Application of Controlled Low-Strength Materials (Flowable Fill), ASTM STP 1331, A.
K. Howard and J. L. Hitch, Eds., American Society for Testing and Materials.
Pierce, C. E., S. L. Gassman, and T. M. Richards, (2002). “Long-Term Strength Development of Controlled Low-Strength
Material,” ACI Materials Journal, Vol.99, No.2, 157-164.
Puppala, A, J., and S. Hanchanloet, (1999). “Evaluation of a new chemical treatment method (sulphuric acid and lignin
mixture) on strength and resilient properties of cohesive soils.” Proc., 78th Transportation Research Board Annual
Meeting, Transportation Research Board, Washington, DC, USA.
Puppala, A. J., P. Balasubramanyam, and R. S. Madhyannapu, (2007) “Experimental investigations on properties of
controlled low-strength material.” Ground Improvement J., 11(3), 171-178.
Puppala, A. J., S. Saride, and R. Williammee, (2012a). “Sustainable reuse of limestone quarry fines and RAP in pavement
base/subbase layers.” J. Mater. Civ. Eng., 10.1061/ (ASCE) MT.1943-5533.0000404, 418-429.
Puppala, A., Chittoori, B., and Raavi, A. (2015). “Flowability and Density Characteristics of Controlled Low-Strength
Material Using Native High-Plasticity Clay.” J. Mater. Civ. Eng., 27(1), 06014026.
Raavi, A. (2012). “Design of controlled low strength material for bedding and backfilling using high plasticity clay.” M.S.
thesis, Univ. of Texas, Arlington, TX, 118.
Rajah S., M. McCabe, and J. Plattsmier, (2012). “Classification and Specification of Bedding and Backfill for Buried
Pipelines,” ASCE Pipelines 2012: Innovations in design, construction and maintenance. pp: 940-951.
Riggs, E. H., and R. H. Keck, (1998). “Specifications and Use of Controlled Low-Strength Materials by State Transportation
Agencies,” The Design and Application of Controlled Low-Strength Materials (Flowable Fill), ASTM STP 1331, A. K.
Howard and J. L. Hitch, Eds., American Society for Testing and Materials.
Snethen, D. R., and J. M Benson, (1998). “Construction of CLSM approach embankment to minimize the bump at the end of
the bridge.” ASTM Spec. Tech. Publ., No. 1331, 165-179.
Wu, J. Y., E. Eng, (2005). “Soil Based Flowable Fill for Pipeline Construction,” “Pipelines Optimizing Design, Operations,
and Maintenance”, Pipelines 2005: pp. 925-938.

78
COMPORTEMENT DES SOLS

SOIL BEHAVIOUR

79
80
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Analysis of the time-dependent deformation behavior of soft soils

L‘Analyse du comportement de déformation des sols mous en fonction du temps

R. Katzenbach1,*& S. Leppla1
1
Technische Universität Darmstadt, Insitute and Laboratory of Geotechnics, Germany
*
Corresponding Author

ABSTRACT - In principle short time compression tests are carried out for the analysis of the deformation
behaviour of soils. For the analysis of the time dependent deformation behaviour of soft soils these short time
compression tests underestimate the creep settlements. Long time compression tests are necessary for the
prediction of the stiffness of soft soils with strongly time dependent deformation behaviour. This aspect is
explained by an example from engineering practice. It was planned to construct a new seaport on very soft silty
clays and clayey silts of high compressibility and of low shear strength. Regarding the stability and the
serviceability of the harbour constructions like quay walls, buildings and cranes the weak subsoil conditions had
to be considered. Especially the defined settlement restrictions lead to a complex design including soil exchange
and surcharges in combination with a vertical drainage system. The paper presents the test results of the long
time compression tests of very soft soils and focuses on the consequences for the design of the planned seaport.

RÉSUMÉ Pour déterminer le comportement de déformation des sols, il est courant de réaliser des essais
oedométriques de courte durée. Dans le cas des sols mous, ces essais sous-estiment les tassements apparaissant
avec le fluage. Pour prévoir la rigidité de ces sols mous dont le comportement de déformation dépend fortement
du temps, il est donc nécessaire de réaliser des tests à l’oedomètre sur une plus longue durée. Cet aspect est
expliqué ici à l’aide d’un exemple concret dans le cas de la construction d’un nouveau port maritime sur de
l’argile limoneuse molle et sur des limons argileux fort compressibles et avec une faible résistance au
cisaillement. Les conditions liées à ce soussol faible sont à considérer lors de l’évaluation de la stabilité et de
l’aptitude au service des éléments de construction du port tels que les murs de quais, bâtiments et grues. La
limitation des tassements en particulier amène à un design complexe incluant remplacements du sol et surcharges
en combinaison à un système de drainage vertical. L’article présente les résultats des tests à l’oedomètre de
longue durée sur des sols très mous et les conséquences sur le design du port maritime.

1 INTRODUCTION

Due to the increasing global trade activities connected with an increasing need of infrastructures more and more
constructions have to be realized in poor subsoil and groundwater conditions. For example roads and railways
are built in areas with creeping slopes or bridges and harbors built in soft soil. The paper focuses on
constructions of infrastructures on soft soils. For a safe design the load deformation behavior has to be
considered. Therefore the correct determination of the stiffness of soft soils is significant.
For determination of soft soils regulation 90 of DGGT 2008 is useful. Soils may be soft if they have one of the
following characteristics:

 soft consistency (0.50 < IC < 0.75)

 shear strength of the undrained soil of 20 kN/m2 < cu < 40 kN/m2
 (nearly) total saturation with water
 affinity to flow
 slight plastic behavior
 thixotropic behavior
 organic components

81
Soils are soft if they have one of the following characteristics:
 mushy to liquid consistency (IC ≤ 0.50)
 shear strength of the undrained soil of cu ≤ 20 kN/m2
 high sensitivity to vibrations
 water content of w ≥ 35 % for soils without organic components
 water content of w ≥ 75 % for soils with organic components

2 STIFFNESS OF SOFT SOILS

Different constitutive laws are used for analytical and numerical analysis of the load deformation behavior of soft
soils (Neher 2008). Normally the stiffness of soil is determined by the oedometer test, which is a laboratory
compression test (von Soos & Engel 2010). Figure 1 shows on the left a simplified cross section of an oedometer
test. Normally the load periods in oedometer tests last several days resp. weeks.
The settlement behavior of soils consists of 3 parts: the instantaneous, the primary and the secondary part
(Terzaghi 1943, Scott 1963, Atkinson & Bransby 1978, Kempfert & Gebreselassie 2006). Figure 1 shows on the
right the 3 parts of the total settlement. The instantaneous settlement is time independent and occurs immediately
after loading. The primary settlement of cohesive, water saturated soil material is time related to the
consolidation process due to the low permeability. The primary settlement of non-cohesive soil material occurs
immediately after loading due to the high permeability. The secondary settlement is time related caused by
creeping and is normally only recognized on cohesive soil material.

Figure 1. Simplified cross section of an oedometer test (left) and time settlement curve of cohesive soil material (right).

3 INFRASTRUCTURE ON SOFT SOILS

3.1 Harbor constructions

The significance of the correct determination of the stiffness of soft soils will be explained by an example from
engineering practice. This example shows that normal short time compression tests with load periods of days or
weeks are definitely not sufficient for the correct determination of the stiffness of soft soils.
The following example of infrastructure construction on soft soil is a planned new sea port. Port authorities
and terminal operators have to construct new and larger sea ports to face the increasing container traffic (Grabe
2011). The harbor facilities grew to enormous dimensions. Especially the ship-to-shore-cranes may have a mass
up to 1,600 t, a lifting capacity of 85 t and a rail gauge of 35 m. Requirements to the harbor facilities are
described by Küster 2011.
In many cases land reclamations from the sea are carried out before construction of the harbor facilities starts.
Regarding the stability and serviceability of the harbor constructions like quay walls, buildings and cranes
subsoil conditions with very soft soil layers may cause problems like instability and settlements. The main
possibilities to avoid or minimize these problems are:

 construction of deep foundation systems

 installation of anchor systems
 improve soft soil layers
 exchange of soft soil
 drainage of the soft soil

82
 surcharge

Instructions for the construction of quay walls and for land reclamation from the sea are given for example by
EAU 2004, DGGT 2008 and Witt 2003. Regarding the strict investor´s requirements of harbour facilities the
correct determination of the stiffness of the subsoil is very important. Often the maximum long-term
deformations of crane tracks are limited to 1 cm in vertical and to 1.5 cm in horizontal direction. Even for the
container storage areas a differential settlement for 20 ft. containers are limited to ± 2 cm. Considering that in
coastal areas the subsoil is often soft and shows long term settlement behavior, the observance of these
restrictions demand significant technical efforts.

3.2 Project overview

For the construction of a new seaport in the Gulf of Finland land reclamation was necessary. Figure 2 shows the
area of the planned harbor. Figure 3 shows on the left the layout of the planned harbor including the positions of
the investigation drillings. In total more than 100 drillings have been carried out (Figure 3 on the right). The
design of the planned quay wall in a first planning stage is shown in Figure 4. The harbor is gained by dredging
up sand up to its final level.

Figure 2. Area of the planned harbor.

Figure 3. Overview of the planned sea port and the investigation drilling (left) and investigation drilling at the construction
site (right).

83
Figure 4. Cross section of the quay wall in preliminary planning.

3.3 Subsoil conditions

The subsoil conditions in the cross section 1-1 marked in Figure 3 on the left are shown in Figure 5.
Approximately the upper 20 m subsoil consists of sand-silt mixtures, partly organic and liquid clays, soft silty
clays and clayey silts. According to DGGT 2008 the silty clays and clayey silts (layers 1 and 2) are classified as
soft soil. The soft soil is underlain by soft to stiff sandy, gravelly silts and clays (layer 3), followed by very stiff
to hard clays and clayey silts (layer 4).

Figure 5. Subsoil conditions on the construction site, cross section 1-1 of Figure 3 on the left.

Figure 6 displays on the left the range of grain size distributions of the soft soil. Figure 6 shows on the right
the classification of the soft soil according to the plasticity chart. The consistency and the pore volume over
depth are drawn in Figure 7 on the left. The basic soil parameters of the described soil layers are shown in Figure
7 on the right.

Figure 6. Grain size distribution (left) and plasticity index (right) of the soft soil.

84
Figure 7. Consistency and pore volume over depth (left) and soil parameters of soil layer 1 - 4 (right).
3.4 Stiffness of the soft soil layers
The oedometer test of a soil specimen, taken 18 m under sea level, shows that at the maximum load step of 380
kN/m2 the compression of the soil sample is more than 10 % of its height (Figure 8).

Figure 8. Oedometer test of a soft soil specimen.

Even after more than 10,000 hours (more than 400 days) of loading the settlement is still increasing.
Apparently the behavior of the time settlement curve correlates to the theoretical curve in Figure 1 on the right
only under small load steps. At high load steps it seems that even with a consolidation time of more than 1 year
the end of the primary settlement is not reached, the consolidation is not complete and the creeping has not
started yet. The test results explain that for projects on soft soils the time for the determination of the soil
parameters may last for a number of years. The determination of the stiffness with normal oedometer tests lasting
only some days will not deliver the correct stiffness of the soft soil in this special case. It is necessary that the
oedometer tests should continue until the deformation of the soil specimen stops. Similar experiences are
described by Möller 2010.

3.5 Analysis and design

In the first planning phase several numerical analyses have been carried out, showing that due to the
necessary 4 m to 7 m high land filling enormous settlements of the whole harbor area in the range of 1m and
large horizontal deformations of the quay walls up to several decimeters would occur, endangering the stability
and the serviceability of the quay walls, its cranes and the container storage and transport systems and other
facilities.
Due to the results of the numerical analysis the planning was adjusted. To ensure the serviceability of the
harbor along the quay wall in a width of about 60 m the soft soil must be replaced up to a depth of 20 m. In the
remaining area, for instance the container storage area etc. the soft soils may remain beneath the harbor surface.
All over the harbor a 2 m high surcharge in combination with a vertical drainage system in a grid of 1.5 m x 1.5

85
m was planned to ensure that the settlement restrictions concerning the paved surface and port operation are met.
Figure 9 shows the effect of the surcharge and the drainage systems, dependent on the different heights of the
surcharge. The vertical axis in Figure 9 shows on the left the degree of consolidation using classic consolidation
theory of the subsoil without surcharge and without a drainage.

Figure 9. Consolidation process of the soft soil dependent on different heights of the surcharge.

Figure 10. Cross section of the quay wall in final planning and calculated deformations.

Without surcharge the consolidation time (red curve) would last up to 80 years, with the surcharge and the
vertical drainage system the consolidation will be completed in an estimate of 250 days. According to the
planning the surcharge will be removed after 90 days. At this point of time a 2 m high surcharge with a vertical
drainage system would have caused an over-consolidation of the subsoil, recognizable at the peak of the yellow
curve.

The final design and the prediction of the deformations of the quay wall are shown in Figure 10. It is
expected that due to the drainage system and the surcharge after the end of construction less than 10 % of the
calculated settlements will occur. Due to the sand fill in the container storage area without replacement of the
soft soil long-term consolidation and creep settlements in the order of several cm are expected during the service
time of the harbor. Additionally due to the container storage loads during the service time settlements in the
range of decimeters will occur, dependent on the loaded area. These settlements can only be avoided by deep soil
exchange or deep foundation systems or further soil improvement of the soft soils in the whole construction area.

86
4 CONCLUSIONS

For the prediction of the time dependent deformation behavior of soft soils long-term compression tests are
necessary. The common recommendations and remarks for the load periods of compression tests are not
applicable for soft soils. The time of testing may last up to years. This fact has to be considered in an early
planning stage. Investor´s settlement restrictions in the range of 1 cm to 2 cm in soft soil can only by realized - if
technically at all - by extraordinary technical efforts. This means that large investments are necessary for a
successful construction and a sufficient serviceability during the servicetime.

REFERENCES

Arbeitsausschuss “Ufereinfassungen” der Hafenbautechnischen Gesellschaft e.V. (EAU) 2009. Recommendations of the
Committee for Waterfront Structures, Harbours and Waterways EAU 2004 – digitized and updated version 2009. Ernst &
Sohn Verlag, Berlin, Germany.
Atkinson, J.H., Bransby, P.L. 1978. The Mechanics of Soils. McGraw-Hill Book Company Ltd., Berkshire, England.
Deutsche Gesellschaft für Geotechnik e.V. (DGGT) 2008. Recommendations on Excavations EAB. Ernst & Sohn Verlag,
Berlin, Germany.
Grabe, J. 2011. Preface. 2nd Workshop of the research training group “Ports for Container Ships of Future Generatioins”,
Veröffentlichungen des Institutes für Geotechnik und Baubetrieb der Technischen Universität Hamburg-Harburg, Heft 22,
VII-VIII, Germany.
Kempfert, H.-G., Gebreselassie, B. 2006. Excavations and Foundations in Soft Soils. Springer-Verlag, Berlin, Germany.
Küster, V. 2011. JadeWeserPort – First German Deep-Water Port – an Experience Report. 2nd Workshop of the research
training group “Ports for Container Ships of Future Generations”, Veröffentlichungen des Institutes für Geotechnik und
Baubetrieb der Technischen Universität Hamburg-Harburg, Heft 22, 67-78, Germany.
Möller, O. 2010. Zum Langzeit-Kompressionsverhalten weicher organischer Sedimente. Veröffentlichungen des Institutes
für Geotechnik und Baubetrieb der Technischen Universität Hamburg-Harburg, Heft 21, Germany.
Neher, H.P. 2008. Zeitabhängiges Materialverhalten und Anisotropie von weichen Böden – Theorie und Anwendung.
Mitteilung 60 des Institutes für Geotechnik der Universität Stuttgart, Germany.
Scott, R.F. 1963. Principles of Soil Mechanics. Addison-Wesley Publishing Company Inc., London, England.
Von Soos, P., Engel, J. 2010. Eigenschaften von Boden und Fels - ihre Ermittlung im Labor. Grundbautaschenbuch Teil 1:
Geotechnische Grundlagen. Ernst & Sohn Verlag, Berlin, Germany.
Terzaghi, K. [Link] Soil Mechanics. John Wiley and Sons Inc., New York, USA.
Witt, K.J. (Ed.) 2003. Geotechnical Engineering Handbook, Volume 2: Procedures. Ernst & Sohn Verlag, Berlin, Germany.

87
88
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Determination of the modulus of deformation using static and dynamic plate

load tests
Détermination du module de déformation à l'aide d’essais à la plaque sous chargements
statique et dynamique

R. Katzenbach1,*& S. Leppla1
1
Technische Universität Darmstadt, Institute and Laboratory of Geotechnics, Germany
*
Corresponding Author

ABSTRACT - The compression of soil layers near the surface e.g. is controlled by the determination of the
modulus of deformation. In many regulations, specifications, guide lines and recommendations a minimum value
of the modulus of deformation is defined for different types of constructions. To determine the modulus of
deformation two strictly different types of plate load tests can be performed in-situ: the static plate load test and
the dynamic plate load test. The results of the different plate load tests are totally different. Generally the
regulations, specifications, guide lines and recommendations give only a minimum value of the static modulus of
deformation. The paper presents the description of the two strictly different plate load tests and the analysis of
the correlation between the test results.

RÉSUMÉ La compression des couches du sol près de la surface par exemple, est recommandée par la
détermination du module de déformation. Dans de nombreuses règlementations, spécifications techniques,
directives et recommandations, une valeur minimale du module de déformation est définie pour différents types
de constructions. Il existe deux types d’essais in situ pour déterminer le module de déformation : l’essai à la
plaque sous charge statique et l’essai à la plaque dynamique. Les résultats provenant de ces deux essais sont très
différents. Généralement, les règlementations, spécifications techniques, directives et recommandations donnent
uniquement une valeur minimale du module de déformation. Cet article présente une description de ces deux
essais à la plaque et une analyse des corrélations entre les résultats obtenus.

1 INTRODUCTION

For the control of the compaction of soil layers near the surface the modulus of deformation, detected by a field
test, may be used. E.g. the quality of base course at street constructions can be determined (Köhler et al. 1998,
Mair 2005). Normally the modulus of deformation is detected by the static or the dynamic plate load test in-situ.
But the results of both different plate load tests are totally different.
The static plate load test gives the static modulus of deformation EV1 (first loading) and EV2 (reloading). The
dynamic plate load test gives the dynamic modulus of deformation Evd. All modulus of deformation are given in
[MN/m2].
Normally in regulations, specifications, guide lines and recommendations a minimum of the static modulus of
deformation EV1 and EV2 and a necessary relation between them is defined, e.g. FGSV 2009a and DB Netz AG
1997. Requirements on a minimum of the dynamic modulus of deformation Evd are rarely given. The reason is
that the experiences for the static modulus of deformation last over several decades while the dynamic modulus
of deformation is only used for about 20 years.
The performance of a static plate load test causes much more effort than a dynamic plate load test. A lot of
research focused on the investigation of the correlation between the results of the static and the dynamic plate
load tests (Adam & Kopf 2003, Kopf & Erdmann 2005, Kopf et al. 2005a, Kopf et al.2005b, Hartmann & Bräu
2003, Brandl et al. 2007).
In this paper both different field tests are described and measurement data are presented. The intention is to
increase the measurement data for the investigation of the correlation between the results of the static and the
dynamic plate load test.

89
2 PLATE LOAD TESTS

2.1 Static plate load test

For the determination of the static modulus of deformation EV1 and EV2 a load plate is loaded step by step by a
hydraulic jack. The settlements at each step are measured and drawn in a load settlement diagram according for
example to DIN 18134:2012. The quotient between EV1 and EV2 is the rate of the compaction of the soil material
near the surface.
As counter weight for the hydraulic jack a loaded truck may be used, which must be heavy enough for a
successful performance of the field test. Figure 1 shows the field test setup of a static plate load test.
To prepare and to carry out one test needs between 1.0 - 1.5 hours. The sensitive measurement equipment has
to be handled by well trained staff. Further information to the test equipment is given in Table 1.

Figure 1. Setup of a static plate load test.

2.2 Dynamic plate load test

For the determination of the dynamic modulus of deformation Evd a drop weight is released on a guide bar over
the load plate. A sensor measures the acceleration and the processor transforms the measurement data into the
dynamic modulus of deformation. There are no standards or codes for the equipment and the performance of this
field tests but a definition is given in FGSV 2003. The field test can be carried out with drop weights of different
masses. Figure 2 shows the field test setup of a dynamic plate load test with the very common light drop weight.
The test equipment is easy to transport and to handle. The test only needs about 5 minutes. A short briefing
for the staff is sufficient. A counter weight is not needed and it can be used in situations with small space to work
like ditches and other excavations. Further information to the test equipment is given in Table 1.

90
Figure 2. Setup of a dynamic plate load test.

2.3 Comparison between static and dynamic plate load test

Plate load tests can be performed on non-cohesive soil and on cohesive soil with a consistency of IC > 0.75 (DIN
18134:2012, FGSV 2003). Under the load plate should be no gravel or stones with a size of more than ¼ of the
diameter of the load plate. The test level has to be plane. Small asperities may be equalized with a thin layer of
dry sand. The measurement devices have to be calibrated regularly. Both types of field tests are very sensitive to
vibrations. In the vicinity of the test points no compaction works can be carried out during the test. The field tests
have to be carried out 24 hours after the compaction of the test area. The differences between the static and the
dynamic plate load test are shown in Table 1.

Table 1. Comparison of the static and the dynamic plate load test.

91
3 CORRELATION BETWEEN STATIC AND DYNAMIC PLATE LOAD TEST

In literature a lot of different correlations between the results of the static and the dynamic plate load test are
given (FGSV 2009a, DB Netz AG 1997, Kopf et al. 2005b, FGSV 2009b, FGSV 2012, Zorn 2002,
Gütegemeinschaft Leitungstiefbau e.V., Schmidt & Rumpelt 2009, Landesbetrieb Straßenbau Nordrhein-
Westfalen 2006, FSV 2008, Hartmann & Bräu 2003). In many cases these correlations are only for certain soil
materials and homogenous subsoil conditions. For example in Germany the relation of the static modulus of
deformation EV2 and EV1 is preferred (FGSV 2009a, DB Netz AG 1997, FGSV 2009b, FGSV 2012, Zorn 2002,
Gütegemeinschaft Leitungstiefbau e.V., Schmidt & Rumpelt 2009, Landesbetrieb Straßenbau Nordrhein-
Westfalen 2006). In Austria the relation of the static modulus of deformation EV1 and the dynamic modulus of
deformation Evd is peferred (FSV 2008, Brandl et al. 2007).
In engineering practice for the static and the dynamic modulus of deformation often a linear correlation is
used (FGSV 2009a). The linear correlation is expressed by equation (1):

E V 2  2  E vd (1)

A lot of research projects defined different correlations using equations and tables. The Figures 3 to 6 show
correlations for sand, gravel, cohesive soil and cohesive soil improved with fine granular chalk. The information
from literature is drawn in diagrams. Own field tests of static and dynamic plate load tests are added to the
diagrams as red points.
The correlations between EV2 and Evd for sand are shown in Figure 3. The range of results that is covered by
10 different correlations is marked in yellow. The correlation according to equation (1) is nearly in the middle
and is equal to the correlations of FGSV 2012, Zorn 2002, Gütegemeinschaft Leitungstiefbau e.V. and Schmidt
& Rumpelt 2009 and is partly similar to correlation of FGSV 2009b. The correlation of FGSV 2009a, DB Netz
AG 1997, Kopf et al. 2005 and Landesbetrieb Straßenbau Nordrhein-Westfalen 2006 are definitely different to
the correlation according to equation (1). The results of own field tests are shown with red points and can be
described for engineering practice quite well with equation (1).

Figure 3. Correlations for sand.

The correlations between EV2 and Evd for gravel are shown in Figure 4. The range of results that is covered by
10 different correlations is marked in yellow. The correlation according to equation (1) is nearly in the middle
and is equal to the correlations of FGSV 2009, FGSV 2012, Zorn 2002, Gütegemeinschaft Leitungstiefbau e.V.
and Schmidt & Rumpelt 2009 and is partly similar to correlation of FGSV 2009b. The correlation of DB Netz
AG 1997, Kopf et al. 2005 and Landesbetrieb Straßenbau Nordrhein-Westfalen 2006 are definitely different to
the correlation according to equation (1). The results of own field tests are shown with red points and can
roughly be described by the correlation of Landesbetrieb Straßenbau Nordrhein-Westfalen 2006.

92
Figure 4. Correlations for gravel.

The correlations between EV2 and Evd for cohesive soil are shown in Figure 5. The range of results that is
covered by
8 different correlations is marked in yellow. The correlation according to equation (1) is nearly in the middle and
is equal to the correlations of FGSV 2012 and Schmidt & Rumpelt 2009 and is partly similar to correlation of
DB Netz AG 1997, FGSV 2009b and Gütegemeinschaft Leitungstiefbau e.V. The correlation of Kopf et al. 2005
and Landesbetrieb Straßenbau Nordrhein-Westfalen 2006 are definitely different to the correlation according to
equation (1). The results of own field tests are shown with red points. Due to the low number of tests no
correlation can be detected for the own field tests.

Figure 5. Correlations for cohesive soil.

93
Figure 6. Correlations for cohesive soil improved with fine
granular calcium carbonate.

The correlations between EV2 and Evd for cohesive soil are shown in Figure 6. The range of results that is
covered by 8 different correlations is marked in yellow. The correlation according to equation (1) is nearly in the
middle and is similar to the correlations of FGSV 2012 and Schmidt & Rumpelt 2009 and is partly similar to
correlation of FGSV 2009b. The correlation of DB Netz AG 1997, Kopf et al. 2005 and Landesbetrieb
Straßenbau Nordrhein-Westfalen 2006 are definitely different to the correlation according to equation (1). The
range of results of DB Netz AG 1997 is marked in grey. The results of own field tests are shown with red points
and are in the range of results covered by equation (1) and Kopf et al. 2005.

4 CONLUSIONS

The paper presents an overview of the up to now developed correlations between the results of static and
dynamic plate load tests. First results of own field tests have been added to the research results.
The comparison of all relations shows, that the relation described by equation (1) is always in the middle.
This means that equation (1) is useful for engineering practice but only gives a rough correlation.
For more precise correlations further research is necessary (Hartmann & Bräu 2003). Regarding the financial
and time efforts for the quality control of compaction works at large construction projects as described e.g. by
Richl et al. 2013 these further investigations are economically important. In Austria for every single dynamic
plate load test device a specified correlation has to be created according to Brandl et al. 2007.
The comparison of the different correlations shows that it is recommended to verify the results of dynamic
plate load tests should be verified by static plate load tests in-situ on the construction area. If there is any doubt
about the application of existing correlations it is reasonable to develop a specific correlation for a certain
project.

REFERENCES

Adam, D., Kopf, F. 2003. Messtechnische und theoretische Untersuchungen als Grundlage für die Weiterentwicklung und
normative Anwendung der dynamischen Lastplatte (Leichtes Fallgewichtsgerät), Mitteilungen des Instituts für Grundbau
und Bodenmechanik der Technischen Universität Braunschweig, Heft 68, 1-20, Germany.
Brandl, H., Adam, D., Kopf, F. 2007. Die Kalibrierung der dynamischen Lastplatte, Bundesministerium für Verkehr,
Innovation und Technologie, Straßenforschung, Heft 561. Vienna, Austria.

94
DB Netz AG 1997. DB NGT 39 – Richtlinie für die Anwendung des leichten Fallgewichtgerätes im Eisenbahnbau.
Germany.
Deutsches Institut für Normung e.V. 2012. DIN 18134:2012 Baugrund – Versuche und Versuchsgeräte –
Plattendruckversuch. Beuth Verlag, Berlin, Germany.
Forschungsgesellschaft für Straßen- und Verkehrswegebau FGSV, Arbeitsgruppe Erd- und Grundbau 2003. Technische
Prüfvorschrift für Boden und Fels im Straßenbau TP BF-StB Teil B 8.3 Dynamischer Plattendruckversuch mit Hilfe des
Leichten Fallgewichtsgerätes. FGSV Verlag, Cologne, Germany.
Forschungsgesellschaft für Straßen- und Verkehrswegebau FGSV, Arbeitsgruppe Erd- und Grundbau 2009a. Zusätzliche
Technische Vertragsbedingungen und Richtlinien für Erdarbeiten im Straßenbau ZTV E-StB 09, FGSV Verlag, Cologne,
Germany.
Forschungsgesellschaft für Straßen- und Verkehrswegebau FGSV, Arbeitsgruppe Sonderaufgaben 2009b. Zusätzliche
Technische Vertragsbedingungen und Richtlinien für Tragschichten im Straßenbau ZTV T-StB 95, FGSV Verlag,
Cologne, Germany.
Forschungsgesellschaft für Strassen- und Verkehrswegebau FGSV, Arbeitsausschuss Kommunaler Strassenbau 2012.
Zusätzliche Technische Vertragsbedingungen und Richtlinien für Aufgrabungen in Verkehrsflächen ZTV A-StB, FGSV
Verlag, Cologne, Germany.
Gütegemeinschaft Leitungstiefbau e.V. Anhaltswerte zur Kontrolle der Verdichtung mit dem Dynamischen
Plattendruckgerät bei der Verfüllung von Leitungsgräben. Berlin, Germany.
Hartmann, K., Bräu, G. 2003. Aufstellung von Anforderungswerten an den dynamischen Verformungsmodul,
Bundesministeriums für Verkehr, Bau- und Wohnungswesen, Forschung Straßenbau und Straßenverkehrstechnik, Heft
863. Bonn Germany.
Köhler, U., Herald, A., Hering, A 1998. Dimensionierung von Oberbauten von Verkehrsflächen und die Einschätzung der
Tragkraft des Erdplanums. Vorträge zur Baugrundtagung 1998 in Stuttgart, 487-498, Germany.
Kopf, F., Erdmann, P. 2005. Numerische Untersuchungen der flächendeckenden dynamischen Verdichtungskontrolle,
Österreichische Ingenieur- und Architekten-Zeitschrift (150), Heft 4-5, 126-143, Vienna, Austria.
Kopf, F., Adam, D., Paulmichl, I. 2005a. Untersuchungen des statischen Lastplattendruckversuchs mit der
Randelementemethode, Österreichische Ingenieur- und Architekten-Zeitschrift (150), Heft 4-5, 144-161, Vienna,
Austria.
Kopf, F., Adam, D., Paulmichl, I. 2005b. Untersuchungen des dynamischen Lastplattendruckversuchs mit dem leichten
Fallgewichtsgerät
unter Verwendung der Randelementemethode, Österreichische Ingenieur- und Architekten-Zeitschrift (150), Heft 4-5, 162-
173, Vienna, Austria.
Landesbtrieb Straßenbau Nordrhein-Westfalen 2006. Planungsgrundsätze & Qualitätssicherung für den Erd- und
Straßenbau. Münster, Germany.
Mair, P. 2005. Der dynamische Plattendruckversuch mit dem leichten Fallgewichtsgerät, Neue Landschaft (50), Heft 2, 42-
43, Pater Verlag, Berlin, Germany.
Österreichische Forschungsgesellschaft Straße Schiene Verkehr FSV 2008. Technische Vertragsbedingungen Vor-,
Abbruch- und Erdarbeiten RVS 08.03.04 – Verdichtungskontrolle mittels dynamischen Lastplattenversuches. Vienna,
Austria.
Reichl, I., Michels, U., Schäfer, R., Spang, C. 2013. Geotechnische Herausforderungen auf der Ausbaustrecke Nürnberg-
Ebensfeld. Bautechnik 90, Heft 3, 162-172, Ernst & Sohn Verlag, Berlin, Germany.
Schmidt, H.-H., Rumpelt, T. 2009. Erdbau. Grundbau Taschenbuch Teil 2, 7th edition, Ernst & Sohn Verlag, Berlin,
Germany.
Zorn Instruments 2002. Bearing capacity test on subsoil and granular layers. Germany.

95
96
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Effects of Matric Suction on the Shear Strength of Highly Plastic Compacted

Clay

Effets de la succion matricielle sur la résistance au cisaillement d’une argile compactée très
plastique

A.M. Elsharief 1* & O.A. Abdulaziz 2

1
Associate prof. Building and Road Research Institute, University of Khartoum, Khartoum ,Sudan,
aelsharief@[Link]
2
Formerly graduate student, Building and Road Research Institute, University of Khartoum ,Sudan,
gadomaima@[Link]
*
Corresponding Author

ABSTRACT: Matric suction is one of the stress state variables that control the shear strength of unsaturated soil,
therefore this work basically investigates its effects on the shear strength of a highly plastic unsaturated
compacted clay soil from Sudan. Matric suction and drained shear strength were independently measured using
the filter paper and the conventional shear box apparatus, respectively. The tested samples were all prepared at
the same dry density but each sample had different moisture content, covering dry & wet of optimum, mainly to
study the effects of variation in air and water phases (matric suction) on shear strength under different vertical
stress values. From the tests results, it has been found that the shear strength of unsaturated soils is dependent on
the soil moisture condition, i.e. soil suction, and under the same applied vertical stress the shear resistance and
peak shear strength increased as matric suction increased and this increment is not linear. The lines representing
the relationship between peak shear strength and matric suction for different values of vertical stress showed
sharp increase in shear strength for the first portion of the curve whereas the lines are almost parallel for high
values of matric suction. Also it is noticed that the lines tend to curve as the matric suction approaches a
specified value (for this particular soil type this value was found to be within the range of 600 to 700 kPa).The
shear parameter ϕb was found to be equal to the angle of internal friction of the soil when the matric suction is
lower than the air entry value of the soil, whereas for higher values of matric suction this value becomes very
small. It was found that the soil cohesion increased as matric suction increased and this increment is linear and
the internal angle of friction of the soil did not experience significant change with change in matric suction.
Hence it could be stated that the increase in soil shear strength due to increase of matric suction is a result of
increase in soil total cohesion.
Keywords: Unsaturated, shear strength, matric suction, shear box, filter paper.

RÉSUMÉ: succion matricielle est l'une des variables d'état de stress qui contrôlent la résistance au cisaillement
des sols non saturés, donc ce travail étudie essentiellement ses effets sur la résistance au cisaillement d'un sol
saturé d'argile compactée très plastique du Soudan. Matrice de succion et de la force de cisaillement ont été
mesurées indépendamment drainé à l'aide du papier filtre et l'appareil de la boîte de cisaillement classique,
respectivement. Les échantillons testés ont tous été préparés à la même densité sèche mais chaque échantillon
avait teneur en humidité différente, couvrant sec et humide de l'optimum, principalement pour étudier les effets
de la variation dans les phases de l'air et de l'eau (la succion matricielle) sur la résistance au cisaillement sous
différentes valeurs de contrainte verticale. D'après les résultats des tests, il a été constaté que la résistance au
cisaillement des sols non saturés dépend de l'état d'humidité des sols, à savoir la succion du sol, et sous la même
contrainte verticale appliquée la résistance au cisaillement et la résistance au cisaillement de pointe a augmenté
comme la succion matricielle augmenté et cette augmentation est non linéaire. On remarque que les lignes
représentant la relation entre la résistance au cisaillement de pointe et la succion matricielle pour différentes
valeurs de contrainte verticale ont montré forte augmentation de la résistance au cisaillement de la première
partie de la courbe alors que les lignes sont presque parallèles pour des valeurs élevées de la succion matricielle.
En outre, il est remarqué que les lignes ont tendance à se courber en tant que matrice de succion se rapproche
d'une valeur spécifiée (pour ce type de sol particulier cette valeur a été trouvée comme étant dans la plage de 600
à 700 kPa). Cette étude et des études antérieures montrent que cette valeur correspond à la valeur du sol de
l'entrée d'air. Le paramètre φb cisaillement a été trouvé égal à l'angle de frottement interne du sol lorsque la
succion matricielle est inférieure à la valeur du sol de l'entrée de l'air, tandis que pour des valeurs plus élevées de
succion matricielle cette valeur se rapproche de zéro. Il a été constaté que la cohésion du sol a augmenté comme
la succion matricielle augmenté et cette augmentation est linéaire tandis que l'angle de frottement interne du sol
n'a pas connu de changement significatif de changement dans la succion matricielle où l'on pourrait dire que

97
l'augmentation de la résistance au cisaillement des sols due à augmentation de la succion matricielle est due à une
augmentation dans le sol cohésion totale.
Mots-clés: insaturés, résistance au cisaillement, succion matricielle, boîte de cisaillement, papier filtre.

1. INTRODUCTION

Recent studies have shown Terzaghi (1925) effective stress equation to be inadequate to account for the behavior
of partially saturated soils, because in partially saturated soils, the water pressure acts over only a part of the area
of any plane through the soil. Hence, the physical meaning of effective stress remains the same and only two
additional factors must be considered, the stress acting through the air phase (i.e., the pore air pressure, ua) and
the difference between the pore air pressure (ua) and the pore water pressure (uw), which is known by matric
suction (ua-uw). The effective stress equation for unsaturated soil can therefore be expressed as:

σ' = σn-ua +χ(ua-uw) ……….. (1)

Where,

σ' : effective stress , (σn-ua ): net normal stress , σn : total stress , χ : parameter related to the degree of
saturation of the soil,

Earth fill dams, embankments, and highways are typical examples of earth structures made of compacted,
unsaturated soils. The theory and measurement of shear strength of unsaturated soils have gained increasing
attention during the past decades (Gan et al (1988)). From practice, the problem in unsaturated soil is not
associated with the shear strength at certain degrees of saturation; it rather arises from change in shear strength
due to change in the degree of saturation. Moisture characteristic curve (MCC) is a characteristic curve for a
certain soil relating changes in suction to changes in water content. Consequently, water content is better
represented by matric suction when characterizing effects of differential changes in water content on shear
strength and compressibility. This is logical because changes in matric suction cause change in the stress state of
a certain soil.
This paper presents the results of a series of direct shear tests on highly plastic unsaturated clay soil to assess the
influence of matric suction on its shear strength.

1.1 Unsaturated Soil Shear Strength

A shear strength equation for an unsaturated soil in which two independent stress state variables are used was
proposed by Fredlund et al. (1978). The two stress state variables most commonly used are the net normal
stress,(σn-ua ) and the matric suction, (ua-uw ). The proposed shear strength equation has the following form:

τf = c'+ (σn-ua)f tan ϕ’ +(ua-uw)f tan ϕb …………….. (2)

Where:
τf:: shear strength of unsaturated soil at failure.
c': soil cohesion at zero matric suction and zero net normal stress.
ϕb: soil internal friction angle associated with matric suction; it describes the rate of increase in shear strength
relative to matric suction.
ϕ′: angle of internal friction associated with the net normal stress variables.

The equation above defines a planar surface, which is called the extended Mohr-Coulomb failure envelope.
We noticed that the shear strength has been written in terms of effective stress state variables for an unsaturated
soil and it is an extension of the equation used for saturated soil. The first two terms on the right-hand side of the
equation describe the conventional M-C criterion for the strength of saturated soil. The third term captures the
increase in shear strength with increasing matrix suction in unsaturated soil. The shear strength of an unsaturated
soil consists of an effective cohesion intercept, c', and the independent contributions to shear strength from net
normal stress, (σn-ua), and matric suction, (ua - uw). The shear strength contributions from net normal stress and
matric suction are characterized by ϕ' and ϕb angles, respectively. The failure envelope on the frontal plane (i.e.,
τf versus (σn-ua) plane) is the same as the Mohr-Coulob failure envelope for saturated conditions.
Experimental results of suction-controlled simple direct shear tests were carried out for Fraser River sand (K.
Shimanda, 1997) and for for Madrid clayey sand (Escario & Saez, 1986), respectively. Both results show
bilinear relationship between shear strength and matric suction. It was reported that for the sand soil the
relationship is almost linear and flat under all values of stress and the effects of matric suction on the shear stress

98
are relatively minor. On the other hand the bilinear relationship was evident and clear for the sandy clay of
Madrid soil (Escario & Saez (1986)).

2. THE STUDY PLAN

Consolidated drained direct shear test was selected for the measurement of the shear strength of the tested
samples. The simplicity of the direct shear test and the less time required for failure to take place was the reason
for its selection. The tests were conducted on specimens dynamically compacted to reach the same value of dry
density (1.48 gm/cm3) but at different moisture content values, i.e. 18, 20, 22, 25 and 28 percent. These values
cover the dry and wet of optimum water contents of the soil. Here the optimum water content of the tested
sample was found to be 22 %. Using the same value of dry density for the samples with varying moisture
contents for each test as stated above allowed observing the change in shear strength of the soil with change in
pore water and pore air values (i.e change in soil suction) for the same amount of soil solids per unit volume.

3. MATERIAL AND TESTING METHODOLOGY

The soil under study was obtained from Al Damazin in south-eastern Sudan. It was chosen because it had been
used as embankment material for road construction and for the construction of the earth dam of the second
largest dam in Sudan. The tested soil was grayish highly plastic silty clay with an approximately 48% clay and
52% (silt + sand). The plastic limits of the soil was reported to be 26 %, the liquid limit 56% and the optimum
moisture content was 22%. The soil is classified according to USCS as highly plastic silty clay (CH) .Substantial
amount of soil for both shear test and suction measurement was obtained and the soil was carefully crushed using
a manual pestle and mixed to ensure uniformity. The crushed soil was allowed to dry in the air for 24 hours while
being covered with very thin cloth to protect it from dust. The soil was divided into five parts and each part was
mixed with specific quantity of distilled water to increase the water content of the soil from the initial value to
the required one. The moist soils were then wrapped in a plastic bag and stored for testing.
Conventional triaxial and direct shear equipment requires modifications prior to their use for testing unsaturated
soils. The modifications must accommodate the independent measurement or control of the pore-air and pore
water pressures. In this investigation normal direct shear box was used for the measurements of shear strength of
the unsaturated samples. The matric suction relative to the same value of moisture content was measured
independently using the filter paper method. Shearing for the soil was done using an approximate shearing rate of
0.02 mm/min to allow drained conditions.
The Filter Paper method was selected for this research study because it is capable of covering the full range of
suction, which is relatively high for this type of soil. The ease of the method, the availability of the required
laboratory equipment justified using the filter paper method in this study. Whatman no. 42 filter paper was used
and the test procedure followed the guides of ASTM D5298-03 with its corresponding calibration curve.

4. TESTING PROCEDURE

Automated direct shear equipment was used in this investigation and the test data was automatically stored in the
data logger. The weight of the soil required to fill the shear box in layers in order to reach the target dry density
(1.48 g/cm3) was calculated, for each water content. All the specimens were compacted into a 6.0cm x 6.0cm
mold in three layers to give a total sample height of 2.5 cm using wooden dolly to the required soil dry density.
Fifteen test specimens were prepared for the shear tests. All the specimens were prepared at the same dry density
but at five different water contents (18, 20, 22, 25 and 28%), i.e. every three samples have the same water
content.
Three normal stresses 100 kPa, 150 kPa and 200 kPa were applied for each water level. Data obtained from the
tests was transferred to Excel sheets in order to determine the peak shear stress at failure for each specimen
(Table 1). The test results were plotted and consequently the cohesion (C) and the angle of internal friction (ϕ)
were determined from Mohr-Columb relationship (Table 1).

99
Table 1. Shear Strength Data

Moisture Peak Shear Strength , τ f , (kPa) Shear Strength Parameters

Content
σ =100 kPa σ =150 kPa σ =200 kPa C (kPa) ϕ⁰
m.c (%)
18 252.00 261.76 273.76 230.2 12.1
20 207.47 215.85 227.32 187.1 11.3
22 177.40 198.39 220.67 133.9 12.8
25 154.46 186.71 200.11 111.2 33.5
28 94.17 102.60 139.45 44.15 24.4

5. ANALYSIS OF TESTS RESULTS:

Figure 1gives representative and typical test data for the shear stress versus horizontal displacement when 150
Kpa vertical stress was applied at different water content or initial matric suction values. It is noticed from the
Figure that under the same applied vertical stress the shear resistance and peak shear strength increased as the
initial matric suction increased. The graphs do not show a clear relationship between failure strain and matric
suction. Figure 2 shows the soil water characteristic curve (SWCC) for the tested soil. The peak shear strength is
plotted in Figure 3 against initial matric suction (ua – uw). The Figure shows that the peak shear strength of the
soil increases as matric suction increases in nonlinear form. Although the well-known shear strength equation
developed by Fredlund et al. (1978) is linear (equation 2), many experimental results {Gan et al (1988), Escario
and Juca (1989), Abramento and Carvalho (1989) and Wheeler and Sivakumar (1992)} show nonlinear variation
of shear strength with respect to matric suction when tested over wide range of matric suction. The nonlinearity
of the failure envelope of unsaturated soil as reported by earlier studies is supporting the experimental results
shown in Figure 3. This could be attributed to the non-linear soil water characteristic curve presented in Figure 2.
Bilinear lines were drawn for each relationship for the different normal stress. It is noticed that all lines tend to
curve as the matric suction approaches a specified value. For this particular soil type this value was found to be
within the range of 600 to 700 kPa. A study carried out by Fredlund and Rahardjo (1985) concluded that this
value corresponds to the air entry value of the soil. According to Thamer et al (2006), the air entry value could
be obtained by plotting the two tangents shown in Figure 2 and the intersection of the two lines will determine
the air entry value. It is found to be equal to 600 kPa It is clear from Figure 3 that envelope lines are almost
parallel for high values of matric suction and there is sharp increment in the shear strength for the first part of the
curves, i.e. between the lowest suction value 109 kPa and 441 kPa (this study did not include saturated
specimens with suction value approaching zero). This sharp increment could be related to the effect of inter-
particle forces produced due to the negative pore water pressure. The inter-particle force depends on the particle
size and it is believed that this force in clay soil is substantially higher than in silt and sand (Lu, and Likos,
2006). It is observed that, for the same value of matric suction, the shear strength of the soil increased as the
normal stress is increased (Figure 4); i.e the shear strength of unsaturated soil is directly proportional to the
increases of vertical stress. However it is not known whether these increments are governed by a specified limit
or not.
The shear parameter ϕb was obtained from Figure 3. It was previously defined as the change in shear strength
with change in suction, which is the slope of the plot of matric suction versus shear strength when the net normal
stress is kept constant. The values obtained for the first part of the lines where the matric suction is lower than
the air entry value is within the range 10.3° to 14.1° with an average value of 11.6°. (10.3 ° for 100 kPa vertical
stress, 14.1° for 150 kPa vertical and 10.3° for 200 kPa vertical stress). It is noticed that these values are
approximately equal to the angle of internal friction, ϕ’. On the other hand, for matric suction higher than air
entry value all the lines gave slope angle less than unity (ϕb ˂ 1.0). Hence we can state that, for matric suction
values lower than the air entry value of the soil an increase in matric suction is very effective in increasing the
shear strength of the soil as does an increase in net normal stress ( ϕb = ϕ’).

100
Figure 1. Shear Strength Vs Horizontal Displacement, σ=150kPa.

Figure 2. Soil Suction– Moisture Content Relationship.

Figure 3. Change of Peak Shear Strength with Matric Suction

101
Figure 4. Normal Stress vs Peak Shear Strength

Figure 5. Relationship between Soil Cohesion and Matric Suction

At matric suctions higher than the air-entry value of the soil, the soil starts to de-saturate and the contribution
of matric suction towards the strength of the soil is less than the contribution of the net normal stress at the same
stress level (as ϕb ˂ ϕ'). In other words the increase in shear strength with respect to matric suction is less than
the increase with respect to net normal stress as shown in Figure 4. The figure plots shear stress versus normal
stress for different suction values; it is observed from the figure that for the same value of the soil suction, the
shear strength increases as the vertical stress increases. This relationship makes the unsaturated soil looks similar
to a saturated one. It is also seen that the lines are almost parallel indicating that the friction angle of the soil is
practically independent of water content and also independent of soil suction. This brings the reader to the points
that the saturated soil is an unsaturated soil with an approximate 100% degree of saturation. Theoretically, φ'
should not vary too much (Fredlund & Morgenstern 1978). In other words, the lines for different values of
suction, (ua-uw), should be roughly parallel. However from Figure 3, tendency for slight divergence at increasing
loads is observed. The results are consistent with those reported by Thamer et al (2006) for residual soil samples
from Kuala Lumpur. A previous study on unsaturated soil carried out by Tuffour (1984) found that the friction
angle values φ, obtained from slow shear tests on unsaturated soils is independent of soil water content and
compatible with the effective friction angle, ϕ', of the saturated soil. This is consistent with the proposition that
unsaturated soils are internally drained during shearing.
The relationship between matric suction and the soil cohesion is linear as presented on Figure 5 giving that
the soil cohesion increased as soil suction increased. The envelope is identical to the relationship between shear
strength and soil suction as given in Figure 3. The sharp increase in the soil cohesion between suction values 109
kPa and 441 kPa as presented in Figure 5 can be related to the same reason as explained earlier. As shown
before the internal angle of friction of the soil did not experience significant change with change in matric
suction and therefore it could be stated that the increase in soil shear strength due to increase of matric suction is

102
a result of increase in soil total cohesion. The increment in total soil cohesion due to suction increment could be
expressed as follows:

ct = c’+(ua–uw)tanϕb …………… (3)

Where:
ct : total cohesion of the soil including the effect of suction increment.
c’ : effective cohesion relative to zero suction.
(ua-uw): matric suction.
ϕb: slope of the failure line representing the change in soil shear strength due to change in soil suction.

CONCLUSIONS:

Consolidated drained direct shear test was selected for the measurement of the shear strength of the tested
samples, while in-contact filter paper test was used for the measurement of matric suction. Both tests were
carried out for a compacted soil samples prepared at the same value of maximum dry density and different
moisture content. From the tests results obtained herein, it has been found that the shear strength of unsaturated
soils is dependent on the soil moisture condition, i.e. soil suction. Also it is noticed that under the same applied
vertical stress the shear resistance and peak shear strength increased as matric suction increased. It was found
that the peak shear strength of the soil increased as matric suction increased, and this increment is not linear. The
nonlinearity of the failure envelope of unsaturated soil as reported by earlier studies is supported by the
experimental results obtained from this study on a highly plastic compacted clay soil. It is noticed that the lines
representing the relationship between peak shear strength and matric suction are almost parallel for high values
of matric suction. There is sharp increment in the shear strength for the first part of the curves, i.e between the
lowest suction value 109 kPa and 441 kPa .This sharp increment was related to the effect of inter-particle forces
produced due to the negative pore water pressure. The inter-particle force depends on the particle size and it is
believed that this force in clay soil is substantially higher than in silt and sand. Also it is noticed that the lines
tend to curve as the matric suction approaches a specified value. For this particular soil type this specified limit
falls between 600 to 700 kPa. A study carried out by Fredlund, D. G., and Rahardjo (1985) concluded that this
value corresponds to the air entry value of the soil. The shear parameter ϕb was found to be equal to the angle of
internal friction of the soil when matric suction is lower than the air entry value of the soil, whereas for higher
values of matric suction this value reached zero. It was found that the soil cohesion increased as matric suction
increased and this increment is linear while the internal angle of friction of the soil did not experience significant
change with change in matric suction. It could be stated that the increase in soil shear strength due to increase of
matric suction is a result of increase in soil total cohesion.

REFERENCES:
Abramento, M. and Carvalho, C.S., (1989), “Geotechnical Parameters for the Study of Natural
Slopes Instabilization at serra do Mar-Brazilian Southeast,” Proceedings, Twelfth International Conference on Soil
Mechanics and Foundation Engineering, Rio de Janeiro, Vol. 3, A. A. Balkema Brookfield, MA, pp. 1599-1602.
Escario, V., and Sáez, J. (1986). “The shear strength of partly saturated soils.” Geotechnique, 36(3), 453–456.
Fredlund, D.G., Morgenstern, N.R., & Widger, R.A. (1978). “The shear strength of unsaturated soils”. Canadian
Geotechnical Journal, 15: 313-321.
Fredlund, D. G., and Rahardjo, H. (1985). “Theoretical context for understanding unsaturated residual soil behavior”.
Proceedings of the First International Conference on Geomechanics in Tropical Lateritic and Saprolitic Soils, Sao Paulo,
Brazil, February 11-14, pp. 295-306.
Gan, J. K.-M., Fredlund, D. G., and Rahardjo, H. (1988). “Determination of the shear strength parameters of an unsaturated
soil using the direct shear test”. Canadian Geotechnical Journal, 25(3), 500-510.
Kiyoshi Shimada. (1997). “Effect of Matric Suction on Shear Characteristics of Unsaturated Fraser River Sand”. Journal of
the Faculty of Environmental Science and Technology, Vol.3, No.1, pp.l27-134, 1998.
Lu, N., and Likos, W. J. (2006). “Suction stress characteristic curve for unsaturated soil.” J. Geotech. Geoenviron. Eng.,
132(2), 131–142.
Terzaghi, K., (1925), “soil mechanics (Erdbaumechanik)”, Franz Deuticke, Vienna.
Thamer, A. M., Faisal, H. A., S. Hashim and Bujang B.K. Huat. (2006). “Relationship Between Shear Strength and Soil
Water Characteristic Curve of an Unsaturated Granitic Residual Soil” American Journal of Environmental Sciences 2
(4):142-145, 2006.
Tuffour, Y.A., (1984). ”Shear strength of partly saturated clays”. MSc. Thesis. Faculty of Civil Engineering, Technion,
Israeli Institute of Technology. Haifa, Israel.
Wheeler, S. J. Sivakumar, V. (1992). “Critical state concepts for unsaturated soil”. Pro. 7th Int. Conf. Expansive Soils,
Dallas, 1992. In press.

103
104
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Comparison of shear strength characteristics of cohesive soils evaluated by the

triaxial mohr’s circle and stress path methods
Comparaison des caractéristiques de résistance au cisaillement de sols cohérents évalués par
l’essai triaxial

A.K. M. Zein
Building and Road Research Institute (BRRI), University of Khartoum, SUDAN

ABSTRACT: The present study aims at comparison of the shear strength evaluated according to the
conventional and modified stress path methods of the Mohr-Coulomb failure envelope criterion. An experimental
program was conducted on clay soils from different regions in Sudan subjected to laboratory tri-axial tests till
failure following the CU and UU test procedures. In general, the study shows that there is a very good agreement
between the results obtained by the two analysis methods for the various shear strength data considered. The soil
angle of internal friction (φ) determined from both methods compared very well, whereas appreciable variations
were indicated for the soil cohesion. The results of regression analysis revealed an overall strong correlation
between the failure shear strength data computed from both methods for the soil types and principal stress state
conditions covered by this study. However, there are variations in shear strength magnitudes evaluated according
to the two comparison methods when the behaviour of individual soil samples is considered. For most soil
samples, the conventional Mohr-Coulomb method tends to slightly overestimate soil shear strength compared to
the modified method for the principal stress levels assumed in analysis. It may be concluded that while both
methods compared very well and can be applied with confidence for determining the shear strength of cohesive
soils the modified Mohr-Coulomb method is the preferred one since it is based on a better rational and facilitates
a simple approach of analysis.

RESUME: La présente étude vise à comparer la résistance au cisaillement évalué selon les méthodes de chemin
de contraintes classiques et modifiées du critère de l'enveloppe de rupture de Mohr-Coulomb. Un programme
expérimental a été mené sur les sols argileux de différentes régions du Soudan soumis à des essais triaxiaux CU
et UU au laboratoire jusqu'à la rupture. En général, l'étude montre qu'il ya un très bon accord entre les résultats
obtenus par les deux méthodes d'analyse pour les différentes données de résistance au cisaillement considéré.
L'angle de frottement interne (φ) déterminé à partir de deux méthodes de sols très bien comparé, tandis que les
variations appréciables ont été indiqués pour la cohésion du sol. Les résultats de l'analyse de régression ont
révélé une forte corrélation globale entre les données de résistance de cisaillement calculées à partir de deux
méthodes pour les types de sols et des conditions d'état de contraintes principales par cette étude. Cependant, il
existe des variations de grandeurs de résistance au cisaillement évalués selon les deux procédés de comparaison
lorsque le comportement des échantillons de sol individu est considéré. Pour la plupart des échantillons de sol, la
méthode de Mohr-Coulomb classique tend à surestimer légèrement la résistance au cisaillement du sol par
rapport à la méthode modifiée pour les principaux niveaux de contraintes assumées dans l'analyse. Il peut être
conclu que si les deux méthodes comparées très bien et peuvent être appliqués avec confiance pour la
détermination de la résistance au cisaillement des sols cohérents la méthode de Mohr-Coulomb modifié est celui
préféré car il est basé sur une meilleure rationnelle et facilite une approche simple de l'analyse.

1. INTRODUCTION

The shear strength of a soil, Ԏf, defined as the maximum stress causing failure in a given direction, is normally
expressed mathematically in terms of the normal stress at failure, σf the cohesion, c and the angle of internal
friction, φ of the soil as follows:

Ԏf = c + σf tan φ [1]

The parameters c and φ are not intrinsic soil properties but are applicable to a specific soil condition and
should only be assumed to be constant within the range of stresses for which they are evaluated (Chin et al,
2001). They are determined from triaxial testing using the conventional Mohr-Coulomb’s criterion. This criterion

105
defines the “failure envelope” as a common line tangential to Mohr’s circles drawn on a shear stress (Ԏ) versus
normal stress (σ) plot for different major and minor principal stresses. The angle φ is the maximum internal soil
friction angle defining the criterion of failure envelope line in this method. In the Mohr-Coulomb circle method,
the stresses at failure plane are normally evaluated using some assumptions related to the normal and shear
stresses acting on the sample failure surface plane. One assumption is that only normal i.e. principal stresses are
fully acting on the plane i.e. no shear stresses exist. Another assumption is that the failure occurs in the sample at
a plane with an inclination angle equal to (45+ φ/2) to the minor principal stress. The claim that the point of
tangency of the critical stress circle with the failure envelope, as constructed on the Mohr diagram, represents the
normal and shear stresses on the failure plane with normal inclined to (45+ φ/2) is not always observed in
experiments. Apart from the problem related to these assumptions the Mohr circles diagrams are tedious to plot
for a variety of reasons.. It is not practical to trace the loading of a soil element or soil sample in a laboratory test
by using Mohr circles because an infinite number of Mohr circles would be required (Amsiejus et al, 2010).
Some methods have been proposed and applied to resolve the problems associated with the above said
assumptions (Parry, 2004). Instead of plotting a Mohr circle to represent the state of stress at a point in
equilibrium, it may be more convenient to represent the state of stress by a point with a maximum value of shear
stress on the circle. Lambe (1964) suggested a type of graphical stress path representation known as the “p-q
diagram” i.e. by plotting the coordinates of the top of the Mohr's circle, p and q, against each other. A “stress
path” is defined as a line that connects a series of points, each of which represents a successive stress state
experienced by a soil specimen during the progress of a test. There are several ways in which a stress path can be
drawn. The highest points of the Mohr circles, defined as the stress points which have the coordinates expressed
in terms of principal stresses as p = (σ1+σ3)/2 and q = (σ1- σ3)/2 are joined together to produce a plot. The locus
of the modified stress points is called a “stress path”, or a “Kf Line” and the criterion is the maximum shear
rather than the soil angle of friction. The difference between the conventional Mohr circles and stress path failure
envelopes are graphically illustrated in Figure 1.
The modified Mohr-Coulomb envelope; also known as “Kf line” is represented in terms of the modified shear
strength parameters “d and ψ” by the following equation:.

qf = d + pf *tan ψ [2]
In equation 2, the p and q values may be expressed in terms of the total or effective stresses. From the values
of “ψ” and “d” the shear strength parameters c and φ may be determined for a given soil through application of
the following relationships:

φ = arcsin(tan ψ) [3]

Figure 1. Graphical representation of the Mohr-Coulomb and Kf shear strength failure envelope lines

c = d/cos φ [4]

Other modified Mohr Coulomb diagrams have been proposed in some previous studies such as those based
on σ1-σ3 coordinate system (Bukharsev, 1988), MIT coordinate system(Craig, 2004) or Cambridge coordinate
system (Parry, 2004) .

106
2. STUDY OBJECTIVE AND SCOPE

The conventional Mohr circles method is rarely employed nowadays worldwide due to the problems stated
previously and the difficultly of test results presentation,. To date, the practice followed in academic and
professional geotechnical institutions in Sudan and perhaps in some other countries is to evaluate the soil shear
strength parameters according to the traditional Mohr-Coulomb diagram drawn in the Ԏ-σ coordinate system
The aim of this study is to examine the usefulness of the modified Mohr-Coulomb method in determining the
soil total and effective shear strength parameters of Sudanese soils of different types and conditions. The scope
of experimental program comprised two test series performed on undisturbed soil samples representing a fairly
wide range with respect to their types and conditions as briefly described below.

3. SOIL MATERIAS AND TESTING PROCEDURES

The studied soils included alluvial fine grained silty or sandy clays of low to high plasticity (CL, CH) and clayey
silts (ML, MH) deposits obtained from several sites located near the banks of the River Nile in three Sudanese
states namely; Khartoum, River Nile and Northern States. Figure 2 shows the representative points of the tested
soil types in the soil plasticity chart.
A standard triaxial compression machine was utilized for performing the tests and determination of shear
strength parameters for undisturbed soil samples tested following the UU and CU triaxial test procedures. In the
CU test the samples were allowed to come to equilibrium under some state of stress and then sheared undrained,
to correspond to an applied loading that occurs too rapidly to allow drainage. The confining pressures applied
during testing varied from 100 to [Link] two test series included UU tests on 50 samples and CU tests on
25 samples and were conducted on soil specimens of 38mm diameter.

Figure 2 . Representative points of clay samples tested on soil plasticity chart

The results obtained from the two test series were plotted on the “Ԏ-σ” and the “p-q” coordinate systems
representing the conventional and modified Mohr-Coulomb failure envelopes. An example of such plots is
shown in Figure 3 for a typical clay sample subjected to CU triaxial testing.

Figure 3. Typical CU test results presented on the conventional (left) and the modified (right) Mohr-Coulomb plots

107
4. ANALYSIS AND DISCUSSION OF STUDY RESULTS

The conventional shear strength parameters φ and c were determined directly from the Mohr’s circles on the Ԏ-σ
plot (denoted as φc and cc). The modified parameters (denoted as φm and cm,) were computed from equations 3
and 4 using the angle ψ and d. values determined from the Kf line or modified failure envelope on the “p-q”
coordinate system plot. Then, the two sets of parameters evaluated from the two analysis methods were plotted as
shown in Figures 4 and 5 for the UU and CU procedures respectively.
A linear regression analysis was carried out to establish the relationships between the parameters c, φ
determined from the conventional and modified Mohr Coulomb methods for the UU and CU test data. Table 1
summarizes the results of the statistical and regression analysis carried out on the shear strength characteristics
derived for soil samples from the UU and CU test results.
For the total and effective angle of internal friction data plotted in Figure 3, the two methods compared very
well for the soils with φ of about 18º and higher with a discrepancy of 12 to 13% and a standard deviation of 3.9
to 5.4% in over 80% of the samples tested. However, as may be noted from this figure as well as the regression
analysis results in Table 1, the Mohr circles method tends to produce slightly higher φ values compared to the
stress path method in samples of relatively low internal friction angles. For samples of very small internal friction
angles (2 to 10º), the discrepancies between the results obtained from the conventional and modified Mohr-
Coulomb methods become very significant reaching 37% and 46% for the computed total and effective φ values
respectively.

Figure 4. Variation of the total and effective soil angle of friction determined from Mohr circles and stress path failure
envelopes

Figure 5. Variations of the total and effective soil cohesion determined from Mohr circles and stress path failure envelopes

108
Table 1. Analysis of results for the total and effective shear strength parameters determined from the two comparison
methods
Standard
Regression Discrepancy
Soil shear strength parameter Values Range R2 Deviation
Equation Range (%)
(%)
Total angle of internal friction φ (º) 2.0 – 47.5 φc = 0.987φm 0.986 -46.0 to 45.0 14.7
Total soil cohesion c (kPa) 2.3 – 124.0 cc = 0.976 cm 0.981 -46.8 to 66.7 14.2
Effective angle of internal friction φ’ (º) 4.4 – 55.3 φc = φm 0.987 -11.3 to 8.4 3.8
Effective soil cohesion c’ (kPa) 0.2 - 100 cc = 0.976 cm 0.969 -60.0 to 22.9 20.3

In Figure 5, the degree of data scatter in the total and effective soil cohesion is more appreciable compared to
that noticed for φ data indicating a lower correlation between the total and effective cohesion values determined
according to the two analysis methods. The discrepancy between the cohesion values determined according to
the conventional and modified Mohr-Coulomb methods were 22 to 46% for the UU and CU tests respectively in
most soil samples. However, very significant differences of up to 67% were observed in few samples of
relatively low total cohesion values. Generally, the agreement between the cohesion parameters pertaining to the
conventional and modified Mohr-Coulomb methods is much better for the data deduced from the CU than from
the UU tests. This discussion shows that some variations in the magnitudes of the soil angle of internal friction
and cohesion may result for a given soil when different coordinate systems are utilized for presentation of the
triaxial test data. Such variations in the soil cohesion parameter determined according to the two comparison
analysis methods have previously been reported in some investigations (e.g. Amsiejus et al, 2010).
The total and effective shear strength parameters determined as described above were subsequently used to
evaluate the soil failure shear strength using a mathematical expression derived in a previous study (Ismail and
Zein 1987) as explained below.
The failure shear strength of a given soil at a point in a particular plane, denoted here as Ԏf, may be expressed
by relationship in equation 1. The normal stress at failure, σf can be expressed in terms of the minor and major
principal stresses, σ1 and σ3 respectively as follows:

σf = 0.5(σ1 + σ3) + 0.5(σ1 - σ3) cos 2θ [5]

Where 2θ = (π/2 + φ), therefore;

σf = 0.5(σ1 + σ3) + 0.5(σ1 - σ3) sin φ [6]

Dividing both sides of equation 5 by σ3, we get the ratio “σ1/ σ3” known as the principal stress ratio (denoted
by letter R) and equation 6 can therefore be rearranged as follows:

σf/ σ3 = 0.5(R+1) – 0.5(R-1)sin φ [7]

Expressing the normal stress at failure, σf in terms of the minor principal stress σ3, major principal stress, σ1
and angle of internal friction, φ and substituting it in equation 1 , the failure shear strength Ԏf , may be computed
for assumed principal stress ratio R (=σ1/σ3) and σ3 from the following equation :

Ԏf = c + 0.5(σ3 tan φ)[R(1-sin φ) + ( 1 + sin φ) [8]

Equation 8 enables the evaluation and comparison of the soil failure shear strength Ԏf using the parameters
obtained according to the conventional and modified Coulomb-Mohr failure envelopes based on the Ԏ-σ and p-q
coordinate systems respectively. Different minor principal stresses (σ3) and stress ratios R values may be
assumed for determination of the total or effective failure shear strength. For analysis purposes, the soil strength
was calculated for σ3 and R combinations representing; a low level (σ3 =100kPa and R = 2), intermediate level
(σ3 =200kPa and R = 3) and a high level (σ3 =300kPa and R = 4) of principal stress states.
The total and effective failure shear strength Ԏf values computed according to the conventional and modified
Mohr-Coulomb failure envelopes for the UU and CU test series were plotted in Figures 6, 7 and 8 which pertain
to the low, intermediate and high principal stress levels respectively. A summary of the regression analysis
results for the total and effective shear strength Ԏf computed according to the two comparative methods for are
given in Table 2.

109
Figure 6. Variations of the conventional and modified Mohr-Coulomb total and effective shear strength at low principal
stress level

Figure 7. Variations of the conventional and modified Mohr-Coulomb total and effective shear strength at intermediate
principal stress level

The trends in Figures 6, 7 and 8 show that the best fit and equality lines are almost coincident indicating a
strong correlation relationships between the shear strength data for the types of soil considered and test
procedures followed. Values of coefficient of determination R2 higher than 0.985were yielded from the analysis
based on the conventional and modified Coulomb-Mohr methods when sets of shear strength data were
correlated as populations. However, when the behaviour of individual soil samples is considered there are
noticeable variations in the analyzed data evaluated from the two comparison methods despite the noted overall
strong correlation as discussed below.
For the total shear strength data, the conventional Mohr-Coulomb method indicated higher values for the
three assumed principal stress levels compared to the modified method in 62 to 66% of the samples tested. For
the effective soil shear strength, the data points appear to be evenly distributed along the best fit lines in all
figures and as may be noted in Table 2, small discrepancies and standard deviations were yielded for the three
principal stress states covered. The conventional Mohr-Coulomb method indicated higher effective shear
strengths than the modified method in 68, 72 and 80% of the samples at the low, intermediate and high stress
state levels respectively. Thus, a consistent pattern was revealed by the both analysis methods for the total and
effective failure shear strengths of the soil types tested.

110
Figure 8. Variations of the conventional and modified Mohr-Coulomb total and effective shear strength at high principal
stress level

Table 2. Analysis of results for total and effective shear strength data determined from the two comparison methods
Standard
Values Regression Discrepancy
Effective soil shear strength parameter R2 Deviation
Range Equation Range (%)
(%)
Total failure shear strength Ԏf in Low 15.7 – 150.0 Ԏfc = 0.981Ԏfm 0.988 -29.8 to + 6.7 6.4
kPa computed at different principal Intermediate 36.8 – 224.5 Ԏfc = 0.983Ԏfm 0.987 -32.5 to +16.3 7.2
stress levels High 66.1 – 511.6 Ԏfc = 0.984Ԏfm 0.986 -33.2 to +21.8 8.2
Effective failure shear strength Ԏ’f Low 48.9 – 172.2 Ԏfc = 0.984Ԏfm 0.995 -5.5 to + 2.9 2.4
in kPa computed at different Intermediate 48.9 – 349.9 Ԏfc = 0.983Ԏfm 0.987 -11.3 to +4.9 3.7
principal stress levels High 48.9 – 558.5 Ԏfc = 0.987Ԏfm 0.990 -18.4 to +5.4 5.0

Chin and Sew (2001) investigated the slope stability problem for the case of residual soils and stated that the
critical slip surfaces are commonly shallow with typically low applied effective stresses and indicated that at
such low effective confining stresses, the conventional Mohr-Coulomb failure envelope might overestimate the
soil strength.
In general, the level of principal stress does not seem to have a significant effect on the relationship between
the magnitudes of total and effective failure shear strength computed according to the conventional and modified
Mohr-Coulomb methods as inferred by the present study results. However, a slightly better agreement between
the two analysis methods was indicated for the shear strength computed at the lower than at the higher stress
levels as reflected from comparison of the discrepancy and standard deviation values in Table 2.
The discrepancies between shear strength parameters and the failure shear strength values determined from
the conventional and modified Mohr-Coulomb methods may be attributed to the different angle inclinations of
the failure plane on which the shear and normal stresses act in the tested samples. On the basis of the findings of
the present study and the experience reported in some previous research works, one should not rely solely on the
conventional Mohr-Coulomb method for the determination of the total and effective soil shear strength
characteristics. The modified Mohr-Coulomb or stress path method has an advantage over the conventional
method as it has a better rational and facilitates a great simplification of data analysis by avoiding the time
consuming and tedious procedure of constructing Mohr’s circles.

5. CONCLUSIONS

The following conclusions may be drawn from the results of the present study on the application of the
conventional and modified Mohr-Coulomb methods for analysis of total and effective shear strength behaviour
for some Sudanese clayey and silty soils:
a) The total and effective angle of soil internal friction determined from the two methods compared very well
for relatively high φ values in most of soil samples tested. However, at relatively low φ levels the
conventional method tends to produce slightly higher values than the modified method.

111
b) The relationships between the total and effective soil cohesion parameters determined from the two analysis
methods were not as good as those obtained for the friction angle as inferred from the significant
discrepancies in cohesion data.
c) In general, strong relationships have been developed between the failure shear strengths evaluated from the
conventional and modified Mohr-Coulomb methods for the soil types covered in this study on the basis of
overall statistical regression analysis. However considering the actually measured or observed behaviour of
individual soil samples, the conventional Mohr-Coulomb method gave higher total and effective failure soil
shear strength than the modified method for most of the soil samples tested.
d) The principal stress state assumed for soil shear strength computations does not seem to have a significant
effect on the relationship between failure shear strength values determined using the conventional and
modified Mohr-Coulomb failure methods.
e) While both analysis methods of soil shear strength considered can produce reliable results, the modified
Mohr-Coulomb or stress path approach is preferred as it relies on better rational and facilitates a great
simplification of data analysis by avoiding the lengthy and tedious procedure of constructing Mohr’s circles.

REFERENCES

Amsiejus [Link] N. and Norkus A., (2010). Analysis of methods for evaluation of soil shear strength parameters .
Proc. 10th International Conferenceon Modern Building Materials Structures and Techniques, Velnius, Lithuania, pp
1077-1082.

Bukhartsev, V. N. (1988). Statistical procccesing of results of stabilometer tests, Soil Mechanics and Foundation
Engineering25(6): pp 562–565. doi:10.1007/BF01721620

Chin, T [Link] Sew, G. S. (2001). The determination of shear strength in residual soils for slope stability analysis. Seminar
Cerun Kebangsaan, Cameron Highlands

Craig, R. F. (2004). Craig’s Soil Mechanics. London: Spon Press. 447 p. ISBN 0-415-32703-2

Ismail H. A. E and Zein A. K. M. (1987) Prediction of the undrained shear strength and standard penetration test using the
static cone penetration test data. Proc. 9th Regional Conference for Africa on Soil Mechanics and Foundation
Engineering,

[Link], T. W., (1964) Methods of estimating settlement. J. Soil Mechanics Div., ASCE, Vol. 90, SM5, Part 1, pp 43-
67

Parry, R. H. G. (2004). Mohr Circles, Stress Paths and [Link]: Spon Press. 264 p. ISBN 0-415-27297

112
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Saturation and Sample reconstitution Effects on the Mechanical Behavior of

Chlef Sand in Northern Algeria-An Experimental study

Effets de la Saturation et la Reconstitution des Echantillons sur le Comportement Mécanique

du Sable de Chlef au Nord de l’Algérie-Etude Expérimentale

N. Della*1, M. Belkhatir1, J. Canou2 and J.C. Dupla2

1
Laboratory of Material Sciences and Environment (LMSE) / University Hassiba Benbouali of Chlef, Chlef,
Algeria
2
Laboratoire Navier Géotechnique (CERMES)/ Ecole Nationale des Ponts et Chaussées, Champs sur Marne,
France
*
Auteur correspondant : Noureddine DELLA, Email : [Link]@[Link]

ABSTRACT – In order to evaluate shear resistance characteristics of sand, which has previously experienced
liquefaction, two series of drained and undrained monotonic triaxial compression tests on medium dense sand
were carried out. In the first test series, the influence of the specimen preparation method and confining pressure
has been studied. It was found that there was a marked difference in the behavior even though the density and
stress conditions were identical. The conclusion was that the soil fabric was responsible for this result. In the
second series of tests, the saturation influence on the shear resistance of the sand was examined. The results
showed that the decrease in Skempton's pore pressure coefficient B improves the shear resistance and increases
the friction angle of the sand.
Keywords: Shear resistance, monotonic, depositional method, drained, undrained, saturation, friction angle.

RÉSUMÉ – A Afin d'évaluer les caractéristiques de résistance au cisaillement du sable, qui a subi précédemment
la liquéfaction, deux séries d'essais de compression monotones drainés et non drainés à l’appareil triaxial sur le
sable moyennement dense, ont été effectuées. Dans la première série d'essai, l'influence de la méthode de
préparation de specimens et de la pression de confinement a été étudiée. Il a été constaté qu'il y avait une
différence très marquée dans le comportement même si les conditions initiales de densité et de contraintes sont
identiques. La conclusion était que le mode de fabrication du sol était responsable de ce résultat. Dans la
deuxième série d'essais, l'influence de saturation sur la résistance au cisaillement du sable a été examinée. Les
résultats ont prouvé que la diminution du coefficient de Skempton B améliore la résistance au cisaillement et
augmente l'angle de frottement du sable.
Mots clés : Résistance au cisaillement, Monotone, mode de déposition, drainé, non drainé, saturation, angle de
frottement.

1 INTRODUCTION

On October 10th, 1980 at 13h25 (local time), the Chlef region in northern Algeria was hit by a disastrous
earthquake of a magnitude of 7.3 according to the calculation of Papastamatiou (1980), followed by strong
aftershocks of magnitude 6 and 6.1 some hours afterwards. Numerous aftershocks were felt during several
months after the quake (Ouyed 1981).
The disaster of October 10, 1980 generated a number of geodynamic phenomena at the ground surface
including: movements of ground of variable nature and size, and especially the liquefaction of the sandy soils
inducing a loss of shearing resistance. The liquefaction in this region was marked by craters of sand, lateral
displacements of ground and settlement of parent soil (figure 1).

113
Figure 1. Sand boils due to the liquefaction phenomenon

2 PRIOR STUDIES

Numerous studies have reported that the behavior of sands can be greatly influenced by the initial state of the
soil. Polito and Martin (2003) asserted that the relative density and skeleton void ratio were factors that seemed
to explain the variation in different experimental results. Yamamuro and Lade (1997, 1998) and Yamamuro and
Covert (2001) concluded that complete static liquefaction (zero effective confining pressure and zero effective
stress difference) in laboratory testing is most easily achieved in sands at very low pressures. Kramer and Seed
(1988) also observed that liquefaction resistance increased with increasing confining pressure.
However the effect of the preparation method of test samples has been the subject of controversy amongst
researchers. Many studies have reported that the resistance to liquefaction is more elevated for samples prepared
by the method of sedimentation than for samples prepared by the dry funnel pluviation and wet deposition
methods (Zlatovic and Ishihara 1997); other studies have found that the specimens prepared by dry funnel
pluviation method tend to be less resistant than those reconstituted by wet deposition method (Mulilis et al. 1977;
Yamamuro and Wood 2004). Other researchers indicated that the tests prepared by dry funnel pluviation are
more stable and dilatant than those prepared by wet deposition (Benahmed et al. 2004; Canou 1989). Vaid et al.
(1999) confirmed this result while showing that wet deposition encourages the initiation of liquefaction in
relation to a setting up by pluviation under water. Yamamuro et al. (2008) have concluded, after their laboratory
investigation, that the method of dry pluviation prmotes the instability of samples contrary to the method of
sedimentation. Wood et al. (2008) found that the effect of the method of deposition on the undrained behavior
decreases, when the density increases. The main focus of this study was to identify the differences in undrained
triaxial compression behavior that can result from using different reconstitution techniques in preparing sand
specimens.
Soil behavior under static or cyclic loading is an important study in geotechnical and earthquake engineering. It
was thoroughly investigated for saturated soils using both postearthquake observations and laboratory tests.
However, less work was devoted to the response of unsaturated soils. Martin et al. (1978) used a mathematical
model to investigate the influence of saturation on the liquefaction potential of soils. The results showed that soil
resistance to liquefaction increases significantly with the reduction of the degree of saturation. Yoshimi et al.
(1989) confirmed this result on the basis of triaxial tests performed on Toyoura sand. More recently, this issue
gained an increasing interest and numerous studies found the same tendancy (Atigh and Byrne 2004; Nakazawa
et al. 2004; Mathiroban and Grozic 2004; Pietruszczak et al. 2003; Yang and Sato 2001; Bouferra 2000; Ishihara
et al. 2001, 2004; Yang 2002; Yang et al. 2004; Bouferra et al. 2007).
This literature survey reveals a real need for experimental data concerning the behavior of unsaturated soils
under monotonic loading. This paper includes a contribution to the experimental data in the arena of unsaturated
soils under monotonic loading conditions. It presents results of drained and undrained monotonic triaxial tests
performed on Chlef sand with various values of Skempton’s pore pressure coefficient B.

3 EXPERIMENTAL METHODS

Presented below is an experimental study of the behavior of medium dense sand under static loading conditions.
Both drained and undrained tests were performed.

114
3.1 Material tested

Sand samples were collected from a liquefied layer of the deposit areas close to the Chlef earthquake
epicentre (October 10th, 1980). All tests in the present study were performed on Chlef sand (Algeria). The sand
contains 5% silt. The grain size distribution curve of this sand is given in figure 2.

Figure [Link] size distribution of the used sand

Its predominant minerals are feldspar and quartz. The silt component is non plastic with a plasticity index of
5.81%. The index properties of the sand used in this laboratory research work are presented in Table 1.

Table 1. Properties of tested soil

γdmin γdmax γs Cu D50 D10

Material emin emax Grains shape
g/cm3 g/cm3 g/cm3 D60/D10 mm mm
O/Chlef 0.54 0.99 1.34 1.73 2.67 3.2 0.45 0.15 Rounded

3.2 Testing equipment

An advanced automated triaxial testing apparatus type Bishop and Wesley (1975) was used to conduct the
monotonic drained and undrained tests (figure 3).

Figure 3. The experimental device used

3.3 Specimen preparation

Triaxial tests were performed on cylindrical specimens measuring 70 mm in diameter and 140 mm in height
(H/D = 2.0).

115
In this study, two methods of sample preparation were used, namely: the dry funnel pluviation and the wet
deposition methods. The methods are briefly described below.
In the first method, the dry soil is poured into the mould with the help of a funnel. By controlling the height of
the funnel, different samples’ densities can be arrived at. The second method consists of mixing the previously
dried sand with a small quantity of water (3%); and then carefully depositing the humid soil in the mould in a
homogeneous manner, as much as possible. The soil was placed in seven successive layers. A constant number of
strokes (25) were applied to arrive at a homogeneous and isotropic structure.

4 RESULTS AND DISCUSSION

4.1 Confining Pressure and Specimens Fabric Effects

The effect of variation of effective confining pressure on the liquefaction resistance of sand is shown in figures
4 and 5. As the confining pressure increases, the liquefaction resistance of sands increases for both dry funnel
pluviation and wet deposition methods. The results in figures 4a and 4b with an initial density of 50% (medium
dense state) for specimens reconstituted by the first method (dry funnel pluviation) at a confining pressure of 50
kPa, show a weaker resistance than those shown at a confining pressure of 100 kPa and 200 kPa. Its resistance
increases at the beginning of loading up to a value of 25 kPa corresponding to an axial strain of 0.5%. Then, it
decreases up to an axial distortion strain of 5% to stabilize passing nearly a quasi steady state (QSS); afterwards,
the sample mobilizes a residual strength increasing the resistance of the sample in the steady state. The stress
path diagram (figure 4b) shows a reduction of the effective mean stress until a value of 10 kPa is reached,
followed by an upward trend characterizing a dilating state. The same trends also exhibited for the samples at
confining pressure of 100 and 200 kPa, with peak deviatoric stress of 48 kPa and 90 kPa respectively.

300 Dry funnel pluviation (RD=50%) 250

P'c= 50 kPa
P'c= 100 kpa
250
P'c = 200 kPa 200
Deviatoric stress q (kPa)

Deviatoric stress q (kPa)

Dry funnel pluviation (RD=50%)

200 P'c = 50 kPa

150 P'c = 100 kPa

P'c = 200 kPa
150

100
100

50
50

0 0
0 5 10 15 20 25 30 0 50 100 150 200 250
(a) Axial strain Eps1 (%) (b) Effective mean stress P' (kPa)

Figure 4. Undrained tests for samples prepared by dry funnel pluviation method:
(a) deviatoric stress-strain curve, (b) stress path

250
100
Wet deposition (RD=50%)

200 P'c= 50 kPa

80 Wet deposition (RD=50%)
Deviatoric stress q (kPa)
Deviatoric stress q (kPa)

P'c=50 kPa P'c=100 kPa

P'c=100 kPa P'c=200 kPa

60 150
P'c=200 kPa

40 100

20 50

0 0
0 5 10 15 20 25 30 0 50 100 150 200 250
(a) Axial strain Eps1 (%) (b) Effective mean stress P' (kPa)

Figure 5. Undrained tests for samples prepared by wet deposition method: (a) deviatoric stress-strain curve, (b) stress path

116
Effective stress paths for undrained triaxial compression tests on Chlef sand for samples prepared by wet
deposition method with initial relative densities of 50% are plotted on the p’-q diagrams shown in figure. 5b. As
can be seen, complete static liquefaction occurred in the test sample with the lowest confining pressure (50 kPa).
Static liquefaction lead to the formation of large wrinkles in the membranes surrounding the specimens. Figure
5b also shows that when the initial confining pressure is increased beyond 50 kPa, the effective stress paths
exhibit a behavior that is characterize by increasing stability or increasing resistance against liquefaction. This is
demonstrated by examining the stress-strain curves in figure 5a. The initial confining pressures and densities are
shown for each tested sample. The stress-strain curves of the 100 and 200 kPa initial confining pressure tests
show that the stress difference does not reach zero as exhibited by the 50 kPa test sample resulting in complete
liquefaction, but rather decreases to a minimum value before increasing again to higher levels. In general, the
effect of increasing confining pressure increases the dilatant tendancies in the soil.

The effect of the specimen fabrication method on maximum deviatoric stress is shown in figure 6a. It can be
noticed from the results that the dry funnel pluviation method (DFP) gives higher values of the maximum
deviator stress: therefore a higher resistance to liquefaction, contrary to the wet deposition method (WD) where
lower values of the maximum deviator stress were obtained. The same tendencies were noticed for the variations
of the values in the peak deviatoric stress given in figure 6b. As it can be seen that the samples prepared by dry
funnel pluviation method exhibit a resistance to the monotonic shearing greater than those made by wet
deposition method.
250 RD=50% 200

Deviatoric stress at peack qpic (kPa)

Maximal deviatoric stress qmax (kPa)

Dry funnel pluviation RD=50%

200 Wet deposition Dry funnel pluviation

150
Wet deposition

150
100

100

50
50

0 0
0 50 100 150 200 250 0 50 100 150 200 250 300
(a) Initial confining pressure (kPa) (b) Effective mean stress p' (kPa)
Figure 6. Effect of the depositional method on maximum deviatoric stress (a) and
on the peak deviatoric stress (b)

4.2 Effect of the saturation

Figure 7 shows the results of the drained compression tests carried out for Skempton’s pore pressure
coefficient B ranging between 13 and 90%. The tests were performed on medium dense samples (RD=50%) at
initial confining pressure of 100 kPa. It is clear from this figure that this coefficient significantly affects the
variation of the deviatoric stress (figure 7a) and the volumetric deformation (figure 7b). The increase in
coefficient B from 13 to 90% induces a reduction in the initial stiffness and resistance of the soil (maximum
deviator). With regard to the volumetric deformation, we note that the increase in Skempton’s pore pressure
coefficient B delays the start up of dilatancy; for comparative purposes, the sample with B=13%, the dilatancy
begins after 3% of axial strain; while the sample with B=90% exhibits an amplification of the contraction phase
of contractancy and the dilatancy is delayed and appears after 12% of axial deformation.

117
250 16

12
200 Chlef sand Chlef sand
Deviatoric stress (kPa)

Volumetric strain (%)

B = 13% 8
B = 13%
B = 21%
150 B = 21%
B = 54% 4
B = 54%
B = 88%
0 B = 88%
100 B = 90%
B = 90%
-4
50
-8

0 -12
0 5 10 15 20 25 30 0 5 10 15 20 25 30
(a) Axial strain (%) (b) Axial strain (%)

Figure7. Influence of Skempton’s pore pressure coefficient B on drained response of Chlef sand

Figure 8 shows the influence of Skempton’s pore pressure coefficient B on the friction angle at the end of
shearing (sinφ= (σ’1- σ’3)/ (σ’1+σ’3)) and the characteristic angle corresponding to the contraction-dilatancy
phase change (transition from the contraction phase to the dilatancy phase). It can be noticed that the two lines
tracing the change of the friction angle and the characteristic angle are almost parallel. We note that these two
angles decrease (according to a quasi-linear relation) from 24° to 19° and from 23° to 19°, when Skempton’s
pore pressure coefficient B increases from 13 to 90% respectively. The reduction of the friction angle and the
characteristic angle with the increase in coefficient B explain the role of saturation in the increase in the
contraction of the studied sand.
40

Friction angle
Characteristic angle
30


0
0 20 40 60 80 100
Coefficient of Skempton, B (%)

Figure 8. Influence of Skempton’s pore pressure coefficient B on the friction angle and the characteristic angle

5 CONCLUSION
The effects of depositional method and saturation on the mechanical characteristics of the Chlef sand were
investigated. Both drained and undrained monotonic triaxial tests were performed at an initial confining pressure
ranging from 50 to 200 kPa. The following primary conclusions were drawn as a result of this study.
1. Specimen forming method in the laboratory has a profound influence on the static undrained behavior of
the sand.
2. At identical void ratio and effective stress, specimens formed by wet deposition tend to be more
contractive and susceptible to liquefaction compared to dry pluviated counterparts, which may even be
dilative.
3. Samples formed by dry funnel pluviation exhibit higher peak shear stress and maximum deviatoric stress
than those prepared by wet deposition.

118
4. Complete static liquefaction occurred at low confining pressures and as confining pressures increased, the
soil became more dilatant and more resistant to shearing.
5. The decrease in Skempton's pore pressure coefficient¦ B leads to a significant reduction in the rate of
development of the pore pressure, and consequently improves the soil resistance to drained static
loading. This, results in a significant effect on the volumetric response inducing an amplification of the
phase of ¦ contraction when Skempton's pore pressure coefficient B increases.
6. The increase of the degree of saturation affects considerably the evolution of: the friction angle and the
characteristic angle of the sand. The friction angle and the characteristic angle have decreased from
24° to 19° and 23° to 19° respectively, when Skempton's pore pressure coefficient B increased from 13
to 90%.

REFERENCES
Atigh, E. and Byrne, P. M. (2004) Liquefaction flow of submarine slopes under partially undrained conditions: an effective
stress approach. Can. Geotechnical J., 41, pp. 154-165.
Benahmed, N., Canou, J. and Dupla, J. C. (2004) Structure initiale et propriétés de liquéfaction statique d'un sable,
Comptes Rendus Mécanique 332, pp. 887-894.
Bishop, A. W. and Wesley, L.D. (1975) A hydraulic triaxial apparatus for controlled stress path testing, Geotechnique, 4,
pp. 657-670.
Bouferra, R. (2000) Etude en laboratoire de la liquéfaction des sols. Thèse de doctorat, Ecole Universitaire des Ingénieurs
de Lille USTLille, pp. 95-111.
Bouferra, R., Benseddiq, N., Shahrour, I. (2007) Saturation and preloading effects on the cyclic behaviour of sand,
International journal of geomechanics, ASCE, pp. 396-401.
Canou, J. (1989) Contribution l'étude et à l'évaluation des propriétés de liquéfaction d'un sable, Thèse de Doctorat de
l'Ecole Nationale Des Ponts et Chaussées, Paris.
Ishihara, K., Tsuchiya, H., Huang, Y. and Kamada, K. (2001) Recent studies on liquefaction resistance of sand effect of
saturation. Proc. 4th Conf. Recent Advances in Geotech. Earth. Engg., Keynote Lecture.
Ishihara, K., Tsukamoto, Y., and Kamada, K. (2004) Undrained behavior of near-saturated sand in cyclic and monotonic
loading, Proc. Conf., Cyclic Behavior of Soils and Liquefaction Phenomena, pp. 27-39.
Kramer, S.L. and Seed, H.B. (1988) Initiation of soil liquefaction under static loading conditions, J. Geotech. Eng 114, No.
4, pp. 412-430.
Lade, P. V. and Duncan, J. M. (1973) Cubical triaxial tests on cohesionless soil, Journal Soil Mechanics and Foundations
Division ASCE 99, No. SM10, pp. 793-812.
Martin, G.R., Finn, W.D.L. and Seed, H.D. (1978) Effects of system compliance on liquefaction tests, Journal of
Geotechnical Engineering Division, ASCE, GT4.
Mathiroban, S. and Grozic, J., (2004) A model to predict the undrained behaviour of loose gassy sand. Proceeding 57th
Geotech. Conf., Session 6G, 16-22.
Mulilis, J. P., Seed, H. B., Chan, C. K., Mitchell, J. K. and Arulanadan, K. (1977) Effects of sample preparation on sand
liquefaction, Journal of Geotechnical Engineering Division, ASCE 103, GT2, pp. 91-108.
Nakazawa, H., Ishihara, K., Tsukamoto, Y., and Kamata, T. (2004) Case studies on evaluation of liquefaction resistance of
imperfectly saturated soil deposits, Proc., Conf., Cyclic Behavior of Soils and Liquefaction Phenomena, Germany.
Ouyed, M. (1981) Le tremblement de terre d’El-Asnam du 10 octobre 1980: étude des répliques, Thèse de 3ème Cycle,
Université de Grenoble.
Papastamatiou, D. (1980) El-Asnam, algéria earthquake of October 10, 1980: field evidence of ground motion in the
epicentral region, Geognosis Ltd., London.
Pietruszczak, S., Pande, G. and Oulapour, M. A. (2003) hypothesis for mitigation of risk of liquefaction, Geotechnique, 9,
pp. 833-838.
Polito, C.P. and Martin II J.R. (2003) A reconciliation of the effects of non-plastic fines on the liquefaction resistance of
sands reported in the literature. Earthquake Spectra 19, No. 3, pp. 635-651.
Vaid, Y. P., Sivathayalan, S. and Stedman, D. (1999) Influence of specimen reconstituting method on the undrained
response of sand, Geotechnical Testing Journal 22, No. 3, pp. 187-195.
Wood, F. M., Yamamuro, J. A. and Lade, P. V. (2008) Effect of depositional method on the undrained response of silty
sand. Canadian Geotechnical Journal 45, No. 11, pp. 1525–1537.
Yamamuro, J.A. and Lade, P.V. (1997) Static liquefaction of very loose sands. Canadian Geotechnical J. 34, No. 6, pp.
905-917.
Yamamuro, J.A. and Lade, P.V. (1998) Steady state concepts and static liquefaction of silty sands, J. Geotech. Geoenv.
Eng. ASCE 124, No. 9, pp. 868-877.
Yamamuro, J.A. and Covert, K.M. (2001) Monotonic and cyclic liquefaction of very loose sands with high silt content, J.
Geotech. Geoenvironmental Engineering ASCE 127, N° 4, pp. 314-324.
Yamamuro, J. A., Wood, F. M. and Lade, P. V. (2008) Effect of depositional method on the microstructure of silty sand.
Canadian Geotechnical Journal 45, No. 11, pp. 1538–1555.
Yang, J. (2002) Liquefaction resistance of sand in relation to P-wave velocity, Geotechnique, 4, pp. 295-298.

119
Yoshimi, Y., Tanaka, K. and Tokimatsu, K. (1989) Liquefaction resistance of partially saturated sand Soils and
foundations, 3, pp. 157-162.
Yamamuro, J. A. and Wood, F. M. (2004) Effect of depositional method on the undained behavior and microstructure of
sand with silt, Soil Dynamics and Earthquake Engineering, 24, pp. 751-760.
Yang, J., and Sato, T. (2001) Analytical study of saturation effects on seismic vertical amplification of a soil layer,
Geotechnique, 2, pp. 161-165.
Yang, J., Savidis, S. and Roemer, M. (2004) Evaluating liquefaction strength of partially saturated sand, Journal of
Geotech. and [Link]., ASCE,9, pp. 975-979.
Zlatovic, S. and Ishihara, K. (1997) Normalized behavior of very loose non-plastic soils: effects of fabric, Soils and
Foundations 37, No. 4, pp. 47-56.

120
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Influence de l’initialisation des contraintes sur les résultats des calculs par la
méthode des éléments finis.
Influence of stress initialization on the results obtained by the finite element method.

M. D. NGOM1, S. BURLON1, N. DRONIUC2 and J. P. MAGNAN1

1
IFSTTAR/Université Paris-Est, Marne-la-Valée, France
2
FUGRO/Nanterre, France

RÉSUMÉ - Les travaux, présentés dans cet article, s’inscrivent dans une thématique de recherche visant à
contribuer à une meilleure appréciation de l’influence des contraintes initiales dans les massifs de sol pour les
calculs réalisés par la méthode des éléments finis. Ce papier comporte une évaluation de la variation des résultats
induits par l’incertitude relative à l’état initial des contraintes. Le cas traité est celui d’une fondation superficielle
chargée verticalement. Trois valeurs de coefficient de pression des terres au repos sont considérées. Les résultats
analysés sont relatifs à la charge limite et aux déplacements de la fondation. L’expansion des déformations
irréversibles est aussi traitée en fonction de la valeur du coefficient de pression des terres au repos. Les
conclusions, à l’attention des utilisateurs de la méthode des éléments finis, porte sur le choix du coefficient de
pression des terres au repos et son influence sur les résultats utiles à la conception des ouvrages.

ABSTRACT – The works, presented in this paper, are a part of a research topic related to the study of initial
stress state in soil mass. This study estimates the impacts of the coefficient of earth pressure at rest on the results
obtained by the finite element method. An example including a shallow foundation submitted to a vertical force
is analyzed in order to understand the relation between the loading, the coefficient K0, and the settlement of the
shallow foundation. Three values of the coefficient of earth pressure at rest are considered. Moreover, the results
are analyzed in order to evaluate the influence of the coefficient K0 on the irreversible deformation of the soil.
This section informs about the extent of irreversible deformation in the soil mass and its evolution in terms of the
value of the coefficient K0. The conclusions of this study aware the users of finite element method to be
conscious of the influence of the coefficient K0 for the design of geotechnical structures.

1 INTRODUCTION

Comparée aux méthodes classiques de dimensionnement à l’équilibre limite, la méthode des éléments finis donne
des informations sur le champ de contrainte, les déplacements et la charge limite. Les informations
complémentaires, qu’elle fournit, font qu’elle est de plus en plus utilisée pour la conception, le dimensionnement
et la justification des ouvrages géotechniques. Son utilisation nécessite la définition du champ initial des
contraintes qui, dans les calculs élastoplastiques, impacte les résultats. Le champ initial des contraintes d’un
massif de sol dépend du type de sol, de l’histoire géologique du massif, du relief etc.…
Dans les massifs de sol normalement consolidés à surface horizontale, le rapport des contraintes effectives
horizontale et vertical est constant. Il correspond au coefficient de pression des terres au repos (K0) et est
introduit par Jaky (1944). De nombreux travaux se sont, par la suite, intéressés à l’approximation du coefficient
de pression des terres au repos. La majorité des expressions empiriques proposées lie le coefficient de pression
des terres au repos à l’angle de frottement interne du sol, au rapport de surconsolidation pour les sols
surconsolidés, et/ou à l’indice de plasticité (Kenney 1959, Hendron 1963, Brooker et Ireland 1965, Mayne &
Kulhawy 1982).
Une incertitude importante affecte les valeurs du coefficient de pression des terres au repos. En effets,
les expressions empiriques proposées sont généralement tirées d’essais réalisés en laboratoire. Ces derniers
doivent satisfaire une condition de non déplacement latérale afin de reproduire les conditions (en déformation)
sur le site. La mesure de la contrainte horizontale avec un déplacement horizontal totalement nul étant
impossible, les valeurs du coefficient K0 trouvées sont entachées d’erreur. A l’erreur faite lors de la mesure
s’ajoute la perturbation engendrée dans le sol par le prélèvement des échantillons. Le cumul d’erreur accentue
l’incertitude relative à la valeur du coefficient de pression des terres au repos.
Les mesures in-situ de la contrainte horizontale se présentent comme une alternative. Cependant
l’introduction des appareils de mesure dans le massif de sol crée des déformations latérales et modifie le champ
de contrainte sur place. L’interprétation des résultats est également une difficulté et représente un risque d’erreur
sur la valeur du coefficient de pression des terres au repos. La détermination, par mesure, du coefficient de

121
pression des terres au repos nécessite la mesure des contraintes verticale et horizontale et de la pression
interstitielle. L’erreur faite sur le coefficient K0 peut être exprimée par l’équation (1) (Leroueil et Tavenas,
1979).

K 0  h  v  v   h u
  
K0  h  u  v  u  h  u  v  u 
K 0 1   h  1  K 0 u
    v   (1)
K0  v  u  K0   h  u K 0

Lors des mesures, une erreur relative de l’ordre de 30% peut entacher la valeur du coefficient de pression des
terres au repos trouvée (Leroueil et Tavenas, 1979).
Le coefficient de pression des terres au repos est d’une importance notable pour le calcul des ouvrages de
soutènement ou pour les travaux d’excavation. Il influence également les calculs de tassement. Les travaux de
Mestat (1998) ont traité le cas d’une excavation faite sur un massif de sol rectangulaire et homogène. Les
résultats de cette étude soulignent l’impact non négligeable du coefficient de pression des terres au repos sur les
chemins de contrainte et de déformation. Le choix d’un coefficient K0 supérieur ou inférieur à 1 peut inverser le
sens des déplacements notés. Les déplacements, l’intensité des contraintes horizontales ainsi que les
déformations irréversibles augmentent avec le coefficient K0. Les mouvements du sol sont également influencés
par la valeur du coefficient de pression des terres au repos. Mestat (1998) montre que le massif de sol a tendance
à essentiellement se soulever pour de faibles valeurs de K0. Pour de fortes valeurs de K0, l’accroissement des
contraintes horizontales entraine un développement des déplacements latéraux.
La valeur du coefficient K0 a de l’influence sur les zones d’apparition des déformations irréversibles au tour
d’un tunnel circulaire excavé en une seule étape. La zone, autour du tunnel, avec un comportement élastique est
d’autant plus grande que le coefficient K0 est petit. (Wong et Kaiser, 1991).
Burlon (2007), lors de ses travaux sur les mouvements de sol induits par des excavations, s’intéresse à
l’influence du coefficient K0. Il montre, à l’issue d’une série de calcul réalisée avec 5 valeurs de coefficient de
pression des terres au repos différentes (0.5, 0.75, 1, 1.25 et 1.5) que :
- la valeur du coefficient K0 impacte les chemins de contraintes.
- la déformée du soutènement est d’autant plus importante que le coefficient de pression des terres au
repos est élevé.
- l’effet du coefficient K0 se fait plus ressentir en plasticité.
- le soulèvement du fond de fouille augmente avec l’augmentation du coefficient de pression des terres
au repos.
Les travaux précités ont mis en relief l’importance du coefficient de pression des terres au repos sur les
déformations subies par les massifs de sol en déchargement (excavation, mise en place de tunnel etc.).
Cet article étudie l’influence de l’initialisation des contraintes sur un massif de sol en chargement. Le cas
simple d’une fondation superficielle est traité. Les travaux présentés dans cet article s’intéressent à l’écart relatif
noté, d’une part, sur les tassements de la fondation et d’autre part, sur l’étendue des déformations irréversibles,
suite à une variation du coefficient de pression des terres au repos.

2 PRESENTATION DE L’EXEMPLE ANALYSE

2.1. Géométrie du modèle et propriétés des sols

Le modèle de cette étude est une fondation superficielle, de largeur B = 2 m et d’épaisseur 0,30 m, posée sur le
massif de sol de 12 m de large et de 20 m de hauteur. Compte tenu des conditions limites, seule la moitié de
l’ouvrage est modélisée (Figure 1). Le massif de sol est supposé homogène et isotrope sur toute sa hauteur.

122
Figure 1. Modélisation de la fondation étudiée.

Tableau 1. Propriétés physico-mécaniques du massif de sol étudié

E (MPa)  (kg/m3)  c’ (kPa) ’ (degré)  (degré)

20 1900 0,3 5 30 0

2.2. Calculs réalisés

L’étude porte sur 3 valeurs du coefficient de pression des terres au repos (K0 = {0.3, 0.5, 0.95}) correspondant
respectivement aux calculs f1-v3-k3, f1-v3-k5 et f1-v3-k9. Le champ initial des contraintes est défini en
équilibrant la contrainte verticale par le poids des terres au dessus du point considéré. La contrainte horizontale
est obtenue avec le coefficient de pression des terres au repos imposé.
Les calculs sont effectués avec le module MCNL (Mécanique en Comportement Non Linéaire) de
CESAR-LCPC. Les modèles sont élastoplastiques parfaits avec le critère de rupture Mohr-Coulomb et
suivent une règle d’écoulement non associée. Le
Tableau 1 donne les propriétés mécaniques retenues pour l’étude.
La fondation étudiée est chargée verticalement d’une pression sur toute sa largeur. Le chargement est
appliqué de façon incrémentale, jusqu’à la divergence du calcul, afin de suivre la réponse du massif à plusieurs
niveaux de chargement. Le calcul diverge quand le chargement appliqué sur la fondation entraine des
déplacements trop importants. Les calculs étudient l’impact du coefficient K0 sur les tassements de la fondation,
d’une part, et sur l’étendue des déplacements irréversibles, d’autre part.

2.3. Principe d’analyse RI et RP

L’influence du coefficient de pression des terres au repos sur le tassement de la fondation est quantifiée par le
rapport « RI » dénommé rapport d’influence. Il correspond, pour un niveau de chargement donné, à l’écart relatif
des tassements calculés avec deux valeurs de K0 différentes. Ce rapport est noté « RKI 0 / K 0  » et est exprimé en
pourcentage. (Voir équation 2)

S K 0  S K 0 
RKI 0 / K 0   (2)
S K0 

Où, S K 0 est le tassement calculé avec K0α , S K 0 est le tassement calculé avec K0β. K0α et K0β font


référence aux deux coefficients de pression des terres au repos comparés.

123
Les rapports RKI 03 / K 05 et RKI 09 / K 05 sont les rapports d’influence associés respectivement à une variation du
coefficient de pression des terres au repos (K0) de 0,5 à 0,3, d’une part, et de 0,5 à 0,95, d’autre part.
Pour étudier l’influence de l’initialisation des contraintes sur l’étendue des déformations plastiques, une
surface carrée de 4m de côté et située sous la semelle est définie comme étant la zone de suivi des déformations
irréversibles (Figure 3a). Dans celle-ci, la norme des déformations plastiques y est affichée. La surface couverte
par les points, ayant subi des déformations irréversibles, est mesurée à chaque niveau de chargement et pour les
différentes valeurs du coefficient K0 prises en compte dans l’étude. Cette surface est notée AP.
Le rapport RP, défini dans l’équation ci-dessous, permet de quantifier l’influence du coefficient de pression
des terres au repos sur la surface AP.
AKP0  AKP0 
R P
K 0 / K 0   (3)
AKP0 

Où, AKP0 est l’aire couverte par les points ayant subis des déformations irréversibles avec K0α, pour un
niveau de chargement donné. Cette aire est appelée surface plastifiée pour la suite de l’étude. AKP0  est la surface
plastifiée avec K0β, pour le même niveau de chargement. K0α et K0β sont les deux coefficients de pression des
terres au repos comparés.

3. ANALYSE DES RESULTATS

3.1. Impact du coefficient de pression des terres au repos sur le tassement de la fondation

Les courbes d’évolution du tassement en fonction de la charge appliquée sont représentées sur la Figure 2a.
Elles montrent que la charge limite de la fondation (QL), qui est la charge maximale appliquée sur la fondation
avant la divergence du calcul, est indépendante de la valeur du coefficient de pression des terres au repos. Elle
vaut 470 kPa avec une variation de plus ou moins 2 kPa selon le calcul. Les trois courbes de la Figure 2a ont une
évolution linéaire jusqu’à un niveau de chargement donné. Ce dernier est noté « QE ». Il varie en fonction du
coefficient de pression des terres au repos. Les valeurs de QE correspondent à un taux de chargement de 20%,
27% et 40% de la charge de rupture respectivement pour des valeurs de K0 de 0,3, 0,5 et 0,95.
Pour un taux de chargement donné, le tassement subi par la fondation varie en fonction de la valeur du
coefficient de pression des terres au repos. Une hausse du coefficient de pression des terres au repos induit une
diminution du tassement subi par le massif de sol. En guise d’illustration, pour un taux de chargement de 85% de
la charge limite (400 kPa), la Figure 2a montre qu’une variation du coefficient de pression des terres au repos de
0,5 à 0,3 entraine une hausse du tassement de la fondation de l’ordre de 24%.
En effet, les contraintes verticales ne dépendant que du poids des couches au dessus du point considéré, une
augmentation du coefficient K0 se manifeste par une hausse de la contrainte horizontale qui augmente le
confinement du sol. Cette hausse du confinement du massif de sol limite les tassements en faisant croître la
rigidité apparente du sol.
La Figure 2b trace l’évolution des rapports RI, définis précédemment, en fonction de la pression appliquée
sur la fondation. Elle met en évidence une forte influence du niveau de chargement du massif de sol sur le
rapport RI.

3.1.1 Evolution de RKI 03 / K 05 en fonction de la charge.

La Figure 2b trace l’évolution du rapport RKI 03 / K 05 en fonction de la charge appliquée sur la fondation. Le
rapport RKI 03 / K 05 diminue de façon monotone avec l’incrémentation du chargement. L’évolution du rapport
d’influence montre que la sensibilité du tassement de la fondation à la variation du coefficient K0 de 0,3 à 0,5,
diminue avec l’incrémentation du chargement. Cette sensibilité est négligeable au voisinage de la charge limite.
En effet, pour un taux de chargement d’environ 20% de la charge limite (100 kPa), la réduction de la valeur du
coefficient de pression des terres au repos de 0,5 à 0,3 augmente le tassement de la fondation de 56%. Pour un
taux de chargement de 95% de la charge limite, la même variation du coefficient de pression des terres au repos
entraine une augmentation du tassement de 9%.

124
3.1.2 Evolution de RKI 09 / K 05 en fonction de la charge.
La Figure 2b trace l’évolution du rapport RKI 09 / K 05 en fonction de la charge appliquée sur la fondation. Elle
montre que l’évolution du rapport RKI 09 / K 05 décrit une courbe en cloche avec un maximum localisé à 60% de la
charge limite. A un niveau de chargement compris entre 20 et 60% de la charge limite, l’augmentation de la
charge appliquée sur la fondation entraine une hausse du rapport RKI 09 / K 05 . Au delà de 60% de la charge limite
de la fondation, le rapport RKI 09 / K 05 diminue avec l’accroissement du niveau de chargement.

(a) (b)

Figure 2. (a) Courbe d’évolution du tassement en fonction de la charge appliquée. (=0,3)

I
(b) Evolution de l’influence de K0 sur le tassement de la fondation ( R K
03 / K 05 et RKI 09 / K 05 ) en fonction de la charge.

En résumé, l’analyse des courbes tracées sur la Figure 2 permet de conclure que :
- la charge limite de la fondation n’est pas impactée par la valeur attribuée au coefficient de pression des
terres au repos ;
- l’influence du coefficient K0 sur les tassements de la fondation étudiée est fortement liée à son niveau de
chargement ;
- pour des taux de chargement de la fondation assez faibles (inférieurs à 50 % de la charge limite), la
variation du coefficient K0 de 0,5 à 0,3 a plus d’effet sur les tassements qu’une variation du coefficient
K0 de 0,5 à 0,95. L’inverse est noté pour des niveaux de chargement plus importants ;
- les tassements calculés sont généralement plus sensibles à l’initialisation des contraintes quand la valeur
du coefficient de pression des terres au repos est inférieure à 0,5.

3.2 Impact du coefficient de pression des terres au repos sur les déformations irréversibles

Les Figure 3 (b et c) illustrent l’étendue des déformations irréversibles dans la zone étudiée pour un niveau de
chargement de 400 kPa. Les résultats obtenus lors des différents incréments de chargement ont permis de tracer
les courbes de la Figure 4a.
Nous relevons une augmentation de la surface AP avec l’incrémentation de la charge appliquée. Une baisse
du coefficient de pression des terres au repos (K0) entraine également un élargissement de la surface AP. Cette
surface est mesurée manuellement par un système de quadrillage réalisé sur les différentes zones plastiques. Une
même échelle colorimétrique est utilisée afin de faciliter la comparaison entre les différentes surfaces mesurées.
L’évolution de l’aire AP est tracée en fonction de la charge appliquée sur la fondation. Les trois courbes
d’évolution correspondant aux trois valeurs du coefficient de pression des terres au repos sont représentées sur la
Figure 4a.
La Figure 4b montre que, quelque soit la valeur du coefficient de pression des terres au repos affecté au
massif de sol, les déformations irréversibles n’apparaissent pas pour un niveau de chargement inférieur ou égal à
20% de la charge limite. Au delà de ce niveau de chargement, l’écart de surface AP noté, pour une variation du
coefficient K0 de 0,5 à 0,3, est plus faible que celui relevé pour une variation du coefficient K0 de 0 ,5 à 0,95.

125
La Figure 4b représente l’évolution du rapport RP, défini précédemment, en fonction de la charge appliquée.
L’allure des courbes d’évolution est quasiment identique pour les deux rapports RP. La Figure 4b montre que les
rapports RP diminuent au fur et à mesure que la charge appliquée sur la fondation augmente. Ils atteignent leur
valeur minimale à un niveau de chargement voisin de 80% de la charge limite. L’augmentation de la charge, qui
conduit à une augmentation des déformations irréversibles, diminue l’effet du coefficient de pression des terres
au repos sur ces déformations. Cette analyse montre que la sensibilité de l’étendue des déformations irréversibles
au coefficient K0 diminue avec l’augmentation de l’étendue des déformations irréversibles. Au delà de 80% de la
charge limite, l’influence du coefficient K0 augmente avec l’accroissement de la charge.
Une analyse parallèle des Figure 4 (a et b) permet de conclure qu’une variation du coefficient de pression
des terres au repos impacte l’étendue de la surface AP. Ce phénomène s’explique, d’une part, par le fait qu’une
augmentation de la contrainte horizontale entraine systématiquement une hausse de la contrainte moyenne.
L’augmentation de cette dernière a tendance à diminuer la plasticité (Burlon, 2007). D’autre part, une baisse du
coefficient de pression des terres au repos engendre un rapprochement du champ de contrainte par rapport au
critère de rupture. En effet, plus le coefficient de pression des terres au repos est faible plus le champ de
contrainte initial est proche du critère de rupture. Par conséquent, le massif de sol, dont le coefficient de pression
des terres au repos K0 est plus faible, rentre plus tôt en plasticité que celui dont le coefficient K0 est plus grand.
L’analyse précédente de la Figure 2a, portant sur l’effet du coefficient K0 sur la charge QE, conforte cette
interprétation. Elle a montré que l’entrée en plasticité du massif de sol dépend aussi de la valeur du coefficient
K0 car le modèle, dont le coefficient K0 est le plus faible, est le premier à subir des déformations irréversibles
(plastiques).

(b) Q = 400 kPa / K0 = 0,3 (c) Q = 400 kPa / K0 = 0,5

(a) AP = 3,503 m² AP = 2,904 m²

Figure 3. (a) zone de suivi des déformations irréversibles et échelle de couleur.

(b) norme de la déformation plastique pour K0 = 0,3. (c) norme de la déformation plastique pour K0 = 0,5.

(b)
(a)

Figure 4. (a) Courbe d’évolution de la surface AP en fonction de la charge appliquée.

P
(b) Evolution du rapport RP ( R K
03 / K 05 et RKP09 / K 05 ) en fonction de la charge.

126
4. CONCLUSION

Les calculs présentés dans cet article permettent de conclure que l’initialisation des contraintes, lors des calculs
par la méthode des éléments finis, a une influence non négligeable sur les résultats. Les tassements calculés,
peuvent dans certaines conditions, varier de plus 70%. Les erreurs commises sur les valeurs du coefficient de
pression des terres au repos se répercutent également, d’une part, sur le développement des déformations
irréversibles au sein du massif de sol et d’autre part, sur leur expansion.
L’intervalle auquel appartient le coefficient de pression des terres au repos joue également un rôle
considérable sur les rapports RI et RP. En effet, les résultats obtenus par la méthode des éléments finis sont plus
sensibles à l’incertitude relative à la valeur du coefficient K0 quand celle-ci est inférieure à 0,5.
Ces analyses permettent de définir des intervalles ou « zone de forte sensibilité au coefficient K0 » :
- quand la valeur du coefficient de pression des terres au repos est faible (< 0,5), son influence sur les
tassements est notable si le niveau de chargement du sol est inférieur à 80 % de la charge limite de la
fondation.
- quand la valeur du coefficient de pression des terres au repos est élevée (> 0,5), son influence sur les
tassements est notable si le niveau de chargement du sol est supérieur à 20 % de la charge limite de la
fondation.
- l’influence du coefficient de pression des terres au repos sur le développement des déformations irréversibles
dans le massif de sol est significative si le niveau de chargement du sol est compris entre 40 et 80% de la
charge limite de la fondation.
Lors de la conception ou de la justification d’ouvrage géotechnique, l’ingénieur doit, si les conditions de
chargement de l’ouvrage coïncident à celles énumérées ci-dessus, prendre en considération l’incertitude qui
affecte la valeur du coefficient K0.

RÉFÉRENCES

Brooker, E. W. and Ireland, H. O. (1965) Earth pressures at rest related to stress history. Can. Geotech. J., 2(1), 1–15.
Burlon, S. (2007) Modélisation numérique des mouvements de sol induits par des excavations et des injections de
compensation. Thèse université des sciences et technologie de Lille, N° 4067.
Jaky (1944) The coefficient of earth pressure at rest. Journal of society of Hungarian Architects and Engineers, pp. 355–
358.
Kenney T. C. (1959) Discussion de l’article de WU (1958). ASCE vol.85, SM3 67-69.
Leroueil S. et Tavenas F. (1979) : Discussion : design parameters for stiff clays. Proc. 7th ECSMFE Brighton 4, 4.44, 167-
169.
Mayne, P. W., and Kulhawy F. H. (1982) K0-OCR relationships in soil. Journal of the Geotechnical Engineering Division,
vol. 108, n°. 6, pp. 851–872.
Mestat P. (1998) Etat de contraintes initiales dans les sols et calcul par éléments finis. Bulletin des Laboratoires des Ponts et
Chaussées – 215- Mai-Juin- ref. 4188 – P.15-32.
Wong R. C. K. et Kaiser P.K., (1991) Performance assessment of tunnels in cohesion less soils. ASCE, Journal of
Geotechnical engineering, vol.117, 12, pp. 1880-1901.

127
128
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Comportement sur chemin œdométrique à succion contrôlée de l’argile du

barrage de Boughrara (Tlemcen).
Behavior on controlled suction œdometric path of Boughrara clay earth dam (Tlemcen).

F-E.M. Derfouf*1, S. Taïbi2, N. Abou-Bekr3 and M-A. Allal4

1
Laboratoire eau et ouvrages dans leur environnement/université de Tlemcen, Tlemcen, Algérie.
2
Laboratoire ondes et milieux complexes/Université du Havre, Le Havre, France
3
Laboratoire eau et ouvrages dans leur environnement/université de Tlemcen, Tlemcen, Algérie.
4
Laboratoire eau et ouvrages dans leur environnement/université de Tlemcen, Tlemcen, Algérie.
*
Derfoufmounir@[Link]

RÉSUMÉ - Dans cet article, on présente un nouvel appareil œdométrique osmotique à double flux. Les premiers
résultats obtenus sur chemin de chargement déchargement à succion contrôlée, sur l’argile du noyau du barrage
de Boughrara (wL=54%, Ip=28%), consolidée à 100kPa, sont représentés. Ces résultats montrent que la succion
influe les paramètres de compressibilité et rigidifie le matériau. L’interprétation des courbes en terme de
contrainte effective de Terzaghi et généralisée, montre que ces derniers se regroupent d’une façon satisfaisante
au tour de la droite NC saturée, justifiant la validité de ce concept pour les sols à succion non nulle.

ABSTRACT – In this paper, a new osmotic oedometer with double flux is presented. The first results obtained
on loading-unloading stress path with constant suction, for the core clay of Boughrara earth dam (wL=54%,
Ip=28%), consolidated at a vertical stress of 100 kPa are presented. These results show that suction influence
greatly the compressibility parameters and contribute to the stiffness of the soil. The interpretation of results with
Terzaghi’s and generalized effective stress shows that curves are gathered satisfactorily around saturated NC
line, justifying the validity of this concept for the non-zero soil suction.

1 INTRODUCTION

Depuis les années soixante du siècle dernier, de nombreux travaux ont été menés afin de comprendre le
comportement hydromécanique des sols non saturés. Ce nombre relativement important de recherches n’a eu lieu
qu’après le développement d’appareils mettant en œuvre les différentes techniques d’imposition de succion.
L’œdomètre est l’un des appareils adaptés assez tôt pour imposer la succion (Escario, 1969). Ainsi, les premiers
résultats obtenus avec de tels dispositifs ont constitué les bases théoriques du développement de modèles propres
aux sols non saturés tels que : les modèles en contrainte effective dans sa version première énoncée par Bishop,
1959, et généralisée (Modaressi et Abou-Bekr, 1994, Loret et Khalili, 2000, etc ), les modèles basés sur le
concept de surface d’état (Matyas et Radhakrishna, 1968, Lloret et Alonso, 1985, etc), et les modèles en
variables d’états indépendantes représentés par le modèle d’Alonso et al, 1990 et ses dérivés.
Á l’université de Tlemcen et en collaboration avec l’université du Havre, un nouvel œdomètre à succion
contrôlée a été développé. Cet appareil permet d’étudier le comportement des sols non saturés suivant différents
chemins de sollicitations hydriques et mécaniques, en utilisant la technique osmotique comme moyen
d’imposition de la succion. Les premiers résultats de caractérisation de l’argile du noyau de barrage de
Boughrara, consolidé à 100kPa, suivant un chemin de chargement déchargement à succion contrôlée sont
présentés. Ces résultats sont ensuite interprétés en terme de contraintes totales et contraintes effectives (Terzaghi
et généralisée).

2 DEVELOPPEMENT D’UN ŒDOMETRE OSMOTIQUE A DOUBLE FLUX

En partant d’un œdomètre pour les sols saturés, les modifications apportées avaient pour objectifs de :
 permettre une imposition de la succion par la base et la tête de l’échantillon. Ceci conduit à une
réduction de la hauteur de drainage et une homogénéisation de la succion dans l’échantillon.
 appliquer des charges élevées sans avoir de problème de circulation du PEG, notamment quand la
concentration de ce dernier devient élevée.

129
 contourner les problèmes d’étanchéité rencontrés dans plusieurs anciens dispositifs, causés par
l’utilisation de la colle pour la fixation de la membrane semi-perméable et ceci par l’utilisation de
joints appropriés.

2.1 Description de la cellule

Sur la figure 1, une coupe transversale de l’œdomètre est représentée. Dans la nouvelle configuration,
l’embase de la cellule (1) est modifiée de façon à permettre l’imposition de la succion par la base de
l’échantillon, et ceci en usinant un réservoir (6) d’un volume de 37.9 cm3, surélevé par une plaque perforée (3)
suffisamment rigide pour supporter les contraintes appliquées à l’échantillon de sol. Cette plaque repose sur un
joint torique (4) logé sur le périmètre du réservoir afin d’assurer son étanchéité. Ce dernier est relié d’un côté au
piston par le biais d’un tube flexible et un robinet (8) et de l’autre, par un deuxième robinet (8) raccordé au
réservoir de la solution de PEG par un autre tube flexible.
Pour la bague de la cellule, un joint torique (7) est logé sur le périmètre assurant ainsi une étanchéité entre la
membrane semi-perméable (5), la plaque perforée (3) et l’échantillon de sol.
L’ensemble de ces modifications permet d’avoir une étanchéité parfaite après le serrage vertical de l’anneau
(15) et la bague par les boulons de fixation (9).
Usuellement, le piston (10) assure la transmission de la charge appliquée à la surface supérieure du sol. Dans
cette cellule, il est confectionné de façon à permettre aussi l’imposition de la succion sur la face supérieure de
l’échantillon. De la même façon que la base de la cellule, un réservoir (11) de 11.6 cm3 est usiné, et sur lequel
repose une plaque perforée rigide (13) de 63 mm de diamètre. Une différence par rapport au système de
l’embase, est que l’étanchéité est assurée par un joint torique (13) logé sur un écrou (14) qui se fixe avec le
piston. Le serrage de ce dernier provoque l’emboitement du joint torique avec la membrane semi-perméable (12)
placée sur la plaque perforée.
Les déplacements verticaux sont mesurés à l'aide d'un capteur de déplacement de type LVDT (Linear
Variable Differential Transformer, N°DCTH1000A, domaine de linéarité  25 mm) avec une précision de l'ordre
de 0,01mm. Les déformations verticales ont été corrigées en prenant en compte les déformations du système
"bâti-cellule-capteur" qui ont été évaluées en utilisant une cale indéformable en acier. En ce qui concerne le
chargement mécanique, ce dernier est réalisé par paliers successifs. Le passage d’un palier à un autre se fait après
stabilisation des déformations verticales. Le type de PEG utilisée au cours de tous les essais est le PEG 6000,
avec une membrane semi-perméable de type Spectra /Por® n°3 avec un MWCO (Molecular Weight Cut-Off ) de
3500 Dalton.

Figure 1. Coupe transversale de l’œdomètre osmotique à double flux ;(1) : Embase de la cellule, (2) : Réservoir de la base,
(3) : Plaque perforée, (4) : Joint torique, (5) : Robinet, (6) : Bague de la cellule, (7) : Joint torique, (8) : Membrane semi-
perméable, (9) : Boulon de fixation, (10) : Piston, (11) : Réservoir du piston, (12) : Plaque perforée, (13) : Joint torique,
(14) : Écrou, (15) Anneau de serrage, (16) bague en plexiglasse.

130
3 MATERIAU ETUDIÉ

Le matériau testé a été prélevé du gisement d’emprunt destiné à la réalisation du noyau du barrage de Boughrara,
situé à l’extrême Nord Ouest de l’Algérie. Il est situé sur le point de confluence entre l’oued Tafna et l’oued
Mouillah à environ un kilomètre, en amont de la localité de hammam Boughrara, et à 60 Km de la ville de
Tlemcen.
Les résultats des essais d’identifications sont représentés dans les tableaux 1 et 2. Ces résultats ont montré
une prédominance de la fraction argileuse, de façon à ce que la valeur de bleu est de 10.11, et plus de 50% des
éléments sont inférieurs à 2µm. Ce matériau est caractérisé par une forte plasticité, et un pourcentage de 5 à 6.5
% de matière organique, déterminée par la méthode de calcination. Cette argile paraît peut active, du fait que le
coefficient d’activité est de l’ordre de 0,39. De la valeur de la teneur en carbonate de calcium et de la surface
spécifique, on peut déduire que c’est une argile marneuse, conformément au constat géologique, composée
essentiellement de Kaolinite et d’illite. D’après la classification USCS/LPC, ce sol est désigné : fO-At, argile très
plastique faiblement organique.

Tableau 1. Paramètres physiques de l’argile de Boughrara.

Granulométrie Plasticité Densité des grains solides

NF P94-056 et 057 NF P94-051 NF P94-054
D60 D10
<80µm <2µm(2) wL wP Ip(1) γs/γw

97% 52% 1µm 0.6µm 54% 26% 28% 2.65

Tableau 2. Paramètres chimiques de l’argile de Boughrara.

Teneur en CaCO3 Valeur de bleu Surface Spécifique Teneur en MO Ac
NF P94-048 NF P94-068 NF P94-068 NF P94-047 (1/2)
20% 10.11 212 m²/g 5-6.5% 0.39

4 PROGRAMME EXPERIMENTAL

L’argile de Boughrara a été initialement préparée sous forme de pâte saturée, ensuite consolidée à une contrainte
de 100kPa sous une pression interstitielle nulle, avant d’être drainée à une succion de 200 kPa, 2000 kPa et 8000
kPa (essais E1, E2, E3). Une phase de chargement-déchargement est entamée en maintenant la succion
constante. La série a été complétée par deux essais saturés classiques : un sur une pâte saturée(E4). et l’autre sur
une pâte consolidée à 100kPa (E5). L’ensemble des chemins suivis et des caractéristiques initiales et finales des
échantillons sont donnés dans la figure 2 et le tableau3.

Tableau 3. : Caractéristiques initiales et finales des échantillons testés

Succion
référence wi(%) ei Sri(%) e100 e100s wf(%) ef Srf(%) Chemin suivi
imposée (kPa)
E1(*) 200 53 1.39 100 0.973 0.826 21.30 0.613 92.08 ABCDE
E2 2000 39.9 1.06 99.68 0.825 0.59 16.12 0.494 86.47 ABFGH
E3 8000 36.07 0.96 99.98 0.815 0.612 15.46 0.48 85.35 ABIJK
Pâte saturée (E4) 0 91 2.48 97.04 / / 30.67 0.870 96.47
Pâte consolidée à /
0 38.88 1.034 99.65 / / 24.78 0.659 99.65
100kPa (E5)
(*) : les caractéristiques initiales correspondent à une pâte saturée, e100 : représente l’indice des vides après chargement à
100kPa, e100s : représente l’indice des vides après mise en succion, sous une charge de 100kPa.

131
Figure 2. Chemin de contraintes suivi.

Dans un premier temps nous avons choisi de consolider nos échantillons directement dans l’oedomètre
osmotique (essai E1, tronçon A-B, figure 2). Ceci permet d’entamer la phase de mise en succion sans
déchargement de l’éprouvette puisque l’état initial (e0) est connu, cependant cet état sera différent pour chaque
échantillon préparé. Cette procédure a été remplacée par consolidation d’une quantité suffisante de matériau dans
un consolidomètre. La pâte consolidée obtenue est conservée dans un sac étanche, ensuite placée dans un endroit
hermétiquement isolé pour être utilisée pour les essais qui restent (E2, E3, E4).
A chaque essai, l’échantillon est extrudé avec la bague de l’oedomètre à partir de la pâte consolidée. Ses
caractéristiques initiales : teneur en eau, indice des vides, degrés de saturation, sont déterminées par pesée des
chutes au Kerdane. Pour les essais E2 et E3, une phase de rechargement en conditions saturées à 100 kPa est
suivie, avant l’application de la succion, afin de retrouver l’état de contrainte initiale (tronçon A-B dans la figure
2).

5 RESULTAS ET DISCUSSION

Les résultats obtenus sont représentés sur la figure 3. Les traits en pointillés correspondent à la phase de
chargement à 100 kPa. La pente de ces droites est comprise entre 0.073 et 0.118. Pendant la phase de mise en
succion sous une charge constante, l’éprouvette se déforme. Cette déformation notée εs est calculée selon la
formule suivante :

e100  e100 s
s  (1)
1  e100

132
Figure. 3. Courbes œdométriques à succion contrôlée de l’argile de Boughrara consolidée à 100kPa, tracées en contraintes
totales.

Pour une succion de 200 kPa, la valeur de εs est de l’ordre de 7.45% est se stabilise au alentour d’une valeur
moyenne de 12.47% pour une succion comprise entre 2000 kPa et 8000 kPa, marquant la présence d’un palier de
retrait identique à celui observé dans les courbes de drainage sous une contrainte extérieure nulle (Fleureau et al,
1993).
En ce qui concerne la phase de chargement déchargement à succion contrôlée, les courbes obtenues mettent
en évidence la présence d’un écrouissage hydrique, ou une surconsolidation induite par la phase de drainage,
résultant dans l’apparition de deux domaines de déformations. Le premier est caractérisé par de faibles variations
de l’indice des vides, et le deuxième par des variations plus importantes. Cette surconsolidation est de plus en
plus observée pour les fortes succions.
Sur la figure 4(a), nous avons tracé l’évolution des pentes de chargement plastique λs et de déchargement κs
calculées dans le plan des contraintes totales en fonction de la succion. Le calcul de λs se fait entre les deux
derniers points de chargement, tandis que celui de κs est réalisé entre le dernier point de chargement et le point
qui correspond au poids du piston (1 kPa). Nous avons aussi ajouté sur la même figure les valeurs des pentes λ et
κ des essais œdométriques saturés E4 et E5. Les résultats obtenus sont en accord avec ceux de la littérature
(Alonso et al, 1990, Cui et Delage, 1996, etc.) et montrent que de l’état saturé à 8000kPa, la pente λs à diminué
de plus de 50 %. Concernant la pente de déchargement κs, il semble que cette dernière est indépendante de la
succion, rejoignant ainsi les observations de Rampino et al, 2000, sur l’argile du Barrage de Metramo. Par
contre, on remarque que l’ensemble des valeurs de κs obtenues après application de la succion sont inférieures à
celles calculée à l’état saturé.
L’examen de l’évolution de la pression de préconsolidation montre aussi que celle-ci augmente avec la
succion (Figure 4b). Sachant que pour l’essai E5, la valeur de la pression de préconsolidation apparente est
estimée à 110 kPa, en passant à une succion de 200kPa, cette valeur à pratiquement doublée, tandis que pour une
succion de 8000kPa, sa valeur est multipliée par plus de cinq.

133
Figure 4. (a) Évolution de la compressibilité λs et κs en fonction de la succion, (b) Évolution de la pression de
préconsolidation en fonction de la succion.

La représentation des résultats dans le plan indice des vides- contrainte effective de Terzaghi (σv+s), conduit
à un décalage des courbes vers la droite (Figure 5). Ainsi, les courbes de chargement des essais E1 et E2 se
trouvent regroupées d’une façon satisfaisante autour de la droite NC saturée (essai E4). Par contre, on observe
que le drainage des échantillons à une succion de 8000 kPa (essai E3) entraine un éloignement de la courbe de
chargement de la droite NC saturée. L’examen des courbes de drainage humidification de l’argile de Boughrara
consolidée à 100 kPa sous contrainte extérieure nulle (Derfouf et al, 2013), a montré que le point d’entrée d’air
correspond à une succion de l’ordre de 2000 kPa. Dans ce cas, on peut dire qu’à partir d’un certain niveau de
succion (S=2000 kPa), ces essais ne sont plus interprétables avec le concept de contrainte effective de Terzaghi.

Figure [Link] œdométriques à succion contrôlée de l’argile de Boughrara consolidée à 100kPa, tracées en contraintes
effectives de Terzaghi.

134
Dans le but de généraliser le principe de contrainte effective aux sols non saturés, Modaressi et Abou-Bekr,
1994, et en se basant sur les travaux de Taibi, 1994, définissent la contrainte effective comme étant la somme de
la contrainte totale et la contrainte capillaire πeq. (Équation 2) Cette dernière est en fonction de la succion, la
densité et la granulométrie du matériau représentée par la valeur de D10. La formulation mathématique de la
contrainte capillaire πeq est donnée par une fonction hyperbolique (Équation 3), caractérisée par une asymptote
quand la succion dépasse la succion d’entrée d’air (c.-à-d., quand le sol devient non saturé).
'    eq (2)

L’expression de la fonction eq :

max S
π eq =π eq th( ) (3)
max
π eq

 eq
max
Caractérise le palier de la succion maximale et « th » représente la fonction tangente hyperbolique.

L’influence de la densité et la granulométrie sur la succion est prise en compte dans le paramètre eq
max
qui s’écrit
sous la forme suivante, déduite du modèle microstructural de Taibi, 1994 :
max 2.π.Ts
π eq = (4)
K(e).D10

Où « Ts » est la tension superficielle à l’interface eau air, et le terme π= 3.141, K(e) est une fonction
d’interpolation pour les arrangements irréguliers de sphères donnée par :
K(e) = 0,32 e2 + 4,06 e + 0,11 (5)
Avec « e » représente l’indice des vides.

Ce principe intervient quand le sol devient non saturé, c.-à-d. au-delà la succion d’entrée d’air, estimée dans
notre étude à 2000 kPa. Dans notre cas, on a constaté que les essais réalisés à des succions supérieures à cette
valeur ne justifient pas une interprétation en contrainte effective de Terzaghi. Les nouvelles courbes sont
retracées sur la figure 6, avec la nouvelle définition de la contrainte effective généralisée, calculée en considérant
D10=0.6µm (cette valeur est déduite de la courbe granulométrique, c.f Tableau 1).

Figure 6 . Courbes œdométriques à succion contrôlée sur l’argile de Boughrara consolidée à 100 kPa, tracées en contrainte
effective généralisée (D10=0.6 µm).

135
On constate que l’ensemble des courbes se situe assez bien avec la droite NC expérimentale déduite du sol
saturé (Essai E4) et celle déduite de la corrélation de Biarez et Favre. Ceci permet de conclure que la contrainte
effective généralisée définie à partir de la contrainte capillaire permet d’interpréter le comportement de l’argile
de Boughrara sur chemin œdométrique.

6 CONCLUSION

L’utilisation d’un œdomètre osmotique à double flux, a permis de présenter les premiers résultats sur chemin de
chargement déchargement à succion contrôlée de l’argile du noyau de barrage de Boughrara consolidée à
100kPa. Cet œdomètre, mettant en œuvre la technique osmotique, permet d’imposer des succions allant jusqu’à
10MPa.
Les résultats obtenus ont montré un certain nombre d’aspects du comportement hydromécanique des sols non
saturés, à savoir : i) le retrait du matériau soumis à un drainage est d’autant plus important que la succion
imposée est élevée, ii) l’imposition de succions croissantes provoque une rigidification de la structure du sol qui
se traduit par une augmentation de la contrainte de préconsolidation et une diminution de la pente de
compressibilité plastique et élastique. Pour cette dernière, les valeurs obtenues restent très proches sur toute la
gamme de succion étudiée
L’interprétation des résultats en fonction de la contrainte effective de Terzaghi a confirmé la validité de ce
principe tant que le sol reste saturé, c.-à-d. pour des succions inférieures à 2000kPa. Le dépassement de cette
valeur de succion entraine une désaturation des échantillons. Dans ce cas-là, on a eu recours à une définition
généralisée de la contrainte effective. Les courbes retracées en fonction de cette contrainte se regroupent autour
de la droite NC saturée, justifiant une interprétation des résultats avec ce principe.

REFERENCES

Alonso E.E., Gens A., Josa.A.(1990) A constitutive model for partially saturated soils. Géotechnique 40, 405-430.
Biarez J., Favre J.L., (1975) Table ronde sur les corrélations des paramètres en mécanique des sols. Ecole Centrale Paris.
Bishop A.W.,(1959) The principle of effective stress. Teknisk Ukeblad 39,859-863.
Cui Y.J., Delage P., (1996) Yielding and plastic behaviour of an unsaturated compacted silt. Geotechnique 46(2), 291-311.
Derfouf F-E.M., Allal M.A., Abou-Bekr N., Taibi S. (2013) Mesure et prédiction des chemins de drainage-humidification.
Application au noyau d’un barrage. Revue Française de Géotechnique 142 (1er trimèstre), 49,60.
Escario V., (1969) Swelling of soils in contact with water at a negative pressure. Proc. 2nd Int. Conf. on Engineering and
Research on Expansive Clays, Texas, 207-217.
Lloret A. Alonso E.E., (1985) State surfaces for partially saturated soils», Proc. of the 11th Int. Conf. on Soil Mechanics and
Foundations Engineering, San Fransisco. 2, 557-562.
Loret B., Khalili N.,(2000) A three-phase model for unsaturated soils»,. Int. J. Numer. Anal. Meth. Geomech. 24, 893-927.
Matyas E.L., Radhakrishna H.S. (1968) Volume change characteristics of partially saturated soils. Géotechnique 18 (4), 432-
448.
Modaressi A., Aboubekr. N., (1994) A unified approach to model the behavior of saturated and unsaturated soils.
Conference on computer methods and advances in geomechanics. [Link] and M.M. Zaman (ed.), Balkema.
Rampino C., Mancuso C., Vinale F., (1990) Laboratory testing on an unsaturated soil: equipment, procedures, and first
experimental results. Canadien Geotechnical journal, 36(1),1-12.
Taibi S.,(1994) Comportement mécanique et hydraulique des sols soumis à une pression interstitielle négative, étude
expérimentale et modélisation. Thèse de Doctorat. École centrale de Paris.

136
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Analyse de la réponse hydromécanique des sols compressibles saturés

Analysis of the hydromechanical response of saturated compressible soils

Abdeldjalil Zadjaoui*1
1 : Faculté de Technologie, Université de Tlemcen (Algérie), Laboratoire Risque Assessment and Management
[Link]@[Link]

RESUME - Cet article décrit un modèle numérique en éléments finis construit sur la base des équations de la
consolidation des milieux poreux élastoplastiques saturés. Les équations du modèle tiennent compte de
l’interaction des déformations du squelette, des écoulements de l'eau. La résolution numérique de ces équations
associe la méthode des éléments finis pour discrétiser l'espace et un schéma d'intégration implicite pour
discrétiser le temps. Les déformations du squelette sont décrites selon l'approche des variables indépendantes.
Une application est décrite, sur un cas réel. Il s’agit de remblai expérimental B de Cubzac – Les- Ponts en
France. Le code de calcul CESAR-LCPC a été utilisé dans cette étude qui vise de faire des comparaisons avec
les travaux de Belkeziz, (1980) et les mesures expérimentaux effectuées sur place. L’auteur donne une piste sur
l’exploitation pratique de ce résultat.

ABSTRACT - This article describes a numerical finite element model constructed on the basis of the equations
of the consolidation of saturated elastoplastic porous media. The model equations take into account the
interaction of skeletal deformities, the water flows. The numerical solution of these equations combines the finite
element method to discretize the space and implicit integration scheme to discretize time. The skeletal
deformities are described according to the approach of the independent variables. An application is described on
a real case. This is experimental embankment B Cubzac - The Ponts in France. The computer code CESAR-
LCPC was used in this study aimed to compare the work of Belkeziz, (1980) and experimental measurements on
site. The author gives the tracks on the practical use of this result.

1. INTRODUCTION

La modélisation du comportement mécanique des sols non saturés ne fait pas l'objet d'un consensus parmi les
spécialistes de la géotechnique. L'extension de la notion de contraintes effectives, telle qu'elle a été proposée par
Bishop (1961) pour servir de cadre à la description des milieux poreux (sols), a fait l'objet de différentes
critiques, souvent étayées par des résultats expérimentaux obtenus en laboratoire (Jennings et Burland, 1962 ;
Matyas et Radakrishna, 1968 ; Fredlund et Morgenstern, 1976). D'autres approches, comme celles d'Alonso et al.
(1988, 1990), ou de Fredlund (1989) semblent apporter une réponse plus complète et mieux décrire la réalité par
une modélisation physique avec deux champs de contraintes indépendants.
Différents modèles théoriques et numériques basés sur les modèles physiques en contraintes effectives ou en
variables indépendantes sont apparus depuis une dizaine d'années, comme ceux d'Alonso et al., (1988), de Nanda
(1989), de Gens et al., (1995) et de Gatmiri et al., (1995).
L'étude présentée dans cet article a été réalisée au moyen du code de calcul en éléments finis CESAR-LCPC
et a cherché à appliquer à la programmation des équations des sols non saturés la même démarche systématique
que pour les calculs relatifs aux sols saturés. Pour décrire le comportement des sols non saturés, nous avons
utilisé une formulation couplée et un comportement élastoplastique avec écrouissage du squelette du sol, qui
permet de prendre en compte l’irréversibilité des déformations.
Cet article rappelle les équations qui représentent le comportement élémentaire des milieux non saturés,
décrit les étapes de leur transformation en équations matricielles adaptées à leur résolution par la méthode des
éléments finis, puis présente quelques résultats de calcul et une étude comparative avec les mesures effectuées
sur place (Belkeziz., (1980)).

137
2. HYPOTHÈSES ET DÉFINITIONS

La représentation mécanique que nous avons adoptée pour la modélisation numérique des sols repose sur la
superposition de trois milieux continus, qui couvrent chacun l'ensemble de l'espace occupé par le sol:
- le milieu global (sans distinguer de phases: on admet que les modifications de la composition de ce milieu
triphasé par suite des mouvements de l'eau et de l'air ont un effet négligeable sur ses propriétés physiques et
mécaniques, c'est à dire que l'on peut définir une loi de comportement à partir de son état initial);
- le milieu continu "air" et le milieu continu "eau" (ces deux milieux peuvent se déplacer l'un par rapport à
l'autre à l'intérieur du milieu global et en sortir ou y entrer).

2.1 Champs de contraintes

Trois champs de contraintes sont définis respectivement dans le milieu global, l'eau et l'air: la contrainte totale ,
la pression de l'eau pw et la pression de l'air pa.
Les conventions de signe utilisées dans les équations sont des conventions "mixtes": contraintes totales
négatives en compression, pressions d'eau et d'air positives en compression.
Pour le calcul, les pressions d'eau et d'air sont remplacées par les charges d'eau hw et d'air ha définies, selon
pw pa
les habitudes de la mécanique des sols, par : hw = +z et ha = +z ,
w a
Avec : γw - poids volumique de l'eau; γa - poids volumique de l'air; z – altitude.
Le champ de contraintes du milieu global dépend directement de la pesanteur et des conditions aux limites et,
indirectement, de la loi de comportement et des autres conditions aux limites. Le champ associé à la pression de
l'eau (respectivement, à la pression de l'air) dépend directement de la pesanteur et des conditions aux limites sur
la charge d'eau (respectivement, la charge de l'air) et, indirectement, de la loi de comportement du matériau
global et des autres conditions aux limites.

2.2 Champs de déplacements et de déformations

Trois champs de déplacements, auxquels sont associés trois champs de déformations, sont utilisés pour décrire
l'état du sol non saturé :
- un champ de déplacements associé au milieu continu global ou « champ principal », noté u;
- un champ de déplacements associé à l'eau, noté uw;
- un champ de déplacements associé à l'air, noté ua.
Au champ de déplacements principal u est associé le champ de déformations totales du milieu global . Le
champ de déplacements associé à l'eau est utilisé sous la forme du vecteur des vitesses moyennes d'écoulement
de l'eau vw et sous la forme d'un scalaire égal au flux d'eau sortant d'un volume unitaire, div vw (et de même pour
les déplacements associés à l’air : va et div va).
Les déformations totales  sont négatives en contraction et les flux sont positifs quand ils sortent du volume
élémentaire.

3. CHOIX D'UN PRINCIPE VARIATIONNEL

Le passage des équations précédentes à une formulation variationnelle permettant d'associer la solution cherchée
(un état d'équilibre local) à un extrémum d'une fonctionnelle (minimum ou maximum) n'est pas une opération
classique dans le cas des milieux triphasiques.
La procédure que nous avons adoptée comporte deux étapes :
 dans la première étape, on fige l'état des fluides et on se préoccupe de trouver l'état d'équilibre du milieu
global en minimisant son énergie potentielle, c'est à dire que l'on écrit un "principe des travaux virtuels"
dans lequel le travail interne (déduit de la loi de comportement) est équilibré par le travail des forces
externes (forces de pesanteur et forces de surfaces, c'est à dire pressions totales) :
  ij   ij d  -  Fi  u i d - S Ti  u i d S  = 0 .

(1)
Cette intégrale correspond au travail de la contrainte totale. Le champ de contrainte  est défini sur le milieu
global et correspond à  ;
 dans la seconde étape, on s'occupe des équations d'écoulement et on écrit des équations correspondant
aux bilans globaux des mouvements des masses fluides (eau et air) présentes dans le sol, compte tenu de

138
l'écoulement, de la compressibilité des fluides, des variations du degré de saturation et des variations
des volumes des pores :
Pour l'eau:

 
a w n S r p w p w d +   w n S r p w d -   w S r  ij  ij p w d -
 

 p 

div  w k w grad  w + z  p w d +   W p w dS w = 0
 w 

S w
(2.a)
Pour l’air :

 a a n 1 - S r + H S r  p a p a d +   a n H - 1 S r p a d +
 

  pa 
  a 1 - S r + H S r   ij  ij p a d  -  div  a k a grad 
 + z  p a d -
  a 
  pw 
 div   a H k w grad 

 w
+ z   p a d +   a dS a = 0


S a
(2.b)
Le couplage entre le solide et l’eau est assuré par la présence de la pression d'eau dans la première intégrale
et de la variation du volume du sol dans la deuxième intégrale. Le couplage entre le solide et l’air est assuré par
la présence de la pression d'air dans la première intégrale et de la variation du volume du sol dans la dernière
intégrale. Le couplage entre l’eau et l’air est assuré par la présence du coefficient de solubilité de l'air dans la
dernière intégrale.
L'application du principe variationnel choisi à l'analyse du comportement au cours du temps des massifs de
sols élastoplastiques non saturés donne le système d'équations suivant, que l'on doit à résoudre dans le domaine
fixe :
dF   De ij ij d    a Fe  1ha ij d    w Fe hw ij d 

   Fi ui d   Ti ui dS   Fe  1 a z ij d   Fe w z ij d  0
 S

  ij h

dFw    w ng1m1 De  S r  ij
T
 t
hw d   n w2  g 2  g1 Fe  w hw d 
t

 h
   n a w  g 2  g1 Fe  a hw d   k w hw, j  whw,i d 
t

  S  w whw dSV  0
Vw w

 ij

 dFa  
 a  a n H  1  m T
1 g 1 D e  1  S r  HS r 
t
 ij ha d 

 h
   a w  a nH  1g1 Fe   a nH  1g 2  w d 
 t

ha h
    a2  a nH  1g1 Fe ha d    a2  a nH  1g 2 a ha d 
 t t
 h
    a2Ca n1  S r  HS r  a ha d 
 t
    a k a ha , j ha ,i d    a H k w hw, j ha ,i d  S  aaha dSV  0
 Va a

(3)
et auquel il faut ajouter les conditions aux limites et les conditions initiales.
L'application du principe variationnel aux équations et la discrétisation dans l'espace (éléments finis) et dans
le temps (méthode implicite) conduisent au système matriciel suivant Kormi (2003):

139
K U  U  = K U  U  + F t + t  + F
H
j
i
j+1
i M
j
i
j
i-1 i
j
 i  t 
j
-F
 i-1

(4.a)
avec
 R -C uw C ua   U 
   
KH = C wu - C ww + t K w C wa   U = H w 
C a u - C aw + t K wa E a + t K a  H 
  a (4.b)
0 C uw - C ua   U 
 
KM = 0 - t K w 0 
 U = H w 
0 - t K wa - t K a  H 
 a (4.c)
 F  t + t   F  t   F   t  
     
F  t + t  = Q w  t + t  ; F  t  =  0  F   t =  0 
 Q  t + t    0   0 
 a      (4.d)
Dans la méthode de résolution décrite ici, le schéma numérique implicite, plus stable numériquement, a été
choisi en raison du caractère non linéaire du problème.
La résolution du problème est réalisée par un procédé itératif. A chaque itération on cherche les valeurs des
inconnues principales, des inconnues secondaires, du résidu et du paramètre d'écrouissage de telle façon que,
pour j (nombre d'itérations) tendant vers l'infini, les champs de valeurs des inconnues soient solution du
problème complet.

4. APPLICATION AU REMBLAI B DE CUBZAC LES PONTS

Le choix de ce cas de remblai B de la commune de Cubzac-Les-Ponts est pleinement justifié par l’existence des
mesures en place des variables pour faire une comparaison avec les calculs du module CSNL (CESAR-LCPC).
Le site expérimental comprend selon l’étude géotechnique :
- Une couche mince de terre végétale d’environ 0,30 m d’épaisseur.
- Une couche d’argile limoneuse de 2 m d’épaisseur approximativement, formant une croute de surface
surconsolidée et altérée par dissection, le niveau de la nappe phréatique variant avec les saisons.
- Une couche d’argile molle grise, pouvant atteindre 6m d’épaisseur, plus ou moins organique et
légèrement surconsolidée.
Le substratum est constitué d’une couche de graves, située au-dessous de 9 m (limite inférieure de la figure
1).
La figure 2 présente le chargement utilisé dans le calcul. Nous avons essayé de reproduire le plus fidèlement
possible le chargement réel surtout la chronologie de la construction de remblai (Belkeziz 1980).

Figure 1: Remblai « B » Cubzac-Les-Ponts (Belkeziz 1980). Figure 2: Calendrier de chargement (Belkeziz 1980).

Les valeurs moyennes de plusieurs mesures effectuées donnent les paramètres de calcul qui sont regroupés dans
le tableau.1. (Ould Amy et Magnan, 1991)
Le choix des paramètres utilisés dans le calcul par élément finis est très important, si l'on tient compte du fait que
les courbes de variation des caractéristiques mécaniques du sol ne représentent que les valeurs moyennes des
mesures effectuées.

140
De plus, des paramètres comme la perméabilité sont difficiles à estimer globalement et ne peuvent être mesurés
d'une manière assez correcte que sur le site.

Tableau 1: Paramètres introduits dans les calculs.

Couche  E (kPa) kx (m/s) ky (m/s) K0

-9 -9
0m - 1m 0,35 1814 0,46.10 0,90.10 0,5
1m - 2m 0,35 3748 1,25.10-9 1,25.10-9 0,5
2m - 4m 0,35 222 2,50.10-9 0,72.10-9 0,5
4m - 6m 0,35 246 2,95.10-9 1,08.10-9 0,5
6m - 9m 0,35 382 1,30.10-9 0,94.10-9 0,5
-9 -9
0m - 9m 0,35 300 1,6.10 0,96.10 0,5

5. RESULTATS DES CALCULS

5.1 Déplacements horizontaux

La figure 3 représente les déplacements horizontaux en pied du talus du remblai, pour différentes valeurs du
temps, le sol de fondation étant considéré comme homogène. On y remarque que les déplacements horizontaux
sont trop importants par rapport aux mesures.
De plus, les déplacements horizontaux maximaux (42,5 cm) sont obtenus à t = 0 (fin de la période de
construction). Les déplacements tendent ensuite à revenir vers l'axe du remblai. Ce sens de variation des
déplacements horizontaux est contraire à toute expérience.
Une explication de ce phénomène est donnée par l'expression de la variation de la déformation latérale Δε3
dans la théorie de l'élasticité linéaire qui reste négative pendant le processus de consolidation.

Figure 3: Déplacements latéraux sous le pied du talus du remblai (sol Figure 4: Déplacements latéraux sous le pied du talus du
homogène). remblai (sol hétérogène).

Pour le calcul, tenant compte de l'hétérogénéité du sol de fondation, les déplacements horizontaux sont moins
importants (figure 4). L'introduction des couches raides de surface contribue à retenir considérablement les
déplacements horizontaux.
Cependant, le sens de variation des déplacements horizontaux reste contraire à l'expérience et la comparaison
avec les mesures n'a pas de sens.
Ce retour des déplacements horizontaux vers l'axe du remblai s'explique simplement par le fait que les zones
supérieures, ne s'étant pas plastifiées, a un comportement élastique, pour lequel ce phénomène est observé. De
point de vue modélisation reste de faire un effort important dans le choix des paramètres permettant une
plastification plus conséquente. Ce résultat montre la limite d’un modèle élastique isotrope dans la prédiction des
déplacements horizontaux.
Une explication de ce phénomène est donnée par l'expression de la variation de la déformation latérale Δε
dans la théorie de l'élasticité linéaire qui reste négative pendant le processus de consolidation.

141
5.2. Tassements

Les théories linéaires isotropes semblent cependant prédire de façon convenable les déplacements verticaux
(figure 5) à 1000j, les tassements mesurés et calculés sont pratiquement identiques.
On constate cependant un écart important entre les courbes mesurée et calculée au début de la consolidation.
Cet écart est plus apparent sur la figure 6, qui représente les tassements en fonction de log(t).
A la fin de la phase de construction, on calcule 34 cm de tassement au lieu des 12 cm mesurés (Belkeziz
32cm). L'écart avec les mesures décroît cependant rapidement, et il est logique de l'attribuer à l'hypothèse de
paramètres de l'élasticité constants.
Les calculs élastiques isotropes supposant le sol de fondation homogène provient la stabilisation des
déplacements verticaux entre 10000j et 20000j à pratiquement 1m.
L'introduction des couches raides de surface permet de réduire les tassements immédiats à 19cm (Belkeziz 17
cm). Le tassement final maximum est également moins important et se stabilise à 86cm à15000j (Belkeziz 80 cm
à 5000j) (figure 7).

Figure 5: Tassement au centre du remblai (sol homogène). Figure 6: Tassement au centre du remblai en fonction
de log(t) (sol homogène).

Figure 7: Tassement au centre du remblai (sol hétérogène). Figure 8: Tassement au centre du remblai en fonction de
log(t) (sol hétérogène).

5.3. Pression interstitielle

La figure 9 représente les surpressions interstitielles pour différentes valeurs du temps. Leur forme semble
convenable et correspond aux conditions aux limites imposées.
La comparaison entre les surpressions interstitielles calculées et mesurées (figure 10) montre une bonne
correspondance.

142
Figure 9: Isochrones des surpressions interstitielles.

Cependant, l’examen de la courbe de dissipation des pressions interstitielles en fonction du temps à -5 m de

profondeur (figure 11) indique une vitesse de dissipation des surpressions interstitielles calculée est plus rapide
que dans la réalité (la valeur mesurée résulte d'une interpolation des mesures effectuées à 5m et à 8m de
profondeur).
Il est cependant important de remarquer, que l'effet Mandel-Cryer n'est pas observé.
L'effet Mandel-Cryer, rappelons-le, se caractérise par des augmentations locales de la pression interstitielle
au-delà de la charge appliquée. Il s'explique physiquement de façon simple.
En raison de la chute brutale des pressions interstitielles et des contraintes totales sur les faces drainées, une
redistribution des contraintes reporte une partie des forces sur les régions centrales où la pression interstitielle
augmente.

(a) (b)

(c) (d)
Figure 10 : Isochrones des surpressions interstitielles, comparaison les mesures (sol homogène). (a) :10 jours, (b) :140
jours, (c) :640 jours, (d) :1040 jours.

143
(a) (b)
Figure 11: Dissipation de la pression interstitielle (sol homogène). (a) : à -5m, (b) : à -8m.

6. CONCLUSION GENERALE

Conscients de la difficulté à la fois théorique et numérique que posent les problèmes de couplage
hydromécanique, nous avons adopté au cours de ce travail une démarche par étapes.
La comparaison entre la modélisation numérique et la résolution des équations analytiques de Terzaghi et de
Gibson nous a permet d’être sûre de la bonne utilisation de ce module d’une part et d’autre part une validation
puisqu’il s’agit d’un module au cours de développements. Une très bonne cohérence entre les deux méthodes
dans le cas d’un comportement élastique isotrope.
La prise en compte du comportement élastoplastique du squelette du sol est possible dans le programme de
calcul par éléments finis des problèmes de consolidation CESAR-LCPC par le module CSNL.
Il reste cependant difficile par la comparaison des résultats des calculs avec ceux de la campagne de mesures,
l’ingénieur est donc confronté à un problème de choix pour l'étude des remblais sur sols compressibles, chacun
ayant ses avantages et ses inconvénients.
Le modèle élastique linéaire isotrope, s'il est attrayant par sa simplicité, présente de sérieuses divergences
avec les mesures, particulièrement en ce qui concerne les déplacements latéraux, très importants en début de
consolidation et revenant ensuite vers l'axe du remblai, et en ce qui concerne les tassements immédiats,
également importants;
Le comportement élastoplastique est certainement le modèle qui représente le mieux le comportement réel
des sols compressibles. Les résultats obtenus à l'aide d'un tel modèle et du critère Cam-Clay modifié restent
cependant sensibles aux valeurs des paramètres. Les critères de plasticité peuvent être améliorés par la prise en
compte de l'anisotropie et le réexamen du cycle itératif dans les zones qui ramollissent.
Malgré l’utilisation de la version développement pour l’application de modèle Cam-Clay modifie au remblai
expérimental, le problème de convergence du modèle ne s’est pas posé. (deuxième trimestre 2014).
Concernant les courbes de dissipation des surpressions interstitielles, relativement écartées par rapport aux
courbes des mesures, peut incriminer, en particulier, l'hypothèse des coefficients de perméabilité constant.

REFERENCES BIBLIOGRAPHIQUES

Abida H. (1992). La modélisation des sols non saturés - analyse numérique. Thèse de Doctorat, Ecole Nationale des Ponts et
Chaussées, Paris, 271 pages.
Alonso A.E., Batlle F., Gens A., Lloret .A. (1988). Consolidation analysis of partially saturated soils. Application to earth
dam construction. Proceedings, 6th International Conference on Numerical Methods in Geomechanics, Innsbruck,
Balkema, pp. 1303-1308.
Alonso A.E., Gens A., Josa A. (1990). A constitutive model for partially saturated soils. Géotechnique 40, n°3, pp. 405-430
Belkeziz A. (1980), Etude par la méthode des éléments finis les problèmes de consolidation des sols fins, Thèse de Doctorat
a l'Université de Pierre et Marie Curie Paris VI (France).
Biot M. A. (1955), Theory of elasticity and consolidation for a porous anisotropic solid, J of applied physics, Vol. 26, pp
182-185.
Bishop A.W., Donald I.B. (1961). The experimental study of partly saturated soils. Géotechnique, vol. 13, n°3, pp. 177-197.
Fredlund D.G. (1989). Discussion leader's report. Proceedings, 12th International Conference on Soil Mechanics and
Foundation Engineering, Rio de Janeiro, vol. 5, pp. 2861-2878.
Fredlund D.G., Morgenstern N.R. (1976). Constitutive relations for volume change in unsaturated soils. Canadian
Geotechnical Journal, n°13, pp. 261-276.

144
Gatmiri B., Tavakoli S., Moussavi S., Delage P. (1995). Numerical approach of elastoplastic consolidation of unsaturated
soils. Proceedings of the first International Conference on Unsaturated Soils, /UNSAT'95 Paris/, France, vol 2, pp 1057 -
1064.
Gens A., Vaunat J., Ledesma A. (1995). Analysis of hydratation of an engineered barrier in a radioactive waste repository
scheme using an elastoplastic model. Proceedings of the first International Conference on Unsaturated Soils, /UNSAT'95
Paris. vol 2, pp 1057 - 1064.
Jennings J.E.B., Burland J.B. (1962). Limitations to the use of effective stresses in partly saturated soils. Géotechnique, 12,
n°2, pp. 125-144.
Kormi T. (2003), Modélisation numérique du gonflement des argiles non saturées. Thèse de doctorat, Ecole Nationale des
Ponts et Chaussées, Paris, (France).
Magnan J. P. (1986), Modélisation numérique du comportement des argiles molles naturelles, Rapport de recherche, Paris,
n° 141, 255p.
Matyas E.L, Radhakrishnan A.S. (1968). Volume change characteristics of partially saturated soils. Géotechnique, vol. 18,
pp. 432-448.
Nanda A. (1989). Finite element analysis of unsaturated coupled flow and deformation. Rapport CERMES, Ecole Nationale
des Ponts et Chaussées, 18 pages.
Ould Amy M., Magnan J.P: (1991). Modélisation numérique des écoulements et des déformations dans les barrages de terre
construits sur des sols mous. Collection «Etudes et recherches des LPC », série Géotechnique, n°10. 145 pages.

145
146
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Des Lois de Comportement des Géomatériaux Pertinentes pour les Calculs aux
Eléments Finis
Relevant constitutive models for geomaterials used in Finite Element analysis

Emmanuel BOURGEOIS 1*, David REMAUD 2

1
Université Paris-Est, IFSTTAR-COSYS, Marne la Vallée, France
2
itech, Saint-Maurice, France

RÉSUMÉ — Une boîte à outils de lois de comportement a été développée et intégrée au logiciel aux Eléments
Finis CESAR-LCPC. Elle synthétise les travaux de recherche récents portant sur une meilleure compréhension
des géomatériaux. Dans cet article, on la met en œuvre sur deux exemples simples de fondation et d'excavation.
On montre ainsi sa valeur ajoutée apportée à l'ingénieur géotechnicien pour une meilleure analyse du
comportement des ouvrages géotechniques.

ABSTRACT — A numerical toolbox of constitutive models has been developed and integrated in the software
CESAR-LCPC. It encompasses the recent research works carried for a better understanding of the geomaterials
behaviour. In this article, the authors apply the tool on two simple examples of foundation and excavation. Thus
they show its value added to geotechical engineers for a better analysis of earthworks behaviour.

1 INTRODUCTION

La modélisation des ouvrages géotechniques par la méthode des éléments finis est maintenant devenue d’un
usage commun en bureau d’études. Les outils logiciels proposés sur le marché permettent une grande complexité
des géométries, des chargements et une analyse fine des résultats. Pour autant, la loi de Mohr-Coulomb est
encore employée dans la grande majorité des applications, alors que le retour d’expérience indique qu’elle ne
répond pas à toutes les problématiques. Aussi le modélisateur doit maintenant consacrer plus d’énergie à la
compréhension du comportement des géomatériaux employés dans ses modèles.
Dans ce but, l’IFSTTAR a conçu pour son logiciel CESAR une boîte à outils de lois de comportement
élastoplastiques mettant à disposition des bibliothèques de lois élastiques, de critères de plasticité et de lois
d’écrouissage (IFSTTAR, 2014). Ce développement permet de capitaliser le grand nombre de lois de
comportements développées pour divers matériaux du Génie Civil, dont les géomatériaux, par les équipes de
recherche depuis l'origine de CESAR. Ainsi en fonction du type de matériau et du contexte du projet, l’ingénieur
pourra mettre en œuvre une nouvelle loi de comportement adaptée.
L’article présente le concept de la boîte à outils lois de comportement, et deux exemples d'utilisation sur des
modèles d’ouvrages géotechniques simples. Le premier met en œuvre le modèle de sol écrouissable HSM sur un
ouvrage de soutènement. Le second met en œuvre le critère de Mohr-Coulomb avec une élasticité non linéaire
pour l’analyse d’une fondation mixte.

2 PRESENTATION DE L’OUTIL

2.1. Développement numérique

Pour élargir la gamme de lois de comportement utilisables avec CESAR, on a introduit une nouvelle structure
des lois de comportement pour les éléments de massif en mécanique, dans laquelle l’utilisateur est invité à définir
le comportement comme la combinaison d’une élasticité choisie dans une liste prédéfinie (élasticité linéaire ou
non, isotrope ou non), et d’un ou deux mécanismes plastiques (définis chacun par un critère, une loi
d’écoulement, et éventuellement une loi d’écrouissage).
On peut aussi modéliser une évolution non drainée (en prenant en compte la contribution du fluide au module
de compression du fluide) ou prendre en compte une amélioration des propriétés élastiques dues à des inclusions
de renforcement, au moyen d'une approche de type homogénéisation (proposée par de Buhan et Sudret, 1999).
Chacun de ces éléments est à choisir dans une liste de modèles préexistants, mais la possibilité de combiner
les éléments du comportement permet de multiplier les lois de comportement utilisables.

147
En pratique, l'utilisateur utilisera les boîtes de dialogues des interfaces 2D et 3D de CESAR-LCPC (cf. §2.2).
La loi de comportement sera alors traduite dans le jeu de données par une succession de lignes, chacune associée
à un aspect de la loi considérée.
Pour un problème mécanique, la seule information indispensable pour le traitement numérique concerne la loi
d’élasticité, qui permet de calculer une matrice de rigidité.
Le logiciel propose une liste de lois élastiques (qui pourrait être étendue pour prendre en compte d'autres lois
non linéaires, ou une élasticité orthotrope générale, ou différents types de variations spatiales des propriétés
élastiques) :
- élasticité linéaire isotrope homogène
- élasticité linéaire isotrope avec E et  variant linéairement avec la profondeur
- élasticité linéaire isotrope avec E variant comme une puissance de la profondeur
- élasticité linéaire isotrope transverse
- élasticité non linéaire de type Cam-Clay (module de compression K variant linéairement avec la
contrainte moyenne)
- élasticité non linéaire de type Fahey et Carter (1993)
- élasticité non linéaire du Hardening Soil Model (Schanz et al, 1999).
Pour la définition des mécanismes plastiques, un certain nombre de fonctions du tenseur des contraintes
peuvent être utilisées comme critères. La plupart des fonctions utilisables comme critère peuvent également l'être
comme potentiel plastique, si l'on souhaite définir une loi d'écoulement non associée.
Les critères de plasticité pris en compte dans la nouvelle structure des lois de comportement sont les suivants:
- les critères "isotropes" classiques : Tresca, von Mises, Mohr-Coulomb, Drucker-Prager, Cam-Clay
modifié, Hoek-Brown (1997), critère "parabolique", critère de Mohr Coulomb avec c et  variant
linéairement avec la profondeur :
- des critères anisotropes (pour tenir compte de la fracturation des roches ou de l'anisotropie de résistance
de certains matériaux, comme la maçonnerie) : critère orienté, critère de Hill (1947) et variante
proposée par Lourenço et al (1997),
- le modèle Hierarchical Single Surface proposé par Desai (1980),
- le critère du mécanisme frottant du Hardening Soil Model (Schanz et al, 1999).
Dans le cas de l'élastoplasticité avec écrouissage, on peut décrire une évolution du domaine élastique, en
faisant varier un ou plusieurs paramètres du critère. Un certain nombre de lois d'écrouissage classiques ont été
introduites pour certains critères (critère de Cam-Clay modifié, critère du mécanisme frottant du Hardening Soil
Model, parmi d'autres).
La plupart des critères dont la surface de charge est régulière peuvent être combinés pour définir un modèle à
deux mécanismes plastiques (figure 1).
La nouvelle structure permet de reconstituer la plupart des modèles élastoplastiques disponibles jusque-là
dans CESAR (Mestat, 1993), et introduit un grand nombre de possibilités nouvelles.

2.2. Interface graphique

La boîte à outil a été programmée dans les interfaces 2D et 3D de CESAR-LCPC par la société itech. De plus, la
gestion des paramètres matériaux a été entièrement revue, de manière à permettre la sauvegarde et le partage de
bibliothèques de paramètres, et la modification aisée et interactive des valeurs de paramètres.
Les possibilités de combinaisons suivantes (figure 2) sont proposées dans une fenêtre de dialogue qui permet
les opérations sur les bibliothèques, et la visualisation des lois de comportement créées et des valeurs des
paramètres associées.

3 EXEMPLES DE MISE EN OEUVRE

3.1. Fondation superficielle

Le premier exemple concerne le calcul du tassement d'une fondation circulaire soumise à une charge verticale.
La thèse de Sophie Coquillay (2005) présente des simulations d'essais de chargement réalisés à Labenne, sur des
fondations encastrées à différentes profondeurs. Le sol est constitué par une couche de sable propre de 6 m. Les
essais ont été effectués sur une demi-douzaine de configurations différentes. Des études antérieures (Mestat et
Berthelon, 2001) avaient montré qu'il était difficile de reproduire les courbes charge-tassement en utilisant le
modèle de Mohr-Coulomb classique (c'est-à-dire avec une loi élastique linéaire) ou avec le modèle de Nova : les
simulations ne permettaient pas de retrouver la courbure observée expérimentalement.

148
Coquillay a étudié l'apport d'une loi élastique non linéaire, celle proposée par Fahey et Carter (1993), dans
laquelle le module de cisaillement dépend du déviateur , demi-différence entre les contraintes principales
σ1  σ 3
extrêmes ( τ  ).
2

 g 2
1   τ  
 f  

τ
 max 
Gt
   (1)
G0   τ  
g
1  f 1  g    
 τ
 max  
 
Dans l'expression précédente, la valeur de référence du module de cisaillement G 0 est donnée par :

0 ,5
G0  p 
 C   (2)
pa  pa 

où p désigne la contrainte moyenne et p a une pression de référence (par exemple la pression atmosphérique).
D'autre part, f et g sont deux paramètres sans dimension, et τ max dépend des caractéristiques de résistance c et 
suivant :

3  p sin   c cos  
τ max  (3)
3  sin 

mécanisme plastique n°1 mécanisme plastique n°2

critère de plasticité n°1 critère de plasticité n°2

Loi élastique + + + c omportement
drainé ou non
potentiel plastique n°1 potentiel plastique n°2

loi d’écrouissage n°1 loi d’écrouissage n°2

Figure 1. Synoptique des possibilités de combinaisons possibles

149
Figure 2. Exemples de boîtes de dialogue de CESAR-LCPC (version 6) pour la saisie des paramètres définissant une loi de
comportement

On complète la formulation du modèle en choisissant le coefficient de Poisson de telle sorte que le module de
compression ne dépende pas du déviateur :

1  ν0   Gt 1  2ν0 
G0
νt  (4)
2 1  ν0   t 1  2ν0 
G
G0

L'utilisation de ce modèle permet de reproduire l'ensemble des essais, en utilisant le même jeu de paramètres
pour toutes les configurations, issu de calages sur des essais triaxiaux. Un exemple de comparaison entre les
simulations avec différents modèles et les résultats expérimentaux est présenté sur la figure 3.

150
Figure 3 . Comparaison des simulations numériques avec les modèles de Mohr Coulomb, de Nova et de Fahey et Carter avec
les résultats expérimentaux sur une des fondations expérimentales testées à Labenne (d'après Bourgeois et al, 2005).

D'un point de vue qualitatif, une analyse détaillée des résultats montre que le modèle concentre les
déformations au voisinage de la fondation, alors que les modèles à élasticité linéaire ont tendance à produire des
déformations à des distances très importantes (et souvent peu réalistes) des ouvrages.
La mise en œuvre de ce modèle est possible très simplement à l'aide de la boîte à outils implantée dans
CESAR, en choisissant au lieu de la loi élastique isotrope de base la loi de Fahey et Carter.

3.2. Excavation

On présente ici à titre d'illustration un exemple (fictif) d'une excavation non soutenue. Dans ce type de
configuration, les simulations conduites avec le modèle de Mohr-Coulomb conduisent à des résultats peu
réalistes, au moins au début de l'excavation : la paroi tend à se déplacer vers le terrain soutenu.
Un des intérêts du Hardening Soil Model (Schanz et al, 1999) est de mieux rendre compte du comportement
réel de l'ouvrage. C'est la raison pour laquelle on a implanté une version simplifiée du modèle dans CESAR, qui
ne prend en compte que le mécanisme plastique frottant.
Ce modèle adopte une loi élastique isotrope non linéaire. Le coefficient de Poisson est constant et noté  ur ,
mais le module d'Young, noté Eur , est donné par :

m
ref 
  3 sin c cos 
Eur  Eur  (5)
 
 pref sin c cos 
où  3 et p désignent respectivement la contrainte principale mineure et la contrainte moyenne.
Le critère de plasticité est défini par :

2  Rf 2  (  1   3 )2
f (  )  q a (  1   3 )   2   p [ qa  (  1   3 )] (6)
 50
E Eur  E ur
où  p
est le paramètre d'écrouissage, et où l'on a posé

2  3 sin  c cos 
qa  (7)
Rf 1  sin 
m

ref   3 sin  c cos 

E50  E50  (8)
 pref sin  c cos  
 

151
Comme pour le critère de Mohr-Coulomb, la surface de charge n'est pas régulière et présente des arêtes. On
trouvera dans (Schanz et al, 1999) des informations sur la signification physique des différents paramètres.
La loi d'écoulement est non associée ; elle est caractérisée par la valeur de l'angle de dilatance  m , dont la
valeur est donnée par une règle relativement complexe (Plaxis, 2014). Il reste à décrire l'évolution du paramètre
d'écrouissage ; elle est donnée par :

d p  d vp / sin m (9)

On notera que, dans le modèle HSM de Plaxis, pour éviter que le sol ne gonfle indéfiniment lorsqu'on le
cisaille, l'utilisateur a la possibilité de définir une valeur de l'indice des vides au-delà de laquelle l'angle de
dilatance est nul. Ce traitement complémentaire de la dilatance n'a pas été reproduit dans la version du modèle
programmée dans CESAR. D'autre part, le modèle HSM original comporte un deuxième mécanisme plastique
(contractant), qui met en jeu une surface de charge qui borne le domaine élastique en compression. Ce deuxième
mécanisme n'a pas été introduit dans CESAR.
Ce modèle est mis en œuvre pour modéliser l'excavation dont la géométrie est représentée sur la figure 4a.
Les contraintes initiales sont géostatiques ( = 20 kN/m3 et K o =0,5), et on modélise l'excavation en déformation
plane, en une seule fois, de la zone rectangulaire en haut à gauche de la figure 4a. Le maillage est représenté sur
la figure 4b : Il comporte 2200 nœuds et 1100 triangles à 6 nœuds.
Le calcul est fait en condition drainée, le sol étant supposé complètement sec.

40 m

10 m zone excavée

20 m

Figure 4 . a) géométrie et conditions aux limites pour le calcul d'excavation ; b) maillage utilisé

Les paramètres adoptés pour le calcul sont les suivants :

c = 50 kPa ;  =  = 30 degrés ; p ref = 100 kPa ; m =0,5 ; R f = 0,9 ;
Eurref = 300 MPa ; E50ref = 100 MPa ;  ur = 0,2
ref
On notera que les autres paramètres du modèle (en particulier Eoed ) ne sont pas précisés, parce que seule
une version simplifiée du modèle HSM a été implantée dans CESAR pour le moment. Cette version simplifiée ne
prend en compte que le mécanisme frottant, l'implantation du mécanisme contractant restant à faire.
Malgré cette limite, on vérifie que l'utilisation du modèle HSM modifie qualitativement l'allure de la
déformée calculée, par rapport à un calcul fait avec le modèle de Mohr-Coulomb habituel (figure 5). Avec le
modèle de Mohr-Coulomb, le déplacement du fond de fouille est en valeur absolue la principale composante du
déplacement, alors qu'avec le modèle HSM le tassement de la surface du sol et le déplacement horizontal de la
paroi verticale sont d'un ordre de grandeur comparable à celui du soulèvement du fond de fouille.
L'allure du déplacement horizontal de la paroi est également nettement modifiée : avec le modèle de Mohr
Coulomb, la paroi de l'excavation se déplace vers la droite ; ce n'est pas le cas avec le Hardening Soil Model.

152
Figure 5 . Comparaison des déformées avec le modèle HSM (à gauche) et avec le modèle de Mohr-Coulomb classique (à
droite) - isovaleurs du déplacement total.

4 CONCLUSION

Le souci de certains utilisateurs de CESAR-LCPC de mieux représenter le comportement des terrains et de

mieux cerner l'influence des différents aspects d'un modèle de comportement sur les résultats des simulations
numériques a conduit à proposer une nouvelle structure pour les lois de comportement utilisables en
élastoplasticité. Elle permet à l'utilisateur de combiner différents éléments pour définir une loi de comportement
adaptée à son problème.
On peut en particulier tester différents types de lois élastiques (linéaires ou non, isotropes ou non). On a
introduit un certain nombre de critères nouveaux pour prendre en compte l'anisotropie des caractéristiques de
résistance des géomatériaux, ou pour obtenir une représentation plus fidèle du comportement en déchargement.
La nouvelle programmation permet également désormais d'effectuer des calculs en condition non drainée (en
prenant en compte le comportement formulé en contraintes effectives).
La multiplicité des modèles de comportement disponibles permet de tester l'influence de certains paramètres
sur les résultats des simulations numériques, mais elle ne résout pas la question de la détermination des
paramètres des modèles ; elle ne permet évidemment pas de s'affranchir, pour un projet donné, d'une campagne
de reconnaissance approfondie. L'utilisation de modèles plus avancés que le modèle de Mohr-Coulomb aurait
même plutôt tendance à exiger davantage d'essais.
Enfin, l'approche est pour le moment limitée aux modèles élastoplastiques : elle pourrait être étendue, par
exemple, pour prendre en compte d'autres phénomènes, comme le fluage ou l'endommagement.

REFERENCES

Bourgeois E, Coquillay S, Mestat Ph. 2005. Exemples d’utilisation d’un modèle élastoplastique avec élasticité non linéaire
pour la modélisation d’ouvrages géotechniques. Bulletin des Laboratoires des Ponts et Chaussées 256-257, 67-84.
IFSTTAR. 2014. CESAR-LCPC version 6 – Reference manual of the solver, IFSTTAR, France.
Coquillay S. 2005. Prise en compte de la non linéarité du comportement des sols soumis à de petites déformations pour le
calcul des ouvrages géotechniques, Phd Thesis, ENPC, France.
de Buhan P, Sudret B. 1999. A two-phase elastoplastic model for unidirectionally-reinforced materials and its numerical
implementation. Eur. J. Mech. A/Solids 18(6), 995–1012.
Desai C.S. 1980. A general basis for yield, failure and potential functions in plasticity, Int. J. Numer. Anal. Meth. Geomech
4, 361-375.
Fahey M, Carter JP. 1993. A finite element study of the pressuremeter in sand using a nonlinear elastic plastic model.
Canadian Geotechnical Journal 30, 348-362.
Hill R. 1947. A theory of the yielding and plastic flow of anisotropic materials, Proc Royal Soc London A193, 281-297.
Hoek E, Brown ET. 1980. Empirical strength criterion for rock masses. J. Geotechnical Engineering Division ASCE, 1013–
1025.
Lourenço PB, De Borst R, Rots JG. 1997. A plane stress softening plasticity model for orthotropic materials. Int J Numerical
Methods in Engineering 40, 4033-4057.
Mestat P. 1993. Lois de comportement des géomatériaux et modélisation par la méthode des éléments finis, Série
géotechnique GT52, ERLPC, France.
Mestat P, Berthelon JP. 2001. Modélisation par éléments finis des essais sur fondations superficielles à Labenne. Bulletin
des Laboratoires des Ponts et Chaussées 234, 57-78.
Plaxis. 2014. Material Models Manual.
Potts D, Zdravkovic L. 1999. Finite element analysis in geotechnical engineering - Theory. Thomas Telford Publishing.
Schanz T, Vermeer PA, Bonnier PG. 1999. The hardening soil model: formulation and verification, Beyond 2000 in
Computational Geotechnics - 10 years of Plaxis.

153
154
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Hydraulic Properties of South African clayey soils as Landfill barriers

Les propriétés hydrauliques des sols argileux d'Afrique du Sud que les obstacles à dechets

F. N. Okonta*1, L. J. Moatlhodi1, O.O. Ojuri 2 and P. Meruti1

1
University of Johannesburg, Department of Civil Engineering Science, Johannesburg, South Africa
2
Federal University of Technology Akure, Akure, Nigeria
* fnokonta@[Link]; Tel: +27 11 559 2318; Fax: +27 11 559 2342.

ABSTRACT: Series of tests on leachate flow through a composite liner with defected geomembrane was
conducted in a fabricated assembly with consolidation and percolation chambers. The composite liner consisted
of 24 mm soil barrier layer, a plastic with 5 mm hole to simulate defect in geomembrane, and 225 mm
attenuation layer of soils. The diameter of both the upper barrier layer and lower attenuation layer was 15.5 cm.
The barrier layer was a mixture of clay soil and 50% coarse sand replacement and the attenuation layer was a
mixture of gravely soil and 50 % coarse sand replacement. Saline water prepared by mixing 10 g of salt per 1L
of deionized water was used to simulate leachate. The Modular Consolidometer-Percolation Column Hybrid was
seated in a 16 metric-tons compression loading frame. Flow rates with and without geomembrane were measured
and compared with Darcy’s and Giroud’s flow equations for imperfect poor contact condition. Flow rates
through the geocomposite without geomemebrane was a thousand times higher than the predicted results and
flow rates through geocomposite with defected geomembrane were also lower than the predicted flow rate using
Giroud equation. Changes in the ions concentration of attenuation layer indicated that, reconstituted natural soils
are capable of attenuating chemicals to some degree and are thus partial candidate for liners.

RESUME: Série des essais sur les flux de lixiviats par un revêtement composite géomembrane défection a été
menée dans un ensemble fabriqué avec consolidation et percolation chambres. Le revêtement composite est
composée de 24 mm de couche barrière dans le sol, d'une matière plastique avec un trou de 5 mm pour simuler
défaut dans géomembrane et 225 mm couche d'atténuation des sols. Le diamètre à la fois de la couche barrière et
la couche supérieure d'atténuation inférieur était de 15,5 cm. La couche barrière était un mélange de sol argileux
et de 50% de sable grossier et la couche d'atténuation était un mélange de sol graveleux et 50% de sable grossier.
L'eau saline préparée en mélangeant 10 g de sel par 1 litre d'eau désionisée a été utilisée pour simuler les
lixiviats. La colonne hybride modulaire consolidomètre-percolation était placée dans un cadre de chargement 16
tonnes métriques. Les débits avec et sans géomembrane ont été mesurés et comparés avec les équations
d'écoulement de Darcy et Giroud pour imparfaite condition de contact pauvres. Les débits à travers le
géocomposite sans geomemebrane étaient mille fois plus élevés que les résultats prévus et les taux d'écoulement
à travers géocomposite avec géomembrane défection étaient également plus faibles que le débit prédit en
utilisant l'équation Giroud. Les variations de la concentration en ions de la couche d'atténuation ont indiqué que,
sols naturels reconstitués sont capables d'atténuer dans une certaine mesure des produits chimiques et sont donc
candidat partielle pour doublures.

1. INTRODUCTION

The challenge facing engineers is to provide a long-term environmental protection against landfill leachate. This
can be overcome by constructing a barrier system and a leachate collection system below the waste body. Barrier
system will normally comprise of composite liner as one of its components. A liner system will typically be
comprised of a protection layer overlying a composite liner. Composite liner will typically be comprised of a
Geomembrane (GM: 1.2 – 2.0 mm thick high-density polyethylene), a Geosythethic Clay Liner (GCL: 5-10 mm
thick bentonite clay encased between two Geotextiles), and a Compacted Clay Liner (CCL: 600 – 1200 mm
thick) [1]. However, liner systems, normally comprise of series of components that can be arranged in a variety
of ways to provide the required protection of ground water. In practice, it is generally not possible to construct
Geomembrane with no holes [1]. Hence it is important to consider incorporating clay liner to underlie
Geomembrane in a composite liner system. [1] gave an indication that seepage through a GM-GCL composite
liner is low that in GM-CCL composite liner. Landfilling is an inevitable waste handling option in all the solid
waste management scenarios. This is evident because about 95% of wastes generated in South Africa are
disposed of in landfills and for the world the figure is estimated at around 85% [2]. Even though landfilling is the
waste handling option that is almost always available for many types of waste, there are still challenges that are
associated with it.

155
These challenges include the availability of land, economic implication of landfilling, environmental impact
and others [2]. Because leachate have the potential of being toxic, it is recommended that all B+ sites (landfill
sites that generate significant leachate), must have composite liners constructed as part of leachate management
[2]. It is generally required that substantial composite liner and leachate management be provided for all
hazardous landfill and lagoon irrespective of the climate water balance. These composite liners usually comprise
of GM and GCL, GM and CCL, or the combination of GM, GCL and CCL. Some researchers [3] stated that
these combinations of liner system are accepted worldwide. However, liner systems comprise of series of
components that can be arranged in a variety of ways to provide the required protection of ground water.

2. MATERIALS AND METHODS

The laboratory tests were performed to assess the primary factors that affect the hydraulic conductivity of CCL,
the leachate flow through composite liner due to geomembrane defects, and the attenuation of leachate
contaminants by natural soils. Local clayey soils, local granular soil for Attenuation Layer, Geomembrane
(HDPE), and Saline water (to simulate leachate) are the list of proposed materials used in this study. Five local
clay soils as well as local gravely soil samples were collected from different places in Johannesburg South
Africa.
2.1 Percolation Test: Using Modular Consolidometer-Percolation Column Hybrid

In practice, defects in geomembrane cannot be completely avoided. Therefore to ensure that important ground
water regime is protected against impacts from landfill leachate, attenuation layer may be installed as part of
composite liner system to underlie compacted clay liner. The aim of this test was to measure the flow rate
through geocomposite liners with geomembrane that has a defect and to investigate the chemical attenuation
capacity of natural soils. A device called ‘Modular Consolidometer-Percolation Column Hybrid’ depicted below
was used to perform this test. Only one clay soil was investigated as CCL in composite liner system. Red clay
from Marie Louise Landfill was used in performing this test. Saline water prepared from mixing 10 g of salt per
1 L of de-ionized water was used to simulate leachate. Since that this project was not intended to investigate the
effect of geomembrane material, polyethylene plastic with a 5 mm hole was used to simulate geomembrane with
defect in landfills . Fig 1 shows the image of the chamber and provides an idea of the experimental set up of the
chamber. Layer of flat stones were primarily for drainage purpose and, on top of stones layer in the attenuation
chamber, a layer of geotextile was provided to improvise for effluent outlet clogging by soil fines. A layer of
coarse sand was place also for drainage; to ensure that the pressure plate does not block the hole in the plastic.
The only load that was applied to the barrier layer came from the pressure plate self-weight (fig.1) which was
only about 0.769 kPa and the leachate head. A 500 ml cylinder was used to collect effluent and flow rate was
measured after every 500 mL of effluent. Ions concentration and conductivity readings of effluent were also
taken at that point using ion meter.

Fig. 1: Gravely Soil compacted in attenuation; chamber as AL

156
3. RESULTS AND DISCUSSION

Dominate minerals identified for the gravelly soil, sand and red soils are quartz and muscovite, but the
percentage of muscovite in the sand is low, in addition, the percentage of quartz and goethite in the red clay is
also low. Hydraulic conductivity, specific gravity and percolation test were also conducted. Values of hydraulic
conductivity of stratified soil, and estimated flow rate calculated from hydraulic conductivity test results were
also presented. Other soil samples were eliminated at the early stage of the project. Hydraulic conductivity test
was conducted for red clayey soil, red clayey soil mixed with coarse sand and gravely soil mixed with coarse
sand. Red clayey soil and red clayey soil mixed with coarse Sand were prepared at OMC and MDD determined
from compaction test. Gravely soil was prepared at moisture content and compaction efforts as used in the
percolation test as shown in table 1.
From the test results the hydraulic conductivity of a barrier layer-AL system was calculated, and used in the
calculation of flow rate through the system. Even though barrier layer and AL are not in direct contact, an
assumption that leachate that flows through a barrier layer went straight into attenuation was made.

Table 1. Hydraulic conductivity of various soils

Red Clayey Soil Red Clayey Soil Gravely Soil mixed barrier layer-AL
mixed with coarse with Coarse Sand system
Sand
4.25 x 10-4 cm/s

3.1 Percolation Test Results

3.1.1. Without Geomembrane

The following figures present plots of Concentration and Conductivity vs. Elapsed Time and, Flow rate vs.
Elapsed Time for a test conducted without Geomembrane.

Fig. 2: Concentration and Conductivity vs. Time plot Fig. 3: Flow rate vs. Time plot (Without Geomembrane)
(Without Geomembrane)

With similar assumptions as those made for the calculation of hydraulic flow rate through a composite liner-
AL system, flow rate can be calculated for test without a geomembrane using Darcy’s equation. Q = 5.12 x 10-9
m3/s.

3.1.2 With Geomembrane

The following figures present plots of Concentration and Conductivity vs. Elapsed Time and, Flow rate vs.
Elapsed Time for a test conducted with Geomembrane.

157
Fig. 4: Concentration and Conductivity vs. Time plot (With Fig. 5: Flow rate vs. Time plot (With Geomembrane)
Geomembrane)

The predicted flow rate for imperfect poor contact condition as expressed in Eq (1) is Q = 6.22 x 10-8 m3/s
and is based on the general assumption that the tests conditions for this project were similar. The predicted flow
rate through composite liner-AL system was calculated using the following equation.

(1)

Cqo = is a constant = 1.15 for poor contact conditions; hw = leachate head = 25 cm, HL = thickness of the
underlying soil layer = 24.9 cm ; ro = radius of geomembrane defect = 0.25 cm, kL = hydraulic conductivity of
the underlying soil layer = 4.25 x 10-4 cm/s. The equation for predicting flow rate for poor contact condition
( ) presented by [4] could not be used since that the test set-up for this project did not meet
the requirements for using that equation.

i. Concentration and Conductivity of AL pore water

After test with Geomembrane was terminated, soil samples were taken from the top and bottom layers of AL.
Samples were taken from the centre and peripherals of these layer. The tables below present readings from soli –
deionized water solution as well as Concentrations and Conductivity of AL pore water calculated from those
readings. The following equation was used to calculate concentration and conductivity of AL pore water.
(2)
Where:
Ci = Initial concentration/conductivity; Cf = Final concentration / conductivity
Vi = Initial volume; Vf = Final volume

Table 2: Concentration and Conductivity readings for soli – deionized water solution
Concentration (g/L) Conductivity (mS)
Layer Centre Peripheral Centre Peripheral
Top 1.43 1.51 2.87 3.02
Bottom 2.09 2.07 4.18 4.15

Table 3: Concentration and Conductivity of AL pore water

Concentration (g/L) Conductivity (mS)
Layer Centre Peripheral Centre Peripheral
Top 11.45 12.10 22.99 24.19
Bottom 14.84 14.70 29.68 29.47

ii. Percolation Test Data

1. Test with Geomembrane:

Initial (source) Concentration reading = 8.49 g/L
Initial (source) Conductivity reading = 17 mS

158
2. Test without Geomembrane:
Initial (source) Concentration reading = 8.49 g/L
Initial (source) Conductivity reading = 17 mS

iii. Hydraulic Conductivity

Conductivity test was performed on Red Clayey Soil, Red Clayey Soil mixed with coarse sand and, Gravely
Soil mixed with coarse sand. From the test results, Red Clayey Soil has the lowest hydraulic conductivity and
Gravely Soil mixed with coarse sand has the highest hydraulic conductivity as expected. Red clayey soil has a
hydraulic conductivity value of 2.99 x 10-6 cm/s which is higher than the required values of 1 x 10-6 cm/s for
general landfills and 1x 10-7 for hazardous landfill according to [2].

Percolation Test

Without Geomembrane ; The highest measured flow rate through a barrier layer-AL system was 4 x 10-6 m3/s,
which is higher that the predicted flow rate of 5.12 x 10-9 m3/s through the system. This could be due to
inaccuracy in compaction of the barrier layer. This implies that leachate was percolating through a barrier layer
at a higher rate than expected. Reason for the continuous decrease of flow rate could be due to the fact that
attenuation layer was only slightly compacted. So, the soil was collapsing as leachate was flowing through it and
causing a reduction of voids volume. This is evident from the observation during the test.
With Geomembrane; The highest measure flow rate was 5.72 x 10-9 m3/s, but this was lower than the predicted
flow rate of 6.22 x10-8 m3/s through the system. The reason for this could also be due to flaws in compaction.
Other reason could be that, the assumptions made for calculating the hydraulic conductivity through the barrier
layer-AL system were invalid. Flow rate could be decreasing due to clogging of pores of the sample holder
caused by soil fines. It could also be due to swelling of clay as it absorbs water.
Chemical Attenuation by Natural Soil (AL)
Table-3 presents the calculated concentrations and conductivity of soil sample taken from different layers of
AL. It can be seen from the table that all values are higher than that of source leachate. This is an indication that
there was adsorption of ions by AL. The concentrations of the samples taken from the centre were expected to be
higher than those from the peripherals of the AL because the defect was created at the centre. This implies that
there was a lateral flow between the barrier layer and the geomembrane. Since the measured flow rates indicated
that there could be inaccuracy in compaction, it could be that the vertical flow rate on the sides of barrier layer
was higher than that at the centre.

4. CONCLUSIONS

From the tests results the following conclusions can be drawn. Without Geomembrane, the highest measured
flow rate through a barrier layer-AL system was 4 x 10-6 m3/s, which is higher that the predicted flow rate of 5.12
x 10-9 m3/s through the system. The maximum flow rate for test with geomembrane was found to be lower than
the predicted flow rate by Giroud equation. If the assumption that the calculated hydraulic conductivity through
the system was valid a further assumption can be made that, flow rate measured at steady state would have been
closer to the predicted value. From the measurements of concentration and conductivity of the samples taken
from the AL, it was evident that there was a simultaneous downward as well as lateral flow at the geomembrane-
soil liner interface. Concentration and conductivity were found to be higher at the peripherals than in the centre.
This is also an indication that there was flow at the geomembrane-soil liner interface. Percolation tests were
terminated before achieving steady state flow. Calculated concentrations and conductivities of the samples taken
from AL indicate that natural soils can adsorb ions.

5. REFERENCE
[1]. Rowe, R. K. (2011): Systems engineering: the design and operation of municipal solid waste landfills to minimize
contamination of groundwater: Geosynthetics International; September, 18 (6), pp. 319-404.
[2]. Department of Water Affairs and Forestry (2005). Minimum Requirements for Waste Disposal by Landfill, Third
Edition. Retrieved 10 July 2012,
[3]. Rowe, R. K., and Fraser, M. J. [n.d]: Composite liners as Barriers: Critical Considerations. Geotechnical research centre;
London.
[4]. Giroud, J.P., Khatami, A., and Badu-Tweneboah, K. (1989): Evaluation of the rate of leakage through composite liners.
Geotextiles and Geomembranes; 8 (4), Elsevier Science Publishers Ltd, pp. 337 – 340.

159
160
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Effect of gravel fractions on direct shear strength of sand

Effet de fractions de gravier sur la résistance au cisaillement directe de sable

F. N. Okonta*1 and L. J. Moatlhodi1

1
University of Johannesburg, Department of Civil Engineering Science, Johannesburg, South Africa
* fnokonta@[Link]; Tel: +27 11 559 2318; Fax: +27 11 559 2342.

ABSTRACT: Naturally occurring and reconstituted granular soils consisting of a matrix of large and small
particles i.e. quarry sand, backfills and saprolithic sands are encountered very often in construction. 100mm
square shear box was used to determine the shear strength of 0.425mm – 9.5mm granular particles. The results
reveal that for the size of shear box used, shear strength increases with average particles size up to 4.25mm and
erratic for higher particle size. Addition of gravel results in decrease in dry density of the mixture and a decrease
in mobilized strength. The Guth (1945) [1] Model is linear and predicted an increase in shear strength with
gravel content in contrast with the laboratory values. The difference is due to the effect of rough textured gravel
on the on the mobilized strength of the lower dry density matrix.

RÉSUMÉ : les sols granulaires reconstituées d’origine naturelle constituée d'une matrice de particules grandes et
petites à savoir carrière de sable, remblais et les sables saprolithic sont rencontrés très souvent dans la
construction. La boîte de cisaillement de 100mm carré a été utilisée pour déterminer la résistance au cisaillement
de particules granulaires de 0.425mm - 9,5 mm. Les résultats révèlent que pour la taille de la boîte de
cisaillement utilisé, la force de cisaillement augmente avec la taille moyenne des particules jusqu'à 4,25 mm et
erratique de la taille de particules plus élevées. L'addition de gravier se traduit par diminution de la densité sèche
du mélange et une diminution de la résistance mobilisée. Le Guth (1945) [1] Modèle linéaire est prédit une
augmentation de la résistance au cisaillement avec une teneur en gravier à la différence des valeurs de
laboratoire. La différence est due à l'effet de gravier texture rugueuse sur la force mobilisée de la matrice de
densité sèche inférieure.

1. INTRODUCTION

Materials forming part of engineered or reconstituted backfills, glacial tills, mudflows, debris flows, residual and
saprolithic soils, have a distinct structure which consists of a mixture of a base soil matrix (sand, silt, or clay or a
combination of these soils) and large particles of gravel. In order to obtain the shear strength of these mixtures in
the laboratory, large representative samples need to be tested in either the triaxial or the direct shear apparatuses.
These large samples require large direct shear or triaxial cells and large loading systems in order to simulate field
stress conditions. The use of large triaxial or direct shear equipment makes the tests very time consuming and
expensive. Direct shear and triaxial compression tests on granular materials with oversized particles conducted
by ([2]; [3]; [4]; [5]; [6]; and [7]) have indicated that that the shear strength of the mixtures depends upon the
relative concentration by volume (or weight) of the oversized particles in the mixtures. These researchers found
that if the concentration by volume was greater than 45% (>70% by weight), the shear strength of the mixtures
was basically that of the large particles alone. If the concentration by volume of the large particles in the
mixtures was between 45% and 31 % (70% and 48% by weight), the shear strength of the mixtures was partially
controlled by the friction between the large particles. If the concentration by volume of the large particles was
less than 31% (<48% by weight), the shear strength of the mixtures was basically that of the matrix of the smaller
particles alone.
The aim of the study is to determine the shear strength of granular specimens of uniform particle sizes of
0.425mm with different percentages of 9.5mm, gravel in a 100mm x 100mm shear box apparatus. It is thus
intended to obtain the shear strength of larger granular soils from that of the mixtures with dispersed oversized
particles from the strength of the smaller granular matrix which will be easily obtained using conventional
laboratory tests and the concentration by volume of the dispersed oversized particles. If the proposed
methodologies are found to be feasible for soils with non-contiguous or dispersed oversized particles, it will be
highly beneficial to the geotechnical engineering community in terms of time and money saved since large
conventional equipment would not be required to measure the shear strength of soil-rock mixtures since the
property of the soil matrix, can easily be measured with conventional geotechnical equipment and the value of
concentration can be easily estimated.

161
2. MATERIALS AND METHODS
It is also important to examine the stress strain behavior for specimens of gravel particles that are contiguous i.e.
the gravel particles touch each other. In the sand-gravel mixtures the particles of gravel will touch when their
concentration increases in the sand-gravel mixtures. For the determination of the shear strength of granular soils
[1] proposed Equa. 1 in terms of peak stress or stress ratio thus
τc = τm (1 + 2.5 C) (1)
In Eq. (1), τc is the shear strength of a granular material with dispersed oversized particles; τm is the shear
strength of the granular matrix in which the large particles are dispersed, and C is the concentration by volume of
the oversized particles. The square shear box had a width of 100 mm and a depth of 40 mm aspect ratio,
H/L=0.40. The samples were sheared at a constant rate of 0.25 mm/min, which is consistent with the standard
rate for drained tests on sands. Tests were conducted in a water bath with the sample completely submerged in
water to assure that the samples had no cohesion. The thickness of the 100 mm shear box did not meet the [8]
minimum criteria of six times the maximum particle diameter for some specimens. Three set of specimens were
tested (1) T1; 0.425-2.36mm, T2; 2.36-4.75mm, T3; 4.75-9.5mm, T4; 9.5-13.2mm, T5; 13.2-19.0mm, (2) Type
1+2, Type (1+2) + (20%T4), Type 1+2 + (40%T4), Type 1+2+ (70%T4), where Type (1 +2) = 50/50 mix of
0.425-2.36mm and 2.36-4.75mm; Type 4 = 13.2-19.0mm and (3) Interface direct shear tests

3. RESULTS AND DISCUSSION

The results of shear strength parameters of granular soils derived from shear tests in a 100mm x 100mm shear
box was presented in Table 1. The frictional angle increased with particle size up to 4.75 - 9.5mm and remained
constant afterwards. Larger particle mobilize shear stress within a large shear band. The maximum particle size
to specimen thickness ratio did not meet the ASTM criterion and thus restrained particle displacement may have
resulted in limited mobilization of direct shear stress. Also in Table 1, the addition of large gravel particles up to
70% to a sand specimen resulted in a decrease in mobilized strength. Addition of rough textured gravel up to 70
% resulted in reduced dry density and thus a slight decrease in mobilized strength. Very few tests designed to
obtain the shear strength of granular materials with dispersed oversized particles have been conducted to date.
The reason for the lack of test results in these materials was to do with the fact that large representative samples
need to be tested in either the triaxial or the direct shear apparatuses. The results of the effect of particle crushing
on shear strength and dilation characteristics of Sand-Gravel Mixtures by [9] is shown in Table 2. The results
were derived from series of tests on sand containing different percentages of 15mm and 25mm particle size
gravels. The results show that irrespective of the size of the gravel, the shear strength remain constants up to 40%
and increased significantly afterwards, it is thus implied that the 40% is the boundary between the near field and
the far field behaviour.

Fig 1 and Fig 2 illustrates the incompatibility of stress and dilatants deformation in relation to large particle
size. The large granular soils exhibited residual stress state at increasing dilatants deformation. [10] conducted
some conventional direct shear tests using a 65 mm shear box on mixtures of sand and gravel. The sand had an
average diameter of 0.3 mm and the gravel with an angular shape had a diameter equal to 5 mm. The result is
shown in Table 3. Using the values of the shear strength of the sand alone, τm, as well as the concentration by
volume, C, of the gravel in the mixtures, the values of the shear strengths of the mixtures, τc, were evaluated
using the Guth’s model represented by Eq. 1 [1]. The values of the shear strength, τc, of the sand-clay mixture
obtained using the Guth’s model was superimposed on the values obtained by [10]. Fig. 3 shows that the Guth’s
model predicts very well the shear strength of the mixtures. Thus, Eq. 2 seems to predict reasonably well the
shear strength of granular material with oversized particles. In this case the gravel content is limited to 30% and
an icrease in shear stress with gravel content was indicated. Fig 4 showed that the Guth model could not predict
the shear stress of granular soils in a 100mm x 100mm shear box both in the near field region (C = 30%) and the
far field region (C > 30%). There was a slight reduction in mobilized shear stress at gravel contents of 20% -
40%. This may be due to reduced dry density, as was also indicated in Table 1. The major differences between
the test conditions of Fig 3 and Fig 4 are specimen thickness to maximum particle size ratio, the surface texture
of the particle.

162
Table 1. Shear strength parameters of granular soils derived from shear tests in a 100mm x 100mm shear box.
Soil Type Phi Soil Type Phi
0.425 – 2.36 25.4 Type 1 + 2 42.1
2.36 – 4.75 40.8 Type 1 + 2 (20%T4) 42.3
4.75 – 9.5 44.1 Type 1 + 2 (40%T4) 41.2
9.5 – 13.2 38 Type 1+2(70%T4) 41.8

Table 2. Shear results in a 65 mm square shear box on mixtures of sand with average diameter of 0.3 mm and gravel with an
angular shape and diameter of 5 mm [10]

% of Large Normal Stress (158kPa) Normal Stress (237kPa) Normal Stress (316kPa)
Particles
0 106 135 210
6 115 160 220
12 140 165 240
18 160 210 280
32 201 308 340

Fig. 1. Stress deformation curves of Type 1+2

163
Fig 2. Strength Envelope of Mixtures of T1 +T2 and T4

Fig. 3. Prediction of [10] shear stress by [1] model.

Fig. 4: Prediction of direst shear stress by [1] model.

164
4. CONCLUSION

The use of Guth model for the estimation of the shear strength of non-dispersed and dispersed sand-gravel
mixtures is sensitive to the test conditions. Based on the limited tests in literature and tests conducted in a
100mm x 100mm shear box, comparisons of the shear strength predicted by Guth (1945)[1] model and the
values determine with the aid of the 100mm square box revealed that the strength of gap graded sand and gravel
soil matrix is dependent on the size of the shear box and that of the particles tested, (b) the degree of roughness
of the oversized particles, (c) dry density and magnitude and range of applied normal stress. The addition of
rough textured granular materials up to 60 – 70% may result in decrease in dry density and the mobilized
strength cannot be predicted by a linear version of the Guth Model but a modified model.

REFERENCES
[1] Guth, E. (1945). Theory of filler reinforcement. J. of Applied Physics, Vol. 16, pp. 20-25.43
[2] Doddiah, D., Bhat, H.S., Somasekhar, P.V., Sosalegowda, H.B., and Ranganath, K.N., (1969). Shear strength
charactyeristics of soil-gravel mixtures. J. of the Indian Nat. Soc. of Soil Mech. and Found. Eng., Vol. 8, No. 1, pp. 57-
66.
[3] Fragaszy, R.J., Su, J., Sidiqqi, F.H., and Ho, C.L. (1992). Modeling strength of sandy gravel. ASCE Journal of
Geotechnical Engineering, Vol. 118(6), 920-935.
[4] Holtz, W.G., and Gibbs, H.S. (1956). Triaxial shear tests on pervious gravelly soils. J. of the Soil Mech. and Found. Div.,
ASCE, Vol. 82 (1), pp. 1-22.
[5] Marsal, R.J., and Fuentes de la Rosa, A. (1976). Mechanical properties of rockfill-soil mixtures. Transactions of the 12th
Int. Congress on Large Dams, Mexico City, Vol. 1, pp. 179-209.
[6] Vallejo, L.E. (2001). Interpretation of the limits in shear strength in binary granular mixtures. Canadian Geotechnical
Journal, 38, 1097-1104
[7] Vasileva, A.A., Mikheev, V.V., and Lobanova, G.L. (1971). How the strength of gravelly soils depend on the type of
state of the sand filling the pores. Soil Mechanics and Foundation Engineering, Vol. 8, No. 3, pp. 167-171.
[8] American Society for Testing and Materials (1990). Standard test methods for direct shear test of soils underconsolidated
drained condition, D-3080-90, ASTM, Philadelphia.
[9] Hamidi, A., Alizadeh, M., Soleimani, S.M., (2009). Effect of Particle Crushing on Shear Strength and Dilation
Characteristics of Sand-Gravel Mixtures International Journal of Civil Engineering. Vol. 7, No. 1,
[10] Yagiz, S. (2001). Brief note on the influence of shape and percentage of gravel on the shear strength of sand and gravel
mixtures. Bulletin of Engineering Geology and the Env. 60 (4), 321-323.

165
166
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Assessment of clays by small-scale suction tests

Évaluation des argiles par des essais de succion à petite échelle

Philip Stott and Elizabeth Theron

Central University of Technology, Bloemfontein, South Africa.

ABSTRACT - The relationship between suction and water content gives crucial information about a soil. The
time and cost of determining the full soil water suction curve make it not feasible for small projects like
economic housing. A considerable range of soil suctions can be determined easily and within a reasonably short
time by using small samples, simple suction control and a high precision balance. It appears that in this way it
may be possible to estimate heave potential and variability of soil properties at reasonable cost in an acceptable
time.

RESUME - La variation de la succion en fonction de la teneur en eau (courbe de rétention) donne des
informations importantes à propos du sol étudié. Toutefois, pour des projets de gamme moyenne, notamment les
maisons économiques, la détermination de la courbe de rétention n’est pas toujours possible en raison du cout et
du temps. L’utilisation d’un dispositif précis de mesure de succion pour des petits échantillons de sols, nous a
permis de contourner ce problème. En effet, il est possible en utilisant ce dispositif de déterminer
raisonnablement le potentiel de soulèvement et les variations des propriétés du sol en termes de cout et de temps
acceptables.

1. INTRODUCTION

The tests here presented aim to address problems with the current commercial methods for foundation indicator
tests. Usually such tests involve procedures primarily intended for road construction materials and they
frequently give poor estimates of heave potential for high clay-content soils. Preoccupation with economy points
to a need for simple tests with minimum skilled labour content, minimum opportunity for short-cuts and
straightforward interpretation of results. Tests should directly relate to heave potential, rather than general
properties. Suction potential gives an indication of how readily a soil can draw in water. Change in water content
is the cause of volume change. Suction potential should be a good indicator of swell potential.

2 INDICATION OF SHRINK/SWELL POTENTIAL

Kassa (2005) found that Atterberg limits are not a reliable indicator of volume change under load. Sridharan and
Prakash (2000) found that Atterberg limits can sometimes indicate heave potential significantly higher or
significantly lower than reality. Atterberg limits remain, however, the most popular indicators of heave potential.
The Linear Shrinkage test performed on raw soil samples and allowed to air-dry slowly, gives a graphic
indication of the extent to which a soil may change volume with water content, but does not indicate the likely
pressures involved in this change of volume.
Clays expand when they draw in water. The force which can be exerted in this expansion depends on the
strength with which they can suck in water. The suction of clays has long been recognized as a cardinal indicator
of heave potential, but it is one of the least convenient indicators to measure. The plot of suction from saturation
to desiccation is known as the soil water characteristic curve (swcc). The swcc is an indispensable part of a full
unsaturated analysis, but it requires much time and skill to measure.

3. RECENT ADVANCES AND CURRENT PRACTICE IN SUCTION MEASUREMENT

Three papers presented at the International conference on Soil Mechanics and Geotechnical Engineering in Paris
2013 deal with advances in techniques of suction measurement. Two other papers deal with measuring suction in
practical engineering projects.
Advances in dew-point potentiometer technology point to extending the range of suctions measurable by this
type of instrument (Macek et al 2013). The use of micro-porous membranes may allow quicker suction

167
measurement in the 0-30 kPa suction range (Nishimura 2013). The use of a centrifuge may speed up
measurement in the 0 to 900 kPa range (Reis et al 2013).
But in practical use, for assessing lime treatment on London Clay (Mavroulidou et al. 2013) and for modeling
the impact of climate changes on embankments and cuttings (Mendes and Toll 2013), Whatman No.42 paper
remained the method of choice. This time-honoured system takes two to six weeks for a suction measurement
and requires careful laboratory technique.

4. SMALL SCALE SUCTION POTENTIAL MEASUREMENT

A more limited indication of suction potential can be found by allowing samples to reach equilibrium at known
temperature and humidity. Soil samples can be brought to equilibrium with saturated solutions of various salts
(Blight 2013). Blight’s tests took typically 90 days. Although the equipment was cheap and unsophisticated, and
the skilled labour component not large, the time frame is not feasible for normal engineering practice. The
authors have modified the procedure to give quicker results, making enquiry into some important questions
feasible.
The procedure involves using small pieces broken from a soil specimen. Breakage usually occurs along planes
of existing weakness; the micro-structure and fabric of the soil are not greatly disturbed. Samples are placed in
small glass weighing bottles with ground-in lids. When firmly closed, little air or water vapour leaves or enters
in the time taken for weighing. Samples are weighed on an analytical balance and then placed with lids open in a
container at controlled temperature and humidity. Equilibrium moisture content is a measure of the suction
potential of the soil under those conditions. Samples can be weighed periodically by closing the ground-glass lid
and removing from the controlled atmosphere to the analytical balance. For most of the tests performed in this
investigation suctions corresponding to saturated solutions of KCl and NaCl were used, though a number of tests
used both higher and lower suctions.

Figure1 shows curves of moisture content against time at constant suction for samples of 30 clayey soils from
housing developments in central South Africa. The masses of the samples range between 25g and 35g. Starting
water contents were arbitrary. The equilibrium moisture contents at 22MPa suction (corresponding to a saturated
solution of KCl at 200C) vary from 3.6% to 12.6%. There appears to be a correspondence between moisture
content and possible heave potential. Those with the highest water contents have Plasticity Index close to 40,
which would normally be considered an indication of high heave potential. Those with the lowest water content
have PI below 15, normally taken as an indication of low heave potential.

Figure 1 30 samples at 22MPa suction

These same samples were used to probe questions concerning differences of behaviour in wetting and drying,
and the possibility of permanent change in properties due to oven drying. Figures 2 to 4 show these 30

168
samples being tracked from equilibrium moisture content at 22MPa, through oven drying and re-wetting by
absorption of water from the air (Figure 2), wetting by the addition of a small amount of die-ionised water and
re-equilibrating at 22 MPa (Figure 3), saturation with de-ionised water and re-equilibrating at 22 MPa (Figure 4).

Figure 2. 30 samples after oven drying

Figure 3. 30 samples after wetting

Figure 4. 30 samples after saturation

169
From initial moisture content the low-suction samples reached equilibrium moisture content after about 4
days, the high-suction samples reached constant water content after about 9 days. After oven drying and
returning to 22MPa suction conditions all samples reached equilibrium after about 5 days. In most cases the
equilibrium water content was significantly lower than that reached before oven drying. After wetting to water
contents close to or somewhat higher than the original values and re-stabilising at 22MPa almost all of the
samples reached moisture contents higher than the value after oven drying, but substantially less than the original
value.

After saturation most of the samples reached equilibrium at slightly higher moisture content, but again lower than
that reached before oven drying.
Figure 5 shows moisture content changes as a bar chart. Samples whose initial water content was very low are
grouped predominantly on the left side.

Figure 5. Water content reached by 30 samples at 22MPa from initial moisture content after oven drying, wetting and
saturation

Where the initial water content of the soil was well above its 22MPa value, oven drying significantly reduced
the soil’s subsequent suction potential, and even saturation did not restore the original suction potential. Where
the initial water content was well below the 22MPa value, the soil was not so greatly affected by oven drying and
saturation could lead to higher retained moisture content at 22MPa. This suggests that Blight’s contention (Blight
2012) that air drying can lead to permanent changes in some residual soils may commonly apply to many clayey
soils. Alternatively it may be a consequence of the well-known hysteresis property of clays and a very high
degree of saturation may be needed to regain the original drying curve.
This test also suggests that in the majority of cases, whatever suction changes take place, the ratios of water
content between the various samples remain substantially the same.
The possibility of consistent relationship over a range of suctions was tested with 20 samples which were
brought to equilibrium at 37 MPa then 10 MPa and then 2 MPa. Figure 6 shows that almost all of the samples
maintained a consistent relationship throughout this range of suctions. They also maintain this relationship after
oven drying. This suggests that a single determination at one suction may allow estimation of suction potential
over a considerable range of suctions.

170
Figure 6. 20 samples : equilibration of water content at 37 MPa, 10 MPa, 2 MPa and at 22 MPa after oven drying

4.1. Reducing the time of testing.

It could be expected that smaller samples would take less time to reach equilibrium. Figure 7 shows plots of
moisture content against time for seven different-sized small samples of the same clay (PI = 40). The sample
masses vary from 0.71g to 8.4g. All seven samples reached equilibrium water content in approximately one day.
The equilibrium value reached by the 7 samples averaged 13.56% with standard deviation of 0.30%. The
correlation coefficient with sample-size was 0.17, suggesting that there is no significant correlation between
sample size and test outcome for this range of sample masses.

Figure 7. Equilibration of 7 samples (8g to slightly less than 1g) at 22 MPa suction.

5 ASSESSMENT OF VARIABILITY OF SOIL PROPERTIES

Soil is highly variable, and testing procedures usually lay stress on the need to take a sample of adequate size
from which a representative portion is obtained by systematic mixing and division. The intent is to give
characteristic median values. Reliability analysis is an increasingly accepted technique for accounting for
uncertainties in various aspects of geotechnical analysis (Phoon 2008). Reliability analysis needs an estimate of
standard deviation as well as a median value of soil properties. The procedure described above allows a direct
assessment of the variability of a soil by using a number of different small samples. The difference in time
involved in testing one, five, or even twenty samples is small, since the samples are just weighed and put into a
container where they are left to reach equilibrium.
Preparation time is minimal - specimens are simply selected from any preferred location in the sample. There is
little scope for short-cuts in preparation. The soil micro-structure remains substantially intact. For samples
smaller than about 10g, a time of only one or two days may be needed to reach equilibrium when suction is 20
MPa or more. The weighing procedure is quick compared to the time taken to reach equilibrium - each weighing
takes only a few seconds.

171
Table 1. Mean and standard deviation for 5 samples of 6 different soils

Average w% at 22 MPa Standard Deviation

S1 Light brown clayey sand 7.12 0.23
S2 Dark brown clay 11.92 0.29
S3 Black clay 12.4 0.18
S4 Brown clay 13.86 0.32
S5 Light yellow silty clay 6.06 1.22
S6 Light brown sandy clay 11.09 1.76

Table 1 shows results for five specimens each of 6 widely different soils. The first four show only small
standard deviation, suggesting that variability is small and a single test result would probably give a good
representation of the properties of the soil. The results for the fifth and sixth samples show substantial standard
deviation. Sample S5 is a kaolinitic clay presenting little problem for heave. Sample 6 was taken from a subsidy
housing project site where results of foundation indicator tests suggested low heave potential but significant
heave was in fact, encountered.
Ten more samples of the soil S6 were tested. The results for the two tests were:
7.98, 12.22, 11.67, 12.02, 11.58 Average 11.09, [Link]. 1.76
11.63, 11.72, 11.37, 11.23, 11.98, 8.11, 11.50, 11.76, 11.13, 11.37 Average 11.13, [Link]. 1.11

Most of these results give equilibrium water content at 22 MPa close to 12%, indicating the probability of
medium to high heave potential. There are, however, two very different results, one in each set. Both are close to
8%, indicating far lower heave potential.
These results suggest that in a very restricted area there is the possibility of major variation in properties for
some soils. They suggest the possibility that a single laboratory determination of heave potential may give a
misleading value and this could be the cause of the damage which occurred in this case.

6. CONCLUSIONS
The testing technique described here, using very small samples subjected to easily-controlled suctions, may
provide a more convenient means of assessing probable heave potential than current commercial methods. Very
little sample preparation is involved and there is little scope for short-cuts. A large enough number of test
samples can be used to gain insight into the variability of the soil concerned without greatly increasing the testing
time or labour cost.

REFERENCES
Blight G.E. (2012) Microstructure, mineralogy and classification of residual soils. Mechanics of Residual Soils. Blight G.E.
and Leong. E.C (Eds.), CRC Press, London, New York, Leiden. 2012, pp. 50-54.
Blight G.E. (2013) Unsaturated Soil Mechanics in Geotechnical Practice. CRC Press, Boca Raton, London, New York,
Leiden. 2013.
Kassa M. Relationship between Consolidation and Swelling Characteristics of Expansive Soils of Addis Ababa. [Link].
Thesis Addis Ababa University 2005.
Macek M., Smolar J., Petkvsek A.(2013) Extension of measurement range of dew-point potentiometer and evaporation
method. Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013. pp.
1137-1142.
Mavroulidou M., Zhang X., Kichou Z. and Gunn M.J. (2013) Hydromechanical properties of lime-treated London Clay.
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013. pp. 1151-
1154.
Phoon K.K. (2008) Reliability-based design in geotechnical engineering. Taylor and Francis Abingdon and New York.
2008.
Reis R.M., Soboya F., Tibana S., Mardiano C.R., Ribeiro A.B. Sterck W.N. Avanzi E.D. (2013) Determination of soil-
water retention curve for a young residual soil using a small centrifuge. Proceedings of the 18th International Conference
on Soil Mechanics and Geotechnical Engineering, Paris 2013. pp. 1175-1178.
Nishimura T. (2013) Application of micro-porous membrane technology for measurement of soil-water characteristic
curve. Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013. pp.
1171-1174.
Mendes J. and Toll D.G. (2013) Influence of initial water content on the water retention behaviour of a sandy clay soil.
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013. pp. 1137-
1142.

172
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Shortcomings in current methods of identification and assessment of expansive

clays

Insuffisances dans les méthodes actuelles d'identification et d'évaluation des argiles

gonflantes

Philip Stott and Elizabeth Theron

Central University of Technology, Bloemfontein, South Africa

ABSTRACT - The problem of damage to buildings caused by heaving foundations remains a challenge in
Africa. Recent developments in unsaturated soil mechanics have led to theoretically convincing tools for dealing
with most unsaturated soils problems, but the parameters required to apply these methods involve tests so lengthy
and expensive that they are not feasible at this stage for economic housing projects. Widely used empirical
methods rely on simple, economical tests, but procedures usually involved in these tests are primarily designed
for road-building rather than housing foundation purposes. There is a need for simple and economic procedures
tailored to the specific needs of light structure foundations.

RESUME– Les dégâts des bâtiments occasionnés par le soulèvement des fondations demeure un défi à résoudre
en Afrique. Les développements récents dans les sols non saturés ont proposé des démarches théoriquement
convaincantes pour résoudre les problèmes rencontrés dans ce contexte. Néanmoins, les paramètres nécessaires
pour appliquer ces démarches nécessitent une expérimentation couteuse de longue durée et ne pouvant pas être
efficace dans les projets de bâtiments économiques. Dans cet article, on propose une procédure simple et non
couteuse visant à résoudre la problématique des bâtiments économiques reposant sur des massifs d’argile non
saturés.

1 INTRODUCTION

The problem of providing economic housing for lower-income communities is widespread throughout Africa. In
South Africa the government is attempting to provide small subsidised houses for the very poor, while the
economically emerging see building a home for themselves as a high priority. Much of Africa has semi-arid and
sub-humid conditions which lead to generally shallow residual soils subject to seasonal alternation of desiccation
and saturation. Such conditions are known throughout the world for giving expansive foundation problems. The
necessity of reserving easily worked, fertile soils for farming, and of avoiding problem soils for essential and
expensive services like roads, and airports may result in problematic soils being allocated to housing by default.
In the 1960s and 70s a great deal of work was done towards understanding and designing for heaving clays.
Much of the development in this field was funded by large mining companies seeking solutions to problems with
the housing they were providing for their work force. Economic housing does not usually have the financial
backing of such companies today.
At the 7th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Pidgeon (1980)
noted 14 methods of foundation design from America, Australia and Africa and presented a comparison of the
five most promising among them. Since that time great strides have been made in the whole discipline of
unsaturated soil mechanics, of which expansive soils have traditionally featured as a major part. It would be
expected that in the 35 years since that time the problems associated with foundations on heaving clay would
have been solved. Lawson (2006) makes it clear that this is not the case. He observed that even among
experience senior engineers who spend most of their time working on expansive soil problems the only design
solution which is considered reliable is removal of the expansive soil and replacement with inert fill.
The situation was even more clearly highlighted by D.G. Fredlund, perhaps the world’s leading authority on
unsaturated soil mechanics, who noted (Fredlund 2009); “I started in unsaturated soil mechanics by looking at
swelling clay problems, where if you get involved you usually lose money, it is not a good problem.” He then
suggested that geotechnical engineers turn to more profitable problems like environmental protection. This
suggests that the problem of heaving foundations is not even close to being solved.
It is therefore not surprising that many economic houses are still being destroyed by expansive clays. The
authors have visited one housing project where almost 30 percent of the houses had to be demolished after
becoming unserviceable due to heaving foundations. At another project a house suffered such serious heave that

173
it lost its structural integrity before construction was finished. It had to be demolished not many days after the
first brick was laid.
Although this may not be financially a good problem to get involved with, it would not be ethical to simply
turn our attention to more profitable geotechnical problems. There is a need to find a solution which can lead to
acceptable foundations in at least a large majority of cases. While the authors’ experience in this is primarily in
South Africa, their findings may be relevant to other countries since similar conditions are quite widespread in
Africa.

2 SHORTCOMINGS IN AVAILABLE METHODS FOR DEALING WITH THE PROBLEM

The most advanced methods available are mathematical. One need only specify the governing partial differential
equations, the boundary conditions of the system, the relevant parameters for the equations and readily available
software will deliver a solution. Unfortunately, as Fredlund has pointed out, (Fredlund 2000) the soil water
characteristic curve (swcc) is the pivotal input parameter, for which he indicated a typical cost of US $10000
(Fredlund 2009). While approximations can be made to the swcc for preliminary design, Blight (2013) noted that
if it is needed at all, then it must be measured, since estimated values can only serve as very preliminary guides.
For some of the seriously expansive clays tested by the authors, more than six weeks in the pressure plate
apparatus was needed to measure just the first (0-900kPa) section of the swcc. Such a test is not feasible for
small housing projects from both cost and time perspectives. Other parameters besides the swcc must also be
determined. They also require considerable time and expense.
Alternatives to a fully mathematical unsaturated soil mechanics analysis are considered in Nelson and Miller’s
standard text on expansive soils (Nelson and Miller 1992), which is widely considered the most comprehensive
and authoritative guide available for expansive foundation problems. Nelson and Miller consider three general
approaches to heave prediction: the use of the oedometer, the use of shrinkage / suction curves, and the use of
empirical methods.
Several procedures for estimating heave potential by means of the oedometer have been devised. But there are
well known shortcomings even for relatively easily-handled soils. It is difficult to get undisturbed samples to
completely fill the odometer ring and imperfections in the cutting of samples lead to spurious results. Tests by
the authors have shown that expansive clays may take up to two months to stop expanding after inundation under
nominal loading. A meaningful oedometer test on such clays is not feasible for a low-cost housing project.
Nelson and Miller‘s examination of the odometer is lengthy, but the conclusion is that it is not the best tool for
solving the heave problem.
Nelson and Miller indicate that the CLOD test, which entails direct measurement of the shrinkage curve, may
be the best method available. Unfortunately it involves coating samples with Saran resin, which is not widely
available in Africa; it is dangerous from a health point-of view, and involves tests spanning six weeks or more.
Less toxic substitutes have been proposed to replace the Saran resin, but the time for the test remains prohibitive.
It has not been widely adopted in the design of foundations in Africa.
Alternative methods for determining the shrinkage curve have been tried. In Morocco Bensallam et al (2012)
devised a method using direct measurement of un-coated soils. No indication was given of the total time for a
test, but they did indicate six hourly intervals between readings and a final stage of 3 days oven drying of the
whole apparatus. This suggests a possible time-frame of the order one or two weeks. This may be more feasible
than the clod test, but the method is skills, labour and equipment intensive, it does not appear to have become
popular and does not seem likely to be adopted by commercial laboratories.
It is therefore not surprising that the procedures widely used in practice depend on empirical methods rather
than on direct measurement of heave-specific parameters. This approach bases estimates of heave potential on
simple, quick, economic tests, particularly the Atterberg limits and particle size analysis. The most popular
method in South Africa, and the only one dealt with in detail in Nelson and Miller’s summary of empirical
methods, is that devised by Van der Merwe in 1963. Two improvements have been proposed - by Williams in
1976 and by Savage in 2007 - but the method remains basically the same. After fifty years of major technological
progress no new method appears to have been devised which has demonstrated its superiority to the point of
convincing a majority of engineers to use it. Paige Green (2004) noted that it is “somewhat worrying” that little
development has taken place in this regard since the 1980s. Despite half a century of experience with these
empirical methods there are still large numbers of houses suffering heave damage, becoming structurally
unsound within a small fraction of their design lifespan.
Investigations by the authors suggest that a major contributing factor may be that the procedures for
determining the basic parameters are seriously flawed. The soils testing procedures currently in use in
commercial engineering materials testing laboratories are primarily designed for road construction. Road
construction is concerned primarily with coarse-grained material; clay is usually rejected or is given stabilization

174
treatment of some kind. Sample preparation and testing procedures which are suitable for granular soils are used
for all samples, even for those not intended for road construction.

3 CURRENTLY USED PARAMETERS FOR FOUNDATION DESIGN

The parameters most frequently used currently in empirical methods of heave estimation are the Liquid Limit
(LL), the Plastic Limit (PL), the Plasticity Index (PI [= LL-PL]) the Bar Linear Shrinkage (LS) and the
percentage clay fraction. The present investigation has found the current procedures for preparation of materials
and laboratory testing to be usually adequate for assessing typical road construction materials, but very
inadequate for assessment of heave potential in clays.

3.1 Liquid Limit, Plastic Limit and Plasticity Index

Sample preparation involves oven drying of soils, crushing, sieving and testing of the fines fraction (the fraction
passing the 425 micron sieve). The LL is found using a one point determination with a mixing time, from oven
dry to testing in the Casagrande cup of 10 minutes. Material remaining is used to fill the linear shrinkage trough
and to perform the PL test. Moisture contents are determined by oven drying at 1050C to 1100C until constant
mass. The PI of the fines fraction is determined by subtracting the PL from the LL. This value is reduced by
multiplying by the fines fraction to give the gross PI of the whole sample.
Road construction involves re-working material, which is commonly transported and re-compacted in a new
position. Current sample preparation methods may be a reasonable representation of such conditions. Economic
housing does not allow for large-scale disturbance of the soil under foundations. The only disturbed and
transported material is normally a layer of selected fill under floor slabs.
There have been warnings (Blight 2012), that inappropriate preparation procedures may significantly reduce
PI, particularly for the residual clays which are common in semi-arid and sub-humid regions in Africa. Results of
tests prepared using the common standard procedures may therefore be relevant for road construction purposes,
but not for foundations of houses.
Several samples of clays from a number of different subsidy housing sites were tested using “raw”, un-dried,
un-sieved samples to which the only treatment was removal of visible large particles and the addition of
sufficient water to enable a LL determination. In each case the “raw” PI for the whole sample was significantly
greater than the PI of the fines alone from the corresponding sample prepared using the standard procedure.
Further tests showed that when dried fines are allowed to absorb water for longer than the allotted time for the
test, the measured PI increases significantly, initially at a rapid rate. Since economic houses are almost always
founded on natural ground with minimal surface preparation, and since they are subject to wet-season rains
which may last for days, rather than minutes, it seems likely that the PI given after minimum preparation on an
un-dried “raw” sample is likely to be a better indication of the heave that will actually occur than that given by
the standard laboratory procedure.

3.2 Linear Shrinkage

The linear shrinkage test currently in use in South African laboratories uses material left over from Liquid Limit
determination. The shrinkage trough is filled and immediately put in the oven at 1050C-1100C. This appears to be
satisfactory for the granular materials common in road building. Clay soils, however, tend to arch and shatter
when dried rapidly, making shrinkage measurement very inaccurate. The same soil left to air-dry usually arches
and shatters to a far smaller degree; the linear shrinkage is often about double that of the oven-dried sample.
Figure 1 shows eight linear shrinkage troughs all filled from one sample of raw, un-dried, un-sieved clay (LL63,
PI40) which had given closure after 22 taps in the Casagrande cup. Four were put into the drying oven at 1050C-
1100C, the remaining four were left to air dry for five days. The oven dried specimens are so shattered that the
linear shrinkage values (4.39, 12.04, 10.17, 7.16) are meaningless. The air-dried values (15.14, 15.78, 15.05,
14.79) are all within 0.6 of the average value, (15.19 ) and give a consistent indication of high heave potential.

175
Figure 1. Linear Shrinkage of a plastic clay. Above – air dried. Below – oven dried.

Clays under houses experience changes in moisture content slowly over a long period. Oven drying gives a very
poor indication of the shrinkage/swell likely to occur in practice. A far more realistic indication would be given
by a sample consisting of the entire soil allowed to air-dry at normal temperature. Current test methods for linear
shrinkage systematically underestimate the heave potential of expansive soils.

3.3 Clay Fraction

Clay fraction is usually determined by hydrometer analysis. The unsatisfactory nature of the hydrometer has been
recognized for some time. Savage (2007) noted that its unreliability could be due to the fact that clay particles
may be plate-like and not spherical, as assumed in the analysis, and also that clay particles may be carried down
rapidly by larger particles. He suggested that estimation of clay fraction by Skempton’s activity formula may be
more reliable. The authors have found that there are often very significant differences between clay fraction
estimated by this method and the hydrometer results given by commercial laboratories.
Precipitation tests have been performed on a number of clay samples from economic housing sites. Soil was
allowed to precipitate in an apparatus with dimensions similar to a hydrometer. Significant clay content has been
found in all but the coarsest fraction precipitated. Estimates based on these observations suggest that the
hydrometer test may significantly underestimate clay content in some soils, which in turn may lead to significant
underestimation of heave potential.

4 GENERAL CONSIDERATIONS OF THE STANDARD METHODS

The only one of the commonly performed simple tests with the potential to be a direct indication of shrink/swell
potential is the linear shrinkage test. When performed on whole samples and allowed to air dry there is a
possibility that it may give a direct indication of the behaviour of the in-situ material under a foundation. The
Atterberg limits and clay fraction may often give a useful indirect indication of shrink/swell potential, but they do
not give a direct value. Sridharan and Prakash (2000) have shown that Atterberg limits can give values
suggesting very low heave potential for some expansive soils and indications of expansiveness in some soils
which actually have low heave potential. If the laboratory techniques currently employed do not give reliable
results for the heave potential of clays, then it is not surprising that failure of foundations for economic housing
schemes remains at an unacceptably high level.

5 CONCLUSIONS

The design of foundations for small buildings on expansive soils is theoretically a very difficult problem, it is far
from solved, and a theoretically-rigorous approach is impractical at this stage due to the cost and time required
for determining the necessary parameters. Removal of a significant amount of expansive material and
replacement by inert fill is also not feasible for most small structures, since the cost may be greater than the
budget will allow. There seems to be little alternative but to use empirical methods of design based on
parameters which are quick and economical to measure. These parameters should be directly relevant to

176
expansiveness of soils and should be determined using procedures suitable for economic housing foundations
rather than for road construction materials. Currently used tests do not adequately satisfy these criteria.

REFERENCES

Bensallam S., Bahi L. Ejjaaouani H. Shakhirev V. A New Shrinkage Curve Model, Applied to Moroccan Clayey Soil.
International Journal of Geosciences 2012. July 2012 ([Link]
Blight G.E. (2012) Microstructure, mineralogy and classification of residual soils. Mechanics of Residual Soils. Blight
G.E. and Leong. E.C (Eds.), CRC Press, London, New York, Leiden. 2012, pp. 50-54.
Blight G.E. (2013) Unsaturated Soil Mechanics in Geotechnical Practice. CRC Press Boca Raton, London, New York,
Leiden, 2013. p.48
Fredlund D.G., Rahardjo H. and Fredlund M.D. 2000. Unsaturated Soil Mechanics in Engineering Practice. John Wiley
and Sons. p. 189.
Fredlund D.G., (2009) Paradigm Shifts to Facilitate the Practice of Unsaturated Soil Mechanics. 8th C.W. Lovel Lecture.
Perdue Geotechnical Engineering. 26 October 2009.
Lawson W.D.(2006) A Survey of Geotechnical Practice for Expansive Soils in Texas. 4th International Conference on
Unsaturated Soils, Carefree, Arizona, April 2006. pp 304-314.
Nelson [Link] Miller D.J., (1992) Expansive Soils - Problems and Practice in Foundation and Pavement Engineering.
John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, 1992.
Paige Green P.(2004), The Geology And Engineering Geology Of Roads In South Africa Proceedings of the 23rd
Southern African Transport Conference (SATC 2004) Pretoria, July 2004
Pidgeon T.J. (1980) A Comparison of existing methods for the design of stiffened raft foundations on expansive soils.
Seventh Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Accra, June 1980. pp. 277-289.
Savage P.F. (2007), Evaluation of possible swelling potential of soil. Proceedings of the 26th South African Transport
Conference (SATC 2007), 9-12 July 2007, DTT, Pretoria.
Sridharan A., and Prakash K. (2000) Classification procedures for expansive soils. Proc. Inst. Civil Engineers
Geotechnical Engineering, 143, October 2000. pp 235-240

177
178
CARACTERISATION DES SOLS
SOIL CHARACTERIZATION

179
180
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Utilisation du CPT'u et du pressiomètre pour la qualification de la classe de sol

et pour l'estimation de Gmax.

Use of CPT'u and PMT test for the assessment of soil class and Gmax

J.C. Gress
Professeur du cours Reconnaissance et Traitement des sols et des Roches à l'Ecole Nationale des Travaux
Publics de l'Etat. Conseiller Technique d'Hydrogéotechnique.

RESUME - Les Eurocodes 7 recommandent le panachage des moyens d'investigations pour une bonne
reconnaissance géotechnique de site. La caractérisation du sol par l'identification physique, le pénétromètre
CPT'u et le pressiomètre rentre dans ce cadre, et dans une région sismique permet une approche pertinente de la
qualification de la classe de sol et une bonne détermination de Gmax pour chaque couche.

ABSTRACT - Eurocode 7 recommands a mix of ways of geotechnical surveys in order to have a good site
geotechnical assessment. The soil characterization through soil physical identification, CPT'u tests and PMT
tests corresponds to this frame, and in a seismic region, allows a relevant approach of site soil class and a good
assessment of Gmax, for each layer.

1. INTRODUCTION

L'étude géotechnique de l'interaction sol structure dans une zone sismique commence par la détermination de la
classe de sol, ce qui va permettre de déterminer les actions sismiques sur l'ouvrage. Pour déterminer la classe de
sol, le paramètre pertinent est la Vs30, moyenne harmonique des vitesses de cisaillement mesurées sur une
épaisseur de sol de 30 mètres.
La meilleure façon de déterminer la Vs30 reste la réalisation d'un dispositif CROSS HOLE. Les difficultés les
plus fréquentes associées à ce dispositif sont : la superposition de couches, de vitesses différentes, non
régulièrement croissantes, le remaniement du milieu encaissant à la foration, le non parallélisme des forages, la
blocométrie.
Les Eurocodes 8, proposent également de déterminer la classe de sol à partir de la connaissance des valeurs
de SPT ou des valeur de Cu. Le SPT reste un moyen d'investigations adapté seulement aux sables saturés. La
référence à Cu pour un géotechnicien expérimenté est plus que difficile, la mesure de Cu se révélant délicate elle
aussi pour les sols grossiers et adaptée aux sols argileux saturés.
L'approche proposée ici consiste à utiliser l'essai au pénétromètre, et l'essai au pressiomètre, les deux couplés
à l'identification du sol (niveau de saturation, essai au bleu de méthylène sur la fraction totale du sol ou limite de
liquidité pour un sol 0-4000 µ).
Les deux essais permettent l'estimation de la valeur de Gmax quand le sol est saturé, celle-ci permettant
l'estimation de la Vs de chaque couche et donc de Vs30.

g G max
VS  (1)
h
Vs : en m/s, G : en kN/m²,
g : accélération de la pesanteur = 10 m/s/s
૪h : densité humide en kN/m3
Non seulement, il est alors possible de déterminer la classe de sol d'une manière pertinente, mais il est
possible d'affecter à chaque couche une valeur de Gmax, celle-ci pouvant être par ailleurs, mesurée quand la
nature du sol le permet, par des essais en laboratoire, à la colonne résonnante ou au triaxial cyclique.

2. IDENTIFICATION PREALABLE

Il sera toujours important de procéder pour chaque couche à une identification préalable. La densité sèche ૪d et
la teneur en eau pondérale WSC sont des paramètres importants puisque permettant de quantifier le niveau de

181
saturation de l'échantillon, en gardant à l'esprit que la teneur en eau mesurée ne sera jamais la teneur en eau in
situ pour un échantillon à saturation, mais celle de rétention de l'échantillon, compte tenu des manipulations de
l'échantillon.
Nous rappelons que nous avons proposée pour les sols 0.400µ la relation suivante :

WL
WRET  WSAT (2)
WL  0,05
et nous proposons, pour les sols OD :

 1
W  WSAT (3)
RET 

avec:  = 1 + 0,2 x (1 - % 1mm) + 2 x (% 1mm - % 400µ) + 4 %400µ (1 - %2µ) + 1,26 VBOD (4)

où VBOD est la valeur de bleu de la fraction OD. On rappelle que (5) :

VBOD x 1 = VB0/400 µ x %400µ = VBod x %d (5)

En situation non saturée, la teneur en eau mesurée est celle in-situ.

Si le sol n'est pas à saturation, il va falloir corriger les paramètres in situ par le niveau de désaturation, pour
raisonner sur des paramètres à saturation.
Fredlund (8) a proposé comme écriture de la résistance au cisaillement, la relation :

ff = C' + (v – ua) tgØ' + (ua – uw) tg Ød (6)

où :

C' : cohésion en situation saturée à long terme

Ø' : angle de frottement interne à long terme
v : contrainte totale
ua : pression d'air
uw : pression de l'eau
Ød : angle régissant l'augmentation de la résistance au cisaillement avec celle de la succion ua – uw

De nombreuses relations ont été proposées pour relier Ød à Ø'

Oberg et Salfours (9) proposent par exemple :

tgØd  S tgØ' où S est le degré de saturation (7)

Nous avons proposé de plus (7), des relations entre ff et pl* pression limite nette :

pl* # (ff)1,41 en kPa (8)

ou entre ff et qT-v, résistance de pointe au pénétromètre statique CPT corrigée de la pression interstitielle
générée par l'essai et la contrainte totale et :

qT # v + (ff)1,78 en kPa (9)

3. UTILISATION DES RESULTATS DU PENETROMETRE STATIQUE CPT'u EN SITUATION SATUREE

L'essai au pénétromètre, par la mesure de qT et fs frottement latéral sur manchon, permet l'estimation des
paramètres :

182
qt  v (10)
Qt 
'v

(11)
fs
Fr 
qT  v
On peut alors évaluer selon Robertson (10), l'indice de comportement du sol SBI, tel que :

SBI ² = (3,47 – log Qt) + (log Fr + 1,22)² (12)

Le tableau ci-après donne la nature probable du sol testé à partir de la valeur de SBI :

Tableau 1.

Zone Soil description SBI VBOD

7 Gravely sand
6 Sand to silty sand 1.31 0.5
5 Silty sand to sandy silt 2.05 1.5
4 Sandy silt to silty clay 2.6 3.3
3 Silty clay to clay 2.95 5.5

Dans le tableau ci-dessus, il apparaît que les valeurs de SBI peuvent être corrélées à celles de VBOD, ce qui
peut être traduit par les relations :

VB OD  0,4
SBI  3,6 (13)
VB OD  1,83
où :
1,83 SBI  1, 44
VB OD  (14)
3,6  SBI

Le diagramme de Robertson ci-dessous permet de retrouver la localisation du point représentatif du couple

(Ft, Qt)

Figure 1.

183
Il est alors possible de positionner le point dans le diagramme ci-dessous, pour estimer Gmax = Go

Figure 2.

Plus simplement, nous avons proposé (11) les relations explicites suivantes avec VBOD :

2 2
G max  2.15 VB OD 2   2 VB OD 5.3
VB OD  5.6 
log       log Qt  (15)
6.6 VB OD 7  VB 1.3   VB 
qT  OD   OD  1.26 

avec SBI :
2 2
G max 1.46 SBI  4.25  3.64 SBI  7.74   28.4  2.9 SBI 
log       log Q t  (16)
qT SBI  3.09  SBI  6.11   SBI  5.44 

Par exemple, à 5 m de profondeur, eau à la surface du sol avec :

૪h = 18,5 kN/m3 qT- v = 1,1 MPa 'vo = 42,5 kPa v = 92,5 kPa

G max
VBOD = 4,45 QT = 23,7 log  1,645
qT
G max
log  44,15 et Gmax = 48572 kPa Vs = 162 m/s
qT

4. ESSAI AU PRESSIOMETRE EN SITUATION SATUREE

L'avantage de l'essai pressiométrique est de ne pas être limité en profondeur comme le pénétromètre par le refus
de l'appareillage.
Les inconvénients sont la réalisation d'essais tous les mètres, dans un forage dont il faut limiter le
remaniement. La technique de mise en œuvre STAF est ici intéressante.

184
Photo 1. Set of operations (STAF of Apageo Cie)

Photo 2. Self bored slotted tube – 2005 (STAF of Apageo Cie)

La mesure de l'augmentation de volume de la cellule de mesure avec la pression dans la cellule, permet
l'estimation du module de cisaillement du sol G pour le niveau de distorsion de l’essai ૪.
V

Phase pseudoélastique
V1

p
p1 p2
Figure 3.

Nous pouvons écrire classiquement :

p  V  V2  p 2  p1
GV   Vs  1  (17)
V  2  V2  V1

R  Vs  V2 
૪ 2   1 distorsion au milieu de la phase élastoplastique (18)
R  Vs  V1 

185
avec VS = 535 cm3 volume initial de la cellule de mesure.

On peut alors utiliser les relations de Ishibashi et Zang (1993) pour estimer Gmax :

avec :
G ( m  mo )
 K'm 'm in kPa (19)
G max

1
et : K  1 0.984
(20)
 0.000102  n 
1   
  

0.544
e 0.0145 Ip
1. 3
et : m  mo  0.8
(21)
 0.000556 
1   
  

en prenant les valeurs données dans le tableau ci-dessous pour n fonction de Ip :

Tableau 2.
Ip % n
0 0
0 < < 15 3,37.10-6 Ip1,404
15 < < 70 7.10-7 Ip1,976
Ip > 70 2,7.10-5 Ip1,115

Par exemple, à 5 m de profondeur, avec ૪h = 18,5 kN/m3 eau à la surface du sol 'm = 28,3 kPa

Vs = 535 cm2 V1 = 60 cm3 à p1 = 0,03 MPa

V2 = 140 cm3 à p2 = 0,14 MPa
G = 0,87 MPa૪ = 0,065
n = 2,6 x 10-4 pour Ip = 20 % k = 0,006 m-mo = 0,26

G
donc  0,0143 Gmax = 60,8 MPa V = 181 m/s
G max

5. UTILISATION DE L'IDENTIFICATION

Toujours dans un souci de croiser l'information, Kramer (5) recommande l'utilisation pour les sols comportant
une matrice fine, de la relation :

Gmax = 625 F(e) OCRk pa1-n ('m)n (22)

 F(e) étant une fonction de l'indice des vides :

- Hardin (1978) a proposé :

F(e) = 1/ (0,3 + 0,7e2) (23)

- Jamiolkowski a proposé en 1991 :

F(e) = 1/e1,3 (24)

 OCR niveau de surconsolidation 'p/'vo

186
Nous avons proposé (7) comme relation permettant d'estimer 'p pour un sol 0/400 µ, WSC étant la teneur en
eau à saturation, en situation surconsolidée, WL la limite de liquidité :

WSC  0,075
log 'p = 5,25 – 0,25 log 'vo – 3,97 (25)
WL  0,075

 Le coefficient proposé par Hardin et Drnevich (1972) pour k, est détaillé dans le tableau 3 :

Tableau 3.

Ip % 0 20 40 60 80 100
k 0 0,18 0,3 0,41 0,48 0,5

 n coefficient souvent pris égale à 0,5

 'm pression verticale effective moyenne : ('vo + 2 'H)/3
 pa pression atmosphérique dans la même unité que 'm et Gmax

Par exemple à une profondeur de 5 m, nappe en surface avec :

૪h = 18,5 kN/m3 ૪d = 13,6 kN/m2 WSC = 0,358 e = 0,95

Ip = 20% WL = 0,473 'm = 42,5 kpa 'm = 28,3 kpa

'p = 105 kpa OCR = 2,47 n = 0,5 pa = 100 kPa

alors Gmax = 41823 kPa et VS = 150 m/s

par Jamiokolvski ou Hardin.

6. CONCLUSION

L'utilisation de plusieurs méthodes de reconnaissance et celle d'un jeu de corrélations permettent de croiser
l'information sur les valeurs de Vs d'une couche donnée et celles de Gmax.
A partir des différentes valeurs de Vs, il est alors possible de faire choix d'une valeur représentative de Vs et
ayant les valeurs pour chaque couche, d'estimer le Vs30 et donc la classe de sol.
Il est aussi possible de construire un modèle pertinent des valeurs de G.
Certes, ces valeurs sont à comparer à celles de dispositifs CROSS HOLE et d'essais de laboratoire, la
comparaison permettant d'améliorer régionalement les corrélations proposées.

REFERENCES

Braja MD, Ramana GV (2011) Principles of soil dynamics, 2ne Ed, Cengage Learning, USA
DMT test without CREEP, Norme AFNOR NF P94-443-1, France
Filiatrault A (1996) Eléments de genie parasismique, Editions de l'Ecole Polytechnique de Montréal, Canada
Kramer SL (1996) Geotechnical earthquake engineering, Prentice Hall, USA
Mouroux P, Margron P, Pinte JC (1988) La construction économique sur sols gonflants, Rexcoop, BRGM, France
Jean-Claude Gress, Corrélations entre les paramètres oedométriques, la pression limite nette au pressiomètre et la cohésion
non drainée pour sols saturés, Param-Géo 2012, Hammamet, Tunisie.
Fredlund, D.G. & Rahardjoh, H. 1993. Soil mechanics for unsaturated soils. New York: Wiley and sons.
Oberg, Salfours (1997). Determination of shear strength parameters of unsaturated silts and sands based on the water
retention curve, Geotechnical testing journal, vol. 28.
T. Lunne, P. K. Robertson, J. J. M. Powell (1992).Cone penetration testing in geotechnical practice.
Gress J.C. & Mouroux P. (2014), PMT and DMT tests, tools for earthquake engineering, Istanbul, European Conferences on
Earthquake Engineering and Seismology, 2ECEES.

187
188
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Undrained Shear Strength of a Libyan Marine Soil

Résistance au cisaillement non drainé d'un Sol Marin libyen

S. Khatrush(1,*)& G. El-Gehani(1)
1
University of Benghazi, Faculty of Engineering. Civil Engineering Department, Benghazi, Libya;
khatrush@[Link]
(*) Corresponding author

ABSRACT - This paper presents the main results obtained from an extensive laboratory investigation on a soft
marine soil . The samples were recovered from sea bed in the offshore near Benghazi city. The testing program
aimed at determination of the basic geotechnical properties of such soils, the use of simple laboratory Strength
index testing devices ( such as; Torvane & Shear vane) and a conventional unconsolidated undrained triaxial
testing (UU) ,to evaluate the undrained shear strength.
Due to the difficulty in preparing and handling specimens of very soft soils in order to test in triaxial apparatus,
the test may not always be successful and hence other laboratory testing techniques can be adopted provided they
can produce comparable results for such cases. The resulting shear strength obtained by the strength index
devices were examined and compared with that obtained during triaxial testing. The effectiveness of adopting
simple strength index devices to evaluate the undrained shear strength is demonstrated, in which shear vane
device was able to predict a comparable results with that obtained during triaxial UU-test although it is about
20% higher. Torvane device was considerably underestimating the undisturbed shear strength by approximately
40%. Furthermore, its prediction even lower than that obtained for remolded samples.

RÉSUMÉ - Cet article présente les principaux résultats obtenus à partir d'une enquête approfondis en laboratoire
sur un sol marin doux. Les échantillons ont été récupérés du lit de la mer dans la mer près de Benghazi ville. Le
programme d'essais visant à déterminer les propriétés géotechniques de base de ces sols, l'utilisation de
dispositifs de tests de laboratoire simples (dispositifs de Strength Index) pour évaluer la résistance au
cisaillement non drainée (tels que; Torvane & Shear palettes) et un essai triaxial non drainéconsolidéeclassique
(UU).
L'essai triaxial est couramment utilisé pour déterminer les paramètres de résistance au cisaillement, toutefois, en
raison de la difficulté dans la préparation et la manipulation des échantillons de sols très mous afin de tester dans
un appareil triaxial, le test peut ne pas toujours être des techniques d'analyse de laboratoire succès et donc
d'autres peut être adoptée à condition qu'ils puissent produire des résultats comparables pour de tels cas. La
résistance au cisaillement résultante obtenue par les dispositifs de l'indice de résistance ont été examinées et
comparées à celles obtenues lors de l'essai triaxial, l'efficacité d'adopter des dispositifs de l'indice de résistance
simples pour évaluer la résistance au cisaillement est mise en évidence, dans lequel dispositif de cisaillement à
palettes a pu prédire un résultat comparable avec celle obtenue au cours de l'essai triaxial UU bien qu'elle est
d'environ 20% plus élevé. Dispositif Torvane sous-estime considérablement la résistance au cisaillement non
perturbé par environ 40%. De plus, sa prédiction encore plus faible que celle obtenue pour les échantillons
remoulés.

1 INTRODUCTION

Soft soils usually a source for many problems to geotechnical engineers, because of their low shear strength and
liability to exhibit large deformation. The correct evaluation of shear strength is very important for achieving
safe design and avoiding the incidence of instability.
Measurement of undrained shear strength in the laboratory is commonly made using triaxial testing on a
cylindrical specimens directly recovered from the field after they have been subjected to certain sample
preparation to fit with test apparatus .Triaxial test is relatively expensive and time consuming, furthermore for
soft soils are susceptible to certain disturbance during sampling, handling and other pretest process (Barros et al,
2009). The adoption of simple testing devices may be a possible alternative for evaluating undrained shear
strength of soft soils, with respect to cost, time and problems of sample disturbance. An examination for their

189
capability to produce results which are comparable with that provided by other approved testing procedure such
as triaxial test is extremely demanded.
A total of 125 cores of marine soil recovered from shallow depths (0 to 6.0 meter), were included in the
testing program. Samples from sea bed of Benghazi offshore were available for laboratory investigation. a total
of The laboratory investigation consists of tests for the determination of basic engineering properties of such soft
marine soils and the evaluation of undrained shear strength by means of unconsolidated undrained triaxial test
(UU) and other simple strength index methods such as Torvane device (Tv) ,packet pentrometer and miniature
shear vane (Sv).Preliminary trials using packet pentrometer indicate the incapability of the device to measure
strength of soft soil, therefore it was excluded from the testing program.
Some researchers utilized strength index devices for measuring the undrained shear strength for cohesive
soils. Nearing (1988) used vane shear, pocket pentrometer and fall cone devices, Leoni (2009) adopted the
Torvane and pocket pentrometer, for measuring shear strength of reconstituted soils, Howard and Badran (2011)
carried out a comparison study of hand held devices (Torvane and penetrometer) with the unconfined
compression test for low strength cementitous material. The laboratory vane shear test was also adopted by
Kayabali (2011), Mohd et al (1997) and Ghosh (2013), for measuring shear strength of remolded soil.
The main objective of the testing program is to examine the capability of the simple strength index devices to
estimate the undrained shear strength of soft cohesive soils by conducting a comparison with the unconsolidated
undrained triaxial test results on the same soil.

2 METHOD OF TESTING

Testing of soft soil is not an easy matter, and therefore it requires extra care and special precautions in order to
minimize sample disturbance. Cores which reserved in 76 mm plastic tubes were cut to the required lengths to
conduct the strength index testing without extraction from the reserved tube. For smaller size samples required
for triaxial testing ,extraction of 35 mm specimens were made according to the method described in ASTM -
D3213 ( 1997) .This method was found relatively successful, although about ( 50 % ) of samples exhibited high
amount of disturbance and therefore have been discarded from strength testing.

2.1. Triaxial Tests

Samples extracted from tubes are trimmed to at least twice their diameter; samples are to be handled very
carefully in order to avoid deforming their original shape. Many samples especially those having high natural
water content and who showed extra softening and hence, require some support against bulging under its own
weight before placing inside the rubber membrane, this is done by wrapping the sample with filter paper similar
to that used for lateral drainage and finally placed in triaxial cell . The supporting filter paper immediately
saturated by absorbing water from specimen surface, it is thought this should not have any significant effect on
the resulting shear strength in the undrained case. Directly upon placing the specimen inside the cell water is
quickly filled in order to allow some hydrostatic pressure to help maintaining specimen from becoming flabby.
For samples recovered from shallow depths usually subjected to cell pressures higher than the corresponding
value encountered in the field generally not less than 50 kPa in order to provide enough confinement and prevent
specimen from buckling during shearing stage (see figure 1a).

Figure 1a: Triaxial test cell and sample Figure 1b: Vane shear test device and sample

190
2.2 Strength Index Devices

Implementing the strength index devices is rapid and easy, it is not involving additional process of sample
preparation as in triaxial test, hence avoiding extra disturbance Holtz and Kovacs (1981).The test is directly
made while sample remains inside sampling tube, For vane shear test, device is shown in figure (1b), the
procedure followed is according to (ASTM - D4648), whereas for other tests the manufacture manual is
followed .The recorded strength is taken as the average of at least 3 measuring trials for each sample. Initial
measuring trials with pocket pentrometer were not successful for soft soil so the test has been excluded.
Measurement by Torvane device performed only for undisturbed sample, whereas the shear vane device was
utilized for measuring both undisturbed and remolded shear strength. The undisturbed shear strength is measured
by pushing the vane blades into the sample to at least twice its height and then rotates at a constant rotation speed
and record the maximum equivalent torque. The remolded strength value is obtained by rewinding the vane back
to its initial position before measuring undisturbed strength and rotates again for the second measurement; hence
the new reading of maximum equivalent torque is recorded.

3 SOIL PROPERTIES

Soil classification tests including grain size analysis, Atterberg limits and physical properties such as water
content, specific
gravity, and bulk density ,were determined for all samples .The range of index parameters and other physical
properties are presented in table (1) .

Table 1: summary of the soil physical properties

Property Minimum Maximum Mean

LL (%) 30.85 90.3 38.7
PL (%) 20.59 56.98 26.71
PI (%) 0.99 38.47 13.0
Wc (%) 29.96 105.78 48.88
Gs 2.4 2.78 2.57
ρb (g/cm3) 1.55 2.3 1.87
Void ratio (e) 0.77 2.65 1.26
Fine fraction (%) 25.85 98.47 82.85

The position of the soil in the plasticity chart is shown in figure ( 2 ) , the majority of the points are located
below A-line , and according to unified soil classification system (USCS),the soil is classified as low plasticity
silt (ML ) and clay (CL) . The soil is containing large amount of fine material of an average of 83 %.the majority
are of silt fraction. The average value of liquid limit is 39 % and only few samples show relatively high liquid
limit, the plasticity index in most cases not having a wide range. The natural water content is generally high with
an average of 49 %, it is in most of the samples exceeding the Liquid limit ( see figure 3).

Figure 2: Distribution of soil plasticity results within the Figure 3: Plots of water content against liquid limit
plasticity chart

191
4 UNDRAINED SHEAR STRENGTH

Attempts were made to test a total of (125) samples for the laboratory evaluation of undrained shear strength,
only (59) samples were managed to be successfully tested in the UU-triaxial test (108) were tested using Torvane
(Tv) device and (84) in the shear vane (Sv) apparatus. Table (2) contains the range of undrained shear strength
values in (kPa) obtained during testing by various methods.
Table 2: Range of undrained shear strength results.

Test Torvane Shear vane Shear vane UU-triaxial

undisturbed Remolded
Minimum 4.5 6.91 3,91 5.0
Maximum 23.5 39.14 24.13 28.0
Mean 8.3 14.76 10.06 11.57
St. Dev. 3.03 7.00 4.33 5.26

4.1 Unconsolidated undrained strength

Samples were tested in a strain controlled UU-triaxial apparatus, typical resulting stress- strain curves are
shown in figure (4) for samples at different initial water content (wc). Because of soft condition of samples the
curves were rather jagged, furthermore samples were exhibiting high axial strain which led to noticeable sample
bulging, and majority of samples were not showing a clear failure plane, hence testing usually terminated at
nearly 15 % strain. The resulting undrained shear strength (Cu) values are plotted against the effective
overburden pressure and presented in figure 5, more scatter of Cu is seen on the top part at shallower depths less
than 3.0 m, with larger average strength of about (13 kPa) compared with (11.5 kPa)in the lower part. The
maximum predicted Cu value is not exceeding 28 kPa. The soil thus categorized as a very soft soil which
behaves as viscous material. It is simply affected even by low level of vibration or any means of mechanical
process and could easily deform during preparation and handling prior to testing, hence resulting inconsistency
in the measured strength.

Figure 4 Typical triaxial stress-strain curves Figure 5 Plots of undrained shear strength with
effective pressure

4.2 Comparison with index strength data

The results of undrained shear strength obtained using strength index devises ( Tv and Sv ) on undisturbed
samples are plotted against sampling depth and shown in figure (6) together with UU-triaxial test data. The data
indicated that the Cu values evaluated by Sv device are generally higher by 27 % than that obtained during
traixial test, but maintaining approximately similar trend, this may be attributed to greater sample disturbance
exhibited by samples tested in triaxial test. This was also reported by other researchers who indicated that
laboratory Sv usually predicts higher values of shear strength than actual field value ( Mohd et al, 1997).This
may also be associated with great disturbance exhibited by samples tested in triaxial test. The shear strength
obtained by Tv device is generally lower than that obtained by other methods. It is only showing small variation
in estimated values but, with lesser scatter. The average value of Cu predicted by Tv is only 71% of that
predicted in triaxial test.

192
Figure 6: Plots of undrained shear strength with depth- obtained by different devices

Figure (7) presents plots of shear strength evaluated by Sv and Tv against that obtained by triaxial testing ,it
is observed that the shear strength measured by torvane device is considerably underestimated .However,
Torvane testing is usually applied by pressing several short blades into the soft material which is probably
causing some disturbance to the sample surface prior to rotating the device, and subsequently affecting the
measured shear strength, therefore measuring shear strength of very soft soils with Tv may not be an appropriate
choice.

Figure 7: Results of Cu from triaxial test against that from Figure 8: Variation of undisturbed with remolded strength
torvane and lab. measured by laboratory shear vane device. Shear vane for
undisturbed samples.

4.3 Remolded strength

The remolded shear strength, are plotted against undisturbed shear strength obtained by Shear vane device in
figure (8). Good correlation can be obtained from the results, the undisturbed strength is approximately 1.5 times
the remolded value, and nonetheless the soil is not showing high sensitivity upon remolding. However soil of
water content equals or above liquid limit should not be considered totally undisturbed and some change in its
original state is inevitable and, the use of remolded shear strength for geotechnical design may be considered a
conservative choice (O`Kelly, 2013). The remolded strength obtained by Sv is generally higher than that
predicted by Tv device by approximately 17 %, this is supporting the previous conclusion that the Tv device is
not the right choice to predict shear strength at such low strength levels.

193
4.4 Correlations with soil properties

Many researchers suggested correlations of undrained shear strength with soil consistency limits. Most of them
come to the conclusion that, both undisturbed and remolded undrained shear strength of cohesive soils is
significantly affected by the change of water content, therefore they relate it to either water content (wc),(Sharma
and Bora, 2000, Leoni, 2009, Ghosh, 2013, Vipulanandan et al, 2007), plasticity index (PI) (Skempton, 1957;
Bjerum and Simons, 1960) or liquidity index (O`Kelly, 2013), (Koumoto and Houlsby, 2001).

Figure 9: Undrained shear strength (triaxial) against Liquidity Index

However ,for the present work ,since most of water content of tested samples are around liquid limit or above
,with no wide range of (wc),it was found logical to relate the undrained shear strength with the liquidity
index(LI) since it is expressed in terms of the water content above plastic limit with respect to plasticity index ,
hence positive value indicate (wc) higher than (PL). The results of triaxial test data are plotted against (LI) and
presented in figure (9) . The results are arbitrarily correlated, despite high level of discrepancy, It is inevitable to
accept such discrepancy when testing natural soil having high water content, the suggested trend is broadly
emulated with other findings for soft soils in the sense that shear strength is inversely proportional to liquidity
index ,even though none of their suggested relationships was found to reasonably fit with the present results.

5 CONCLUSION

A series of laboratory tests were carried out on soft marine deposit utilizing several methods of testing involving
UU-triaxial, Shear vane and Torvane. The following are the main conclusions;
Testing soils of high water content equals to liquid limit or higher is generally a difficult matter. The use of
triaxial test to determine the undrained strength is possible provided extra care during sample preparation and
testing stages. It is admitted that certain degree of sample disturbance could not be avoided.
The effectiveness of adopting simple strength index devices to evaluate the undrained shear strength is
demonstrated in this work, in which shear vane device was able to predict a comparable results with that
obtained during triaxial UU-test although it is about 20% higher. Torvane device was considerably
underestimating the undisturbed shear strength by approximately 40% . Furthermore, its prediction even lower
than that obtained for remolded samples.
Attempt to relate the undrained shear strength with Liquidity index was only arbitrarily made possible despite
of high scatter of the data. It is generally comparable just with the trend reported in other research workers.
ACKNOWLEDGEMENT

The authors wish to thank the Libyan Technical Consultancy Co. in Benghazi for making the laboratory test data
available.

194
REFERENCES

Barros J. M. C. Amaral C. S & Silveira R. M. S (2009) Geotechnical properties of Brazilian marine clays Proceedings of the
17th International Conference on Soil Mechanics and Geotechnical Engineering Alexandria
Bjerrum, L., & N.E. Simons. (1960) Comparison of Shear Strength Characteristics of Normally Consolidated Clay.
Proceedings. Research Conference on Shear Strength of Cohesive Soils, 1771–1726.
Ghosh R. (2013) , Effect of soil moisture in the analysis of undrained shear strength of compacted clayey soil , J. civil Eng.
& construction technology, vol. 4(1), pp 23-31
Holtz, R.D. & Kovacs W.D. (1981), Introduction to geotechnical engineering. Prentice Hall, Inc., Englewood Falls, NJ.
Howard, I. & Badran, W. (2011) Comparison of Hand Held Gage and Unconfined Compression Results in Low Strength
Cementitious Stabilized Materials. Geo-Frontiers 2011: pp. 2574-2583.
Kayabali K. (2011), Consistency limits – A reappraisal, EJGE, vol. 16
Koumoto, T. & Houlsby, G. T. (2001). Theory and practice of the fall cone test Geootechnique 51, No. 8,
Leoni A.J (2009) Characterization of post pampean clays Proceedings of the 17th International Conference on Soil
Mechanics and Geotechnical Engineering Alexandria
Mohd J. ,Mohd A.,Taha,R. Ahmed J Abu Kassim A. Jamaludin A,J. & Jaaadil A (1997) Prediction and Determination of
Undrained Shear Strength of Soft Clay at Bukit Raja , Pertanika [Link]. & Technology,(5),pp111-126.
Nearing, M.A and West L.T (1988) Soil strength indices as indicators of consolidation, American society of agricultural
engineers, vol. 31(2)
O`Kelly B.C. (2013) Atterberg limits and remolded shear strength – water content relationships ASTM Geotechnical testing
journal, vol.36, No 6
Sharma, B & Bora B.K. (2003) Plastic Limit, Liquid Limit and Undrained Shear Strength of Soil-Reappraisal ASCE J
geotechnical and geoenviromental engineering .
Skempton, A.W. (1957) “The Planning and Design of New Hong kong Airport”. Proceeding. London: Institute of Civil
Engineering 7 305–307.
Vipulanandan C., Kim M.S.& Harendra,.S. (2007) Microstructural and geotechnical properties of Houston-Galveston Area
soft clays , ASCE, Advances in Measurement and Modeling of Soil Behavior.

195
196
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Application of the nchrp-1-28 resilient modulus model for characterizing

granular material in Sudan

Application du modèle de module élastique NCHRP-1-28 pour la caractérisation de

matériaux granulaires en Soudan

Omer G. Omer1, Ahmed M. Elsharief2

Building and Road Research Institute (BRRI), University of Khartoum, Khartoum, Sudan

ABSTRACT
This paper displays an attempt to examine the suitability of the NCHRP-1-28 prediction model for estimating the
Resilient Modulus (MR) of blended unbound granular materials from MR test data. The blending process was
mechanically done using two natural coarse Wadi sand and pure crushed basaltic stones. An experimental testing
program was performed on these typical blends involving the basic tests as well as the dynamic triaxial tests for
resilient modulus (MR). The predicted MR values showed good agreement with the measured ones. The predicted
MR values are about 8 % to 10% lower than the measured ones indicating conservative estimates for MR using
the NCHRP-1-28 model for the blended unbound aggregates of Khartoum.
Key words: Unbound gravel, gradation, gradation correction, prediction, resilient modulus,

RESUME : Ce papier montre une tentative pour étudier la pertinence du modèle de prédiction NCHRP-1-28
pour estimer le module réversible (MR) de matériaux granulaires non liés reconstituées à partir de données
d'essai MR. La procedure de melange a été effectuée mécaniquement à l'aide de deux naturelle sable grossier
Wadi et de pierres basaltiques écrasés. Un programme d'essais expérimentaux a été réalisé sur ces mélanges
typiques impliquant des essais de base ainsi que les essais triaxiaux dynamiques par module élastique (MR). Les
valeurs prédites MR ont montré un bon accord avec celles mesurées. Les valeurs de MR prévues sont environ
8% à 10% inférieures à celles mesurées indiquant des estimations prudentes pour MR en utilisant le modèle
NCHRP-1-28 pour les agrégats non liés mélangés de Khartoum.

1. INTRODUCTION

The resilient modulus is the elastic modulus under dynamic loading. The resilient modulus (MR) can be
numerically quantified as the ratio of the applied deviator stress ( to the recoverable vertical strain “resilient
strain” ( . The measured laboratory MR under the optimum compaction condition could reflect the actual
resilient behavior of granular material in a flexible pavement [1]. The resilient modulus (MR) testing provides a
mean of characterizing pavement construction materials, including subgrade under variety conditions [2].
According to AASHTO 307-TP 46-1999; the resilient modulus test procedure for base/subbase material
comprises 15 sequences each sequence has 100 cycles (repeated load application) with variable deviator stress
values ( ) at 5 different values of confining pressure ( ). The obtained resilient material properties are
important input parameters for the mechanistic procedures for designing pavement structures.
Numerous researchers suggested and developed various regression equations for predicting resilient modulus
values for unbound granular materials. Several attempts were carried out for developing models correlating the
resilient modulus and the cyclic test parameters such as stress states. The recent and the most validated one is the
generalized constitutive model that was earlier suggested by Uzan [3] and developed under NCHRP 1-28 [4] and
used within the context of the 2002 Mechanistic-Empirical Pavement Design Guide.
This paper focuses on reviewing NCHRP-1-28 model used to estimate resilient modulus for unbound
granular base material and verifying its application on mechanically blended unbound gravelly materials.

2 REVIEW OF THE RESILIENT MODULUS (MR) PREDICTION MODELS

Several attempts were carried out for developing models correlating the resilient modulus and the cyclic test
parameters such as stress states [5]. The advantage of the application of the regression equation or model that is
used for determining the resilient modulus of any material (NCHRP project 1-37 A) is to provide a way of

197
documenting the resilient modulus of the materials tested; all final values of the resilient modulus data will be
represented by the materials respective regression constants. The material dependent regression constants can
then be used to determine the resilient modulus for each material at any stress condition [6]. For clarification
certain models will be illustrated hereunder:
Uzan can be considered as the focal developer of the NCHRP model. At the beginning he studied the bulk stress

model (K- ) which was expressed as

MR= k1 ( ) k2 ………………………………………………………. (1)

Where: MR = Resilient modulus

= Bulk stress
k1& k2 = Regression analysis constants

Further, Uzan made a major modification by considering both bulk stress and deviator stress (ᆓd) to account
for the effect of shear stress. The Uzan model is defined by the following relationship:

MR= k1 ( ) k2 (ᆓd) k3 ……………………………………………….. (2)

Where: MR = Resilient modulus
= Bulk stress
k1, k2 & k3 = Regression analysis constants
ᆓd = Deviator stress
Further improvement to the Uzan model (2) was made by replacing the deviator stress term with a more
fundamental parameter, the octahedral shear stress [7]. The produced new model has been accepted as a
universal material model:
MR ………………… ………............. (3)
Where: MR = Resilient modulus
= Bulk stress = First stress invariant (ᆓ1+ 2 ᆓ3) or (ᆓd+ 3 ᆓ3)

= Octahedral shear stress =1/3((ᆓ1- ᆓ2)2+ (ᆓ1- ᆓ3)2+ (ᆓ2- ᆓ3)2)1/2

k1, k2 & k3 = Regression analysis constants
The NCHRP model model was suggested for predicting MR values by NCHRP Project 1-37 [8], and was recently
validated in NCHRP project 1-37 A [9]. The model is the same as the octahedral shear stress model in equation

(3) but it was further modified by adding a “+1” term to avoid the absurd calculation of modulus when the

tends to zero (the modulus tends to zero when tends to zero).

…………………………………….. (4)
Where: MR = Resilient modulus
= Bulk stress

= Octahedral shear stress

k1, k2 & k3 = Regression analysis constants
All final values of the resilient modulus data will be represented by the materials respective regression constants
(k1, k2, k3). The three material dependent regression constants can then be used with the given equation to
determine the resilient modulus for each material at any stress condition [10].
The new Mechanistic-Empirical Design Guide Method (M-EPDG) is utilizing the resilient modulus
constitutive abovementioned equation and it is referred to K1-K3 or universal model and its main advantage being
the consideration of the stress state (i.e., change of normal and shear stress) of the material during testing [11].
Buchanan [11] explained that “k” coefficients for the given model are determined through regression analysis.
The k1 is proportional to the elastic modulus of the material and will be positive; k2 is the exponent of the bulk
stress term and will be positive since an increasing bulk stress results in a higher modulus; k3 is the exponent for

198
the shear stress term and will typically be slightly negative since an increasing shear stress will likely weaken the
material resulting in a lower modulus.

3 THE MATERIALS AND TESTING PROGRAM

The materials used in this investigation are blended natural unbound materials from Khartoum state. The
gradation correction was done using wadi sand and crushed stone. The natural materials are colluvial deposits
originally conglomerates belonging to Nubian Sandstone Formation. The formations were deposited by braided
streams under semi-dry tropical climate. These natural gravels often satisfy subbase requirements but seldom
satisfy base ones.
For this investigation, three different natural unbound gravels were used; Umm Ketti (C1), Al-Silait (M2) and
Huttab (F3) to represent the three ranges of gradation, coarse (C1), medium (M2) and fine (F3), respectively.
Figure 1 shows the three gradations plotted with the Transport Research Laboratory (TRL) standard envelope for
GB3 base material. The coarse aggregates (gravels) are about rounded or elongated for F3 and M2 materials
whereas they are rounded to angular for C1 material.
The Wadi sand was obtained from the seasonal water sources locally known as Wadis. Its gradation is shown
in Figure1. Fresh crushed basaltic stone with three different sizes (19-12 mm, 12-5 mm, and 5-0 mm) was
provided from a crushing plant in south- west Khartoum for the gradation correction process.
Trial mixes were made for the three natural unbound gravels in an attempt to produce materials that satisfy TRL
GB3 base course material requirements [12]. The grain size distribution results from the trial mixes are given in
Figures 2 and 3 for Wadi sand and crushed stone mixes, respectively. The results of the routine laboratory tests:
sieve analysis, Atterberg Limits and Linear Shrinkage of the fines particles, Modified Proctor, California Bearing
Ratio (CBR) and Los Angeles Abrasion for the six blends are presented in Table (1). The CBR values greater
than 100 % are presented for research purpose.

Figure 1: Grain Size Distribution of the Three Natural Unbound Gravels, and Wadi Sand Compared to TRL GB3

Figure 2: Grain Size Distribution for the Three Gravels Blended with 25% Coarse Wadi Sand (TRL GB3)

199
Figure 3: Grain Size Distribution for the Three Natural Blends with Crushed Stone (TRL GB3)

Table 1: Engineering Properties for the Natural and Blended Gravels that Nearly Satisfying or Satisfying TRL GB3 Base
Requirements.

Type of 100% Natural Gravelly Unbound Material 75% Natural+25% WS Natural+ CR Blend (TRL/GB3 Base)
Material CBR% PI% FC% LA% CBR% PI% FC% CBR% PI% FC% LA%
C1 68 11 9 32 79 7 7.25 168 9 7 32
M2 56 15 18.5 37.24 98 16 14.38 128 10 10.3 35.54
F3 32 19 21 29.16 74 14.5 16.25 109 13 14.1 20.04

Note: PI % is plasticity Index, FC % is Fines Content % (% passing sieve No. 200) and LA% Los Angeles Abrasion Value

The blended materials were prepared at their optimum moisture contents (OMC), compacted in the resilient
test mold (300 mm high & 150 mm diameter) to their corresponding maximum dry densities (MDD) using
standard vibratory compactor, afterwards the resilient modulus (MR) test was performed on each of the samples
in accordance to AASHTOT-307 TP 46 – 99. The MR test results compared to their corresponding predicted
values are given in Fig. 4-9.

Figure 4: Measured and Predicted Resilient Modulus for 75% Umm-Ketti+25% Wadi Sand Blend

200
Figure 5: Measured and Predicted Resilient Modulus for 75% Al- Silait+25% Wadi Sand Blend

Figure 6: Measured and Predicted Resilient Modulus for 75% Huttab Gravel+25% Wadi Sand Blend

Figure 7: Measured and Predicted Resilient Modulus for 61% Umm-Ketti+39% Rock Dust (5-0 mm) Blend

Figure 8: Measured and Predicted Resilient Modulus for 50% Al-Silait+7% (19-12 mm)+8% (12-5 mm)+35% (5-0 mm)
Blend

201
Figure 9: Measured and Predicted MR for 65% Huttab+20% (19-12 mm) + 15% (12-5 mm) blend

5 RESULTS ANALYSIS AND DISCUSSIONS

From the measured resilient modulus values of the six treated samples in the study an excel program was used to
solve for the material regression constants (K1, K2 and K3 in equation (4)). The obtained values are given in
Table 2.
The predicted resilient modulus values were calculated using the NCHRP 1-28 (equation (4)) and the
regression constants given in Table (2) for each corresponding measured MR value. The measured and predicted
resilient modulus values are plotted as shown in Figures 4 to 9 for the six different blends.

Table 2: Resilient Modulus Regression Constants

Blends Resilient Modulus

Satisfied
TRL GB3 Generalized Constitutive Model NCHRP 1-28
Specs
Material Dependent Regression Constants Material Dependent Regression Constants for
for Blended gravel with WS Blended gravel with CR
K1 K2 K3 K1 K2 K3
C1 908.211 0.18047 0.10584 637.6701 0.228877 0.055543
M2 638.454 0.25525 -0.00334 683.945 0.22828 0.109538
F3 760.276 0.16388 0.05295 826.91 0.226058 0.08757

Figures [4 - 9] depict the good agreement of the predicted to the measured resilient modulus for the blends
satisfying or closely satisfying TRL GB3 base requirements.
It is observed that the predicted resilient modulus for the six treated gravelly materials reported values slightly
lesser with slight small differences when compared to the measured ones. This slight difference/or tolerance can
be considered as a factor of safety when using the predicted resilient modulus in the mechanistic- empirical
design.
The average reported tolerance compared to the measured resilient values for the given blends ranges
approximately between 7.8% and 10%. Therefore NCHRP 1-28 regression equation can be used with confidence
to predict the resilient modulus values for blended unbound natural materials from Khartoum.

The computed R2 values that displayed in the six plotted figures for the given blends in the study showed
values greater than 0.7 which indicated the fitness of applying NCRHP model for characterizing these typical
granular materials in Sudan.

6 CONCLUSIONS

The main objective of this investigation was to study the application of NCHRP-1-28 model for predicting
resilient modulus for mechanically blended natural unbound base material in Khartoum State. Three selected
unbound gravels from Khartoum state representing coarse, medium and fine gradations were treated with wadi

202
sand (WS) and crushed stone (CR) to obtain blends satisfying TRL GB3 base requirements. Dynamic triaxial test
was used to measure their resilient (MR). A simple excel soft program was used for solving the NCHRP 1-28
model and the materials regression constants (K1, K2 and K3) were obtained; then the predicted resilient modulus
values of the treated samples were computed and plotted with their respective measured ones. Some conclusions
can be drawn as follows:

 The predicted and the measured resilient modulus values showed good agreement, therefore NCHRP 1 – 28
model suits the treated natural unbound aggregates.
 It was observed that the predicted resilient modulus for the six treated gravelly materials reported values
slightly lesser with slight small differences when compared to the measured ones. This slight difference/or
tolerance can be considered as a factor of safety when using the predicted resilient modulus in the mechanistic-
empirical design. The average reported difference for the six blends ranges between 8% and 10%.
 The NCHRP 1-28 regression equation can be safely used to predict the resilient modulus values for the treated
unbound gravelly materials from Khartoum state.
 It was found that the displayed R-square values on the presented charts for the six treated gravels gave values
> than 0.7 which might reflect the application suitability of NCHRP 1-28 for characterizing these typical
granular materials in Sudan.

REFERENCES

[1] W. Virgil Ping [2001]. “Implementation of Resilient Modulus in the Florida Pavement Design Procedure,” Summary of
Final Report, WPI # 0510780 August, 2001, Florida A&M University- Florida State University College of Engineering.
[2] Stolle, D., F., L., Guo, P., Liu, Y. [2006], “Resilient Modulus Properties of Typical Granular Base Materials of Ontario”
Sea to Sky Geotechnique 2006, Department of Civil Engineering Mc Master University, Hamilton, Ontario, Canada, pp.
236-243.
[3] Uzan, J. [1985] 'Characterisation of granular material.' Transportation Research Record, 1022, 52-59.
[4] NCHRP Web Document 14 (Laboratory Determination of Resilient Modulus for Flexible Pavement Design) for
Estimating Stiffness of Subgrade and Unbound Materials for Pavement Design. Final Report (1997), Transport Research
Board (TRB), 464 pages.
[5] Rowshanzamir, M. A.,” Resilient Cross-Anisotropic Behavior of Granular Materials under Repetitive Loading,”
University of New South Wales, School of Civil Engineering, 1995-Anisotropy
[6] Tao, M., Abu-Farsakh, M., and Zhang, Z, [2008].”Characterization of Unbound Aggregates Revealed through
Laboratory Tests”, Technical note in International Journal of Pavement Research and Technology. Vol. 1 No. 2 April
2008. Pp. 72-75.
[7] Witczak, M. W. & Uzan, J. [1988] the Universal Airport Pavement Design System: Granular Material Characterization,
University of Maryland, Department of Civil Engineering, MD.
[8 NCHRP project 1-37 “Development of the 2002 Guide for the Design of New and Rehabilitated Pavement Structure”
Design Inputs. National Cooperative Highway Research Program (NCHRP), 2002.
[9] NCHRP 1–37A. Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures. Final Report,
Chapter 2, Material Characterization, Design Inputs. National Cooperative Highway Research Program (NCHRP),
March, 2004, pp. 65-74.
[10] Tao, M., Abu-Farsakh, M., and Zhang, Z. ”Characterization of Unbound Aggregates Revealed through Laboratory
Tests”, Technical note in International Journal of Pavement Research and Technology. Vol. 1 No. 2 April 2008. Pp. 72-
75.
[11] Buchanan, S. (2007).”Resilient Modulus: What, Why, and How? ,” Vulcan Materials Company, 08/31/07 No. of pages
13.
[12] TRL Overseas Road Note 31 – A guide to the structural design of bitumen-surfaced roads in Tropical and Sub-Tropical
countries pp. 1-31

203
204
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Geotechnical properties of polymer stabilised backfill soils

Propriétés géotechniques du Remblai stabilisé par sol polymère

F. N. Okonta*1; L. J. Moatlhodi1; T. Makharamedzha1; and M.T. Nhlapho1

1
University of Johannesburg, Department of Civil Engineering Science, Johannesburg, South Africa
* fnokonta@[Link];

ABSTRACT - Relatively little is known about polymer as a non-traditional stabiliser, but opportunities exist in
determining their effectiveness. Commercially available polymers other than Dry Powder polymer (DPP), have
not been extensively investigated. Tests were conducted to determine the geotechnical properties of polymer
treated backfill red soils. Polymer treated red soil was stabilized with 2% to 10% polymer content in increments
of 2%. The samples were cured for 3, 7, 14 and 28 days at 40ºC and 80% relative humidity. The results indicated
that waterborne polymer significantly improved the strength behaviour of unsaturated red soils. However
polymer treated red soils exhibited significant reduction in strength when soaked. In addition, increase in soil
fines and plasticity caused a reduction in strength and bearing resistance.

RESUME : On sait relativement peu sur le polymère comme stabilisateur non-traditionnel, mais il existe des
possibilités pour déterminer leur efficacité. Les Polymères disponibles dans le commerce autres que le polymère
de poudre sèche (DPP), ne sont pas étudiés. Des essais ont été effectués afin de déterminer les propriétés
géotechniques de remblai traité des sols rouges polymère. De polymère terre rouge traité a été stabilisée avec
2% à 10% de teneur en polymère par incréments de 2%. Les échantillons ont été cuits pendant 3, 7, 14 et 28
jours à 40 ° C et 80% d'humidité relative. Les résultats ont indiqué que le polymère à base d'eau
considérablement améliore le comportement de la résistance des sols rouges non saturés. Cependant polymère
traité terres rouges présentaient une réduction significative de la force lorsqu'il est mouillé. En outre,
l'augmentation des fines et de la plasticité du sol a provoqué une réduction de la résistance et de la portance.

1. INTRODUCTION

Paper compiled by [1] in which in-situ polymer treated test panels were conducted to check the effectiveness of
polymer additives. Four different soil samples and stabilisers were used to conduct field tests and determine the
performance of these treated soils. The stabilizers were Polymer A (South Africa); Polymer B (UK), Enzyme,
(USA) Sulphonated oil, (South Africa). The in-situ soil from Daveyton consist primary of silty sand; whereas in-
situ soil from Benoni consists of well graded silty sand with gravel and both Quantam and Putfontein contain
well graded sand with gravel. Daveyton soils is windblown sand, Benoni soils is weathered dolerite, Quantam
soils is well graded gravel and Putfontin material is a Ferricrete. The results indicate that polymer treated in-situ
soils exhibit a slight increase in strength in dry conditions. However under soaked conditions, polymer treated
soils exhibit significant strength improvements; Quantam treated soils displayed peak strength when stabilised
with polymer-A, similarly to Benoni treated soils when stabilised with polymer-B. It was determined that the
polymer B stabilised panels showed strength improvement on all the materials, but did not perform well under
wet conditions as polymer A.

Dry powder polymer (DPP) stabilisation is most effective in treating moderate to very poor quality gravels that
lose considerable strength upon wetting. Soaked CBR testing of the polymer-treated samples are significantly
higher than those of untreated samples. The difference is expected to be greater for higher plasticity materials
(clays and silts). The equilibrium moisture condition of the subgrade is anticipated to be lowered as this would
justify an increase in design subgrade CBR and consequently a reduced design pavement thickness [2]. The
mechanistic pavement design method [3] is not well suited for DPP stabilisation, as the method does not take
into account the direct benefit that drier stabilised layers produce. To date, decisions to use DPP polymers have
been based on the accumulating field evidence of their effectiveness [2]. Commercially available polymers other
than DPP, have limited conducted research in determining their engineering behaviour. Mixed results have been
published regarding the success of these polymers as soil stabilisers.

205
Locally available red sandy soil was used in this study. The soil is derived from the Maraisburg and/or Florida
formation which falls under the Jeppestown Subgroup. The Lithology of the area consists of green fine. The top
of the Maraisburg formation is considered as the base of the Central Rand Group because it is forms part of an
upwards coarsening facies succession, which originates from the upper shale of the underlying Roodepoort
formation [4]. The Roodepoort formation comprises of basal quartzite and shaley units in the lower member, fine
quartzite and wackestone in the central member, and a shale unit in the upper member [4].

2. MATERIALS AND METHODS

The samples were tested for their UCSs using a recently calibrated UCS testing machine that was equipped with
a dial gauge to measure the corresponding deformation/strains at failure. More than 350 samples were casted.
Specimens was compacted in 5 layers with 25 taps per layer using proctor energy at fixed OMCs. Series of CBR
tests were also conducted:
The constitution of the specimen are thus:
(1) Red soil + 6% Lime Fly Ash; Red soil + 6% Lime Fly Ash (Sealed) ;
(2) Red soil + 6% Polymer ; Red soil + 6% Polymer (Sealed);
(3) (10 Kaolin: 90 Red soil); + 6% polymer ;
(4) (80 Kaolin: 20Red soil) ; + 6% polymer

These samples were cured under the following conditions: Dry condition: (7days cured at 50ºC and 100%
Relative Humidity); and Wet condition: (Dry condition + 4 additional days soaked). The average results between
two specimens is taken as the interpreted results below. From the grading curve, it is evident that the red soil is
predominately sand and has very little fines as unexpected compare to the surrounding red soil found in the
Brixton formation straight down to Parktown formation which has higher fine content based on literature [5].

3. RESULTS AND DISCUSSIONS

The physical and atterberg limits of the soil specimens are presented in Table 1

Table 1: Engineering properties of the soil

Properties Mariasburg Red Sand

Specific gravity (Gs) 2.67
Fine Gravel (4.75 – 19.00 mm),(%) 2
Coarse Sand (0.075 - 4.75mm),(%) 7
Medium Sand (0.425 – 2.00mm),(%) 20
Fine Sand (0.075 – 0.425mm),(%) 68
Silt (0.002 - 0.075mm),(%) 3
Clay (<0.002mm),(%) 0
Liquid limit,(%) 31
Plastic limit,(%) 23
Plasticity Index (PI), (%) 8
Linear shrinkage (mm) 2
OMC (%) 7.4
MDD (Kg/m^3) 2167
AASHTO Classification A-2-4(0) (Sand)
UCS Classification SP (Poorly graded sand)
Cu 2.7
Cc 1.6
Compaction Results (MOD AASHTO) Untreated red soil = 2170 kg/m3; 2% polymer = 2120 kg/m3; 6% =
2075 kg/m3; 10 % = 1980 kg/m3 OMC increased marginally with polymer from 7.5% to 8.1%

206
Fig. 2: The effect of polymer content and curing period on Fig. 3: The effect of polymer content and curing period on
UCS of red soil UCS of bentonite red soil mixture

Fig. 4: The Effect of polymer content and curing period on Fig. 5: The effect of polymer content and curing period on
UCS of bentonite red soil mixture UCS of kaoline- red soil mixture

14 day strength
8
7
2% Polymer
6
4% Polymer
UCS (MPa)

5 6% Polymer
4
3
2
1
0
0 20 40 60 80 100
Kaolin content (%)
Fig. 6: The effect of polymer content and curing period on UCS of Kaoline- red soil mixture

Fig. 7 below illustrates a relationship between elastic modulus vs. curing time of polymer treated kaolin - red
soil blends (80:20). It is observed that the modulus increases with increasing curing time. However, note should
be taken that the dips exhibited are due to the high strains obtained from the dial gauge used. 6% polymer
content yielded the maximum elastic modulus with a value of 63 MPa after 28 days of curing.

207
Fig. 7: The Effect of polymer content and curing period on Modulus of Kaoline- red soil mixture
California Bearing Ratio

The specimens used for the CBR tests was presented in fig 8. From fig. 9 it is evident that there is considerable
reduction in the CBR when specimens are soaked, especially for polymer treated specimen. Lime fly ash
(unsealed) showed unexpected early CBR strength reduction compared to the unsealed samples. The CBR (%)
value of the unsoaked lime fly ash specimen is equivalent to a G6 natural gravel material which is generally used
as an upper selected layer for high volume pavements or for poor in-situ material with a CBR (%) value of less
than 7. It is evident that polymer treated soils failed under soaked conditions with a CBR value generally less
than 2. In practice, soils with CBR values < 3 are generally regarded as in need of special treatment as they are
not competent materials or suitable subgrade material. It is noted that studies conducted by [6] and [7] showed
that speciemens treated with other brands of polymers actually exhibit reduction in CBR values Some polymer
treated soils under unsoaked conditions exhibit a CBR value range between 3 and 7 which is equivalent to a G10
material in accordance to [8]. G10 materials are the least suitable as in-situ materials; however such materials can
be either mechanically or chemically treated to obtain permissible strengths.

.
Fig. 8: Images of CBR specimen post curing

208
20
CBR (%) Values
18
16 Unsoaked Condition
14
Soaked Condition
CBR (%)

12
10
8
6
4
2
0
Red soil + 6% Red soil + 6% Red soil + 6% (10% Kaolin: (80% Kaolin: Red soil + 6%
LFA (40:60) LFA (40:60) Polymer 90% Red soil) 20% Red soil) Polymer
(Sealed) + 6% Polymer + 6% Polymer (Sealed)
Samples
Fig. 9: Summary of CBR values for soaked and unsoaked conditions

4. CONCLUSIONS

This study was undertaken to investigate the influence of soil properties and polymer content and curing time on
the unconfined compressive strength and bearing capacity of stabilised soils. The conclusions are thus
 Polymer treated soils exhibit an increase in OMC and a decrease in MDD with increasing polymer
content. However, for high plastic (high bentonite content) treated soils, the OMC decreases and the
MDD increases with increasing polymer content.
 The UCS values of polymer treated red soils increased rapidly within the first 7 days and gradually
decreased thereafter. It is evident that high temperatures (+40ºC) may increase the early day strength but
gradually reduce the strength in the long-term. 8% polymer content is concluded as an optimum dosage
for treated red soil, however 6% is recommended as the most economical dosage for the obtained
strength in most soils of similar properties as the ones tested in this dissertation.
 It was observed that polymer treated specimens mobilized lower CBR values than lime fly ash treated
specimens. This is partially due to the fact that polymer treated specimens displayed a more ductile
behaviour while upon deformation , while lime fly ash treated specimens were brittle.
REFERENCES

[1] Van Veelen, M. & Visser, A.T. (2007). The performance of unpaved road material using soil stabilisers. Journal of the
South African institution of civil engineering= joernaal van die suid-afrikaanse instituut van siviele ingenieurswese, 49
(4):2-9.
[2] Rodway, B. (2001). Polymer stabilisation of clayey gravels. 20th ARRB Conference. Melbourne.
[3] Austroads (1992). Pavement Design. A Guide to the Structural Design of Road Pavements, Sydney 1992.
[4] Blane, C.H. (2013). Composition and provenance of quartzites of the mesoarchean witwatersrand supergroup, south
africa,
[5] Brink,A.B.A. (1979). Engineering geology of Southern Africa. V. 1 South Africa:Building Publication.
[6] Santoni RL, Tingle JS, Webster SL (2003) Stabilization of silty sand with non-traditional additives, transportation
research record 1787, TRB, National Research Council, Washington, DC, pp 33–41
[7] Little,D.N. & Nair,S. (2009). Recommended Practice for Stabilization of Subgrade Soils and Base MaterialsNational
Cooperative Highway Research Program, Transportation Research Board of the National Academies.
[8] Technical Recommendations for Highways (TRH 13). (1986). Cementitious Stabilizers in road construction. Pretoria:
South Africa.

209
210
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Efficient Geophysical Survey for Geotechnical Site Investigation

L’Efficacité d une Étude géophysique pour l’Investigation du site géotechnique

Fathi S. Elburas1, *, Bahia M. Ben Ghawar2 and Mokhtar T. Shakshuki3

1
Faculty of Sciences, Geophysics Department, University of Tripoli, Tripoli, Libya
2
Faculty of Engineering, Geological Engineering Department, University of Tripoli, Tripoli, Libya
3
Faculty of Sciences, Geophysics Department, University of Tripoli, Tripoli, Libya
(*) Corresponding author: abuoras@[Link]

Abstract: Geophysists , geologists and civil engineers have used geophysical survey methods for many decades.
Application of these methods provided very useful information concerning the physical properties and
parameters of the shallow subsurface layers. The objective of this paper is to discuss and clarify the best
integration of the geophysics in geotechnical site investigation, where seismic and electrical resistivity methods
applied, and provided as a practical example, at a construction site, near Tripoli. Drilling of 5 boreholes with
geotechnical drilling machine was performed. STP and borehole data are obtained. Soil laboratory tests also
conducted on the selected samples.
Geoelectrical cross – sections in perpendicular directions present four layers with clearly the subsurface are
characterized by lateral homogeneous in the soil properties, along each layer except in some test locations;
below 4 to 5 meters where expected cavity, above the water table.
Seismic refraction survey consisting of 6 lines was also conducted in the area. The results of their interpretation
yielded a first layer thickness ranging between 6 to 15 meters ( 10m average ), while the second layer thickness
was estimated to be in the range of 7 to 17 meters (13m average). All the above information is integrated to
establish good assessment for the construction site.

RESUME: Les Geophysiyiens, géologues et ingénieurs civils ont utilisé des méthodes de levés géophysiques
pendant de nombreuses décennies. L'application de ces méthodes a fourni des informations très utiles en ce qui
concerne les propriétés physiques et les paramètres des couches souterraines peu profondes. L'objectif de cet
article est de discuter et de clarifier la meilleure intégration de la géophysique en étude géotechnique, où les
méthodes de résistivité sismiques et électriques appliqués, et à condition que d'un exemple concret, à un chantier
de construction, près de Tripoli. Le forage de 5 trous avec une sondeuse a été réalisé. STP et données de forage
sont obtenus. Des essais de laboratoire de sol également menées sur les échantillons sélectionnés.
Géoélectriques - sections transversales dans des directions perpendiculaires présentent quatre couches avec
clairement la sous-surface sont caractérisés par latérale homogène dans les propriétés du sol, le long de chaque
couche sauf à certains endroits de test; ci-dessous 4-5 mètres où devraient cavité, au-dessus de la nappe
phréatique.
Enquête de sismique réfraction composé de six lignes a également été menée dans la région. Les résultats de leur
interprétation ont donné une première épaisseur de couche comprise entre 6 et 15 mètres (moyenne de 10 m),
tandis que la seconde épaisseur de couche a été estimé est dans la gamme de 7 à 17 mètres (moyenne de 13 m).
Toutes les informations ci-dessus sont intégrées pour établir une bonne évaluation pour le site de construction.

1 INTRODUCTION

The most widespread site investigation techniques, in geotechnical, are involved the drilling of holes in the
ground, sampling at discrete points, and in situ or laboratory testing. Given the relatively small sums of money
involved in ground investigations of this type, only a very small proportion of the volume of soil and rock, that
will affect construction, can be sampled and tested. Geophysical techniques offer the chance to overcome some
of the problems inherent in more conventional ground investigation techniques. Many methods exist with the
potential of providing profiles and sections (Elburas and Ben Ghawar, 2012) so that (for example) the ground
between boreholes can be checked to see whether ground conditions at the boreholes are representative of that
elsewhere.
Geotechnical geophysics is the application of geophysics to geotechnical engineering problems; such
investigations normally extend to a total depth of less than several hundred feet, or up to the aimed layer of
foundation, and often the most cost-effective and rapid means of obtaining subsurface information, especially
over large study areas. Geotechnical geophysics can be used to select borehole locations and can provide reliable
information about the nature and variability of the subsurface between existing boreholes. An isolated geologic

211
structure such as a limestone pinnacle might not be detected by a routine drilling program, as at Janzuer tested
site. Other advantages of geotechnical geophysics are related to site accessibility, portability, noninvasiveness,
and operator safety. Geophysical equipment can often be deployed beneath bridges and power lines, in heavily
forested areas, at contaminated sites, in urban areas, on steeply dipping slopes, in marshy terrain, on pavement or
rock, and in other areas that might not be easily accessible to drill rigs or cone penetration test (CPT) rigs.
Geophysical techniques also can be of help in locating cavities, backfilled mineshafts, and dissolution features in
carbonate rocks.
Most geophysics carried out during ground investigations is controlled by geologists or nongeophysicts;
generally; such people often have considerable expertise in the geophysical techniques they offer, but they have
very little idea of the contractual constraints within which civil and construction engineers must work. However;
that an understanding of geophysics is not beyond the capabilities of most civil engineers, who generally have a
good education in physics. This paper therefore, provides an engineer’s view of engineering geophysics. It
deliberately concentrates on identifying those situations where geophysical techniques are likely to be of most
help to engineers. With support from practical site application of two mean geophysics techniques (seismic
refraction and geoelectrical survey methods); that clearly discussed by Neil Anderson and Neil Croxton, 2008.

2 GEOLOGY OF THE TEST SITE

The Gargaresh Formation was introduced by Burollet (1960) as the “Gargresc Sandstone” which is quarried at
Qargarish (32◦ 49' N, 12◦ 55' E) west of Tripoli. El – Hinnawy and Cheshitev (1974) selectd the type section of
the Gargaresh Formation along the shore to the west of Janzur village (32◦ 49' N, 12◦ 55' E) where the formation
is approximately 14 m (46ft) in thickness, but the base is not exposed. They dated the exposed part as Late
Pleistocene – Holocene. This studied formation has textural characteristics of a relatively soft and porous
limestone called calcarenite. Calcarenite is a highly variable material consisting of grains of different sizes held
together by varying degrees of cementation, with both factors depending on the exact conditions of the material’s
formation. Therefore, this rock type may cause a surface decay and partial collapse of some parts of any
construction or structure. The tested site for this work is presented in Fig. 1((a) the location map and (b) is the
excavation of the investigated site; with clear presentation of the rock type).

Nort

Figure 1. Location of the geophysical field layout of seismic profiles, electrical resistivity tests, and borehole locations and
site condition.

3 GEOPHYSICS SURVEY METHODS

The more appropriate geophysical methods for geotechnical site investigation are:
1- Seismic Refraction Method:
Rapidly become more popular in geotechnical investigations because of its ability to give valuable information
on the stiffness variations in the ground. The seismic method relies on the differences in velocity of elastic or
seismic waves through different geological materials. An elastic wave is generated in the ground by impactive
force (a falling weight or hammer blow (see Fig. 2a) or explosive charge). The resulting ground motion is
detected at the surface by vibration detectors (geophones , Fig. 2b). Measurements of time intervals between the
generation of the wave and its reception at the geophones enable the velocity of the elastic wave through

212
different media in the ground to be determined (Fig. 2c). A seismic disturbance in elastically homogenous
ground, whether natural or artificially induced, will cause the propagation of four types of elastic wave, which
travel at different velocities. These waves are longitudinal waves (‘P’ waves), transverse or shear waves (‘S’
waves).where the maximum velocity of an S wave is about 70% of the P wave velocity through the same
medium, Rayleigh waves; these waves travel only along the ground surface. In general, Rayleigh waves travel
slower than P or S waves, love waves; these are surface waves which occur only when the surface layer has a low
P wave velocity with respect to the underlying layer (Elburas, 2013 and Richard Van Blaricom, 1980).
Soils generally comprise two phases (the soil skeleton and its interstitial water) and may have three phases
(soil, water and air). P-wave energy travels through both the skeleton and the pore fluid, whilst S-wave energy
travels only through the skeleton, because the pore fluid has no shear resistance. Traditionally, the geophysical
industry has made almost exclusive use of P waves. These are easy to detect, since they are the first arrivals on
the seismic record (Fig. 2d) (Elburas, 2013).

(a) (b) (c)

Figure 2 . (a) Geophone layout, (b) Energy source (hammer), (c) Seismic recorder

Profile Layer Depth Thick. Velocity

No. No (m) ( m) (m/sec)
1 8.81 8.81 771.73
SP1
2 23.34 14.52 1037.92
1 8.75 8.75 742.53
SP2
2 23.26 14.50 1009.91
1 7.93 7.932 525.28
SP3
2 23.24 15.30 903.40
1 13.90 13.90 309.49
SP4
2 23.34 9.43 711.29
1 15.96 15.96 535.13
SP5
2 23.14 7.17 1082.02
1 5.31 5.31 379.7
SP6
2 23.30 17.98 938.97

Figure 3 . (a) Record trace and (b) Quantitative results.

213
Practical example of seismic refraction investigations

Spread Configuration: The selected profiles were designed to cross the tested site and to obtain observations,
which might have affected the data quality. They were selected to cover the possible distance in the construction
site at different directions. (see Fig. 2(a))The seismic profiles consisted of 24 – geophones from first spread over
an interval distance of 2 meters. Shot points were stationed 0.5 meters from geophone position. The long profile
consists of 24 geophone spread over an interval distance of two meters (24 geophones x 2 m = 48 meter length).
Energy Source and Recording: The energy source used in this survey consisted of a hammer and a plate (Fig.
2.(b)). The energy is generated by striking the plate, which is situated at each end of the selected profile and at
fixed distance (0.50 meter) from the first geophone. The transmitted energy through the subsurface is detected by
geophones positioned on the good surface. While the preparation of laying out the cables, and connecting the
geophones was being carried out, the recording system was tested and adjusted.(Fig. 2.(c )) The amplifier and the
cable were tested for continuity to ensure that all the geophones were connected. When the recording unit was
ready to receive the signals, the plate was hit. Immediately after the strike, the record traces were recorded on the
computer screen and evaluated to decide if a second shot was required. If the second shot was not necessary, the
received signals were recorded and saved in doc file. See Fig. 3 (a).
Data Analysis: The first arrival times were picked for all data records, which were adjusted and filtered to
improve their quality due to the importance of the intensity and clarity of each individual trace on the picking of
first arrival time at each geophone is indicated on the corresponding trace by a pronounced rice in amplitude
above the background level after which the level of ground vibration decreases somewhat, but still remains much
more active than before. The records show that each arrival time is characterized by a break followed by a trough
and subsequent peak.

Picking procedures
a. First arrival times were generally chosen at first breaks on troughs. The troughs were easier to
correlate from trace to trace the records than the actual first breaks. They were also very important
for velocity measurements, because the seismic energy, which formed them, was the energy that
passed through the refractor boundary i.e. the calculated velocity would be representing the velocity
of actual layer.
b. The flattened troughs were chosen as first breaks. If troughs were squared off they were very
difficult to be assessed hence first breaks were picked. The picking of first breaks is used for depth
calculation because it occurred immediately as the seismic wave reaches the interfaces between two
different rock units. The interpretation techniques were based almost entirely on use of first arrival
times.

Summary of Seismic Results: six seismic refraction profiles conducted at the site, the summary of the
interpreted results are tabulated below in the associated table in Fig 3b. Two seismic refracted layers were
detected characterized with different velocities and thicknesses, from the attached seismic refracted signal
spreads; shown homogeneity along the subsurface, with some distortion at geophones (19 and 21) at seismic
profile 1, (SP1). This week signal could be due to cavity or voids and low compaction.

2. Electrical Resistivity and Conductivity Methods:

Rely on measuring subsurface variations of electrical current flow which are manifest by an increase or decrease
in electrical potential between two electrodes. This is represented in terms of electrical resistivity which may be
related to changes in rock or soil types. The electrical resistivity method is commonly used therefore to map
lateral and vertical changes in geological (or man-made) materials. The method may also use to; assess the
quality of rock/soil masses in engineering terms; determine the depth to the water table (normally in arid or semi-
arid areas); map the saline/fresh water interface in coastal regions; locate economic deposits of sand and gravel;
and locate buried features such as cavities, pipelines, clay-filled sink holes and buried channels (Elburas, 2012).
The electrical resistivity of a material is defined as the resistance offered by a unit cube of that material to the
flow of electrical current between two opposite faces. Most common rock forming minerals are insulators, with
the exception of metalliferous minerals which are usually good conductors. In general, therefore, rocks and soils
conduct electricity by electrolytic conduction within the water contained in their pores, fissures, or joints. It
follows that the conductivity of rocks and soils is largely dependent upon the amount of water present, the
conductivity of the water, and the manner in which the water is distributed within the material (i.e., the porosity,
degree of saturation, degree of cementation, and fracture state). These factors are related by Archie’s empirical
equation (Zohdy and Robert, 1990).

214
4 PRACTICAL EXAMPLE OF ELECTRICAL RESISTIVITY MEASUREMENTS

Schlumberger configuration array of vertical electrical sounding (8 VES’s), were conducted at selected locations.
See Fig. 1a. This method is based on a direct current generated in to the ground, through two Stan steel
electrodes (A and B). The voltage in the ground was measured between two other Stan steel electrodes (M and
N). The electrical resistivity was measured by using the updated digital CA6471 Earth Resistivity Meter survey
instrument see Fig. 4a. The apparent resistivity is given by (Dobrin and Savit. 1988, Telford, W.M. and others.
1976.):

R = π/4{ (AB²/MN) (ΔV/I)} (1)

Data Interpretation
IX1D computer soft ware of a 1-D Direct current (DC) resistivity sounding inversion program, was used in the
interpretation stage, the obtained quantitative results; e.g. VES location (Loc), Number of layers (L. No),
Resistivity (Ω-m) (Res.), Thickness in (m) (Thic), and depth (m) (Dep); as an example; presented in Fig. 4b and
4c. This processing also could be seen at Tang, W. M., 1986.
The obtained results for the site, are clearly shown that the subsurface is characterized by vertical heterogeneity
in the soil properties, that represented by different electrical resistivity values. Also, the processes of the
geological deposition environment are reflected on the variation of the thickness of the subsurface layers. Two
geo-electrical sections constructed in North south and East West Directions showing the electrical resistivity
variations in four layers, an example presented on Fig. 4d.
Soil Stratigraphic Cross – Sections:Based on the quantitative interpretation of the measured resistivity data; 2
soil stratigraphic cross sections are constructed along East West direction, and soil stratigraphic cross sections
along North South direction, are constructed as in Fig. 4e. this approach discussed clearly by Elburas and Ben
Ghawar , 2014.

RJZ 2
T echno Systems Engineering L td.
1000 1

L. Res. Thic. Dep.

Loc.
No. Ω-m (m) (m)
A pparent R esistivity (ohm -m )

D epth (m )

100 10
RJZ1 1 91.4 2.50 2.50
2 240.0 7.50 10.0
3 76.7 2.00 12.0
4 38.3 23.0 35.0
10 100

10 100 1000
1 10 100
Spacing (m) R esistivity (ohm -m )

Figure 4.(a)Field layout, (b) field curve (Left) and estimated resistivity versus depth (right) for VES RJZ1,(c) quantitative
interpretation results, (d) geoelectrical cross – section (West East) and (e) soil stratigraphic cross – section (North South).

5 PRACTICAL EXAMPLE DRILLING INVESTIGATION

The Drilling investigation at the site was conducted by drilling of 5 boreholes at the site 35 meters depths . The
drilling was performed using a geotechnical rig as in Fig. 5(a) equipped with augers of 1.5 m in length. Once the
drilling approached the desired depth, standard penetration test (SPT) was performed using the split spoon
sampler (Fig. 5 (b)).and soil samples were recovered for laboratory test program. The main characters and
properties of the soil layers at the site have been determined based on the visual description at the field. This is
combined with results of index properties. The soil layers as indicated by the drilling excavation are classified
for every location.
Standard Penetration Test Results; the raw data of the Standard Penetration test ( N field ) obtained from the in
situ measurements (see Fig. 5) has been corrected taking in to account the effects of overburden pressure, length

215
of drilling rod, the soil type and the sampler and the borehole diameter. The procedure adopted for the correction
presented in Bowles (1996).
The summary of the drilling reports could be summarized in the following table;

No. of No of Depth in Average SPT Water table

Soil type
boreholes layers meters blows meter

Silty sand with

5 4 35 5 to 50 12
various density

BH5 BH4 BH1 BH2 BH3 averag

Depth SPT SPT SPT SPT SPT SPT
[m] blows blows blows blows blows blows
0.75 25 12 18 24 12
1.75 17 12 50 41 12
5.75 13 20 33 15 13 13
6.75 13 21 20 17 13 13
10.75 11 49 11 24 19 11
11.75 10 50 7 22 24 7
18.75 50 32 44 46 28 28
19.75 44 29 49 44 26 26
20.75 50 33 25 50 31 25
21.75 49 36 47 36 39 36
22.75 45 48 42 39 27 27
23.75 34 50 50 35 28 28
30.75 32 31 33 32 26 26
31.75 33 33 32 32 29 29
Figure 5. (a) geotechnical rig,(b&c) Standard Penetration test (STP) field measurements and results.

6 DISCUSSION, CONCLUSION AND RECOMMENDATIONS

Discussion: Generally; the physicists in charge of the exploration anticipated and promised
satisfactory results. That could be built on practical and theoretical experience and knowing the
objective of the geotechnical and civil engineers at each site investigations also the economy of the
survey if compared with the coverage area and data obtained during the investigation.
The two practical examples used in this paper are to prove and show how we could use them in any
geotechnical investigation and could be used as guide steps for the geotechnical and civil engineers.

Conclusion: The results of the scheduled strategy were achieved, the site investigations were
performed, and the soil resistivity tests, using Schlumberger vertical electrical sounding method, were
conducted, at 8 locations, the results of each test were obtained. The selected test locations; where
sited along the area where more information are needed, by dividing the construction area in to east
west sections and north south sections.
The interpretation of the electrical resistivity using especial computer software provided valuable
results, the electrical resistivity, thickness and depths of four geoelectrical layers are shown; that the
subsurface conditions along the construction site, has vertical heterogeneity, that produced by the
depositional environments. 2 geoelectrical section and 2 stratigraphic cross sections were constructed
to show the thickness and depths variation, and any voids or cavity within 35 meter depths in deferent
directions.
The obtained results of soil electrical resistivity (ohm-m) investigation conducted at the
construction site are important for any earthing design.
Six seismic refraction profiles conducted at the site, the summary of the interpreted results are
tabulated as shown in (Table (b) of FIG. 3). Two seismic refracted layers were detected characterized
with different velocities and thicknesses, from the attached seismic refracted signal spreads; shown
homogeneity along the subsurface, with some distortion at geophones (19 and 21) at seismic profile
1, (SP1), near BH2. This week signal could be due to cavity or voids and/or low compaction material.
The results of the drilling investigation at 5 boreholes; shown that the subsurface soil of the
construction site characterized with 4 layers of silty sand various in density, consolidation, and
moisture content. The water table measured at the 5 boreholes and the static water level at 12 meter

216
depth. BH2 at average depths between (3.75 to 4.75 meters); where sudden changes in the SPT (from
50 to 12) values exist.

Recommendations; from the obtained results the site generally consists of four subsurface layers
various in thickness, and physical properties. Below 4 to 5 meters depth, around and at BH2 existing
low consolidation material with voids; this voids openings, ranges between (2m to 3 cm). Therefore;
at this location of surface area; about (6m x 6m) should be use back fill material and compacted
until 98% degree of compaction taken as layer thickens 35 cm. (Bowles, 1996).

REFERENCES

Anketell, J . M. and Ghellali, S. M . (1991). A palaeogeological map of the pre- Tertiary surface in the Jifarah plain and its
implication to the structural history of northern Libya, in: Salem, M. J., Sbeta, A. M. and Bakbak, M. R. (eds.), The
Geology of Libya, v. VI, p. 2381-2406.
Burollet, P. F. (1960) - Lexique stratigraphique international, Vol. IV, Afrique, fascicule IVa, Libye : 1-62.
Bowles, J.E., 1996, Foundation Analysis and Design, 5th edn, McGraw-Hill, New York.
Dobrin and Savit. 1988. Introduction to geophysical prospecting. McGraw Hill.
EL HINNAWY, M. & G. CHESHITEV (1974) - Geologic Map of Libya, 1 : 250ʼ000, Sheet : Tarabulus, NI 33-31,
Explanatory Booklet. Tripoli, Industrial Research Centre.
Elburas F. S. 2012. Geoelectrical Investigation at Ain Al-Faras Spring Site of Gadames City, Libya, Hydrol Current Res
2012, S12, HYCR, an open access journal,
Elburas F. S. 2013. Seismic Refraction Investigation at Ain Al Faras of Gadames City, Libya. Annals Of Valahia University
of Targoviste. Geographical Series.
Elburas F. S. and Ben Ghawar B. M. 2013. Estimation of Geotechnical Parameters and Properties From Geoelectrical Survey
Method, Second Symposium on the Geological Resources in the Tethys Realm, Aswan University,
Elburas F. S. and Ben Ghawar B. M., 2014. Mapping Abushaiba Dyke Structure by Geoelectrical Resistivity Method,
Gharian, Libya. The thirteen Arabic international conf. for mining resources. Marrakesh
Neil Anderson and Neil Croxton, 2008. Introduction to Geotechnical Geophysics. Transportation Research Board,
Washington, DC 20001,[Link]
Richard Van Blaricom, 1980, Practical Geophysics for the Exploration Geologist, Northwest Min. Ass., WA99201,
Tang, W. M., 1986. A method of reducing the lateral effects in Schlumberger geo-sounding Geoexplo. Vol. 24.
Telford, W.M. and others. 1976. Applied geophysics. Combridge university press.
Zohdy, A. R. and Robert, J. B., 1990. Schlumberger sounding near Medicine Lake, California. Geophy. Vol. 55, No. 8, p.
956-964.

217
218
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Determination of Jet Grout Column Diameter Using Seismic Crosshole

Methods

Détermination du diamètre de la colonne du Jet Grouting à l'aide des méthodes de sismiques

crosshole

Th. Fechner(1,*), S. Mackens(1), J.C.G. Guerreros(2) & E. Niederleithinger(2)

1
Geotomographie GmbH, Am Tonnenberg 18, 56567 Neuwied, Germany, info@[Link]
2
BAM Federal Institute for Materials Research and Testing, Unter den Eichen 87, 12205Berlin, Germany
(*) Corresponding author

ABSTRACT: The jet grouting technique creates columns of grouted soil by eroding and mixing the in-situ soil
with a thin cement suspension. The columns are generated by rotating and raising the drill stem while a cement
suspension is injected under high pressure. The target column diameter to be reached and the resulting material
strength can be predicted just approximately. Both are dependent on various process parameters as well as the
subsurface soil material and its inhomogeneity. In this paper a seismic method is proposed to determine the
diameter of a jet grout column. The method uses crosshole seismic applied between boreholes drilled outside the
column. The testing is carried out before and after jet grouting. Seismic P- and S-waves are generated at different
depths between the boreholes where the related arrival traveltimes are going to be determined for the different
states. The traveltime differences of different states can be directly related to the influence of the jet grout
injection and used to calculate the diameter of the column at specified depths. Downhole measurements can be
used to improve the results. The new method is going to be evaluated by field trials using newly developed
instruments. It is meant to be a quality control tool to visualize columns under complex site conditions such as
high heterogeneity or high water table.

RÉSUMÉ La technique du jet grouting crée des colonnes de sol injecté par l'érosion et le mélange du sol in situ
avec du ciment. Les colonnes sont générées en faisant tourner la tige de forage et l'élever tout en injectant du
ciment sous haute pression. Le diamètre de la colonne cible à atteindre et la résistance du matériau qui en résulte
peut être seulement prédite. Les deux dépendent de divers paramètres du procédé, ainsi que du matériau de la
couche inférieure du sol et de son hétérogénéité. Nous proposons une nouvelle méthode sismique pour
déterminer le diamètre d'une colonne du jet grouting. La méthode utilise la sismique crosshole appliquée entre
forages extérieures à la colonne. Le test est effectué avant et après jet grouting. Des ondes sismiques P et S sont
générées à différentes profondeurs entre les forages où les temps d'arrivée de trajet sismiques connexes vont être
déterminés pour les différents états. La différence de temps de trajet peut être directement liée à l'influence de
l'injection de jet grouting et utilisée pour calculer le diamètre de la colonne à des profondeurs spécifiques. Des
mesures de fond de puits peuvent être utilisées pour améliorer les résultats. La nouvelle méthode va être évaluée
dans des essais sur le terrain à l'aide de l'instrumentation nouvellement développé. Il est destiné à être un outil de
contrôle de la qualité de visualisation des colonnes dans des conditions de sites complexes tels que
l'hétérogénéité élevée ou nappe d'eau élevée.

1 INTRODUCTION

The use of geophysical methods is a supplementation to geotechnical investigations for a spatial continuous
characterization of the subsurface. Seismic methods are especially suitable for the characterization of soil
dynamic properties of construction sites. Within the last years several seismic methods have been introduced to
measure soil dynamic parameters. In this paper borehole seismic methods and their applications to engineering
tasks are presented (Nolet, 1987, Lines and Newrick, 2004).
The measurement of physical properties of in-situ mixed soils through jet grouting and their geometrical
shape after hardening is an important topic in geotechnical engineering. The jet grouting process constructs
panels, full columns or anything in between (partial columns) with designed strength and permeability. Jet
grouting can be used to support existing foundations, construct excavation support walls and slabs to seal the
bottom of planned excavation sites. Anyhow, the target column diameter to be reached and the resulting material
strength can be predicted just approximately. Both are dependent on various process parameters as well as the
subsurface soil material and its inhomogeneity.

219
In order to identify the diameter of grouted columns more precisely a seismic method has been developed.
The method requires boreholes to be drilled outside the column in order to perform seismic crosshole testing.
Seismic crosshole methods are seismic tomography and S-wave crosshole testing according to ASTM D4428
standard (ASTM D4428, 2014). To evaluate the new seismic method several columns were installed at the BAM
- Federal Institute for Materials Research and Testing test-site close to Berlin, Germany. The estimated diameter
of the columns is supposed to be about 1.20 m. Seismic measurements were carried out before the grouting
process and after a setting time of 1, 7 and 28 days. First results are going to be presented.

2. DESCRIPTION OF THE TEST-SITE

A test site has been prepared to allow the installation of three jet grout columns under controlled conditions (see
figure 1). The investigation program was performed at a dedicated area on the BAM test site south of Berlin
(town Horstwalde about 60 km south of Berlin). The test site is located within a main testing area utilized by
BAM for technical safety issues. The main area covers almost 12 km2 but most parts are forest only. The test site
has formerly been used as a military testing area and training facility since 1928. In 2003 the BAM’s division 8.2
which is focused on non-destructive testing in civil engineering has started to set up a test and validation facility
for various investigation purposes and techniques (Niederleithinger et al., 2009).
Geologically, the site belongs to the northern German Basin. It consists of various sediments with a thickness
of several thousand meters affected by salt tectonics. The site itself is formed by a glacial valley (“Baruther
Urstromtal”) and the near surface is dominated by post glacial sediments of the Nuthe-Nieplitz lowland. The
main part of the site (including the test site discussed in this paper) consists mainly of sandy layers of varying
grain size and admixtures of silt and organic material. Peat lenses are known to exist locally. The groundwater
table is about 3 ± 1 m below surface and varies seasonally. The southern part of the area is covered by
palaeodunes with a height up to 15 m whilst the northeast encloses some swamp areas.

A
B

Figure 1: Aerial image of the test site at Horstwalde (courtesy BAM 2006)
A: Grouting site. B: NDT-CE test and validation center.

At the test site four boreholes were drilled and plastic cased (3 inch) with a center borehole and three boreholes
in 120° order around (see figure 2). The outer borehole annuli were filled with sand to enable a sufficient soil
contact. Distance between the boreholes and the center borehole is about 3 m. Borehole depth is about 15 m. The
deviations of all boreholes were measured and used for the determination of the position of the sensors deployed
in the boreholes. Seismic measurements as described in chapter 3 were carried out late June 2014 to acquire field
data for the natural soil state, i.e. reference state.
The jet grouting was carried out on 15th /16th September 2014. Each of the column was built under different
pressure and mixing conditions from about 3 to 10 m depth. One column was built according to the standard
grouting procedure in order to produce an equal diameter column. The other two columns are supposed to show
irregularities due to different mixing and pressure regimes. The columns were placed half way from the center
borehole BH1 to the three outer boreholes.

220
BH 3

Col 2 N
BH 1

Col 3
Col 1
BH 2

BH 4
Figure 2: Sketch of the borehole locations. Seismic borehole sections are shown in red.

3. SEISMIC METHODS AND DETERMINATION OF THE COLUMN DIAMETER

Seismic methods are in principle able to image grout injections or jet grout columns (Majer, 1989; Ramirez and
Daily, 1990; Axtell and Stark, 2008). Several prerequisites have to be fulfilled to achieve such kind of resolution,
i.e. a “before/after” seismic survey and the knowledge of the sound velocity of the grout injection material itself
as well as an excellent high-frequency seismic signal transmission. If one of these conditions cannot be fulfilled
the desired resolution will decrease as one has to estimate average soil parameters or has to use methods with
lower resolution.
Four seismic testing methods have been applied at the test site to determine their sensitivity to the grout
injection at site. The methods are seismic tomography, crosshole testing, downhole testing and surface wave
seismic (MASW). Seismic measurements were carried out before the grouting process and after a setting time of
1, 7 and 28 days. Testing was carried out from 1 to 14 m below ground.
Using borehole seismic measurements before grouting and after grouting the diameter can be calculated. The
diameter determination is based on the assumption that the travel time of the seismic wave propagating from one
borehole to the other is the sum of the travel time of the wave traveling through the column and the travel time
for crossing the natural soil on both sides of the column. The calculation requires the knowledge of the distance
between the boreholes dBH, the velocity of the natural soil without column vsoil, the average velocity with column
in place vmean and the velocity of the grout column itself vcol. The distance between the boreholes is known. The
velocity vsoil is the taken from the measurement of the reference state before grouting. The average velocity can
be measured after grouting and the wave velocity of the column vcol can be measured using a downhole test. For
the downhole test a pipe has to be installed shortly after grouting in the middle of the column and wave velocity
can be measured along the borehole following the downhole test procedure (ASTM D 7400, 2008). Thus the
accuracy of the diameter determination only depends on the accuracy of the velocity determination.
In this paper the principles of the borehole seismic methods, i.e. seismic tomography and crosshole testing,
are described.

a. Seismic Tomography

Seismic tomography is usually conducted using the properties of the induced P-waves (compressional waves) in
order to obtain structural information with high resolution between boreholes. The determined parameter, the
rock P-wave velocity, depends mainly on porosity and compressional bulk modulus. The combination of seismic
tomography data together with borehole logging and coring information can lead to an almost complete
subsurface characterization.
Seismic waves are generated by an impulsive source in one borehole. The propagation of the seismic waves
is influenced by the varying rock properties between the boreholes leading to slower or faster wave arrivals in the
second borehole to receive the seismic signals. Hydrophones built up in a vertical multichannel string with 24
sensors are used in the second borehole. The signal data quality was improved by signal stacking. During seismic

221
data processing first arrival travel times of the waves related to the P-wave are picked for each source-
hydrophone pair. Using an iterative numerical inversion procedure a seismic velocity distribution between the
two boreholes can be calculated based on the travel times and known source and hydrophone locations.
During seismic field testing the spacing between each shot position in a borehole was 0,5 m. The hydrophone
interval was 0,5 m. Signals were stacked 3-4 times. An impulse source type BIS-SH (Geotomographie brand)
was used to generate high frequency seismic signals. This source produces P-waves but also S-waves and was
also used to carry out the later crosshole testing (please refer to chapter 3.2). A hydrophone string type BHC4
(Geotomographie brand) was used to receive the signals. Data were recorded by a DMT Summit seismograph.

b. Crosshole Testing

The crosshole testing procedure is described in the ASTM D4428 paper (ASTM D4428, 2014). The seismic
crosshole test measures the P- and S-wave velocities between boreholes. At least two boreholes are necessary.
The first borehole is used for installation of a seismic source. As the focus is on the determination of the S-wave
velocity the preferences are on a mechanical source which is able to produce S-wave impulses with reversible
polarity. In the second hole a receiver is installed at the same depth as the source. The measurement is triggered
at the source and the arrival time of S- respectively P-wave is obtained by visual interpretation of the signal from
the receiver borehole. By testing at various depths a velocity profile can be drawn.
For the seismic field tests the spacing between each shot position in a borehole was 0,5m. A borehole
geophone was placed in the second borehole at same depth and pneumatically clamped to the borehole wall. S-
wave signals of SH-type were generated by the borehole source BIS-SH and data were recorded using a
seismograph.

4. FIRST RESULTS

Seismic data are available for the reference state, i.e. the measurement before grouting, as well as for times of 1
day (1+) and 7 days (7+) after grouting. Surveying at a time of about one month after grouting was carried out on
15th/16th October 2014. Preliminary results are available showing the grouting effect on the seismic traveltimes
and the seismic signals.
Figure 3 shows a comparison of signals obtained by seismic tomography. Seismic traces in black color
belong to the reference state before grouting and seismic traces in red color correspond to the 1+ survey. Two
features can be observed. First, the signal amplitude and signal frequency of the 1+ survey are lower compared to
the reference state. Second, the arrival times seem to be same or even slightly later for the 1+ in comparison to
the reference state. It is most likely that the seismic signals measured shortly after grout injection are strongly
attenuated by the suspension whereas the effect on traveltimes is comparable small as soil and suspension may
have the same velocity speed.

Figure 3: Comparison of seismic signals (slice 2-1, 7m depth) between

reference state (black) and the 1+ survey (red)

222
Considering the same setup for the 7+ survey there is a completely different situation. Figure 4 shows a
comparison of signals obtained by seismic tomography. Seismic traces in black color belong to the reference
state before grouting and seismic traces in red color correspond to the 7+ survey. Signal amplitudes and
frequency of the 7+ survey seem to be recovered and more high frequent compared to the 1+ survey. Further, the
arrival times of the 7+ survey are now much faster in comparison to the reference state. Obviously, the hardening
of the grout suspension caused this effect.
.

Figure 4: Comparison of seismic signals (slice 2-1, 7m depth) between

reference state (black) and the 7+ survey (red)

The crosshole test data have been processed for the reference state so far. Arrival times for the P-wave as well as
for S-wave have been determined. Figure 5 shows the arrival times of the S-wave for all three crosshole slices on
the right side and the corresponding velocities on the left side for the reference state. It can be observed that the
variability of the soil properties above about 6 m is higher compared to variations below 6 m. Results for the
wave speed after grouting are pending.

Figure 5: S-wave crosshole arrival times and wave velocities for

slices 1-2 (blue), 1-3 (red) and 1-4 (green) for the reference state

223
3. CONCLUSIONS

A study of the use of seismic borehole surveys carried out at a jet grouting site has been presented. At the site
three jet grout columns have been grouted using different pressure and mixing regimes. Boreholes were installed
to image the grout columns using seismic borehole methods as well as surface seismic methods. First results from
seismic borehole tomography and crosshole testing indicate a clear influence of the grout suspension on the
propagation of the seismic signals. At the time of writing in October 2014 a third seismic survey was carried out
to study the time dependent effect of grout hardening. The proposed seismic methods are supposed to be used for
determination of the diameter of jet grout columns.

ACKNOWLEDGEMENT

The results presented in this work were obtained within the frame of the ZIM-project KF3163801AT3. The
financial support of the German Federal Ministry for Economic Affairs and Energy is gratefully acknowledged.

REFERENCES

ASTM D7400 (2008) Standard Test Methods for Downhole Seismic Testing. American Society for Testing and Materials
(ASTM).
ASTM D4428/ D4428M-14 (2014) Standard Test Methods for Crosshole Seismic Testing. American Society for Testing and
Materials (ASTM)
Axtell, P. and T.D. Stark (2008) Increase in Shear Modulus by Soil Mix and Jet Grout Methods. DFI Journal, Deep
Foundations Institute, Vol. 2, No. 1, November, pp. 11-22.
BAM (2011) Testgelände Technische Sicherheit (BAMTTS, test site technical safety) [Link]
Lines, L.R. and Newrick, R.T. (2004) Fundamentals of Geophysical Interpretation. SEG, Geophysical Monograph Series.
No13.
Niederleithinger, E., Wiggenhauser, H. and Taffe. A. (2009) The NDT-CE test and validation center in Horstwalde. In
NDTCE `09–7th International symposium on nondestructive testing in civil engineering (Proceedings), 1–6. Nantes,
France: Laboratoire Central des Ponts et Chaussées, Juni 30.
Niederleithinger, E., Baessler, M., Herten, M., Rumpf, M. and Tronicke, J. (2013): Geotechnical and geophysical
characterization of a pile test site in post-glacial soil, Geotechnical and Geophysical Site Characterization 4 – Coutinho
& Mayne (eds), Taylor & Francis Group, London, ISBN 978-0-415-62136-6, pp 459 – 464.
Nolet, G. (1987) Seismic Tomography: With Applications in Global Seismology and Exploration Geophysics (Modern
Approaches in Geophysics). Springer; 1987 edition (June 30, 1987)
Majer, E.L. (1989) The Application of High Frequency Seismic Monitoring Methods for the Mapping of Grout Injections.
Int. J. Roch Mech. Min. Sci.&Geomech. Vol. 26, No. 3/4, pp. 249-256.
Ramirez, A.L. and Daily, W.D. (1990) Geotomography of grout penetration in fractured granite. Int. J. Roch Mech. Min.
Sci.&Geomech. Vol. 27, No. 1, pp. 57-63.

224
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

An Experimental Study on the Identification and Classification of Collapse

Potential of Some Soils in Sudan

Etudes Experimentales sur l'Identification et Classification du Potentiel d’Effondrement de

certains Sols au Soudan

M.A. Osman1 E.M. Ali Ahmed2

1Engineering Services & Design, Khartoum, Sudan,
[Link]@[Link]

ABSTRACT: In Sudan the problem of collapsible soil is not new. Collapse problems was reported in northern
Sudan since 1976 Omer. Recently collapse problem was encountered in Omdurman city, the national capital of
the country and in some areas in central, Northern and Western Sudan. This study summarizes a research
conducted on this subject. Both field and laboratory tests were conducted to evaluate the collapse potential at
different depths. Disturbed and undisturbed soil samples were obtained and tested in the laboratory. Plate load
test, insitu density and natural moisture content were obtained in the field. Various collapse criteria reported by
different authors were used to access the collapse potential, based on the laboratory and field tests. Some of these
criteria were successfully used in identifying the collapse potential.

RÉSUMÉ : Au Soudan, le problème du sol pliant est récemment rencontré à Omdurman, la capital nationale du
pays et dans certaines régions du nord et de l'ouest. Cette etude résume celles menées sur le terrain et en
laboratoire pour évaluer le potentiel d’effondrement àux différentes profondeurs. Échantillons de sol perturbés et
non perturbés ont été obtenus et testés en laboratoire. Test de Charge sur Plaque, measures de densité in situ et
d’humidité naturelle ont été obtenues pour le domaine. Critères d’effondrement divers rapportés par différents
auteurs ont été utilisés pour accéder au potentiel de l’effondrement, basés sur les tests de laboratoire et sur le
terrain. Certains de ces critères ont été utilisés avec succès pour identifier le potentiel de l’effondrement.
KEYWORDS: Collapse, Potential, Density, Plate Load, Oedometer

1. INTRODUCTION

Collapsible soils are unsaturated soils undergoing radical particle rearrangement and considerable volume loss
upon wetting with or without additional loading. Such soils are creating an increasing problem to different civil
engineering structures worldwide. In Sudan the problem of collapsible soils are noted in urban as well as rural
areas. The collapse problem was reported by Omer (1981) in Berber town in northern State in Sudan.
Recently Omdurman city, which is part of Sudan capital, has experienced problems associated with collapse of
foundationsoils. Buildings and infrastructure both showed different kinds of distress. Auger holes, test pits as
well as block soil samples were obtained. Soil tests were conducted in the laboratory as well as in the field.
Collapse potential was assessed using direct tests such as plate load and oedometer laboratory tests.
Collapsibility was estimated from the physical properties of the soil and the previously reported collapse criteria.
The results showed that of the reported collapse criteria presented by different authors about four methods were
identified as being suitable for theindirect evaluation of collapse potential of soils in Sudan.

2. LITERATURE REVIEW

In Sudan the problem of collapsible soil was studied by Omer m. o.(1981). His work was focused on the loess
soils in Berber town in the northern part of the country. The soils encountered are mainly clayey soils of low
plasticity. The studies published worldwide on collapsible soils covered the origin distribution. identification and
estimation of degree of collapsibility on such soils. Several methods for the indirect evaluation of collapsible
soils was reviewed by (Das B.M. 2001, Al-Rawas A.A. 2000).
The direct evaluation of collapse potential could be achieved by performing double or single oedometer test in
the laboratory (Jennings J. & Knight K. 1957) or by performing plate load test in the field. The field and
laboratory cone penetrometers were also used for the identification of collapsible soils (Khelifa et al 2010).
Other tests based on the soil suction were reported in the literature as direct methods for the evaluation of the
collapse potential (Fredlund and Rahardjo 1993). Rogers (1994), presented most important characteristics of
collapsible soils. Clemence etal (1981) studied other types of soils that can present collapsible behavior upon
wetting. These include volcanic tuff, gypsum, loose sands cemented by salts, dispersive clays, and sodium rich
montmorillonite clays.

225
3. FIELD SURVEYS:

Omdurman is the largest city of the three cities comprising the greater Khartoum capital of Sudan. It lies on the
west bank of the White Nile and the River Nile. The soils in Omdurman city geologically belong to what is
called Omdurman formation. This formation is a basement complex that out crops all over the region lying west
of White Nile and River Nile Fig. (1). The sedimentary sequence of the Omdurman formation which comprises
sandstones, mudstone and conglomerates belong to late cretaceous (Albian - Cenomanian) age, (Schrank and
Awad 1990). The climate is semi arid with low precipitation <150mm per annum with temperature about 45°C
during summer.
The civil engineering structures built earlier in old districts of Omdurman city were founded on the traditional
Nubian formation, generally showed good performance . The city expanded both horizontally and vertically and
the buildings covered areas with problematic soils, such as collapsible and expansive soils. Pronounced
distresses were noted in different newly built districts. The affected structures included residential buildings,
mosques, schools, and health centers. Selected areas for investigations are shown in the satellite map of
Omdurman Fig. (1).

Fig. (1): Location of Selected Sites in Omdurman City.

Two pits each 3 m depth, designated as TP (1) & TP (2) were excavated in the north side of the city. Test pit
TP (3) of 3 m depth was excavated in the Khartoum new Airport site, located in the Southern part of Omdurman.
Soil samples were taken. Two boreholes 10m depth and one test pit were realized at area 4 lying in the middle
and west side of the city. This is designated as TP (4). The fifth location is lying in the middle of the city and
designated as area TP (5).

4. GEOTECHNICAL PROPERTIES OF THE SOILS

Laboratory and field tests conducted included particle size distribution test, compaction test, Atterberg limits,
specific gravity, one dimensional oedometer test, natural moisture content, field density and plate load tests.
The test pits TP (1) & TP (2) in area (1) are dominated by clayey Gravel with gravel content about 72%.
TP (3) and area 4 are dominated by clayey sand. Fig (2) depicts the variation of the coarse material with depth.
The variation of liquid limit and plasticity index with depth are shown in Fig. (3).

226
Fig. (2): %Age of Fine Soil Fraction and Coarse Fig. (3): Liquid Limit & Plasticity Index vs. Depth
Fraction in TP (1) & TP (2) & TP (3) vs. Depth

5. EVALUATION OF COLLAPSE BY DIRECT TESTS

Plate load tests were conducted near TP (1) in the north side of the city and at TP (5) near the middle of the city
using plate 500*500 mm size. The test results showed the average settlement of the plate under 262 kN/m2 load
is 1.5 mm before wetting, and 8.75 mm after wetting.
The one dimensional single oedometer test was performed on undisturbed soil samples taken from TP No.3 at
four depths.
The test was performed according to (ASTM D5333 – 96, 2000). At the maximum pressure (200 KPa) the
sample was flooded for 24 hours, and the deformation versus time was monitored and recorded. The typical test
results are shown in Fig (4). Summary of the oedometer tests are shown in Table (1). The field tests as well as
the oedometer tests demonstrated that the soils in the five sites in Omdurman are collapsible with varying
degrees of collapse potential.

Fig. (4): Typical Oedometer Test

227
6. ANALYSIS & DISCUSSIONS OF THE RESULTS

The collapse potential was calculated from single oedometer test results and the severity of collapse
corresponding to collapse potential was determined according to (ASTM D 5333 – 96, 2000). The relevant
results are shown in Table (1).
The collapse potential is found to be ranging from 4.2% to 8.3 % and the severity of collapse is moderate.

Table (1): Summary of Single Oedometer Test Results

Depth (m)
Parameter
1.0-1.5 1.5-2.0 2.0-2.5 2.5-3.0
Moisture Content
6.32 7.59 5.75 9.65
(%)
Dry Density
1.692 1.655 1.673 1.707
(gm/cm3)
Collapse
6.1 4.2 6.0 8.3
Potential (%)
Severity of Moderately Moderately
Moderate Moderate
Collapse Severe Severe
Moisture Content
at End of Test 10.36 10.88 11.56 13.43
(%)

7. INDIRECT EVALUATION OF THE COLLAPSE POTENTIAL

The relationship between the physical properties of the soil and the collapsibility of the soil was investigated. In
Fig (5) the relationship between % age finer than sieve 200 is drawn against the dry density (Mock and Pawlak
1983). Almost all of the data lies on the collapse side of the curve. Hence this relationship can give indication of
the collapsibility of Omdurman soil. The Liquid limit versus the dry density advanced by (Holtz and Hilf 1961)
is shown in Fig. 6. It is clear from this figure, this relationship does not hold for the soils in Omdurman.

Fig. (5): Relationship between Dry Density vs. Fine Passing No.200 & C.P.

228
Fig. (6): Relationship between Dry Density vs. L.L & C.P.

An attempt was made to examine the proposed criteria for prediction of collapsibility of soil advanced by
previous investigators using the physical soil properties of soils at the five sites. The results of using these
criteria are shown in Table (2). The criteria that predicted collapse of Omdurman soils included (Priklonski
1952, Handy 1973, Gibbs and Bara 1962), and to lesser degree (Zur and Wiseman 1973).
Failure of most of the reported criteria to predict collapse for Omdurman soil probably due to different soil
types mineralogy and placement conditions. The Omdurman soil contains an appreciable amount of course
material while most of the reported criteria were developed in areas dominated by silt and loess soils.
Additional to this the investigated soils from Omdurman districts have appreciable amount of gravel causing it
difficult to obtain undisturbed samples thus making it difficult for laboratory testing.

Table (2): Applicability of Different Method to Omdurman Soil

Applicability of the Method

T.P No. T.P(3) T.P(4)
Collapse criteria

Sample Depth

Applicability
(m)

Matching
0.5 1.0 1.5 2.0 0.5 1.0 1.5

Abelev,1948 N.C N.C N.C N.C N.C N.C N.C 0/7

Denisov, 1951 N.C N.C N.C N.C N.C N.C N.C 0/7
Priklonski, 1952 N.A C C C C C C 6/7
Clevenger, 1958 N.C N.C N.C N.C N.C N.C N.C 0/7
Gibbs and Bara, 1962 C C C C C C C 7/7
Soviet Building Code,1962 N.C N.C N.C N.C N.C N.C N.C 0/7
Feda 1964 , 1966 N.C N.C N.C N.C N.A N.A N.A 0/7
Handy, 1973 C C C C N.A N.A N.A 4/7
Zur and Wiseman 1973 C N.C N.C N.C N.C N.C N.C 1/7

C.P - Collapsible Potential C - Collapsible Soil

N.C - Non Collapsible N.A - Not Applicable

8. CONCLUSION

Omdurman city and its suburbs is characterized by semi-arid climate. The geologically belongs to what is called
Omdurman formation comprising sedimentary formations. The soils have appreciable amount of fine to medium
gravel with low silt and clay content.

229
Generally the Omdurman city is dominated by collapsible soils with varying degree of collapsibility ranging
between moderate to moderately severe according to ASTM classification. The soils are generally loose
containing appreciable amount of gravel, making difficult to obtain undisturbed soil samples.
The field testing e.g. plate load tests can successfully be carried for identification of collapse potential.
About four reported collapse potential criteria were identified for prediction of collapse potential.

9. ACKNOWLEDGMENT

Help provided by engineers Mohannd A.M. Fadolalla, Ahmed M.A. Osman, and Mr. Mohammed O. Abdallah
are appreciated.

10. REFERENCES

Abelev Y.M. 1948. The essentials of designing and building on microporous soils. Stroitel'naya Promyshelmast, No. 10.
Al-Rawas A.A. 2000. State-of-the-Art Review of Collapsible Soils Science and Technology, Special Review 115-135
SultanQaboos University.
ASTM 2000. Standard Test Method for measurement of collapse potential of soils, D5333-92. Annual Book of ASTM.
Clemence, S. P., and Finbarr, A. O. (1981) “ Design Consideration for Collapsible Soils” , Journal of the Geotechnical
Engineering Division” Vol. 107 (No. GT3) 305 - 317.
Clevenger W. 1958. Experience with loess as foundation material. Transactions, American Society of Civil Engineers,
ASCE, 123: 151-170.
Das B.M. 2011. Principles of Foundation Engineering, Seventh Edition, Global Engineering – Christopher M. Shortt.
Denisov N.Y. 1951. The engineering properties of loess and loess loams, Gosstroiizdat, Moscow.
Feda J. 1964. Colloidal activity, shrinking and swelling of some clays. Proceedings of Soil Mechanic Seminar, Loda, Illinois,
531-546.
Fredlund D. G. and Rahardjo H. 1993 Soil Mechanics for Unsaturated Soils. New York: Wiley.
Gibbs H.J. and Bara J.P. 1962. Predicting surface subsidence from basic soil tests. Special Technical publication No. 322,
American Society for Testing and Materials (ASTM), 231-247.
Handy R.L. 1973. Collapse loess in Iowa. Proceedings of the Soil Science Society of America, 37: 281-284.
Holtz W.G. and Hilf J.W. 1961. Settlement of soil foundations due to saturation. Proceedings of the 5th International
Conference on Soil Mechanics and Foundation Engineering, Paris, 1: 673-679.
Jennings J.E. and Knight K. 1957. The additional settlement of foundations due to a collapse of structure of sandy subsoils
on wetting. Proceedings of the 4th International Congress on Soil Mechanics and Foundation Engineering, London, 1:
316-319.
Khelifa A.,Salah L. M. and Farid M. 2010 Prediction of Collapsible soils by Cone Penetrometer and Ultrasonic Test Studia
Geotechnica et Mechanica vol XXXI I No 2.
Mock R.G. and Pawlak S.L. 1983. Alluvial Fan Hazards at Glenwood Springs in Geological Environment and Soil
Properties. ASCE Conference Proceedings, Houston Texas.
Omer M. O. (1981) “ Properties of a Collapsing Clayey Soil from Sudan “ Proceedins of the first National Conference of
the science and Technology of Buildings 8 - 14 December 1981, Khartoum.
Schrank E. and Awad M.Z. 1990 Palynological evidence for the age and depositional environment of the cretaceous
Omdurman formation in the Khartoum area, Sudan, Berliner Geowissenschaftliche Ablandungen, Reihe A, 120, 169-
182.
Wail O. M. Hamid (2012), “ An Experimental Study on Collapsible Soils in Omdurman” A Thesis submitted for [Link].
BRRI , Univ. Of Khartoum.
Zur A. and Wiseman G. (1973). A Study of Collapse Phenomena of Undisturbed Loess. proc. 8th ICSMFE Moscow.

230
FONDATIONS PROFONDES ET SUPERFICIELLES

DEEP & SHALLOW FOUNDATIONS

231
232
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Optimisation and safety of urban geotechnical construction projects

Optimisation et sécurité des projets urbains des constructions géotechniques

R. Katzenbach*1, S. Leppla1
1
Technische Universität Darmstadt, Institute and Laboratory of Geotechnics, Germany
*
Corresponding Author

ABSTRACT - An optimized and safe design of geotechnical structures in difficult soil and groundwater
conditions is based on a reduction of construction material used, construction time spent, energy consumed and
the adequate consideration of the soil-structure interaction. For bridges and high-rise buildings the Combined
Pile-Raft Foundation (CPRF), in-situ load tests and non-linear numerical simulations are the basis for an
optimized design. For optimization and safety of underground structures appropriate construction techniques are
necessary. In any case at difficult projects the observational method has to be applied for verification and quality
control of the design. Regarding the 4-eye-principle as basis of the quality control management an independent
peer review of the planning, design and construction is the guarantee for optimization, safety and serviceability.
The paper presents several successful realized project examples from engineering practice in urban areas. The
project examples are high-rise buildings, constructions of new tunnels and new structures over existing tunnels.

RÉSUMÉ - Dans des conditions difficiles des sols et des eaux souterraines, une conception optimisé et sûr des
structures géotechniques est basée sur une réduction des matériaux de construction utilisé, la durée de
construction, l'énergie consommée et la prise en compte adéquate de l’ interaction de la structure du sol. Pour les
ponts et les immeubles de grande hauteur aux fondations mixtes semelle-pieux, des tests de charge in-situ et des
simulations numériques non linéaires sont la base d'une conception optimisée. Les techniques de construction
appropriées sont nécessaires pour l'optimisation et la sécurité des structures souterrains. Dans tous cas de projet
difficile, la méthode d'observation doit être appliquée pour la vérification et le contrôle de qualité de la
conception. En ce qui concerne le principe des quatres yeux comme base de la gestion du contrôle de la qualité
d'une évaluation par les pairs indépendante de la planification, la conception et la construction est la garantie de
l'optimisation, de la sécurité et de la fonctionnalité. Le document présente plusieurs success réalisés, des
exemples de projets internationaux de la pratique de l'ingénierie dans des zones urbaines. Les exemples de
projets sont des fondations profondes de bâtiments de grande hauteur, des constructions de nouveaux tunnels et
de nouvelles structures sur des tunnels existants.

1 HIGH-RISE BUILDING IN DIFFICULT SOIL AND GROUNDWATER CONDITIONS

1.1 Project overview

A more than 75 m high-rise building in settlement sensitive soil has been constructed at the coastline of West
Africa. The high-rise building has up to 16 storeys. The foundation system is designed as a Combined Pile-Raft
Foundation (CPRF). The annex buildings are up to 60 m high and include apartments and parking levels. All
structures have one basement level. The whole structure has a total load of more than 700 MN. Due to its
complexity the project is categorized into the Geotechnical Category GC 3 of the Eurocode 7 (CEN 2008). The
Geotechnical Category GC 3 is the category for the most difficult projects.

1.2 Soil and groundwater conditions

The soil investigation was carried out up to a depth of 80 m below the surface:

 0 m to 30.5 m: alternating sequence of loose, medium dense and dense sand layers
 30.5 m to 80 m: alternating sequence of medium dense and dense sand layers and clay and silt layers with low
to high plasticity
 groundwater level is close to the surface

233
1.3 Foundation system

The foundation system is designed as a Combined Pile-Raft Foundation (CPRF). The CPRF is a hybrid
foundation system that combines the effects of a foundation raft and deep foundation elements like piles and
barrettes. The bearing capacity and the deformation behavior are affected by the interactions between the deep
foundation elements, the foundation raft and the subsoil. For an optimized and safe design of a CPRF the
calculation method has to consider these interactions (Hanisch et al. 2002, ISSMGE 2013).
For the determination of the bearing capacity and the load-settlement behavior and to calculate the internal
forces of the CPRF 3-dimensional simulations using the Finite-Element-Method (FEM) are suitable. The
simulations have to consider the non-linear behavior of the soil and have to be calibrated on back analysis of
laboratory tests and in-situ load tests.
At the current project the FEM-simulations for the CPRF are calibrated on a pile load test carried out on the
construction site. For the pile load test Osterberg Cells (O-Cells) were used. The test pile is divided into 3 parts:
the upper pile segment 1, the middle pile segment 2 between the upper and the lower O-Cell and the lower pile
segment 3 (Figure 1 left).
For the determination of the base resistance and the skin friction of pile segments in different soil layers the
O-Cells are activated differently. For determination of the skin friction and the base resistance of pile segment 3
only the lower O-Cell is activated using the segment 2 as abutment. For determination of the skin friction of pile
segment 2 the upper O-Cell is activated and the lower O-Cell is released. Pile segment 1 is the abutment for this
test phase. For determination of the skin friction of pile segment 1 the upper O-Cell was loaded and the lower O-
Cell was closed. The pile segments 2 and 3 are used as abutments. Figure 1 shows on the right the mesh of the
FEM-simulations and the principle arrangement of the pile load test equipment with the 3 pile segments and the
upper and lower O-Cells.

Figure 1. Setup of the pile load test (left) and FEM-Simulation (right).

The results of the back analysis by FEM-simulations are the basis for the adjustment of the estimated soil
parameters and were used to verify the developed, simplified stratigraphy for the analysis of the whole
foundation system. The results of the pile load test in-situ and of the back analysis are drawn in Figure 2 on the
left. The comparison of the results shows a good accordance.

234
Figure 2. Results of the in-situ pile load test and the back analysis (left) and the final CPRF (right).

The whole foundation system was designed by 3-dimensional, non-linear FEM-simulations. The length, the
diameter and the number of the piles were optimized by the FEM-simulations and adapted to the bearing
capacity and the load-settlement behaviour of the CPRF. Figure 2 shows on the right the final CPRF design
illustrated by the FEM-simulation. The pile diameter is 1.5 m. The maximum pile length is 37 m. The CPRF
coefficient is αCPRF = 0.8. During the construction phase and for the first years of service time the loads of the
piles, stresses under the raft and the deformation behavior of the CPRF are measured by a monitoring program
according to the requirements of the observational method (Katzenbach et al. 2010).

2 TUNNEL CONSTRUCTION IN URBAN AREA

2.1 Project overview

Due to the continuously growing traffic volume in most metropolitan areas large infrastructure projects are
accomplished, mainly in order to improve the public transport (bus, metro, train, tram) and the individual traffic
(cars, pedestrians). That means underground constructions in high density urban areas realized in almost every
big city in Europe, e.g. the metro in Vienna (Moritz & Koining 2011), the metro in Rom (Hofmann et al. 2010),
the metro in Budapest (Bäppler 2009), the road tunnels of the M-30 in Madrid (Marqués & Lorenzo Romero
2010), the metro in Frankfurt am Main (Schmitt et al. 2011). In the following the construction of the new high-
speed railway line in Barcelona will be explained.
For the new Spanish high-speed railway line (AVE) connecting Madrid and Paris a new tunnel with a length
of 5.6 km was constructed under the city centre of Barcelona. The tunnel passed directly next to the famous
church of Sagrada Familia and the famous building called Casa Milà, both belonging to the World Heritage
Properties of the UNESCO. The tunnel has an outer diameter of 11.55 m. The bottom of the tunnel is located in a
depth of at most 40 m under the surface. The tunnel is built with a tunnel boring machine (TBM), using an earth
pressure balanced shield (EPB). Figure 3 shows the EPB shield on the right. With the EPB shield the soil is
conditioned with water and foam injections at the cutter head. The homogenized, excavated earth slurry is used
as support medium. The gap between the TBM shield and the excavated soil is injected with bentonite. The gap
behind the tail of the shield is permanently grouted with mortar to provide a compression grouting procedure
(Kastner et al. 2010).

2.2 Soil and groundwater conditions

The project is located in the comparatively plain area in the City Centre of Barcelona. Most of the soil layers
passed by the TBM are tertiary layers (Figure 3 left). In the first kilometer of the tunnel the TBM passed tertiary
clay. Then a section of tertiary silty sands follow. In this section the World Heritage Properties Sagrada Familia
and Casa Milà have been passed. In the vicinity of Sagrada Familia and Casa Milà the layers are given as
follows:

 0 m to 2 m: filling
 2 m to 12 m: alternating layers of quaternary sandy silts and silty sands
 12 m to 60 m: tertiary sands, intercepted by tertiary clay layers of various thickness between 0.4 m and 2.0 m

235
Because of the dense clayey interceptions various aquifers exist in the city of Barcelona. The highest aquifer
is a free aquifer, the lower ones partly have confined groundwater conditions. In the vicinity of Sagrada Familia
and Casa Milà the free groundwater level is about 16.5 m below the surface.

Figure 3. Soil and groundwater conditions (left) and the TBM with EPB shield (right).

2.3 Tunnel constructions in the vicinity of World Heritage Properties

As shown in Figure 4 the new tunnel passed the Sagrada Familia and the Casa Milà in a horizontal distance of
only 4 m in a depth of more than 30 m. In order to guarantee the safety and to avoid settlements soil
improvements were carried out and a bored pile wall was constructed between the foundations and the new
tunnel.
Regarding the requirements of the Geotechnical Category GC 3 (EC 7) for maximum safety the observational
method was applied. The monitoring was realized with a dense grid of geodetic and geotechnical measurement
devices in the surrounding of the tunnel on one hand and with a permanent monitoring of the most important
operational parameters of the TBM on the other hand. The measured settlements on the surface in the axis of the
tunnel are shown in Figure 5. The maximum measured settlements were less than 0.5 cm. The biggest settlements
occurred at the start of the TBM in the tertiary clay. It was possible to minimize the settlements by the gained
experiences in the first section of the tunnel.

Figure 4. Cross section of Sagrada Familia (left) and Casa Milà (right).

236
Figure 5. Settlements on the surface in tunnel axis.

3 CONSTRUCTIONS OVER EXISTING UNDERGROUND STRUCTURES

3.1 Project overview

The city of Frankfurt am Main, Germany, plans to redesign the historic centre by reconstruction of historic
facades and buildings. To create the necessary space on the surface an up to 14 floors high-rise building was
deconstructed down to the sublevels. The high-rise building and its underground parking overlay 2 tunnels and
an underground station of the urban metro system. The loads of the superstructures are directly transferred onto
the tunnels and the underground station. Figure 6 gives an overview on the situation before the deconstruction.
The sealing of the structures was made of outside layers of bitumen-based materials. It must be guaranteed that
during the deconstruction of the existing high-rise building and the construction of the new buildings the sealing
of the underground structures and the sublevels remain intact. For this purpose especially the uplifts due to the
deconstruction and the deformations of the underground structures and the sublevels had to be monitored
according to the observational method.

Figure 6. Overview of the project area (left) and cross section (right).

3.2 Soil and groundwater conditions

The soil and groundwater conditions are as follows:

 0 m to 7 m: quaternary sands and gravel

 7 m to 30 m: Frankfurt Clay
 below 30 m: Frankfurt Limestone
 groundwater level in a depth of 6 m

The groundwater level is influenced by the river Main which is 180 m south of the project area. In the course
of the geotechnical survey two aquifers have been encountered. The top aquifer is located in the noncohesive
soil. The lower confined groundwater is located in the Frankfurt Clay and the Frankfurt Limestone.

237
3.3 Measurement data

According to the classification of the project into the Geotechnical Category GC 3 an extensive geodetic
monitoring program with 580 measuring points was installed. 220 measuring points are located around the
deconstructed building, 110 are located in the underground parking and in the sublevels of the deconstructed
building, 30 are in the underground station and the remaining 220 measuring points are located in the tunnels.
The existing buildings were deconstructed down to the sublevels. The uplift that occurred due to the unloading of
the soil is shown on selected points (Figure 7).

Figure 7. Selected measuring points (left) and measurement results (right).

The selected measuring points 1 to 4 are in the sublevel of the former high-rise building. Measuring point 5 is
at the transition of the underground station to the tunnel. At the measuring points 1 to 4 an uplift between 1 cm
and 5 cm was detected in the deconstruction time of 8 month. The measured uplift of measuring point 5 is less
than 0.5 cm. After the deconstruction down to the sublevels the modification of the sublevels began. In that
phase the loads only were changed insignificantly.
The maximum uplift of the whole project area and the neighborhood is drawn in Figure 8. The uplift due to
the reduced stress level of the Frankfurt Clay is continuously raising due to the consolidation processes. A
maximum uplift of 8.5 cm was measured in the area where the most floors were deconstructed. The uplifts fade
down related to the distance very quickly. So no dangerous deformations of the neighborhood were measured.

Figure 8. Maximum measured uplift [cm].

238
4 CONCLUSIONS

All presented urban construction projects belong to the Geotechnical Category GC 3 and were realized
successfully. The success is based on a high-level design considering the soil-structure interaction, the
application of the observational method and an independent peer review and supervision during the planning
stages and the construction stages. Using numerical simulations it is possible to predict precisely the
deformations and stresses of soil material and construction elements. Nevertheless the predictions have to be
controlled by geodetic and geotechnical measurements. The time period and the time interval have to consider
the load-deformation behavior of the soil, especially when the soil has a significant time dependent behavior.

REFERENCES

Bäppler, K. (2009) EPB Shields for Building Metro Line 4 in Budapest/H. Tunnel 3/2009), 20-24.
CEN European Committee of Standardization (2008) Eurocode 7: Geotechnical design - Part 1: General Rules.
Hanisch, J., Katzenbach, R., König, G. (2002) Kombinierte Pfahl-Plattengründungen. Ernst & Sohn Verlag, Berlin,
Germany.
Hofmann, A., Resto, A., Kraft, O. (2010) Mechanized Tunnelling for two new Metro lines in Rome. Tunnel 2/2010, 30-35.
ISSMGE International Society of Soil Mechanics and Geotechnical Engineering (2013) Combined Pile-Raft Foundation
Guideline.
Kastner, R., Emeriault, F., Dias, D. (2010) Assessment and control of ground movements related to tunneling. International
Conference GEOMOS, Geotechnical Challenges in Megacities,7.-10. June, Moscow, Russia, 92-119.
Katzenbach, R., Bachmann, G., Leppla, S., Ramm, H. (2010) Chances and limitations of the observational method in
geotechnical engineering. 14th Danube-European Conference on Geotechnical Engineering, 2.-4. June, Bratislava,
Slovakia, 13 p.
Marqués, M.F., Lorenzo Romero, J. (2010) Modelization, back-analysis and sensitivity analysis of the north tunnel of the M-
30 south bypass (Madrid). Ingeniería Civil 159/2010, 19-31.
Moritz, B., Koining, J. (2011) Geotechnische und messtechnische Herausforderungen beim Auffahren großer,
oberflächennaher Querschnitte im Lockergestein. Taschenbuch für den Tunnelbau, 105-142.
Schmitt, J., Fromm, H., Krajewski, W. (2011) Tunnelbau im Europaviertel Frankfurt. Mitteilungen des Institutes und der
Versuchsanstalt für Geotechnik der Technischen Universität Darmstadt, Heft 88, 119-128.

239
240
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Interest of data bases in geotechnical study of foundations

Intérêt des bases de données géotechniques dans l’étude des fondations
Fatiha DEBICHE *1, Ratiba MITICHE KETTAB 2, Fadila NAOUN 3 and Ammar NECHNECH 4
1
University of Sciences and Technology Houari Boumediene/ Bab zzouar/ ALGERIE
2
Ecole Nationale Polytechnique/ Alger/ ALGERIE
3
University of Sciences and Technology Houari Boumediene/ Bab zzouar/ ALGERIE
4 University of Sciences and Technology Houari Boumediene/ Bab zzouar/ ALGERIE

ABSTRACT - The foundations are the part of the structure which transmit all the dead and live loads to the
underlying soil or rock. Whether the foundations are simply laid or restraint in the ground, they must ensure
stability and sustainability of the structure. The design and dimensioning of this part of the structure require prior
geotechnical investigations for the sugared soil characterization. This communication is a presentation of a
conceptual model Geo-base which consists of an information system analysis of the results of geotechnical
studies.
The design of our system is based on extensive research that is related to the everyday practice of geotechnical
investigation. This research allows us to have large operating and heuristic knowledge in the field of subsurface
investigations, more precisely in the interpretation of in situ and laboratory tests, as well as all stepsof the
recognition concluding with an application with the software Access.

RÉSUMÉ - Les fondations sont la partie dùn ouvrage qui transmet au sol toutes les charges et
surcharges de cet ouvrage. Posées ou encastrées dans le sol elles assurent la stabilité et la pérennité de
lòuvrage. La conception et le dimensionnement de cette partie dòuvrage nécessite au préalable des
études géotechniques pour la caractérisation du sol support. La présente communication consiste en la
présentation d’un model conceptuel Géo-base qui est le noyau d’un système d’informations pour l’analyse des
résultats des études géotechniques par niveau d’avancement des étapes de la reconnaissance avec une
application du logiciel Access.
La conception de notre système s’est basée sur une recherche approfondie qui est en relation avec la pratique
quotidienne de la reconnaissance géotechnique.

1 INTRODUCTION

La reconnaissance géotechnique et la spécificité de chaque sol, chaque projet renferme un volume d’informations
très large et très riche. Malheureusement cette richesse est stockée soit dans un laboratoire, ou dans un centre
d’archive d’où la réflexion sur la mise en place d'une base de données géotechnique est d'un intérêt scientifique,
technique et économique assez considérable. Les principaux objectifs de la base sont de faciliter la circulation de
l'information du géotechnicien avec son environnement interne et externe, l’optimisation des coûts et les délais
des campagnes d'investigations géotechniques, et la dotation des ingénieurs d’un outil d’aide es dans la
reconnaissance et dans la prise de décision sur les problèmes complexes de géotechnique. Dans le cadre de cette
communication sera présenté un modèle conceptuel développé par la méthode Merise.
Un aperçu historique est présenté d’abord expliquant le développement de bases de données géotechniques
informatisées qui s’est est fait en fonction des moyens matériels disponibles et de leurs évolutions dans
différents pays. Les géotechniciens ont formulés selon les moyens existants de façon à se doter d'un outil d'aide
dans leur travail quotidien. Le stockage de leurs connaissances, de leur savoir-faire, dans un but d'utilisation, de
transfert aux générations d'aujourd'hui ou de demain a été le labeur des plusieurs années de travail. Ensuite sera
présenté le Modèle Conceptuel Géo- base développé en Algérie.

1.1 Développement des bases de données géologiques (1970-1990)

Les premiers travaux des géotechniciens ont concernés le stockage de l’information sur fichier, le stockage de
l'information géotechnique sur ordinateur a concerné le domaine du sondage carotté; les données concernées
étaient les coupes lithologiques et leurs implantations sur carte. Le but cherché était de transférer la
documentation géologique sur ordinateur, les données d'archives ont été contrôlées et classées manuellement
d'abord dans des fiches ; puis un nouveau contrôle se faisait à partir du logiciel utilisé. Après avoir corrigé les
fautes éventuelles, les données sont transférées sur disque ensuite sur bandes magnétiques et ont été classées
dans des fichiers comprenant les caractéristiques géologiques, hydrogéologiques et géo mécaniques ainsi que les
analyses des eaux. Chaque objet est localisé par ses coordonnées, en général les objets concernés par le stockage

241
étaient :l a-nappe phréatique, le substratum rocheux ,la stratigraphie et les positions des points de sondages. Pour
illustrer le travail du transfert de l'information géotechnique sur fichiers, quelques exemples pratiques réalisés
dans différents pays sont présentés dans les paragraphes suivants.

1.1.1. Base de données géologiques de la ville de Prague (1976/1988)

La documentation géologique du territoire de Prague a été transférée sur ordinateur de1976 à 1988. L'analyse
du codage des points de sondages sur cartes géologiques a été effectuée par l'intermédiaire du logiciel (IDMS)
sur ordinateur ICL 2904. Le remplissage de la base de données en termes de coupes de sondages et les
vérifications du système étaient faits au fur et à mesure. Dans chaque carte géologique sont implantés les
numéros des objets qui ont toutes les caractéristiques du sondage. Les cartes utilisées étaient à l'échelle
1/5000ème. Le travail de remplissage a démarré en 1981 et s'est terminé en 1988. Un logiciel a été intégré à la
base de données pour la construction tridimensionnelle du site à partir des données des sondages. La
municipalité de Prague a fourni les finances nécessaires pour préserver les données (ZADINOVA et al, 1992).

1.1.2 Base de données géologiques de la ville de Liège.

La banque de données réalisée dans le but d'archivage des données géotechniques de la ville de Liège contient
un volume important d'informations. Les données brutes de type numérique, tels les essais de pénétration
statique, font l'objet d'un traitement statistique. Les données de type alphanumérique (description des forages par
exemple) sont abordées sous l'aspect syntaxique tandis que les données graphiques sont intégrées à la banque de
données sous forme de fichiers vecteurs ou rasters.
La gestion des données est assurée par l'ensemble de programmes évolutifs autorisant un encodage classique
par masque de saisie, adaptés aux types d'informations (forages, affleurements, galeries,..). Il sont obtenus par
simple digitalisation en coordonnées (X, Y et Z) sont utilisés sous forme de fichiers de références. Ces
renseignements sont consignés dans la banque de données sous forme de fichiers pixels de type raster. Les
documents photographiques, cartographiques et les dessins ou schémas réalisés par le géologue (coupes de
terrain, logs de sondages) forment l'essentiel des données graphiques nécessitant l'utilisation d'un scanner; ils
sont utilisables comme fond de plan ( MARCHAND et all, 1992).

1.2. Conception des systèmes d’informations géotechniques

A partir des années quatre vingt dix (1990) les systèmes d'informations géotechniques ont été développés grâce
à l'intégration des méthodes de conception de systèmes d’informations qui se basent sur l'étude de l'existant et sur
la capacité de pouvoir séparer la structure des données et des traitements. Un rapprochement entre concepteurs
de systèmes d'informations et experts géotechniciens a donné naissance à l'approche « Systèmes Informations
géotechniques ». Les premiers travaux ont commencé vers les années 1988 par Gnewood aux USA, où le
concept de dictionnaire de données structurées en fiches a été utilisé. D'autres travaux ont été lancé dans divers
pays (Royaume Unis, Canada et France ). La démarche de travail des spécialistes de la conception des Systèmes
d'information et les experts en géotechniques était basée sur la pratique géotechnique. La méthode d'analyse pour
la modélisation des données est basée sur la théorie des méthodes des systèmes d'information (GREENWOOD et
al , 1992). Pour illustrer le travail de conception et de modélisation des données, quelques exemples de
conception de bases de données géotechniques sont présentés dans les paragraphes suivants.

1.2 .1. Base de données GÉOTECH.

La base de données « GEOTECH » est née en 1985 d'une collaboration université (École Polytechnique de
Montréal, Génie Civil, Canada) et industrie. Le modèle de données couvre toutes les informations spécifiques à
la géotechnique, des essais de laboratoire jusqu'aux paramètres de conception. Le modèle de données de «
GEOTECH » est un système intégré de gestion et de traitement des données, peut être utilisé aussi bien par le
laboratoire exécutant la campagne d'exploration que par l'expert conseil responsable de la conception des
ouvrages (LESSARD et al ,1992).
La pratique géotechnique a été à la base de la modélisation des données qui a conduit à l'établissement des
tables ORACLE. L'entité de base du modèle de données géotechniques est le sondage. Les essais in - situ
constituent une famille d'entités liées au sondage, tandis que les essais de laboratoire forment une autre famille
d'entités liée au sondage via l'entité échantillon. Le modèle prévoit quelques entités comme des codes
stratigraphiques.

1.2.2. Base de données BDSOL.

BDSOL représente une structure de bases de données pour le stockage des informations sur le sol. Elle est
élaborée par AL-HADAR J., BOULEMIA à l'université de Lille, laboratoire méthode et habitat, France. Elle est
reliée à un outil d'aide KB SOL qui est le système à base de connaissance permettant de vérifier la cohérence des

242
données avant d'être stockés dans BDSOL. Ce prétraitement permet de mettre à la disposition des différentes
applications géotechniques des données relativement fiables (AL-HAJJAR et all, 1992). La base de données «
BDSOL » a été crée dans le but de l'utiliser comme base de données commune à différents projets (fondations,
terrassements, réseaux). Elle est développée dans un environnement relationnel (ORACLE sous DOS-(4/Oracle).

1.2 .3 Base de données MODELISOL.

La base de données « MODELISOL » est issue d'un projet développé vers les années1988 à L'Ecole Centrale
de Paris avec des subventions des Ministères et le Bureau de Recherches Géologiques et Minière(B.R.G.M).Les
buts visées étaient de promouvoir en ingénierie géotechnique les développements faits en recherche sur les lois
de comportement des sols, les méthodes statistiques d'analyse (MARCHAND et a1,1992).

2. DEVELOPPEMENT DU SYSTEME D’INFORMATIONS GEOTECHNIQUES Géo-Base (ALGERIE)

Le modèle Géo -Base est élaboré en 2003 dans le cadre des travaux de recherches d’une thèse de Magistère à
l’Ecole Nationale Polytechnique à Alger (F. DEBICHE,S .BELKACEMI ,2003).En se basant sur l’expérience
internationale évoquée précédemment, un système d’informations a été réalisé en concertation avec les équipes
de recherches universitaires et la communauté des géotechniciens en Algérie. La méthode de conception du
model est Merise qui propose une approche de conception du système d'informations séparant l'étude des
données de celle du traitement, en avançant progressivement par niveaux. Chaque niveau a pour objectif de
fournir un certain nombre de documents permettant la synthèse textuelle d'un processus de réflexion.

2.1 Vision de Géo-Base

Les données modélisées dans la base sont les données géotechniques directement mesurées avec toutes leurs
caractéristiques et celles recueillies après traitements. Les résultats d'essais et les rapports d'études seront ensuite
stockés. La démarche de conception de la base de données s'appuie directement sur la pratique du travail
quotidien du géotechnicien; le stockage de l'information s'étendra du bon de commande au recouvrement de
l'étude. Cette conception est justifiée de manière à ce que la base ne doit pas être modifiée quelques soit les
modifications des méthodes de traitements.

2.2 Architecture

Cartes Essais Essais

géologiq de géophys
ues et laborato iques
Topograp ires et

NOYAU
DE LA
BASE DE
Moyens DONNEES Essais
de hydrauli
prélève ques
ments

Essais Modules Normes

divers de Univers
Calculs elles
d’ouvrag

Figure 1. Architecture du système Géo-Base

La base de données géotechnique est construite d'un noyau et d'un ensemble de logiciels systèmes
satellites entourant le noyau. Dans le noyau sont définis tous les paramètres géotechniques. Les données de la
base sont définies et rangées dans un dictionnaire. L'ensemble des logiciels regroupe tous les procédés de
gestion et techniques d'interprétations et de calcul de fondations. L'architecture choisi est du type fédérée;
l'utilisateur peut extraire des données du serveur, les traiter à son niveau sans être gêné par les autres
utilisateurs. Dans cette architecture le (Description Manipulation and Control Sub — system) DMCS est

243
inclus à l'intérieur du noyau du système de gestion au niveau du serveur et autour duquel s'articulent des
processus supportant les outils et les interfaces externes. Il en résulte pour les utilisateurs une simplicité pour
la consultation multi — usages de la base de données. Dans la pratique, cette architecture permet aux
géotechniciens d'avoir un système informationnel utilitaire pour les études géotechniques.

2.3 Le modèle Conceptuel Géo-Base

Le modèle conceptuel est construit à partir de l'étude de l'existant et du flux d'informations. La nature
d'information échangée entre les différents acteurs a guidé l'élaboration des relations entre objets. Ainsi le
modèle est constitué des objets suivants : Objet Dossier qui rassemble toutes les propriétés relatives à la
partie administrative, et financière de l'étude. Les objets Client, Projet et Site rassemblent les informations
techniques relatives à la description et à la classification suite aux observations soit du géotechnicien, du
géologue et du client. Objet échantillon est relié directement par les essais de laboratoire au des moyens de
prélèvements ou de sondage. Les objets des essais de laboratoires identifications et mécaniques
rassemblent toutes les informations techniques et tous les Mesures des essais de laboratoires. Les objets des
essais de laboratoire sont: des Essais de mesure de densité, teneur en eau, limites Atterberg, essai de
cisaillement à la boite de Casagrande, essai triaxial et essai oedométrique. Les Objets des essais in — situ
rassemblent toute l'information technique et toutes les mesures. Les Objets relatifs à ces essais sont les
essais au pénétromètre statique, essai au pénétromètre dynamique et essai préssiométrique. Les objets
relatifs au sondage et aux essais dans le sondage sont le Sondage carotté et couche de sol. Les essais
géophysiques, les essais hydrauliques constituent une famille indépendante d'objets.

Figure 2. le modèle Conceptuel du système d’information Géo Base

244
3. REALISATION D’UN MODELE PHYSIQUE DE Géo-Base
Le Modèle Physique de Données est réalisé à partir du Modèle Conceptuel des Données, on présente
quelques exemples d’interfaces concernant les objets de notre base de donnée ; tels que la saisie des
données du projet, chantier et un exemple d’essai de laboratoire (figure 3, figure 4 et figure 5).

Figure 3. interface saisie projet

Figure 4. Saisie des données chantier

245
Figure 5. Essais de laboratoire

4. CONCLUSION

Le modèle conceptuel élaboré Géo-base présente un noyau d'un système intégré en géotechnique c'est-à-dire
qu’il stocke les informations du site relative à la géologie , hydrogéologie et les niveaux altimétriques. On
retrouve les données brutes des essais de tous les essais in -situ et les essais de laboratoires. La particularité de
Géo-Base c’est le stockage de l’information
géotechnique à partir du poste de l'opérateur jusqu'au poste de l'ingénieur; de façon à stocker l'information
relative à la mesure pour pouvoir faciliter l'interprétation des résultats dans un contexte normatif universel et ne
pas subjectiviser et cloisonner l’interprétation à l’échelle e individuelle ou spécifique à un laboratoire d'autres
part en utilisant par exemple une bibliothèque hypertexte de normes (un rangement de toutes les normes
universel de géotechnique sera inséré dans le système de gestion du système d'informations). Cette intégration du
système d'information dans les postes de travail permet de systématiser, -de sécuriser et de fiabiliser le stockage
et l'accès aux informations. L'encadrement technique peut suivre l'avancement des études et apporter les
corrections nécessaires sans attendre la fin de l' é[Link] système pourrait servir comme outil d'aide à la décision
en géotechnique et intégré dans les systèmes experts et aux chercheurs.

5. REFERENCES

AFNOR FNBTP., DAEI. (1986) La réglementation française et étrangère en géotechnique. ctes des journées d'étude
organisées les 16 et 17 Décembre 1986 par l'école nationale des ponts et chaussés.
AFNOR., (1995) Géotechnique Essais sur ouvrages géotechniques, Dimensionnement, Exécution, Tome 2 ., Association
Française de normalisation, Recueil de normes Françaises.
AFNOR. (1995)., Géotechnique Essais de reconnaissance des sol Tomel , Association. Française de normalisation, Recueil
de normes Françaises.
AL — HAJJAR J., BOULEMIA C, (1992). Couplage d'un logiciel d'évaluation des fondations à une base de données
relationnelle pour le sol , pp.633-640, Actes du colloque organisé par ' Ecole Nationale des Ponts et Chaussées à Paris,
29 — 30 septembre, ter octobre 1992.
BENOIT J., DE ALBA P.A., SAWYER CRAIG P., (1992) Un logiciel de base de données pour les sites d'essais en
géotechnique aux Etats Unis d'Amérique », pp 649-656, Actes du colloque organisé par 1 ' Ecole Nationale des Ponts et
Chaussées à Paris, 29 — 30 septembre, 1er octobre 1992.
BENTABET A., EEG SIMECSOL, (2001) Développement d'un système d'information géographique , pp 122-132, Actes du
3àme Congrès Algérien de la route, Alger 17-19ctobre2001.

246
DEBICHE F., BELKACEMI S, (2001) Architecture d'une base de données géotechniques pp.91-97, Actes du 3' Congrès
Algérien de la route, Alger 17-19ctobre2001
DEBICHE F.,BELKACEMI (2003) Elaboration du modèle conceptuel de la base de données géotechniques Géo Base »,
Thèse de Magistère, Ecole Nationale Polytechnique, Département Génie Civil. Alger
DJEFAL, H . DEBICHE F (2005) Développement d’un système d’informations géotechniques Géo-Base Projet de fin
d’étude d’Ingénieur à l’Ecole Nationale Polytechnique Alger
DURV1LLE J.L, LACUBE J, (1992) Développement d'un fichier informatique sur les mouvements de terrain pp.657-664,
Actes du colloque organisé par 1 ' Ecole Nationale des Ponts et Chaussées à Paris, 29 —30 septembre, 1er octobre 1992.
FAVRE J.L, HERITER. B, IvIECHALSKI. R, (1992) : « MODELISOL Banque d'essais de référence pour l'Ingénierie de
demain », pp. 665-676, Actes du colloque organisé par 1 ' Ecole Nationale des Ponts et Chaussées à Paris, 29 — 30
septembre octobre 1992.
GILES D.P, (1992) : « Le poste de travail géotechnique avec ordinateur - Lien entre la base de données géotechniques et le
système d'information géographique », pp 685-690, Actes du colloque organisé par I ' Ecole Nationale des Ponts et
Chaussées à Paris. 29 — 30 septembre,1 octobre 1992.
GREENWOOD IR., (1992) : « Collecte, Transfert, Stockages et présentations des données d'étude de sol », pp.691-698,
Actes du colloque organisé par 1 ' Ecole Nationale des Ponts et Chaussées à Paris, 29 — 30 septembre, 1er octobre
1992.
LESSARD G., DUCROT C., SUOLIE M (1992) « Logiciel intégré pour la gestion des données géotechniques », pp. 715-
698, Actes du colloque organisé par 1 ' Boole Nationale des Ponts et Chaussées à Paris, 29 — 30 septembre, 1er octobre
1992.
LOUNI, H., (1993) : «Devellopement d'un système expert d'aide à la définition d'un programme de reconnaissances pour les
projets de tunnels », pp 321-326, Revue tunnels et ouvrages souterrain — N°120 —Novembre — Décembre 1993.
MARCHAND L., MONJOIE A., POLO — CHIAPOLINI C., SHROEDER C, (1992) : « Une banque de données
géologiques pour la structure du sous-sol », pp. 725-732, Actes du colloque organisé par 1 Ecole Nationale des Ponts et
Chaussées à Paris, 29 — 30 septembre, ler octobre 1992.
MESTAT P., (2001) : « MOM1S : Une base de données sur la modélisation numérique des remblais sur sols compressibles
et sur la confrontation calculs — mesures in situ », pp.43-58, bulletin de liaison des laboratoires des ponts et chaussés —
232 — Mai —Juin 2001.
STATEN P.M., Scott Wilson kirkpatrick, UK, CARONNA S, (1992) « Gestion des données géotechniques d'un vaste
programme routier en Grande —Bretagne », pp. 741-748, Actes du colloque organisé par 1 ' Ecole Nationale des Ponts
et Chaussées à Paris, 29 — 30 septembre, ler octobre 1992.
THREADGOLD L., HURCIUSON R3, ( 1992 ) Transmission électronique de données géotechniques à partir d'études de sol
pp. 749-756, Actes du colloque organisé par 1 ' Ecole Nationale des Ponts et Chaussées à Paris, 29 — 30 septembre,1'
octobre 1992.
Toll, D.G., 011iver, A, (1993) : SIGMA : «A knowledge-based system for the interpretation of geotechnical information »,
Proc. of conference on informing technologies for construction civil engineering and transport, pp. 245-254, London,
Brunei University & Science and Engineering Council.
ZADENOVA V., JANOVA J., SOKOL A, PUDIS, (1992) Le système d'information GEO », pp. 757-762, Actes du
colloque organisé par I ' Ecole Nationale des Ponts et Chaussées à Paris, 29 — 30 Septembre, ler octobre 1992.

247
248
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Comportement d’un pieu isolé sous sollicitation axiale cyclique–Modélisation

par éléments finis et construction des boucles hystérétiques de frottement latéral

Behaviour of a Single Pile under Cyclic Axial Loading – Finite Elements Modeling and
Construction of Hysteretic Skin Friction-Settlement Loops
Mohamed KHOUAOUCI1 et Ali BOUAFIA*2
1
Université des Sciences et Technologie Houari Boumediène Alger, Algérie
2
Université des Sciences et Technologie Saâd Dahleb, Blida, Algérie
* Auteur correspondant

RÉSUMÉ- La communication se propose de présenter les principaux résultats d’analyse du comportement d’un
pieu isolé sous une force verticale monotone par le biais d’une modélisation par éléments finis (logiciel Plaxis),
ainsi que la procédure de construction des boucles hystérétiques de frottement à la base des règles fondamentales
de Masing. Une étude paramétrique par la suite a été lancée en vue de dégager l’effet de l’élancement du pieu et
de la compressibilité relative pieu/sol sur la raideur de la boucle et le coefficient d’amortissement radiatif de
l’interface pieu/sol.

ABSTRACT – The paper aims at presenting the main results of the analysis of a single pile behaviour under
monotonic axial loading on the basis of a finite elements modeling (software Plaxis), as well as the procedure of
building the hysteretic skin-friction-settlement loops on the basis of fundamental Masing’s rules. A parametric
study was subsequently carried out in order to investigate the effects of the pile slenderness ratio and the pile/soil
compressibility ratio on the frictional stiffness and the geometrical damping coefficient of the pile/soil interface.

1 INTRODUCTION

Suite à l’essor mondial considérable de la technologie des pieux et sa large application aux génies civil et
maritime, la réponse cyclique d’un système de fondations sur pieux est devenue une préoccupation importante
aussi dans le domaine de l’ingénierie que de la recherche.
L’étude du comportement d’un pieu isolé dans un sol homogène sous une sollicitation verticale harmonique
sert de support fondamental à l’étude sous sollicitations non périodiques telles que le séisme et les chocs
(battage, impact, etc).
La mobilisation du frottement latéral le long du fût du pieu contribue à la reprise des efforts supportés par le
pieu, d’où l’intérêt évident de la quantification d’une telle composante en vue du dimensionnement dans un
projet ou dans le cadre de la recherche sur le comportement d’un pieu chargé verticalement. Outre les études
expérimentales sur des pieux instrumentés en vraie grandeur ou en modèles centrifugés, la modélisation
numérique par éléments finis ouvre des perspectives intéressantes quant à l’analyse des facteurs clés régissant la
mobilisation du frottement au niveau de l’interface d’une tranche élémentaire du pieu.
La présente communication se propose de mener une étude paramétrique détaillée, via une modélisation non
linéaire par la méthode des éléments finis, en vue d’une part de dégager les paramètres d’influence et d’autre part
la proposition d’une formulation simple des paramètres de transfert de charge relatifs au frottement latéral. Ces
paramètres, mentionnés à la figure 1, sont la raideur initiale B0 de la courbe monotone du frottement latéral et le
frottement latéral limite, soit qs.
Les règles de Masing ont par la suite été appliquées à la courbe monotone de frottement latéral, en vue de
construire la boucle hystérétique généré par un chargement cyclique harmonique. La procédure de construction a
été expliquée et l’effet des paramètres de l’interaction pieu/sol sur la boucle a été analysé.

2 PRÉSENTATION DU MODELE D’ÉLÉMENTS FINIS

2.1. Analyse dimensionnelle

La relation générale décrivant la mobilisation du frottement latéral aux petits déplacements est de la forme
suivante :
f1(τ, v, B, D, Ep, Es, υs) =0 (1)

249
Elle se transforme, à l’aide du théorème des π de Vashy-Bukingham, sous la forme adimensionnelle suivante :

Figure 1. Schéma de courbe et boucle hystérétique du frottement latéral

g1(π1, π2, π3, π4 )= 0 (2)

D est l’élancement du pieu, E

1   2  p  K est la compressibilité relative pieu/sol, π3 = υs est le coefficient
B Es
de Poisson du sol et   B0 .B est la raideur adimensionnelle de l’interface pieu/sol. Il s’agit donc de trouver
4
Es
une formulation simple de la forme :
B0 B D E
 F ( , p , s ) (3)
Es B Es

En grands déplacements, la relation décrivant le frottement latéral limite est comme suit :

f2(qs, B, D, Ep, Es, σv0’) = 0 (4)

Elle se transforme alors en :

g2(π1, π2, π3) = 0 (5)

Les termes π1 et π2 sont les mêmes qu’avant, et π3, appelé facteur de frottement latéral, est donné par :

qs
3  (6)
 v0 '
Ce terme est à relier aux autres comme suit :

qs D E
 G( , p ) (7)
 v0 ' B Es

Les symboles τ, v, B, D, Ep, Es et σv0’ désignent respectivement la contrainte de frottement latéral à la

profondeur étudiée, le tassement correspondant, le diamètre du pieu, sa fiche dans le sol, le module d’Young du
pieu, le module d’élasticité du sol, et enfin la contrainte verticale effective à la même profondeur.

2.2. Description du modèle d’éléments finis

Une modélisation rigoureuse des états de contraintes et de déformations générées dans le sol dues à la
pénétration d’un objet cylindrique est difficile à mettre en œuvre, car celle-ci doit faire intervenir les grands
déplacements et les glissements à l’interface pieu/sol.
Le comportement du pieu isolé chargé verticalement peut être modélisé par un système axisymétrique dont
l’axe de symétrie est celui du pieu, traduisant le fait que les points diamétralement opposés subissent les mêmes
états de contraintes et de déformations. On considère le sol composé d’un massif d’argile saturée par une nappe

250
en surface, homogène et normalement consolidée, dont le comportement non drainé est décrit par une loi
d’élastoplasticité parfaite, la rupture suivant le critère de Mohr-Coulomb. Une loi de comportement élastique
linéaire a été considérée pour modéliser le pieu. Enfin, des éléments d’interface ont été introduits à l’interface
pieu/sol en vue de tenir compte d’une interface mi-rugueuse.
Le maillage est constitué des éléments triangulaires à 15 nœuds prédéfinis par le logiciel Plaxis. Le
chargement en tête du pieu est appliqué sous forme d’incréments de déplacements verticaux imposés jusqu’à
10%B, valeur considérée conventionnellement en pratique comme correspondant à la capacité portante du sol
autour du pieu.
Les limites verticale et horizontale du maillage doivent être fixées de telle façon qu’au-delà de ces limites, les
déplacements des nœuds en tête du pieu se stabilisent et ne varient plus avec les dimensions du maillage. Mestat
et Prat (1999) ont proposé de fixer les limites verticales du maillage à au moins deux fois la longueur du pieu et
la limite horizontale inférieure à au moins 2.5 fois sa longueur en dessous de la pointe, soit 3.5 D, D étant la
fiche du pieu. Yaich-Achour (2004) a fixé ces limites à 25.5B dans la direction horizontale et (D+h) dans la
direction verticale, h étant égale à 2 fois la fiche D du pieu.
Par ailleurs, on s’intéresse surtout aux comportements de la pointe, de la tête et de l’interface sol/pieu. Il
serait judicieux de serrer le maillage particulièrement dans ces zones (voir figure 2).

3 FROTTEMENT LATÉRAL SOUS CHARGEMENT MONOTONE

3.1. Analyse en petits déplacements

Il a été constaté que la raideur B0 ne varie pratiquement pas avec la profondeur, autrement dit la raideur
adimensionnelle B0B/Es est constante le long du pieu. En outre, comme l’illustre la figure 3, ce terme est le
même pour trois matériaux argileux étudiés (Argile molle, moyenne ou ferme) et ne dépend ainsi pas de la
consistance de l’argile. On remarque de la même figure que la compressibilité K n’a aussi pratiquement aucun
effet sur la raideur. Enfin, une étude antérieure a montré que le coefficient de Poisson du sol n’a pas d’influence
sur les valeurs de la raideur, laissant à penser qu’en comportement drainé ou non drainé l’interface pieu/sol
mobilise la même raideur (Yaich-Achour, 2003).
L’élancement du pieu a par contre un effet réducteur de la raideur, conformément à une loi logarithmique
obtenue par ajustement des résultats au sens des moindres carrés, avec un coefficient de régression de 96.4% :

B0 B D 1
 F( )  (8)
Es B D
2.91  1.21ln 
B

Figure 2. Schéma du maillage du système pieu/sol

(A: Maillage serré, B: maillage automatique)

Les valeurs obtenues du rapport B0B/Es ont été comparées à celles recommandées en littérature et issues
d’une analyse rationnelle ou numérique, à la figure 4. On constate une bonne concordance avec celles de
Randolph et Wroth (1978), issues d’ailleurs d’une analyse en éléments finis, mais un certain écart par rapport à

251
celles de Cassan (1978), issues d’une analyse rationnelle à la base des équations de Mindlin d’une force enterrée
dans un continuum élastique.

3.2. Analyse en grands déplacements

Un résultat fondamental concernant la fonction adimensionnelle G, définissant le facteur de frottement latéral en

équation (7), est qu’elle ne dépend ni de la profondeur le long du pieu, ni de la consistance de l’argile. En outre,
G ne dépend pratiquement pas de l’élancement D/B du pieu ainsi que de la compressibilité K, et prend une
valeur moyenne égale à 0.203, avec un coefficient de variation de 8.9%.

Figure 3. Variation de la raideur adimensionnelle à l’interface pieu/sol

Figure 4. Comparaison des valeurs de la raideur adimensionnelle

Une comparaison avec les méthodes courantes de calcul du frottement latéral est récapitulée au tableau 1, en
se fixant trois profondeurs le long du pieu. Ce tableau montre une excellente concordance avec la méthode de
Burland (1973), bien que les valeurs trouvées sont faibles par rapport à celles données par les méthodes 1 à 4,

252
issues d’ailleurs de l’analyse empirique des essais de chargement des pieux. Il est à noter que le mode
d’installation du pieu est un facteur clef gouvernant le frottement latéral, et les méthodes 1 à 4 ne précisent pas
pour quel mode d’installation elles sont applicables. Le modèle d’éléments finis élaboré de ce travail ne
considère en fait que le cas d’un pieu foré en refoulant pas le sol lors de son installation.

Tableau 1. Comparaison des méthodes de calcul du frottement latéral

No. de la Auteur(s) Valeurs de qs/Cu à Valeurs de qs/Cu Valeurs de qs/Cu à

méthode z= 3 m à z= 6 m z= 9 m
1 Coyle & Reese (1966) 1.50 1.00 0.83
2 Aschenbrener & Olsen (1984) 1.00 1.00 1.00
3 API (1984) 1.00 0.88 0.71
4 Vijayvirgiya (1977) 1.45 1.10 0.92
5 Burland (1973) 0.32 0.33 0.33
6 Formule proposée 0.32 0.33 0.33

4 BOUCLE HYSTÉRÉTIQUE DE FROTTEMENT LATÉRAL

La courbe de frottement latéral, décrivant la relation entre la contrainte de frottement latéral τ et le tassement v
du pieu à la profondeur considérée, a souvent une allure hyperbolique qui peut être formulée comme suit :

v vB0
  (9)
1 v v
 1 R
B0 qs v

Le terme vR égal au rapport qs/B0 est appelé tassement de référence, et le rapport v/vR représente le niveau de
tassement du pieu. En divisant l’équation (9) par qs on obtient :
v
 vR
 (10)
qs 1  v
vR
On a τ = Bs.v, Bs étant la raideur sécante du frottement latéral, ce qui donne:

v vB0
  Bs v   (11)
1 v 1 R
v

B0 qs v

On obtient alors le rapport suivant appelé raideur normalisée, soit k :

Bs 1
 k (12)
B0 1  v
vR
v
Bs vR 
1   (13)
B0 1  v qs
vR
Les règles de Masing en cas d’un chargement cyclique alterné se traduisent par l’équation suivante :

  v  vc 
 c

   c  v  v B0 1  J 
 n


(14)
  


253
J est une fonctionnelle à obtenir par calage avec les résultats expérimentaux d’un essai de chargement
cyclique du pieu, et n est un coefficient qui vaut 1 pour la courbe du premier chargement et 2 pour les courbes de
déchargement et rechargement successives. L’indice c est relatif au point de changement du sens de chargement.
La courbe de premier chargement s'obtient en imposant vc = 0, τc = 0 et n = 1:

  vB0 1  J v  (15)

On obtient alors la fonctionnelle J :

v
 
J v   1 
vB B vR
 1 s  1 s   (16)
vB0 vB0 B0 qs v
1 R
v
D'une manière analogue, on utilise l'équation (14) en imposant n= 2, ce qui donne:
v  vc
 v  vc  2v R
J  (17)
 n 
 1 v  v
c

2v R

   c  B0
v  v 
c
(18)
v  vc
1
2v R

Selon la même démarche, la fonctionnelle J en cas de rechargement à partir du point [-vc, -τc] s'écrit, en
imposant n= 2:

v  (v c ) v  vc
 v  (v c )  2v R 2v R
J   (19)
 
 1  v  (v ) 1  v  v
c c
n

2v R 2v R

   c  B0
v  v  c
(20)
v  vc
1
2v R
Les équations (18) et (20) permettent ainsi une description complète de la boucle du frottement latéral à partir
des caractéristiques de la courbe de premier chargement. Cette boucle, comme le schématise la figure 1, peut être
caractérisée par une raideur sécante Bs définie comme étant la pente de la droite cb joignant les deux points
anguleux de la boucle, ainsi que le coefficient d’amortissement radiatif ξ du système pieu/sol, défini par
l’équation suivante:

1 W
 (21)
4 W

Du point de vue géométrique, W et W sont respectivement les aires de la boucle (énergie absorbée par
amortissement radiatif par cycle) et du triangle oac (énergie élastique stockée par cycle). On montre que le
coefficient ξ dépend entre autres de l’amplitude vc comme suit :

4   2
ln k  
k
 1 (22)
 1  k   1  k  

Les paramètres Bs et ξ ainsi définis varient en fonction de l’amplitude vc comme l’illustre la figure 5. Ainsi,

254
sous l’effet d’une charge verticale harmonique à amplitude croissante, dans le domaine des petites amplitudes, la
pente Bs est pratiquement confondue avec la raideur initiale B0 et l’amortissement radiatif est pratiquement nul.
Au fur et à mesure que l’amplitude vc du déplacement à la profondeur étudiée augmente, l’amortissement
augmente rapidement et la raideur sécante diminue. Dans le domaine des grands déplacements, l’amortissement
tend vers la valeur de 2/π, soit de 63.7%. La raideur normalisée k chute à 0.5 lorsque l’amplitude vc atteint le
tassement de référence vR, et diminue à 10% de la raideur initiale lorsque vc est égal à 10vR, ce qui est le signe
d’une dégradation cyclique de la raideur.
Du fait que la pente initiale B0 ne dépend pas de la compressibilité pieu/sol et diminue avec l’élancement du
pieu, on voit d’après l’équation (23) que la pente Bs de la boucle de frottement est la même pour deux pieux l’un
étant incompressible et l’autre compressible, l’amplitude de chargement vertical étant supposée identique.
En outre, un pieu long mobilise une raideur de frottement latéral plus petite que celle d’un pieu court, toutes
choses étant par ailleurs égales.
Enfin, l’équation (22) nous enseigne que l’amortissement radiatif du frottement à l’interface pieu/sol ne fait
pas intervenir l’élancement du pieu ni la compressibilité pieu/sol.

B0
Bs  (23)
v
1  Rc
v

Figure 5. Variation des paramètres k et ξ de la boucle hystérétique du frottement latéral

5 CONCLUSIONS

La communication a présenté les résultats d’une modélisation par éléments finis de la mobilisation du frottement
latéral le long du pieu installé dans une argile saturée normalement consolidée. Il a été constaté que la raideur de
l’interface pieu/sol ne dépend pas de la compressibilité pieu/sol et diminue selon une loi logarithmique avec
l’élancement du pieu. Le frottement latéral limite présente un profil linéaire ne dépendant ni de la compressibilité
pieu/sol ni de l’élancement du pieu.
Les règles de Masing ont permis la construction de la boucle hystérétique de frottement latéral, ce qui offre la
possibilité d’une analyse simple et performante des déplacements du pieu au cours d’un cycle de chargement
vertical alterné.
Il a été enfin constaté que la dégradation cyclique de la raideur de l’interface se manifeste rapidement en
fonction de l’amplitude du déplacement, et que l’amortissement radiatif augmente avec cette amplitude.

REFERENCES

Aschenbrenner, T. B., and Olson, R. E. (1984) Prediction of settlement of single piles in clay, Analysis and design of pile
Foundations, American Society of Civil Engineers, J. R. Meyer, ed.
Burland, J. B (1973) Skin friction piles in clay- A simple fundamental approach, Ground engineering, 6(3), 30-42.

255
Cassan, M (1978) Les essais in-situ en mécanique des sols, Tome 2: Applications et méthodes de calcul, éditions Eyrolles,
331 pages.
Coyle, H. M., and Reese, L. C. (1966) Load transfer for axially loaded piles in clay, Journal Soil Mechanics and
Foundations
Division, Proceedings Paper 4702, American Society of Civil Engineers, Vol. 93 (SM6).
Heydinger, A. G. (1984) Recommendations: Load-transfer criteria for piles in clay, Technical Report ITL-87-1, U.S. Army
Engineer Waterways Experiment Station, Vicksburg, MS.
Randolph, M. F., and Wroth, C. P. (1978) Analysis of deformation of vertically loaded piles, Journal of Geotechnical
Engineering
Division, American Society of Civil Engineers, Proceedings Paper 14262, Vol 104 (GT12).
Vijayvergiya, V. N. (1977) Load-movement characteristics of piles, Proceedings, Ports 77, American Society of Civil Engineers,
Vol. 2, pp : 269-286.
Yaich-Achour N. (2004) Paramètres de transfert de charges des fondations profondes-Analyse d’une base de données,
Thèse de
Magister en Génie Civil, Université de Blida, 134 pages.

256
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Numerical Analysis of Shallow Foundations on Purely Cohesive Soil under

Eccentric Loading

Analyse Numérique des Fondations Superficielles sur un Sol Purement Cohérent soumises à
un Chargement Excentré

N.E.H. Khitas*1, A. Mabrouki1, D. Benmeddour1 and M. Mellas1

1
Mohamed Khider University, Biskra, Algeria
*
Corresponding Author

ABSTRACT - This paper evaluates the performance of the effective width approach in predicting the ultimate
load of footings subjected to vertical eccentric loading. A numerical analysis is performed to estimate the
undrained bearing capacity mobilized in a purely cohesive soil under a strip footing. The results of this analysis
show that the effective width approach provides a good approximation of the bearing capacity for this kind of
problems.

RÉSUMÉ - L’objectif de ce travail est d’évaluer la performance de la méthode de la largeur effective dans la
prédiction de la charge limite des fondations soumises à un chargement excentré. Une analyse numérique est
effectuée pour estimer la capacité portante non-drainée mobilisée dans un sol purement cohérent sous des
fondations filantes. Les résultats montrent que la méthode de la largeur effective fournit une bonne
approximation de la capacité portante.

1 INTRODUCTION

The bearing capacity of shallow foundations is, usually, calculated using Terzaghi’s equation (Terzaghi (1943))
based on Prandlt’s solution (Prandtl (1920)) and the superposition principle. This equation was derived for a
strip footing resting on the soil surface in conditions of symmetry in geometry and loading. But in the presence of
eccentricity of loading (see Figure 1), the problem becomes more complicated because of the detachment at the
interface soil-foundation.

V V M

e
(a) (b)
Moment reference point

Figure 1. Eccentric loading of a surface footing; (a): eccentric loading, (b): equivalent V-M loading.

In order to solve this problem, the area of the foundation remaining in contact with the soil needs to be found.
Seeking a solution for the problem, Meyerhof (1953) conducted series of laboratory model tests and proposed an
empirical procedure which is the effective width method. In this method, if the load is eccentric in the direction
of one of the footing dimensions, this dimension is reduced by a double of eccentricity.
To take in consideration the effect of eccentricity of loading in Terzaghi’s equation, the effective width
method is highly recommended (e.g. Meyerhof (1963); Hansen (1970)). Equation (1) represents the formula of
the bearing capacity after introducing this method, for a purely cohesive soil. For this same type of soil,
Ukritchon et al. (1998) used numerical upper and lower bound methods to examine the problem of a strip footing
on both uniform and non-uniform soil layers under eccentric- inclined loading. The results showed that the use of
the bearing capacity factors leads to an underestimation of the limit load by more than 25%, for the case of
homogeneous clay. And, those factors are not reliable in cases of clays with undrained shear gradient.
Houlsbay and Purzin (1998) used plasticity theorems to predict the limit load of footings in the presence of
the detachment at the soil-foundation interface. This later causes the violation of the normality condition at the
interface which affects the validity of those theorems. As a solution for this problem, he used a scaling concept

257
and derived "apparent upper and lower bound solutions". The results were presented in terms of failure
envelopes. It was found that no exact solution can be derived for the general loading combination case (vertical,
horizontal and moment loading). Taiebat and Carter (2002) studied the bearing capacity mobilized under strip
and circular footings using the finite element method. It was demonstrated that the effective width method gives
a good approximation of the bearing capacity under eccentric loading.
Michalowski and You (1998) used the upper bound theorem of the limit analysis method to study the bearing
capacity of cohesive, frictional and cohesive-frictional soils with different types of interface. It was found that the
effective width approach gives the same bearing capacity as the upper bound solution for the case of smooth
footings and underestimates the bearing capacity of cohesive and cohesive-frictional soils with less than 8 % if
e
the interface is not bonded and if the eccentricity ( < 0.1) is small. Equation (2) represents the solution
B
developed for the case of purely cohesive soil carrying a strip rough footing.

V e
 5.14(1  2 )cu (1)
A B

V e
 5.33(1  2 )c u (2)
A B

M
Equations (1) and (2) have the same form with different bearing capacity factors Nc . e is the
V
eccentricity of loading, M is the moment acting on the center of the footing base, V is the vertical loading, A
is the foundation’s area, cu is the undrained cohesion of soil.
In this paper, the undrained bearing capacity of strip footing resting on purely cohesive soil under eccentric
loading is investigated numerically using the finite difference code FLAC3D(2006) (Fast Lagrangian Analysis of
Continua in 3 dimensions). Basing on these results, the shapes of the failure mechanisms and the failure envelope
in the loading space (V, M) are investigated.

2 NUMERICAL MODEL

The choice of the code FLAC3D is justified by its efficiency in treating the problems of the bearing capacity of
shallow foundations. This efficiency appears clearly in the analyses conducted by Yousssef Abdel Massih and
Soubra (2007), Michalowski and Dawson (2002) and Puzakov et al.(2009).
The geometry of the numerical model is shown in Figure 2. The footing is strip, rough, rigid and of width B
resting on the soil surface. To insure that the foundation stays planar during the time of analysis, the height of the
foundation is within the range of [0.2B, 0.5B] (Loukidis et al.(2008)) and its undrained Young’s modulus is 103
times the one of the soil. In order to satisfy the plane strain conditions, the foundation is considered infinitely
long.

e
H B
z
y
x
L

Figure 2. Geometry of the numerical model.

The soil is weightless, homogeneous and purely cohesive. It is modelled as an elastic perfectly plastic
material obeying the Tresca failure criterion. The transmission of the load from the footing to the soil is insured
by interface elements. These elements have the soil cohesion to model a rough soil-footing interface. They don’t
have any resistance to tension and they behave according to Coulomb shear strength criterion.

258
The modeling parameters are recapitulated in Table 1. Since the resulting limit load can be normalized by the
foundation’s width B and the soil’s cohesion cu , there is no significance in the choice of their values. According
to Mabrouki et al. (2010), the values of the elastic properties of soil have a negligible effect on the bearing
capacity.
The height h and the width L of the discretized domain are 36B and 9B, respectively (Figure 2). It was
verified that the boundaries have no effect on the limit load and the development of the failure mechanism. The
movement in both directions horizontal (x) and vertical (z) is not permitted for both lateral boundaries and the
bottom of the model. In order to enable plane strain conditions, the displacement in the y direction is fixed for
the entire model.
The numerical model is discretized into elements with different sizes. The meshing is refined in the vicinity
of the foundation’s edges because they are considered as singularity points (Figure 3). This singularity is caused
by the abrupt changing in the direction of the displacement in this vicinity (Loukidis et al. (2008)).

Table 1. Modeling parameters

Parameters Soil Foundation Interface

Young’s modulus E (MPa) 25 25000
Poisson’s ratio υ 0.49 0.3
Cohesion cu (Pa) 100000 100
Normal stiffness kn (Pas/m) 109
Tangential stiffness ks (Pas/m) 109

Figure 3. Mesh of the numerical model in vicinity of the foundation.

The eccentric loading is applied by maintaining two vertical velocity profiles increasing linearly from zero
with a defined gradient and having opposite directions until plastic yielding is achieved.

3 RESULTS AND DISCUSSION

The purpose of this analysis is to study the form of the failure envelope in loading plane (V,M), the failure
mechanism and the normal stress distribution under a strip footing subjected to an eccentric loading.
The numerical bearing capacity factor Nc is equal to 5.27 overestimating the exact solution of Prandlt (1920)
(Nc =5.14) by 2.53%.
Figure 4 shows the numerical non-dimensional failure envelope. Every eccentricity is represented by a point
with coordinates (V/Acu, M/BAcu). V and M are the vertical load and the moment acting on the center of the
foundation base, respectively. They are calculated by integrating the normal stresses generated in the interface
elements using a FISH function. The non-dimensional failure envelopes based on equations (1) and (2) and the
one based on the results of Taibat and Carter (2002) are also represented in Figure 4. All the failure envelopes
are in good agreement for small magnitude of vertical load. However, it can be seen that for larger magnitude the
effective width rule underestimates the limit load and represents a lower bound for the other methods. This same
conclusion was found by Ukrichon et al.(1998) and Houlsby and Purzin (1999) . It can also be observed that the
curve of the present study is very close to the one of Michalowski and You (1998).

259
Figure 5 represents the normalized normal stress distribution at the soil-foundation interface with respect to
the x/B ratio, with x being the coordinate (in the (x) direction) of interface points. The normalized normal stress
is calculated as the ratio between the normal stress σ and the maximum normal stress  max corresponding to
each loading configuration. In case of non-eccentric loading, the stress distribution is symmetrical with respect to
the centerline of the foundation with the maximum values being under the two edges of the footing. Comparing
all cases together allows noticing that the increase of eccentricity changes this distribution in terms of shape and
size. The normal stress is reduced to zero in the vicinity of the left edge of the foundation. It is equal to zero in
this area because of the absence of the transmission of loading from the footing to the soil, which can be
explained by the detachment at the interface soil-foundation. With increasing eccentricity, this detachment
extends over a larger area.

0,7

0,6

0,5

Me ye rhof (1953)
0,4
M/BAcu

Michalowski and You

0,3 (1998)

Taibat and C arte r (2002)

0,2
curre nt study

0,1

0
0 1 2 3 4 5 6
V/Ac u

Figure 4. Non dimensional failure envelopes in the (V, M) plane of loading.

e /B = 0

0,2
e /B = 0,10

0,4 e /B = 0,20
σ/σ

e /B = 0,30
0,6

e /B = 0,40
0,8

e /B = 0,45

1
-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5
x/B

Figure 5. Normalized normal stress distribution at the soil-foundation interface.

260
V
The evolution of the ratio (ultimate load for different eccentricities by the ultimate load for e=0) with
V0
e
respect to the ratio is shown in Figure 6. It appears that the effective width method is in excellent agreement
B
V e
with the numerical results. Basing on equation (1), it is clear that is equal to (1  2 ) which is a linear
V0 B
relationship. Equation (2), as well, gives the same relationship.

current study
0,8
Meyerhof (1953)

0,6
V/V0

0,4

0,2

0
0 0,1 0,2 0,3 0,4 0,5
e/B

Figure 6. Normalised limit load V/V0 versus e/B ratio.

Displacement vectors represent the failure mechanisms of three loading cases (indicated schematically in
Figure 7). In case of non-eccentric loading (see Figure 7.a), one can see the formation of a rigid elastic wedge
under the base of the foundation (where the displacement vectors are vertical). It may, also, be noted that the
numerical failure mechanism extend over a larger soil mass than the one mobilized by the Prandlt’s failure
mechanism, which might be the reason for the overestimation in the numerical bearing capacity factor Nc.
Previous researchers reported the same remark and attributed it to the use of the elasto-plastic model (see
Yousssef Abdel Massih and Soubra, 2007). Regarding the remaining cases (Figures 7.b, 7.c), the failure
mechanism is characterized by the formation of a wedge which is due to the effect of the vertical loading in
addition to a scoop reflecting the moment effect. As long as the eccentricity increases, the size of the failure
mechanism becomes more and more smaller, the scoop is more and more formed close to the ground surface and
its pivot point of rotation is shifted towards the centerline of the footing explaining the decrease in the bearing
capacity. A similar mechanism was found by Ukritchon et al.(1998). The same remark raised previously on the
mobilization of a larger soil mass by the mechanism of the present study is present for the case of eccentric
loading as well.

261
(a)

(b)

(c)

Figure 7. Displacement vectors ; (a) non-eccentric loading, (b) : e/B = 0.1, (c) : e/B = 0.25.

4 CONCLUSION

A numerical simulation of the behavior of a strip footing resting on the surface of a purely cohesive soil and
subjected to eccentric loading has been conducted using the finite difference code FLAC3D. This code proved to
be a very powerful, efficient and precise tool for the treatment of this kind of problems. The distribution of the
normal stress at the interface soil-foundation confirms that the decrease in the bearing capacity with the increase
in eccentricity is due to the loss of contact pressure (detachment) between the foundation and the soil. This
detachment begins at the farthest foundation’s edge from the point of application of the eccentric loading and
extends over a larger area with increasing eccentricity. The decrease in the numerical failure mechanism with
respect to the increase in eccentricity, also, explains the decrease in the limit load. It was found that the effective
width method gives a reasonable estimation for the collapse load.

REFERENCES

FLAC3D (2006) Fast Lagrangian Analysis of Continua in 3 dimensions, Version 3.10. Minneapolis; ITASCA Consulting
Group, Inc.
Hansen, B (1970) A revised and extended formula for bearing capacity. Bulletin No. 28. Danish Geotechnical Institute, 5-
11.
Houlsby, G. T., Purzin, A. M.(1999) The bearing capacity of a strip footing on clay under combined loading. Proceeding of
the Royal Society of London Series A(455), 893-916.
Loukidis, D., Chakraborty, T., Salgado, R. (2008) Bearing capacity of strip footings on purely frictional soil under eccentric
and inclined loads. Can. Geotech. J. 45 (6), 768-787.
Mabrouki, A., Benmeddour, D., Frank, R., Mellas, M. (2010) Numerical study of the bearing capacity for two interfering
strip footings on sands. Computers and Geotechnics 37(4), 431-439.
Meyerhof, G.G. (1963) Some recent research on the bearing capacity of foundations. Can Geotech. J. 1 (1), 16-26.
Meyerhof, G.G.(1953) The bearing capacity of foundations under eccentric and inclined loads. Proc. 3rd Int. Conf. on Soil
[Link] Found. 1, 440-445.
Michalowski, R. L., Dawson, E. M. (2002) Three dimensional analysis of limit loads on Mohr-Coulomb soil. Foundations of
Civil and Environmental Engineering 1, 137-147.

262
Michalowski, R. L., You, L. (1998) Effective Width rule in bearing capacity of shallow footings. Computers and
Geotechnics 23(4), 237-253.
Prandtl, L. (1920) Über die Härte plastischer Körper. Narchrichten von der Königlichen Gesellschaft der Wissenschaften,
Göttingen, Math.-phys. Klasse, 74–85.
Puzakov, V., Drescher, A., Michalowski, R. L. (2009) Shape factor Sγ for shallow footings. Geomechanics and Engineering
1(2), 113-120.
Taibat, H.A., Carter, J.P. (2002) Bearing capacity of strip and circular foundations on undrained clay subjected to eccentric
loads. Géotechnique 52 (1), 61-64.
Terzaghi, K. (1943) Theoretical soil mechanics. J. Wiley, New York.
Ukritchon, B., Whittle, A. J., Scott, W. S. (1998) Undrained limit analysis for combined loading of strip footings on clay.
Journal of Geotechnical and Geoenvironmental Engineering, 265-276.
Youssef Abdel Massih, D., Soubra, A. H. (2007) Numerical simulations for the bearing capacity of strip footings. Advances
in Shallow Foundations.1-10.

263
264
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Comparaison entre la capacité portante in situ et théorique de différents pieux

battus dans les terrains alluvionnaires.

Comparison of Bearing Capacities for practical and theoretical driven Pilesin alluvial soils.

M. Emmanuel KENMOGNE (1),Mlle. Emmanuelle WEMEYI(2), M. Jean Rémy MARTIN

1
Directeur de l’Ingénierie à GEOFOR S.A, Douala, Cameroun.
2
Ingénieur géotechnicienne à GEOFOR S.A.

RESUME- Le calcul de la capacité portante des pieux battus se fait de manière théorique en utilisant soit le
Fascicule 62, Titre V, soit les Eurocodes, soit le DTU(Document Technique Unifié)ou sur le plan pratique en
utilisant la formule du constructeur des marteaux notamment ceux de marque DELMAQ. Ce qui permet de
ressortir le paramètre K qui peut être définit comme le rapport entre la capacité portante théorique (Qadm) et
réelle (Gmin/max). Ce paramètre K peut être un indicateur du type et de la géométrie du pieu en fonction de la
lithologie des terrains à traverser.

ABSTRACT-The bearing capacity of the driven piles can be obtained theorically using Fascicule 62, TITRE V
according to the French standard or using hammerconstructor formulae during field operations. This article
present the K parameter considered as the report Qadm/ Gmin( between the theoretical value of bearing capacity
and the practical value obtained in the field) , This parameter depend on the geometry of the pile and the
lithology of the soil.

1. INTRODUCTION

Le calcul de la capacité portante des pieux battus se fait de manière théorique en utilisant soit le Fascicule 62
Titre V ou , soit les règles de l’Eurocode, le DTU sur le plan pratique en utilisant la formule du constructeur des
marteaux de marque DELMAQ. Cependant l’on remarque que les résultats de portance pratique sur le site ne
reflètent pas toujours les données obtenues en Bureau d’Etude. Ce qui nous amène à établir un rapprochement
entre les capacités portante réelles (Gmin/max) sur site et théorique (Qadm) des pieux battus dans les terrains
alluvionnaires (sable et argile). C’est dans ce contexte que s’inscrit le présent article dont l’objectif fondamental
est de faire ressortir une relation entre (Gmin/max) et (Qadm) obtenues respectivement par la formule modifiée de
GATES des marteaux type DELMAG (D22 et D46) et par les formules classiques du fascicule 62 titre V. Pour y
parvenir, nous avons collecté des données de terrain (essais pressiométriques, nombre de coups, caractéristiques
du marteau et la géométrie des pieux) sur plusieurs sites de battage au Cameroun et dans la sous région d’Afrique
Centrale. L’approche méthodique utilisée est le traitement des données basée sur les différentes formules et
suivant la répartition ci après en fonction de la géométrie des pieux:

Nombre Linéaire Valeur moyenne de K en fonction de

Catégorie Type de pieu
de pieux totale (m) la lithologie
Sables : 0,92<K<1,82
Catégorie I pieux béton 35cmX35cm 74 1110
Argile : 0,2<K<0,82
Catégorie II Pieux acier HEB 400 52 406 Sables : 0,21<K<1,5
Sables : 1,53<K<1,92
Catégorie III pieux acier tubulaire 30’’ 21 441
Argile : 1,52<K<1,82

Ce tableau fait ressortir le paramètre K qui peut être définit comme le rapport entre les capacités portantes
théoriques(Qadm)
et réelle (Gmin/max). Il montre que pour les pieux de catégorie III, le coefficient K est supérieur à 1 aussi bien dans
les sables que dans les argiles. Pour la catégorieI,K est supérieur à 1 dans les sables et inférieur à 1 dans les
argiles et dans la catégorie II K est inférieure à 1 dans les sables. Ce paramètre K peut être un indicateur du type
et de la géométrie du pieu en fonction de la lithologie des terrains à traverser.

265
2. MATERIELS ET METHODES

2.1. Prévision de battage et justification du matériel de battage

Une étude de battage comporte deux étapes distinctes et complémentaires, suivie d'une étape de synthèse.
 Etape (1):Détermination de la résistance du sol au battage
La résistance du sol au battage (ou SRD, de l'anglais "Soil Resistance to Driving") tient compte des propriétés
mécaniques (caractère frottant) du matériau préalablement déterminées (reconnaissance géotechnique), du type
de pieu (ouvert ou fermé, diamètre et section, présence éventuelle d'un sabot, ...) et des phénomènes particuliers
tels que, pour les pieux ouverts, la formation d'un bouchon. En pratique, on est amené à raisonner sur une
fourchette de valeurs et à définir un profil minimal et un profil maximal de résistance en battage continu, ainsi
qu'un profil après cicatrisation du sol suite à un arrêt de battage
La détermination de la SRD fait appel au savoir-faire du géotechnicien.
Les procédures utilisées sont en particulier inspirées des recommandations de Toolan et al (1977), Stevens et al
(1982) et Puech et al (1990). Elles sont bien évidemment d'autant plus efficaces qu'elles s'appuient sur une bonne
connaissance du sol considéré, sur une longue expérience du battage dans le type de sol rencontré et qu'elles ont
pu être recalées sur des cas réels sérieusement documentés.
 Etape (2) : Simulation du comportement du complexe marteau-pieu-sol
Le résultat s'exprime sous la forme d'une relation, pour le système considéré, entre la résistance du sol au battage
et le nombre de coups nécessaires pour enfoncer le pieu de 50 cm (courbes SRD-N/50cm,) confère figure 1 ci-
dessous. Ces courbes sont établies pour différentes profondeurs de pénétration, et, pour quelques valeurs
représentatives du rendement global du marteau considéré et de la répartition relative de l'effort résistant entre la
pointe et le frottement latéral.

Figure 1 : Courbe SRD pour pieux tubulaires

 Etape (3) : Courbes prévisionnelles de battage et choix du marteau.

La comparaison des résultats de la simulation et des courbes de résistance permet d'établir les courbes donnant le
nombre prévisionnel de coups par 50 cm en fonction de la pénétration et, ainsi, de juger de la faisabilité du
battage. On admet généralement que le refus effectif se situe entre 250 et 350coups par 50cm, selon le type de
marteau utilisé.
Cependant,Il est prudent de choisir le marteau, de manière à ce qu'en battage continu, le nombre de coups par 50
cm demeure inférieur à 150. Pour les sols argileux, où le risque de cicatrisation rapide est élevé, la limite ci
dessus pourra être revue à la baisse.

266
En conclusion la procédure suppose que l’étude de battage soit réalisée avant la commande et la mise en
oeuvre des pieux. Si aucontraire la fabrication des pieux est lancée avant l’étude du battage, celle-ci revient
alors à choisir le marteau et à déterminer les procédures de mise en place. Les contraintes dynamiques
maximales en cours de battage sont limitées à 60% de la limite élastique de l’acier. Dans certains cas les
contraintes de 70 à 80% de la limite de l’acier sont admissibles.
Pour notre cas les marteaux DELMAG pris en compte sont le D46-13 et D22-13 avec une énergie minimale de
6700m/kg soit 52.200 [Link].

2.2. Méthode utilisée

Pour aboutir à un résultat cohérent lors de notre étude comparative, nous avons procédé à une synthèse des
données sur chaque site. Les différents sites étudiés ne présentant pas une uniformité lithologique, nous avons
considéré que les sables présents sont de classe A et B, lâches et moyennement compactes selon le Fascicule 62.
Les argiles présentes sont de classe B.
Les sites présentant les caractéristiques ci-dessus sont NGATTA, MVIA et OLAM. Cependant nous allons
étudier simultanément les sites de NGATTA et MVIApuisqu’ils comportent les pieux de catégorie III. Par
ailleurs le site d’Olam est classé dans la catégorie I et enfin les sites MAYIDOMBE et LUFIMI dans la catégorie
II avec la présence des sables de classe A.

3. RESULTATS OBTENUS

3.1. Courbes de portance selon les formules classiques et la formule du constructeur

Les résultats des différentes capacités portantes sont représentés sur les schémas ci-dessous. La figure
2représente la capacités portante à ELU( Etat limite Ultime) calculées en fonction de la profondeur par les
formules statistiques et par celle du constructeur sur les cinq chantiers regroupant les trois catégories de Pieu.
Nous noterons Qadm la capacité portante calculée à partir de Fascicule 62 et Gmoy la moyenne des capacités
portantes maximales et minimales obtenues par la formule de GATES.

Figure 2. Valeurs de la capacité portante (Qadm) dans les sables et les argiles pour les pieux de catégorie III et I.

267
Profil 1

Profil 2

Profil 3

Figure 3. Valeurs de la capacité portante (Gmoy) dans les sables et les argiles : Pieu de catégorie III
Légende
Profil 1 : correspond à un accroissement rapide de Gmoy dans les cinq premiers mètres de profondeur
Profil 2 : zone de faible augmentation de la portance entre 600 et 800 KN sur une dizaine de mètres entre 5 et 15 m.
Profil 3 : caractéristique de la proximité de zone de refus entre 20 et 25 m pour un Gmoy entre 1000 et 1400 KN
correspondant aux argiles compacts.

Figure 4. Valeurs de la capacité portante (Gmoy) dans les sables : Pieu de catégorie I
Légende
Profil 1 : pour de profondeurs inférieures à 10 m et Gmoy inférieur à 700 KN

268
Profil 1

Profil 2

Figure 5. Valeurs de la capacité portante (Gmoy) dans les sables et argiles : Pieux de catégorie II
Légende
Profil 1 : caractérise des zones de remblais et de sables limoneux de surface
Profil 2 : caractérise des sables compacts et très frottant.

3.2. Lithologie des sites

La lithologie des sites est présentée dans le tableau ci-dessous :

Tableau 1 : Coupe lithologique du terrain pour les différents sites des pieux
74 0,00m -9,00m Argile sableuse
échantillons
9,00m -13,50m Sable argileux
testés
13,50m -19,50m Argile
Catégorie I
19,50m -24,50m Sables argileux
24,50m -28,50m Sables moyens
28,50m -35,50m Argile sableuse

52 pieux 0,00m -3,00m Remblai de sable

testés
3,00m -4,00m Sables limoneux
Catégorie II
4,00m -6,50m Sables grisâtres
6,50m -26,00m Sables blancs très compacts

21 pieux
testés 0,00m -2,00m Sables fins
Catégorie III 2,00m -5,00m Argiles
5,00m -17,00m Sables
17,00m -26,00m Argiles

269
3.3. Présentation du rapport K = Qadm/Gmoy

Tableau 2 : Valeur du coefficient K en fonction de la catégorie du pieu sur les différents sites
Pieux de categorie I
ELS ELU
Dans les sables Moyenne 1,47 1,57
Dans les argiles Moyenne 0,81 0,71

Pieux de catégorie II
0,94
Dans les sables Moyenne 0,94
0,35
Dans les argiles Moyenne 0,35

Pieux de catégorie III( concerne 17 pieux)

ELS ELU
Dans les sables
Moyenne 1,84 1,60
Dans les argiles
Moyenne 1,85 1,58
Pieux de catégorie III (concerne 4 pieux
ELS ELU
Dans les sables
Moyenne 0,63 0,82
Dans les argiles
Moyenne 1,38 0,98

Pour les pieux de catégorie I, le coefficient K obtenu aussi bien à l’ELS qu’à l’ELU est supérieur à 1 dans les
sables et inférieur à 1 dans les argiles. Tandis que pour les pieux de type III, K est supérieur à 1 dans les sables et
argiles. Les pieux de catégorie II par contre ont une valeur Kinférieur à 1 dans les sables.

4. DISCUSSIONS

Les résultats obtenus sur les sites montrent que la capacité portante réelle Gmoy du sol varie avec la profondeur
et le type de pieu (figure 3 à 5 ) avec des profils différents.
Il est important de remarquer que pour les pieux tubulairesde catégorie III employés sur le site de NGATTA et
LUFIMI, nous avons les battus fermés et ouverts. Cependant, les valeurs obtenues sur le chantier deNGATTA
sont plus élevées que celles de MVIA. Cette différence est due à l’emploi des pieux battus fermés de 30’’ qui
selon les auteurs tels que (Kyuho Paik; Rodrigo Salgado; et al. 2003) démontrent dans leur article
intitulé « Behavior of Open-and Closed-Ended Piles DrivenIntoSands »qu’après plusieurs tests dynamiques et
statiques, la capacité portante des pieux ouverts est plus faible que les pieux fermés.
Pour les pieux fermés l’accroissement de la résistance en pointe lors du battage et la résistance le long du fut
(frottement latéral) concourent à une augmentation de la capacité portante. Par contre dans les pieux ouverts
l’ouverture ou l’absence de la pointe lui assure une ouvrabilité plus facile (sans compression ou bouchon du sol)
diminuant ainsi sa résistance.
Ceci étant dans les pieux de catégorie III, le choix est donc beaucoup plus orienté vers les battus fermés qui
portent plus que les battus ouverts. Nous avons de 0 à 26m (Qadm) compris entre 0 et 2500KN pour les battus
fermés et 0 à 1200KN pour les battus ouverts. Et (Gmoy) compris entre 0 et 1500KN pour les battus fermés et 0 à
1300KN pour les battus ouverts.
Par ailleurs les valeurs de (Qadm) et (Gmoy) des pieux deCatégorie II sont respectivement comprises entre 712
et 900KN puis 0 et 1500KN pour des profondeurs de [0-9m]. Tandis que pour les pieux de catégorie I(Qadm)
varie de 0 à1700KN de 0 à 16m puis(Gmoy)varie de 0 à 500KN de 0 à 9m et 500 à 900KN de 9,5 à 16m.
Nous remarquons que les valeurs de (Qadm) sont plus optimistes que celles de(Gmoy) quelle que soit le site
étudié. Il ressort ainsi qu’en plus de la lithologie, le choix du Marteau, la méthode d’exécution et la nature du
pieu ont un impact directsur la portance des pieux de fondation.
Les études réalisées nous montrent que les capacités portantes augmentent avec la profondeur quelle que soit
le diamètre ou le type de pieu mis en place. Suivant le rapport K, elle est plus importante dans les sables que

270
dans les argiles par phénomène de frottement latéral et aussi de reconsolidation du sol (telle que expliquée par
certains auteurs précités ci-dessus).
Cependant bien que les résultats obtenues corroborent avec les avis de plusieurs scientifiques, il ne sera pas
évident pour nous d’établir des corrélations directes entre (Qadm) et (Gmoy).Car pour parvenir à une quelconque
relation, il serait indispensable d’analyser plusieurs échantillons sur plusieurs sites. C’est pour cette raison que
nous nous sommes appesantis uniquement sur les comparaisons des différentes capacités portantes pour
différents type de pieu.

5. CONCLUSIONS

Cette étude nous a permis de constater que les valeurs des capacités portantes énumérées ci-dessous peuvent être
influencées par divers facteurs à savoir :
 Le choix du type de Marteau et la méthode d’exécution
 La lithologie du site
 Les caractéristiques géométriques des pieux

L’objectif de notre article étant de ressortir une étude comparative entre la capacité portante théorique et celle
calculée par la formule du constructeur, nous constatons que quelque soit le sol concerné la capacité notée (Qadm)
théoriqueest plus grande que celle notée (Gmoy).Notons que cette simulation n’est qu’estimative et à titre
indicative compte tenu de la situation géographique des sites (qui n’a pas été prise en compte) et du nombre
d’échantillons étudiés assez restreint. Le rapport K obtenu permet de néanmoins caractériser les différents pieux.
Il apparait que les pieux de section circulaire (catégorie III) offre une portance théorique plus élevée par
rapport à la capacité portante réelle obtenue sur le terrain,aussi bien dans les sables que dans les argiles par
opposition aux pieux en béton préfabriqués de section 35 X 35(catégorie I) et aux HEB (Catégorie II).

REFERENCES

Afnor, (2000), Norme NF P94 110-1: essais pressiométriques Ménard. Partie 1, essais sans cycle
Ermias.H.M,Lars.O.G( july 2012), A comparison of friction piles bearing capacity based on theorical and empirical
mathematical models,Norveigian University of Sciences and Technology,116 pages
Franki ,(2008).A Guide to Practical geotechnicalengineering in Southern Africa: fourth edition.440 pages
Fascicule N° 62-Titre V.,Règles techniques de conception et de calcul des fondations des ouvrages de Génie Civil.
Fondation profonde par pieu H /Etat de l’art : PROFILARBED S.A /Arcelor Group, document de synthèse, 13 pages
Kyuho P., Rodrigo S., et al. (2003).Behaviour of open and closed – ended piles driven into [Link] of Geotechnical and
Geoenvironmental Engineering ASCE/April 2003.

271
272
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Développement d’une loi t-z cyclique à partir d’essais en laboratoire et

d’expérimentations en vraie grandeur

Proposal of t-z cyclic model by means of laboratory tests and full-scale experimentations

Z. Abchir1, S. Burlon1, R. Frank2 et A. Le Kouby1

1
Université Paris-Est, IFSTTAR, Paris, France
2
Université Paris-Est, Laboratoire Navier, CERMES, Paris, France

RÉSUMÉ – Les pieux supportant des éoliennes ou des structures off-shore sont soumis à des charges cycliques
et leur dimensionnement requiert des approches spécifiques. Actuellement, les normes et les codes de calcul en
application comme l’EUROCODE 7 et les règles API soulignent l’importance de considérer ces sollicitations
sans proposer des méthodes de dimensionnement simples et applicables aux structures on-shore. Le calcul de
ces déplacements nécessite de prendre en compte les variations de résistance et de rigidité de l’interface sol-pieu
au cours du chargement. Le présent article propose une loi t-z pour l’estimation des déplacements d’un pieu
soumis à un chargement cyclique axial. Cette loi inclut un paramètre contrôlant la dégradation du frottement
axial au fur et à mesure des cycles. Ce dernier est estimé selon une approche de dégradation du frottement axial
issue de l’analyse d’essais en laboratoire. Les résultats numériques obtenus au moyen de cette nouvelle loi sont
comparés aux résultats expérimentaux issus d’essais de chargement cyclique en traction effectués en vraie
grandeur dans le sable de Dunkerque en France par l’Imperial College.

ABSTRACT – Piles of wind turbines or offshore structures are subjected to cyclic loads and their design
requires specific approaches. Currently, design codes such as Eurocode 7 and API rules emphasize the
importance of considering these loadings without suggesting simple design method; in particular that can be
applied to onshore structures. The calculation of these displacements requires taking into account the variations
in strength and stiffness of the soil-pile interface during the cyclic loading. This article proposes a t-z law to
estimate displacements of a pile subjected to axial cyclic loading. This law includes a parameter controlling the
degradation of the axial friction during this kind of loading. This parameter is calculated from laboratory tests
and using a specific approach of degradation of axial friction. The numerical results obtained by this new law are
compared with experimental results coming from full scale tension cyclic loading tests carried out in the sand of
Dunkerque in France by Imperial College.

1. INTRODUCTION
Les fondations profondes d’une large gamme d’ouvrages géotechniques subissent une forte composante de
sollicitation cyclique. Les structures soumises à l’action du vent, de la houle ou des charges roulantes en sont
quelques exemples. Malgré le nombre d’ouvrages concernés, peu de solutions pratiques au niveau de l’ingénierie
ont été apportées pour résoudre le problème complexe de la prévision du comportement des pieux sous
chargement cyclique. Les règlements actuels de dimensionnement des fondations profondes (NF P 94-262,
Eurocode 7) n’apportent que peu de recommandations pour prendre en compte ce caractère cyclique des
sollicitations. En particulier, ces recommandations ne permettent pas d’identifier le niveau de sécurité des
ouvrages subissant ce type de chargement.
Le chargement cyclique impacte le comportement du pieu. En effet, les cycles successifs de charge et de
décharge provoquent une accumulation des déplacements permanents en tête de pieu (Chan et Hanna, 1980 ;
Van Weele, 1979). Ce mécanisme conduit à des désordres au niveau de la structure portée. Afin de prendre en
compte les effets de cette accumulation de déplacements, plusieurs travaux, développés notamment dans
l’ingénierie pétrolière, ont été réalisés (Chin et Poulos, 1992; Randolph, 1986). Par ailleurs, il existe trois
grandes méthodes pour calculer les déplacements d’un pieu isolé (Christoulas et Frank, 1991) : la méthode des
courbes de transfert (méthode t-z), les méthodes des milieux continus comme la méthode des éléments finis et la
méthode d’élasticité utilisant les équations de Mindlin (Poulos et Davis ,1968). Le présent article privilégie la
méthode des courbes de transfert. Elle permet de relier le frottement axial mobilisé au tassement vertical du pieu
d’une part, et d’autre part la résistance en pointe mobilisée et le tassement vertical de la pointe. Les courbes de
transfert sont données pour l’ensemble des sections du pieu. Certaines lois t-z cycliques ont été développées pour
calculer les déplacements des structures off-shore (Randolph, 1986). Le modèle sur lequel se base le présent
article est celui présenté par Burlon et al. (2013). Il représente une extension de la loi t-z formulée par Frank et

273
Zhao (1982) qui permet de calculer les tassements d’un pieu uniquement sous charge statique axiale. Ce modèle
t-z cyclique (modèle TZC) est apte à rendre compte des principaux effets des chargements cycliques observés en
laboratoire : l’adaptation, et le rochet (Lemaitre et Chaboche, 2009). Il permet également d’imposer un
durcissement ou une dégradation du frottement axial et de la résistance en pointe au cours du chargement
cyclique. Cette amplitude de durcissement ou de dégradation cyclique peut être calculée selon différentes
méthodes. Dans cet article, la dégradation est calculée en s’inspirant de la méthode ABC développée par
l’Imperial College dans le cadre du projet GOPAL pour des pieux installés dans du sable (Jardine, 1998). Cette
méthode relie la variation du frottement axial à la variation de la contrainte radiale effective au cours du
chargement cyclique. Une comparaison entre des résultats expérimentaux issus de trois essais de chargement
cyclique en vraie grandeur effectués dans le sable de Dunkerque et des résultats numériques obtenus au moyen
du modèle TZC combiné à la méthode ABC est proposée. Une discussion est menée afin d’analyser la validité de
la comparaison effectuée.

2. MODELE DE CALCUL DES DEPLACEMENTS

2.1 Méthode des courbes de transfert

Dans le présent article, les déplacements sont calculés au moyen de la méthode des courbes de transfert. Elle
consiste à discrétiser le pieu en une série de tronçons et à assimiler le sol autour de la fondation à des ressorts
non-linéaires agissant indépendamment les uns des autres. Au cours du chargement, le frottement axial et la
résistance en pointe sont mobilisés progressivement en fonction du tassement du tronçon de pieu. En écrivant
l’équilibre d’un élément de pieu supposé élastique, l’équation différentielle d’équilibre notée (1) est obtenue.

d ² wz
Eb Ab  Bq s ( wz )  0 (1)
dz ²
avec Eb le module d’Young du pieu, Ab l’aire de la section, wz le tassement du pieu à la profondeur z, B le
diamètre du pieu et qs le frottement axial mobilisé. La résolution de cette équation différentielle se fait en
générant des courbes de mobilisation du frottement axial en fonction du tassement du pieu ainsi que la courbe de
mobilisation de la résistance en pointe en fonction du déplacement vertical de la base du pieu. Des conditions
initiales supplémentaires sont imposées en tête du pieu en termes de déplacement ou de charge. A la base du
pieu, la loi de réaction joue le rôle de condition limite.

2.2 Modèle TZC (Burlon et al., 2013)

Le modèle TZC présenté par l’équation (2), relie le frottement axial qs au déplacement vertical wz et rend compte
des phénomènes de radoucissement et de durcissement cyclique.

w z  w zi
R
ncyl 1 
q s ( wz )  q s ;i  A(1) (q s 0  q s )(1  e ) (2)

Cette équation regroupe six paramètres qui sont présentés dans le tableau 1. Le terme Δqs peut être associé à
différents paramètres tels que le nombre de cycles, la vitesse de chargement, l’amplitude cyclique ou encore aux
phénomènes de contractance et de dilatance.
Les paramètres statiques qs0 et  peuvent être déterminés par analogie avec le modèle de Frank et Zhao. Les
paramètres cycliques (ρ et ξ et Δqs) ne peuvent pas être déterminés à partir des essais usuels. Ils sont à caler sur
des essais de chargement cyclique. Le paramètre Δqs contrôlant la dégradation du frottement axial au cours des
cycles est déterminé dans cet article selon la méthode ABC qui est présentée dans le paragraphe suivant. Les
paramètres ρ et ξ sont égaux à 1 et 0 respectivement.
Pour modéliser le comportement de la base du pieu, un modèle similaire à l’équation (2) relie la résistance en
pointe qb au déplacement vertical de la pointe wb. Le paramètre qb0 correspond à la résistance en pointe
maximale et est déterminé selon la norme nationale française d’application de l’Eurocode 7 relative aux
fondations profondes (AFNOR 2012).

274
3. METHODE DE CALCUL DE LA DEGRADATION DU FROTTEMENT AXIAL

Tableau 1. Paramètres de la loi t-z cyclique

Paramètre Description
i Nombre d’inversion de cycle de chargement
ncyl Nombre de total charges et de décharges (nombre de ½ cycles)
qs0 (kPa) Frottement axial limite
Δqs (kPa) Amplitude de dégradation ou de durcissement cyclique
α Raideur de la courbe de mobilisation du frottement axial pour un chargement monotone
( ncyl 1)
A qs ;i  (1) (q s 0  q s )
A
q s 0  qs
qs ;i Frottement mobilisé à l’inversion de charge (qs ;i = 0 pour i = 1)
wzi Déplacement relatif de l’interface sol-pieu au cycle i (wzi = 0 pour i = 1)
 ( ncyl 1)  ( ncyl 1)
R Re   (1  e )
ρ Paramètre contrôlant l’adaptation, l’accommodation et le rochet ou la relaxation
ξ Paramètre contrôlant la vitesse d’adaptation, d’accommodation et de rochet ou de relaxation

3.1 Méthode ABC

Cette méthode a été développée par l’Imperial College pour rendre compte de la diminution ou l’augmentation
du frottement axial le long du fût du pieu au cours d’un chargement cyclique (Jardine, 1998). Cette méthode
suppose que la variation de frottement axial est liée à une variation de la contrainte radiale effective autour du
pieu et est basée sur l’analyse d’essais de laboratoire (essais de cisaillement simple). La variation relative de
contrainte radiale effective Δσ’rcyclic est exprimée dans l’équation (3) en fonction de l’amplitude de cisaillement
imposée τcy et de la résistance ultime de cisaillement de l’échantillon τmax static. Le terme τmax static est défini comme
étant le produit de σ’r0 par tanδ où σ’r0 est la contrainte radiale effective initiale et δ est l’angle de cisaillement à
l’interface.

 'rcyclic  cy
 A(  B) N C (3)
 'r 0  max static

L’équation (3) est remplacée par l’équation (4) lorsque les valeurs de contraintes radiales et d’angle de
cisaillement à l’interface δ ne sont pas accessibles. Dans l’équation (4), le terme  'r 0 tan  est remplacé par la
résistance au frottement qs0 (appelé aussi frottement axial limite) calculé à partir de la norme NF P 94-262. La
formule exprimée dans l’équation (4) est celle utilisée dans les calculs et résultats présentés dans la suite de cet
article.

qs 
 A( cy  B) N C (4)
qs 0 qs 0

Dans la méthode ABC, le chargement cyclique est modélisé comme une série de paquets. La figure 1a
indique que cette sollicitation est définie par une charge moyenne Qmoy et une amplitude de chargement Qcy
constante durant un nombre de cycles Ntot. Ces deux paramètres Qmoy et Qcy découlent des valeurs maximales
Qmax et minimales Qmin du chargement et ce comme l’indiquent les équations (5) et (6). La figure 1b montre que
le frottement axial est modélisé d’une manière similaire. Pour chaque profondeur z donnée du pieu, le frottement
axial mobilisé est défini par un frottement moyen τmoy et une amplitude cyclique τcy. Ces deux paramètres sont
donnés dans les équations (7) et (8). Ce frottement mobilisé est limité par les valeurs de frottement axial limite
en compression τmax static comp et en traction τmax static trac. Le nombre total des cycles appliqués est divisé en une série
de paquets de Ns cycles Les valeurs de τmoy, τcy, τmax static comp et τmax static trac sont actualisées à la fin de chaque Ns
cycles. Cette actualisation traduit la diminution du frottement ou son augmentation qui peuvent être
respectivement associés à des phénomènes de radoucissement ou de durcissement cyclique.

275
Qmax  Qmin Q  Qmin
Qmoy  et Qcy  max (5) et (6)
2 2

 max   min  max   min

 moy  et  cy  (7) et (8)
2 2
Les valeurs du frottement axial limite en compression τmax static comp et en traction τmax static trac sont actualisées au
bout de chaque Ns cycles Cette actualisation se fait suivant les équations (9) et (10). Les nouvelles valeurs de τmax
static comp et τmax static trac sont réinjectées dans le calcul des Ns cycles suivants.

Figure 1. a) Modélisation de la charge cyclique appliquée, b) Modélisation du frottement axial mobilisé au cours du chargement cyclique
à une profondeur z donnée du pieu

 max staticcomp   max staticcomp ( N s )  qs (9)

 max statictrac   max statictrac ( N s )  qs (10)

3.2 Paramètres A, B et C

Les paramètres proposés dans la méthode ABC ont été obtenus à partir de cinq essais de cisaillement simple
cyclique effectués par FUGRO (Jardine, 1998). Les cinq échantillons de sol utilisés dans ces essais sont
identiques et proviennent du sable très dense de la mer du Nord (Dr proche de 100%). Chaque échantillon a subit
quatre chargements cycliques à basse fréquence (f=0,05Hz) avec respectivement des amplitudes Qcy croissantes.
L’interprétation a été établie en comparant la variation de la contrainte normale effective Δσ’ncyc après
respectivement 5, 10, 50 et 100 cycles, à la contrainte normale effective initiale σ’n0. Le jeu 1 des paramètres A,
B et C, présenté dans le tableau 2, a été adopté au départ car il permettait d’obtenir l’incertitude minimale du
modèle de dégradation. Afin de tester la validité du jeu 1, la dégradation calculée à partir du modèle utilisant les
paramètres A, B et C a été comparée à la dégradation observée lors de chargements cycliques de traction
effectués in-situ. A l’issue de cette comparaison, une surestimation des cycles nécessaires pour atteindre la
rupture a été notée. Afin de corriger cette surestimation, deux essais de traction à Dunkerque on été réalisés (WS
Atkins Consultants Ltd, 2000). Le jeu 2 dans le tableau 2 donne les nouvelles valeurs des paramètres.
Cependant, dans les calculs réalisés dans cet article, l’utilisation du jeu 1 ou du jeu 2 a conduit à une rapide
accumulation des déplacements en tête de pieu et donc à une rupture plus rapide que celle observée lors des
essais expérimentaux. Le jeu 3 a été utilisé dans toute la suite de l’article afin de corriger cette tendance et pour
s’approcher des résultats observés in-situ.
Cette modification a été effectuée après avoir étudié l’influence de chaque paramètre sur le comportement du
pieu au cours du chargement cyclique. En effet, le coefficient A contrôle la vitesse d’accumulation des
déplacements en tête de pieu. Plus sa valeur absolue est élevée, plus les déplacements s’accumulent rapidement.
Le coefficient B, quant à lui, contrôle le nombre de couches dégradées au cours du chargement cyclique et le
nombre de couches subissant un durcissement. Plus le coefficient B est élevé en valeur absolue, plus le nombre
de couches subissant une dégradation est faible. Le coefficient C, quant à lui, n’influence que faiblement
l’accumulation des déplacements en tête de pieu.

276
Tableau 2. Paramètres de la méthode ABC
Jeu Paramètres
A -0.083
1 B -0.060
C 0.355
A -1.245
2 B -0.060
C 0.355
A -0.010
3 B -0.150
C 0.355

4. DONNEES EXPERIMENTALES

Les données expérimentales utilisées dans cette étude sont issues d’une campagne d’essais in-situ de
chargements cycliques de pieux en vraie grandeur. Ces expérimentations ont été effectuées au cours de l’année
1998 dans le cadre du projet GOPAL et des recherches de HSE (Jardine et Standing, 1998). Le site expérimental
se situe dans le port de Dunkerque au nord de la France. Le sol de ce site est constitué de sable dense similaire à
celui trouvé dans la mer du Nord. Les pieux utilisés sont des pieux en acier, ouverts et mis en place par battage
en août 1998. Les caractéristiques de ces pieux sont les suivantes : longueur approximative D égale à 19 m, un
diamètre extérieur B égal à 457 mm, une épaisseur en tête de 20mm et une épaisseur en pointe de 13.5mm.
Chaque pieu a subit un essai monotone afin de déterminer sa résistance ultime en traction Qs, puis deux
séquences de chargement cycliques pour le pieu R3 et uniquement une séquence de chargement cyclique pour le
pieu R4. Si la rupture n’est pas atteinte à l’issue du chargement cyclique, un nouvel essai de traction est effectué
pour mesurer la nouvelle résistance ultime en traction post-cyclique Qs post cyc. Le tableau 3 résume les
caractéristiques des essais de chargements cycliques analysés par la suite.

5. RESULTATS

5.1 Accumulation des déplacements en tête

Tableau 3. Caractéristiques des essais de chargements cycliques réalisés in-situ.

Résistance
Résistance ultime
Nom de ultime à la Nombre Rupture/ Pas
Pieu Type de l’essai Qmoy/Qs Qcy/Qs finale (post-cyclique)
l’essai traction Qs de cycles de rupture
Qs post cyc (KN)
(KN)
2.R3.CY2 Traction cyclique 2317 0.30 0.30 200 Pas de rupture 2050
R3
2.R3.CY3 Traction cyclique 2050 0.46 0.46 13 Rupture --
R4 2.R4.CY2 Traction cyclique 2963 0.34 0.34 221 Rupture --

Pour chacun des pieux R3 et R4, les paramètres de calcul ont été calés sur les essais de traction illustrés sur la
figure2. Le module d’Young de l’acier est supposé égal à 200 GPa. La figure 3 propose, pour les essais de
tractions cycliques 2.R3.CY2 et 2.R3.CY3 la comparaison entre les résultats expérimentaux et les résultats
numériques. Pour l’essai 2.R3.CY2, une charge moyenne et une amplitude cyclique égales à 30% de Qs sont
appliquées. Les mêmes tendances d’accumulation des déplacements au fur et à mesure des cycles sont observées
pour les calculs et les mesures. En effet, les calculs et les mesures montrent que la rupture n’est pas atteinte après
200 cycles de traction. Le soulèvement expérimental maximal en tête de pieu est de 12,10 mm, tandis que le
soulèvement maximal calculé est de 10,08 mm. De plus, le phénomène de dégradation de la résistance ultime à la
traction est noté dans les deux cas numérique et expérimental : Qspostcyc est de 1940 KN pour les calculs et pour
les mesures Qs post cyc est égale à 2050 KN.
Une nouvelle séquence de chargement est appliquée sur le pieu R3 (essai 2.R3.CY3). Dans cette séquence, la
charge moyenne et l’amplitude cyclique sont augmentées (prises égales à 46% de Qs). La figure 3 montre que
dans les cas expérimentaux et numériques, la rupture est atteinte au bout de 13 cycles et 15 cycles
respectivement. La rupture observée est due à la dégradation de la résistance ultime en traction au cours du
chargement cyclique. L’augmentation de Qmoy et Qcy accélèrent alors la dégradation du frottement et par
conséquent la rupture.
Le pieu R4 possède une résistance à la traction initiale Qs plus élevée que celle du pieu R3 (2963 KN).
L’essai 2.R4.CY2 lui est alors appliqué. On observe que la rupture est atteinte dans les calculs et in-situ. En effet,
le critère de rupture fixé à un déplacement en tête supérieur à B/16 est noté au bout de 221 et 121 cycles
expérimentalement et numériquement, respectivement. Cette différence traduit une vitesse d’accumulation des

277
déplacements plus élevée dans les calculs avec le modèle TZC et les paramètres choisis. Cependant, le
déplacement résiduel est équivalent : 19.31 mm mesuré et 20 mm calculé.

Figure 2. Modélisation de l’essai de traction pour les pieux R3 et R4

Figure 3. Modélisation des essais de traction cyclique 2.R3.CY2, 2.R3.CY3 et 2.R4.CY2

5.2 Dégradation du frottement axial limite qs0

La dégradation progressive du frottement axial limite en fonction de la profondeur prédite par le modèle TZC
pour l’essai 2.R3.CY2 est illustrée sur la figure 4a. Le facteur de dégradation noté Fdég est donné par l’équation
(11) où qs0dégradé correspond au frottement axial limite dégradé après N cycles et qs0 est le frottement axial limite
initial.

qs 0 dégradé ( N )
Fdég  1  (11)
qs 0
D’après cette figure, le facteur de dégradation des couches situées au dessus de 4/5D (environ 15 m de
profondeur) augmente avec le nombre de cycles. Ceci traduit une dégradation du frottement axial limite au cours
du chargement. Pour les couches de sol situées en dessous de 4/5D et plus particulièrement au voisinage de la
pointe du pieu, le facteur de dégradation est négatif. Ceci s’explique par un durcissement cyclique dans cette
zone. Cette figure montre également que les couches superficielles se dégradent plus rapidement que celles
situées en profondeur. En effet, les couches situées au voisinage de la moitié supérieure du pieu ont un facteur de
dégradation plus élevé que les couches situées au niveau de la moitié inférieure. Une légère diminution de la
vitesse de dégradation au fur et à mesure des cycles est également notée.
Afin de mieux visualiser l’évolution du frottement axial limite au cours de cette essai 2.R3.CY2, les calculs
ont été poursuivis jusqu’à atteindre la rupture théorique. Elle s’est produite au bout de 370 cycles. A la rupture,
même les couches situées au-dessous de 4/5D ont été dégradées et uniquement le voisinage de la pointe du pieu
subit un durcissement. La dégradation affecte d’abord les couches supérieures pour se généraliser au fur et à
mesure des cycles à toute la longueur du pieu.
La figure 4b représente la répartition des efforts dans le pieu R3 prédite par les calculs au cours du
chargement de traction cyclique 2.R3.CY2. et ce pour le premier cycle (N=1) et le dernier cycle (N=200). La
variation de la distribution des efforts est importante entre le premier et le dernier cycle. Au cours du chargement
cyclique, les couches profondes sont de plus en plus sollicitées. Le cumul des cycles provoque un transfert des
charges des couches superficielles aux couches profondes. Ce transfert des charges est causé par la dégradation
du frottement axial limite au fur et à mesure des cycles.
Ces observations issues de la figure 4 doivent être confrontées à des données expérimentales afin de discuter
de leur validité.

278
Figure 4. a) Modélisation de l’évolution du facteur de dégradation du frottement axial limite pour l’essai 2.R3.CY2, b)
Modélisation de la répartition des efforts pour l’essai 2.R3.CY2

6. CONCLUSION

Le présent article propose un modèle TZC adapté au calcul de déplacement des pieux soumis à une sollicitation
cyclique axiale. Ce modèle est combiné à une méthode de calcul de dégradation du frottement axial au cours de
ce type de chargement. Des paramètres A, B et C ont été fixés et une comparaison entre des résultats
expérimentaux d’essais de traction cyclique, réalisés dans du sable, et des résultats numériques issus du modèle
TZC a été effectuée. A l’issue de cette comparaison, le modèle TZC a permis de rendre compte des principaux
effets du chargement cyclique qui sont : l’accumulation des déplacements en tête de pieu et la dégradation de la
résistance ultime au fur et à mesure des cycles. Les déplacements calculés sont comparables aux déplacements
mesurés. Cependant, la vitesse d’accumulation de ces déplacements est élevée pour l’essai 2.R3.CY2. Cela peut
être expliqué par les valeurs des paramètres A, B et C qui doivent être testés sur une plus large gamme d’essais
expérimentaux en vraie grandeur. L’étude de l’influence de certains paramètres tels que l’historique de
chargement, l’amplitude et la charge moyenne appliquées peut être également envisagée. De plus, le travail de
modélisation mené sur les pieux installés dans du sable peut être étendu à des pieux situés dans de l’argile.

REFERENCES

Burlon S., Thorel L., Mroueh H. (2013) Proposition d’une loi t-z cyclqiue au moyen d’expérimentations en centrifugeuse.
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris.
Chan S., Hanna T.H. (1980) Repeated loading on single piles in sand. Journal of Geotechnical Engineering Division, 106
(GT2)
Chin J.T., Poulos H.G. (1992) Cyclic axial loading analyses : a comparative study. Computers and Geotechnics, 13, 137-
158.
Christoulas S., Frank R. (1991) Deformation parameters for pile settlement. Proceedings of the Tenth European Conference
on soil mechanics and foundation engineering. Florence.
Frank R., Zhao S. (1982) Estimation par les paramètres pressiométriques de l’enfoncement sous charge axiale de pieux forés
dans des sols fins. Bulletin Liaison Laboratoire Ponts et Chaussées 119, 17-24
Jardine R. J. (1998) Interim report on cyclic loading model and synthetic soil profile for HSE funded pile cyclic loading
study. Department of Civil Engineering, Imperial College, London
Jardine R. J., Standing J. R. (1998) Pile load testing performed for HSE Cyclic Loading Study at Dunkirk, France, Volume
1. Dept of Civil & Environmental Engineering, Imperial College of science & Technology.
Lemaitre J., Chaboche J. L. (2009) Mécanique des Matériaux Solides. Dunod (éd.2ème), Paris
Poulos H., Davis E. (1968) The settlement behavior of single axially loaded incompressible piles end piers. Géotechnique
(18), 351-371.
Randolph M.F. (1986) RATZ: Load transfer analysis of axially loaded piles, Report Geo: 86033, Department of Civil
Engineering, The University of Western Australia.
Van Weele A. (1979) Pile bearing capacity under cyclic loading compared with that under static loading. Second
International Conference On Behaviour of Off-Shore Structures, 475-488
WS Atkins Consultants Ltd (2000) Cyclic Degradation of offshore piles. Health Safety Executive HSE

279
280
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Deep foundations and underground structures – Tailor-made solutions for soft

soil

Fondations profondes et structures souterraines - Solutions sur mesure pour les terrains
meubles

Manfred Schoepf*1
1
BAUER Maschinen GmbH, Schrobenhausen, Germany

ABSTRACT - The presence of soft and mostly unconsolidated soil in lowlands or in river plains may pose a
challenge to geotechnical engineers, designers as well as to foundation contractors. It needs special attention to
select appropriate foundation systems. The article will describe selected methods of transferring structural loads
in soft soil, based on decades of worldwide practical experience. When applying advanced technologies and
equipment it is possible to reach a technically sound and economically acceptable solution. For areas where a
competent stratum is available below the soft soil, the most common solution is “bridging” the soft strata with
piles. Two modern piling techniques (CFA-Continuous flight auger, and CCFA-Cased continuous flight auger)
will be explained in more detail as a modern technologies of rotary drilling with special emphasis on quality
control methods and the use of latest state-of-the-art equipment.
For supporting widespread, medium-sized loads “reinforcing” of the subsoil is a widely used method. A
typical method is Vibrodisplacement or Stone columns. It refers to columns of compacted, gravel size stone
particles constructed vertically in the ground to improve the performance of soft or loose soils.
A third method which will be described is Soil mixing. With this system the natural soil is mixed
mechanically with a binder to create in-situ circular or rectangular elements of cemented soil. The binder is
injected as the mixing tool is advanced down to assist in penetration and to take advantage of this initial mixing.
The soil and binder are mixed a second time as the tool is extracted. While the system was originally developed
for cohesionless soil, it is now also possible to improve soft cohesive soil material efficiently and economically.

RESUME: La présence de sol mou et la plupart du temps non consolidée dans le sol profond ou dans les plaines
de la rivière peut représenter un défi pour les ingénieurs géotechniques, les concepteurs ainsi que les
entrepreneurs de fondation. Il nécessite une attention particulière pour sélectionner les systèmes de fondation
appropriés. L'article décrit les méthodes sélectionnées de transfert de charges structurelles dans le sol mou, basée
sur des décennies d'expérience dans le monde pratique. Lors de l'application des technologies et équipements
avancés, il est possible de parvenir à une solution techniquement fiable et économiquement acceptable. Pour les
zones où une strate compétentes est disponible ci-dessous, la solution la plus commune est "pontage" les couches
tendres avec des piles. (Tarière CFA-continu, et CCFA-tubés tarière continue). Deux techniques d'empilage
modernes seront expliquées plus en détail comme les technologies modernes de forage rotatif avec un accent
particulier sur les méthodes de contrôle de la qualité et l'utilisation de dernier état-of-the- équipement de pointe.
Pour supporter les charges généralisées, moyennes "renforcement" du sous-sol est une méthode largement
utilisée. Une méthode typique est Vibrodisplacement colonnes ou en pierre. Elle se réfère à des colonnes de
compactée, gravier particules de pierre de taille construites verticalement dans le sol pour améliorer les
performances des sols mous ou lâches. Une troisième méthode qui sera décrit est un mélange de sol. Avec ce
système, le sol naturel est mélangé mécaniquement avec un liant pour créer in situ circulaire ou rectangulaire
éléments de sol cimenté. Le liant est injecté sous la forme de l'outil de mélange est avancé vers le bas pour aider
à la pénétration et à profiter de ce mélange initial. Le sol et le liant sont mélangés un deuxième temps que l'outil
est extrait. Bien que le système a été initialement développé pour le sol sans cohésion, il est maintenant
également possible d'améliorer matériau de sol cohésif souple efficacement et économiquement.

1 GENERAL

For geotechnical engineers soft soil is typically defined as soft, unconsolidated cohesive soil. A wider definition
may also include loose, noncohesive to slightly cohesive soils.
Whenever structures have to be erected on soft soil it may pose a challenge to geotechnical engineers, designers
as well as to foundation contractors. It needs special attention to select appropriate foundation systems.
Typical foundation systems are dependent on: the thickness of the soft stratum, the type of loading, the
availability of equipment and techniques.

281
The article describes several selected methods of transferring structural loads in soft soil. The main concepts can
be grouped in categories such as
Bridging soft soil to transfer loads to deeper competent strata => Piling systems
Reinforcement - techniques that typically construct a reinforcing element within the soil mass without
necessarily changing the soil properties. The performance of the soil mass is improved by the inclusion of the
reinforcing elements. => Vibrodisplacement, also referred to as Stone columns.
Fixation - techniques that fix or bind the soil particles together, thereby increasing the soil strength and
decreasing its compressibility and permeability.

2 PILES FOR BRIDGING SOFT STRATA

For areas where a competent stratum is available below the soft soil and concentrated loads have to be
transferred, the typical solution is the installation of piles.
Standard piling methods, such as driven piles with Diesel- or hydraulic hammers, or rotary bored piles
installed with the Kelly system are not described. Emphasis is given to two new piling methods (CFA and
CCFA)

2.1 CFA Continuous flight auger

A continuous flight auger (CFA) pile is a deep foundation technology whose importance has increased
significantly in recent years in many areas of the world.
The CFA Pile is a non-displacement pile used where fast and environmentally friendly installation is
required. It can provide great advantages for many types of projects and ground conditions. These piles are
typically installed with diameters ranging from 0.3 to 1.0 m and lengths of up to 30 m. (Brown 2005). The
gradual increase of diameter and length is a result of the availability of larger and more powerful drilling rigs
with increased engine power, large torque capacity, high pulling forces and long masts.

2.1.1 Description of the system

Continuous flight auger (CFA) piles are cast-in-situ drilled foundations in which a hollow stem continuous flight
auger is drilled into the ground at a controlled speed to the specified depth in a continuous operation. The auger
flights of the auger are filled with soil as the auger is drilled into the ground, thus ensuring continuous lateral
support of the borehole wall. As the concrete is placed, while the auger is removed, the hole is never left open or
unsupported.

Figure 1. Drilling rig with CFA attachment

Concrete is then placed by pumping under continuous positive pressure through the hollow stem of the auger
to fill the borehole as the auger is withdrawn.
Once the auger is completely out of the borehole and the borehole is completely filled with fluid concrete, a
reinforcing cage can be inserted and driven into the fresh concrete to the designed depth. The working sequence
is shown in Figure 2.

282
Beton
Concrete

Figure 2. CFA – Working sequence (Bauer Maschinen, 2010)

During the drilling operation care must be taken to ensure that the speed of rotation of the auger is not
excessive in respect to its rate of penetration. If the speed of rotation is too high there is a risk of "Overflighting"
i.e. a removal of soil from the sides of the borehole and conveying it to the surface, thus reducing the
geotechnical characteristics of the pile/soil interface. The effect may cause decompression of the load bearing
soil strata and even cause ground subsidence at the surface. This phenomenon becomes especially important
when the CFA auger meets hard layers of soil or rock after drilling through a weak stratum and the speed of
penetration slows down.
The problems can be avoided by proper controlling of the rate of penetration. Modern equipment can be
equipped with monitoring systems, which allow the operator to follow and react in real time. It is a good aid for
the operator, but he still has to make the right decisions.

2.1.2 Automatic control systems (“Assistants”)

A new and exciting topic for specialist foundation engineering equipment is active intervention in work
processes by process control software programs. After setting basic parameters by the operator, the electronic
“Drilling Assistant” is able to undertake the execution of drilling processes. This ensures a consistently high-
quality execution and enables continuous process monitoring. As a result, the strain on the machine operator is
effectively reduced and his role becomes more and more that of a controller.
The operator has to enter the geometric data of the CFA auger on the touchscreen of the drill rig (OD auger,
OD hollow stem, pitch of the flights). He also sets an initial target penetration rate. Based on these data, the
assistant calculates the optimum filling ratio of the auger flights. The computer will automatically regulate the
penetration speed and the down force in relation to the auger revolutions.
Once the final depth is reached, concrete is pumped under pressure through a line to the top of the rig and
delivered into the hollow stem of the auger string, while the auger is lifted smoothly and in one continuous
operation. A pressure sensor measures the concrete pressure. An additional sensor records the pump strokes.
Concrete pressure and concrete volume are continuously displayed on the touchscreen in the operator´s cab.
The extraction of a long, soil filled auger with big diameter requires high capacity rigs with large pulling force.
Under certain circumstances it may be necessary to activate a pulling force of 100 t.
The coordination of auger extraction and concreting speed is a critical operation. Insufficient coordination
may result in severe defects of the concreted column. The use of another electronic device (“Pulling assistant”)
ensures a continuous pile quality. After activating the system, the Pulling Assistant controls and adjusts the
extraction speed based on the actual quantity of pumped concrete.

283
2.2 CCFA system (Cased Continuous Flight Auger)

The basic principle of double rotary drive systems is a continuous flight auger in combination with an outer
casing, that are drilled simultaneously, but in opposite directions, into the ground. These characteristics led to the
common term CCFA - Cased Continuous Flight Auger Pile. (Gerressen F.-W., Platzek P., 2009)

Figure 3. Drilling rig with CCFA attachment

The spoil is transported upwards by the auger flights surrounded by the casing and exits through openings at
the top of the casing. The system allows relative vertical movement between the casing and the auger of +/- 300
mm.
After reaching the final depth, concrete is pumped under continuous positive pressure through the hollow
stem of the auger to fill the borehole as the auger is withdrawn.
Once the auger is completely out of the borehole a reinforcing cage is inserted and driven into the fresh
concrete to the designed depth.
Following the development of high torque rotary heads and drilling rigs with extended masts and high crowd
and extraction forces, the range of diameters and depths for CCFA piles has increased significantly. Typical
depth capacity ranges from 17 – 24 m and diameters up to 1000 mm.
It is possible to install CCFA piles not only in soft soil but also through a wide variety of soils including
clays, silts, peat, sand, gravel and even stones.
There are some significant advantages of the system in comparison to CFA piling: When working in soft
ground, there is a minimized risk of bulging during concreting. When constructing secant pile walls, the key
advantages are excellent verticality of the piles and the capacity of trouble-free cutting into the concrete of
adjacent primary piles.

3 VIBRODISPLACEMENT

For supporting widespread, medium-sized loads “reinforcing” of the subsoil is a widely used method. A typical
method is Vibrodisplacement or Stone columns. It refers to columns of compacted, gravel size stone particles
constructed vertically in the ground to improve the performance of soft or loose soils.
The method is used to increase bearing capacity (up to 240 to 480 kPa), reduce foundation settlements,
improve slope stability, reduce seismic subsidence, reduce lateral spreading and liquefaction potential.
Stone columns improve the performance of soils in two ways - densification of surrounding granular soil and
reinforcement of the soil with a stiffer, higher shear strength column.

Table 1. Expected densification and reinforcement (Hussin, 2006)

Soil description Densification Reinforcement
Gravel and sand <10% silt, no clay Excellent Very good
Sand with between 10 and 20% silt and <2% clay Very good Very good
Sand with >20% silt and nonplastic silt Marginal Excellent
Clay Not applicable Excellent

284
3.1 Work procedure
The construction of stone columns by the bottom feed method consist of 2 main working steps.
Step 1:The depth vibrator is positioned over the compaction point on the working platform. Material transfer
pipe and material storage container are filled with granular backfill material. Air flushing is activated.
Step 2: The bottom-feed vibrator is lowered to the specified design depth. After reaching the design depth
the vibrator is retracted in steps of around 0.5 m allowing the granular backfill to discharge into the annular
space formed by the vibrator. Repeated penetration and retraction of the vibrator will compact the backfill
material and displace it laterally into the surrounding soil. This process is repeated until either the hydraulic
pressure of the vibrator has increased to around 270 to 290 bar or the volume of backfill material required for the
specified column diameter has been placed.

Figure 4. Bottom feed method – Work procedure (Bauer Maschinen, 2013)

The densification is performed by a depth vibrator. It consists of a cylindrical pipe with a nose cone wear
shield at the lower end. Horizontal vibrations are created by a rotating eccentric weight which is powered by a
hydraulic motor. The vibratory energy is applied directly to the ground through the cylindrical pipe and therefore
its output remains constant whatever the depth of penetration. As the drive for the eccentric weight is
hydraulically powered, the vibration frequency is easily changeable and varies normally between 25 and 50 Hz.
An elastic coupling separates the vibrating probe from the follower tubes, which are connected to the upper
part of the coupling and are adapted in length according to the required penetration depth.
The follower tubes house the hydraulic power line and allow flushing air to flow to the vibrator. The flushing
air which exits at the tip of the vibrator is used to aid penetration of the vibrator.
A stone skip is filled with stone on the ground with a front end loader and a separate cable raises the skip to a
chamber at the top of the tremie pipe. The stone is typically a graded crushed hard rock, although natural gravels
and pebbles have been used. The greater the friction angle of the stone, the greater the modulus and shear
strength of the column. The volume of gravel installed can be regulated by the hydraulic material gate.
Compressed air is also used at the material transfer pipe to aid the material transport.

3.2 Quality control and quality assurance

All process-specific production data are monitored and displayed on the B-Tronic monitor inside the operator
cab for quality assurance purposes. Electronic data acquisition of all relevant production data for documentation
purposes is also carried out by the B-Tronic system.
Relevant production parameters generally include: Production time, Depth attained below working platform
taking into account any empty bore, Time and speed for each compaction stage, Energy consumption (hydraulic
pressure) of depth vibrator, consumption of backfill material.
The stored data can be downloaded onto a storage device or transferred by radio transmission. The stored
data can subsequently be processed and evaluated. For documentation reasons, a production report can be
produced for each column or compaction point. The data can be presented over both depth of penetration and
time.
After completion of the stone column installation, post-treatment penetration testing can be performed to
measure the improvement achieved in granular soils. The bearing capacity of the improved ground can be
determined by plate loading tests with vertical test loads up to 150% of the design load.

4 SOIL MIXING

285
The basic idea of all soil mixing systems is improving the properties of the soil by mechanically mixing a binder
in the ground. The common way is to use slurry, made of water and cement containing binders. Ideally the slurry
is filling up the pore volume of the present soil, thus creating a concrete-like structure. The achievable properties
of a soil mix structure strongly depend on the prevailing soil conditions and the composition of the slurry.
The following soil parameters are influencing the soil mixing procedure and the final product: silt and clay
fraction, homogeneity, original moisture content, depth of water table.
The cement and bentonite content and the water/cement ratios of the mixing slurry are determined by the strength
and/or permeability requirements of the project and the properties of the soil being mixed. In general, the cement
content is increased and the water/cement ratio is lowered if a higher strength of the mixed element is required.

4.1 Main systems

In general there are two different principles of mixing tools in use. The first concept is based on mixing tools
which turn around single or multiple vertical axes. Since the introduction of Cutter Soil Mixing (CSM) in 2003
by Bauer Maschinen, this cutter based system with its horizontal rotating mixing tools becomes more and more
popular.
Soil mixing by single or multi vertical axis systems is using augers or mixing paddles that are connected to rotary
heads. A special 3-axis system called 'Soil-Mixing-Wall' System (SMW) allows the use of smaller base carriers
by reducing the necessary power during the mixing process. The soil is cut by each auger that contains two
blades with flat teeth and a pilot bit in the center position. The two outside axis cut in the opposite direction.
During the cutting process the soil is transported to the middle auger where slurry exit nozzles are placed. Hence
the soil is mixed with the slurry directly after cutting. The mixing paddles of the SMW tool sit above the augers.
Due to interlocking mixing paddles a proper homogenization of the soil slurry mix in the complete panel width
can be achieved.
The system is equipped with inclinometers in the two outside augers. Additional information about the slurry
flow, the actual depth, the actual penetration speed the rotation speed of the augers are displayed in real time on
the operator´s screen. All process data are also stored by the B-Tronic, which enables a separate evaluation of the
processing data.

4.2 CSM System

The CSM system is based on the trench cutter technology with two mixing wheels rotating around horizontal
axes. (Fiorotto et al. 2004)
The wheels are driven by hydraulic motors which are located in a water-tight housing. The slurry is introduced
into the soil directly between the mixing wheels. During construction, the counter-rotating mixing wheels and
vertically mounted plates are effectively acting like a forced action mixer. For loosening and mixing the soil
different types of mixing wheels were developed.
The mixing unit is mounted on a guided Kelly bar. The system is suitable of reaching depths up to 43 m. The
CSM system is also equipped with the B-Tronic control and visualization system.

Figure 5. CSM – Cutting and mixing head

4.2.1 Construction steps

286
Liquefaction of the soil mass during penetration to the final depth as an appropriate slurry is simultaneously
introduced. The volume of slurry injected is determined by the rate of penetration. Usually cement slurry is used
in this construction step.
During withdrawal, the precise quantity of slurry, which is required for producing the final column or wall
element, is injected.
To utilize the wall as a structural retaining wall, steel columns (IPB sections) are inserted into the freshly mixed
wall panels.
A two phase system is used when deeper panels are mixed or the expected production time is higher due to
difficult soil layers or present rock. Instead of using self-hardening slurry from the beginning onward as in a one-
phase system, just bentonite slurry is used on the down stroke. Once the final depth is achieved, the cement slurry
is introduced and mixed on the upstroke. This method prevents the mixing tool from being trapped in the panel if
the panel construction time exceeds the initial set time for the cement slurry.
The size of individual panels is determined by the type and size of equipment being deployed. Panels can be
constructed in lengths ranging from 2.4 m to 2.8 m and wall thicknesses of 0.55 m to 1.2 m.

5 CONCLUSION

The use of soft soil for transferring structure loads poses a challenge to geotechnical engineers, designers as well
as to foundation contractors. There are, however, several proven techniques and equipment available for
installing safe, reliable and economic foundations.
A selection of some state-of-the-art techniques has been outlined. They include piling solutions for areas with
high loads and competent strata below the soft layers. They also include ground improvement techniques like
stone columns or soil mixing for supporting widespread, medium-sized loads.
All systems have one thing in common: The selection of suitable drilling equipment is of utmost importance.
When choosing an unsuitable technique or equipment, considerable economic or quality problems can be
created.
Well trained, knowledgeable site staff, from engineer, foreman to rig operator, are the other key prerequisite
for a successful project performance.

REFERENCES

Bauer Maschinen (2010) Method Statement Continuous Flight Auger Piles. Company publication
Bauer Maschinen (2013) Method Statement Vibro Displacement Columns. Company publication
Brown Dan A. (2005) Practical considerations in the selection and use of Continuous Flight Auger and Drilled Displacement
Piles. Geotechnical Special Publication No 132, ASCE 2005, 1-11.
Fiorotto R. et al. (2004) Cutter Soil Mixing an innovation in soil mixing for creating cut-off and retaining walls. Proc. 16th
Int. Conf. on Soil Mechanics and Found. Eng.3, Osaka, 1185-1188.
Gerressen F.-W., Platzek P. (2009) Double Rotary Head Piling Systems -Cased CFA. Proc. 34th Annual Conf. DFI, London,
pg 333
Hussin James D. (2006) Methods of soft ground improvement. The Foundation Handbook, Taylor&Francis Group, pg 541
Photographs Bauer Maschinen GmbH

287
288
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Étude de l’influence des charges verticales sur la réponse latérale des pieux
dans des sols argileux

Investigation of the influence of vertical loads on the lateral response of pile foundations in
clayey soils

L. Hazzar*1, M.N. Hussien1 and M. Karray1

1
Department of Civil Engineering, Faculty of Engineering, Sherbrooke University, 2500 Boulevard Université,
Sherbrooke (Quebec/Canada) J1K 2R1
*
Auteur correspondant

ABSTRACT – Although the applied loads on pile foundations are usually combinations of vertical and lateral
components, there have been few investigations on the response of piles subjected to combined loadings. Those
few studies led to inconsistent results with regard to the effects of vertical loads on the lateral response of piles.
In this study, a series of three-dimensional finite differences analyses is conducted to evaluate the influence of
vertical loads on the lateral response of pile foundations in clayey soil. The pile material is modelled as linearly
elastic and the soil is idealized using the Mohr-Coulomb constitutive model with a non-associated flow rule.
Vertical loads ranging from 25 to 100 % of the ultimate vertical capacity of the pile are applied prior to the
application of the lateral loads. Numerical results have shown that the vertical load leads to a significant decrease
in lateral capacity of piles in homogeneous clay layer.

RÉSUMÉ - Bien que les charges appliquées sur les pieux sont généralement des combinaisons de composantes
verticales et latérales, il y a eu un peu d’études sur la réponse de pieux soumis à des charges combinées. Ces
études ont conduit à des résultats contradictoires en ce qui concerne les effets des charges verticales sur la
réponse latérale du pieu. Dans cet article, une série d’analyses numériques tridimensionnelles en différences
finies est effectuée afin d'évaluer l'influence de charges verticales de la réponse latérale des pieux dans un sol
argileux. Le matériau du pieu est modélisé comme élastique linéaire et le sol est caractérisé en utilisant le modèle
de comportement Mohr-Coulomb. Des charges verticales de 25 à 100 % de la capacité verticale ultime du pieu
sont appliquées avant l'application des charges latérales. Les résultats numériques ont montré que la charge
verticale conduit à une diminution significative de la capacité latérale des pieux dans un sol argileux.

1 INTRODUCTION

Pile foundations are extensively used in many civil structures to support vertical and horizontal loads which
result from the weight of the structure and dynamic load (wind, vibration, etc.) or seismic forces. In conventional
design methods, the axial and lateral responses of piles are often evaluated separately neglecting their possible
interaction. This would lead to unnecessary over-conservatism in the design, as pile foundations for several types
of structures are often subjected to simultaneous vertical and lateral loading. The separate consideration of the
vertical and lateral loading therefore cannot be expected to account properly for the pile behavior (Georgiadis
and Saflekou, 1990; Zhang et al., 2002; Hussien et al., 2014a, 2014b).
The behavior of pile foundations under either axial or lateral loads has been investigated for more than a
century through full-scale tests (e.g., Brown et al., 1987, 1988; Rollins et al., 1998, 2005), centrifuge model tests
(e.g., McVay et al., 1995, 1998; Tobita et al., 2004), and analytical (e.g., Matlock and Reese 1960; Randolph
and Wroth, 1978; Zhu and Chang, 2002) as well as numerical solutions (e.g., Ottaviani, 1975; Hussien et al.,
2010, 2012; Hazzar et al., 2013) among others. The procedures used for modeling soil range from rigorous soil
continuum discretization such as finite element (FE) or finite differences (FD) formulation to simplified
interaction models such as the subgrade reaction approach. In the conventional subgrade reaction approach, soil
is modelled by spring elements attached to the pile at different depths. These springs generally have nonlinear
load-displacement characterizations called (t-z) and (p-y) curves for axial and lateral loading, respectively. These
two types of springs are generally uncoupled and therefore soil reactions along the corresponding degrees of
freedom are also uncoupled. In other word, the influence of load acting in one direction on the characteristics of
the spring in the other direction is neglected (Hussien et al., 2014a). Anagnostopoulos and Georgiadis (1993)
have reported through model tests supported by two-dimensional (2D) FE analysis that the modified status of soil
stresses and local plastic volume changes in the soil continuum under combined vertical and lateral loads cannot

289
be accounted for in general by the subgrade reaction and elastic half space methods of analysis. Therefore, they
suggest using a nonlinear three-dimensional (3D) FE or FD technique for analyzing this problem.
Achmus and Thieken (2010) used the 3D FE to investigate the behavior of piles in non-cohesive soil under
combined horizontal and vertical loading, and they reported that the combined loading on pile induces
interaction effects due to simultaneous mobilization of passive earth pressure due to horizontal loads and pile
skin friction due to vertical loads. Karthigeyan et al. (2007) showed through a series of 3D FE analyses on piles
that the presence of vertical loads decreases the lateral capacity of piles in clayey soil. In fact, the scopes of the
previous attempts examining this problem using a 3D FE model have been limited to the behavior of piles
installed in homogeneous clayey soils. Little work has been devoted to the behavior of piles subjected to
combined effects of vertical and lateral loads in inhomogeneous soils which is often encountered in geotechnical
projects.
In view of the above mentioned issues, this paper presents and discusses the results of a series of 3D FD
analyses carried out using FLAC3D (Itasca, 2009) in order to evaluate the influences of vertical loads on the
lateral capacity as well as internal forces of piles installed in a clayey layer with constant undrained shear
strength. Numerical models are first validated and then analyses were carried out to investigate the influences of
vertical loads on the lateral capacity of piles as affected by typical soil characteristics such as undrained shear
strength. Combined load analyses were performed for vertical loads equal to 25 %, 50 %, 75 %, and 100 % of
the ultimate vertical load capacity of the pile, Vult.

2 FINITE DIFFERENCES MODELING AND PARAMETERS IDENTIFICATION

2.1 Finite differences

The 3D FD program FLAC3D (Itasca, 2009) was employed to study the behavior of piles under horizontal and
vertical loading. Full 3D geometric models were used to represent the coupled soil-pile system. Taking
advantage of symmetry, only half of the actual model was built, thus significantly reducing the computational
effort. Figure 1 shows the general layout and meshing of the FD half model used for the soil-pile system. A
floating pile with a diameter, B was embedded in the soil to a depth of L while the total thickness of the soil
stratum was selected at L+6B. The soil-pile system was meshed with 8-node brick elements, and the soil elements
are fairly small adjacent to the pile and gradually increase in size as they move away from it. The soil element
size was kept uniform at 0.5 m in the vertical direction. The total mesh size was extended to a horizontal distance
of 16B from the center of the pile. This distance was decided after performing a number of initial trial analyses
with several horizontal distances until the displacements and stresses of the pile did not change significantly with
further increasing of the distance. All displacements were restrained at the bottom of the meshes while those at
the vertical “external faces” were fully fixed in the x- and y-directions. The symmetry face were fixed against
displacement normal to the symmetry plane, but were free to move on the surface of the plane. The top and
bottom of the pile were set as displacement and rotation free.

Figure 1. The general layout and meshing of the FD half model used for the soil-pile system.

290
2.2 Soil model

The Mohr-Coulomb model, extensively used in geotechnical engineering practice, was adopted in this study to
simulate the nonlinear behavior of soil. In FLAC3D, six parameters are required to effectively define the soil
behavior. These parameters are the elastic bulk modulus, K, the elastic shear modulus, G, the mass density, , the
friction angle, , dilatancy angle, , and the cohesion, c.

2.3 Pile model

The pile is modeled as linear-elastic material. Three parameters are required to define the pile material behavior.
These parameters are the elastic bulk modulus, Kp, the elastic shear modulus, Gp, and the mass density, p.

2.4 Soil-pile interface model

The constitutive model of the soil-pile interface is defined in FLAC3D by a linear Coulomb shear-strength
criterion that limits the shear force acting at an interface node. The shear-strength criterion that limits the shear
force acting at an interface node is given by:
Fs max  ci A  tan i  Fn  pA  (1)

where Fsmax is the limiting shear force at pile-soil interface, Fn is the normal force, ci is the cohesion along the
interface, i is the angle of friction of the interface surface, p is the pore pressure (interpolated from the target
face), and A is the area associated with an interface node. The shear strength was defined with zero cohesive
strength and 2/3 of the friction angle for sandy soils. In the case of clayey soils, the interfaces were assumed to
have zero frictional strength and the same cohesive strength of the surrounding soil. Separation is able to cause a
significant increase in displacements and therefore the interface elements are allowed to separate if tension
develops across the interface and exceeds the tension limit of the interface. Once gap is formed between the soil-
pile interfaces, the shear and normal forces are automatically set to zero (Itasca, 2009).
The normal and shear forces at the interface nodes are determined by the following equations:

Fn(t t )  kn un A   n A (2)

F ( t t )
si F(t )
si  k s u ( t  0.5 t )
si A   si A (3)

where Fn and Fsi are the normal and shear force, respectively, kn and ks are the normal and shear stiffness,
respectively, usi is the incremental relative shear displacement vector, un is the absolute normal penetration of
the interface node into the target face, n is the additional normal stress added due to interface stress
initialization, and si is the additional shear stress vector due to interface stress initialization (Itasca, 2009).
In the current study that involves nonlinear analysis; high soil-pile interface stiffness is assigned to minimize the
contribution of soil-pile interface elements to the accumulated pile displacements. According to the results of
trial numerical analyses conducted to identify an appropriate stiffness value, a value of 108 Pa/m for both kn and
ks was found to be sufficient to ensure that no additional deflections were attributed to the pile due to the
deformation of the springs representing the interface. The use of such considerably higher values is tempting as it
could be considered as more appropriate, but in that case the solution convergence would be very slow. In that
way, the interface elements behave practically as a slider with a rigid/plastic behavior.

2.5 Analysis scheme

The model is first brought to an equilibrium stress-state under gravitational loading before the installation of the
pile. In the next stage of analysis, the model is brought into equilibrium after the installation of the pile. The
installation is modeled by changing the properties of the pile zones from the properties representing the soil
material to those representing the pile material. It is helpful to know the addresses of the grid-points at the pile,
to facilitate both the loading of the pile and the monitoring of the pile response.
The ultimate vertical capacity (Vult) of the pile is evaluated by applying a vertical velocity at the pile head
while the pile load and settlement are monitored. According to CGS (2013), the value of Vult is defined as the
vertical load corresponding to the point with maximum curvature on the vertical load-vertical displacement
curve.
After the model is brought to equilibrium for the vertical loading, the pile top is then displaced laterally for a
deflection of 0.1B. To minimize the number of parameters, this value is fixed in all studied cases. The vertical

291
load is kept constant while applying the horizontal displacement. It is fair to mention that the pile is assumed to
be in a stress-free state at the beginning of the analysis, and the effect of the pile installation is ignored in the
analysis.

3 VALIDATION OF THE FD MODEL

The validity of the FD model to the analysis of pile in clayey soil and subjected to both vertical and lateral loads
was also verified by back predicting the pile load test data from a published case reported by Anagnostopoulos
and Georgiadis (1993). This case study pertains to laboratory model tests on aluminum closed-ended piles of 19
mm outside diameter and 1.5 mm wall thickness, jacked 500 mm into a prepared soft clay bed (cu = 28 kPa). The
laboratory test was performed on a single pile under both vertical and lateral loads applied to the pile head at
ground elevation through dead weights. The combined vertical and lateral loads were applied in two stages, in
the first stage a vertical load of 160 N was applied and in the second stage the lateral load of 130 N was applied
incrementally while the vertical load was kept constant. The soil shear modulus considered by Anagnostopoulos
and Georgiadis (1993) (2500 kPa) was used in the current analysis. The Poisson’s ratio of the clayey soil was
selected at 0.49 assuming an undrained response during the load test. The sequence of the load application used
in the current FD analysis is the same as that followed during the laboratory tests. The comparison between the
test data and the predicted results of piles under pure vertical load and combined vertical and lateral loads are
shown in Figs. 2b and 2c. The FD prediction in both the cases matched well with the test data. Hence, it could be
concluded that the numerical scheme adopted in the present investigation is capable of modeling the soil-pile
interaction under vertical load, lateral load, and combination of vertical and lateral loads.

Figure 2. Comparison of FD predicted results with: (a) vertical loaded pile test data of Anagnostopoulos and Georgiadis
(1993) and (b) combined vertical and lateral loaded pile test data of Anagnostopoulos and Georgiadis (1993).

4 PARAMETRIC STUDIES

FLAC3D was employed to study the behavior of piles under combined vertical and lateral loading in a clayey
layer with constant shear strength. A floating concrete pile, conformed to a grade M25, with a diameter, B = 1.0
m was embedded in the soil to a depth of L = 10 m while the total thickness of the soil stratum was selected at 16
m. The elastic modulus (Ep) and Passion’s ratio (υp) of the pile were set to be 25 GPa and 0.15, respectively. Soil
parameters considered in the analyses are summarized in Table 1. The elastic shear modulus G for cohesive soil
was determined using undrained shear strength cu based on the correlation provided by Poulos and Davis (1980):
G  300  cu (4)
The response of piles to pure lateral load was first evaluated for each case considered. Then, the response of
piles to combined vertical and lateral loads is examined for different values of vertical loads ranging from 25 to
100 % of Vult. The combined loads are applied in two stages. In the first stage, vertical loads were applied and
then in the second stage, lateral loads were applied while the vertical load was kept constant.

Table 1. Model parameters of clayey soil used in parametric study.

Clay type Mass density Shear modulus Bulk modulus Undrained shear Angle of friction
 (kg/m3) G (MPa) K (MPa) strength cu (kPa)  (°)
Soft 6.00 58.0 20
Medium1 9.00 87.0 30
Medium2 1600 11.70 113.1 39 0
Stiff 19.20 185.6 64

292
5 RESULTS AND DISCUSSIONS

Figure 3 shows the variation of the lateral capacity as a function of lateral deflection of piles in clayey soils. In
the presence of vertical loads, the lateral capacities developed at all deflections are less than the corresponding
load developed under pure lateral load. Karthigeyan et al (2007) have reported similar findings for piles in
clayey soils through FE analysis of single pile under combined loads.
The percentage of decrease in lateral capacity (DLC) has been defined to measure the influence of vertical loads
on the lateral capacity of piles:
P P (5)
DLC  100  V 0 V v
PV 0
where PV=v is the lateral capacity with vertical loads and PV=0 is the lateral capacity under pure lateral loads.
Figure 4 shows that up to cu = 16 kPa, the DLC is equal to 0 (the vertical load has no effect below this value).
After cu = 16 kPa, the DLC values increase with the increase in cu as well as with the increase in vertical loads.
The comparative results shown in Fig. 4 confirm that the undrained shear strength is important in the design of
pile foundation under the combined action of vertical and lateral loading.
The mechanism of the decrease in lateral capacity of pile installed in clayey soil under the action of vertical
loads has been also examined by plotting the stress state (Mohr circle) of a soil element adjacent to the pile and
at a depth of 3 m. The ultimate shear stress ult, corresponding to the failure, is calculated by the following
equation:

1   3
 ult   cu (6)
2

Figure 3. Lateral load–lateral deflection curves of piles installed in clayey soil for (a) cu = 20 kPa, (b) cu = 30 kPa, (c) cu =
39 kPa and (d) cu = 64 kPa.

293
Figure 4. DLC versus cu at 0.1B lateral deflection

The 1 and 3 corresponding to stress state of the soil element before and after the application of a vertical
load on a pile installed in clay are plotted in Fig. 5 for both the cases with (V = Vult) and without (V = 0) and for
all cu considered. When Vult is applied to the pile, it is clear that the Mohr circles will have larger radiuses than
those corresponding to (V = 0). Thus, the presence of the vertical load decreases the lateral resistance of soil and
subsequently will lead to development of lower resistance to lateral pile deformation.
Figure 6a shows the stress paths of soil elements attached to the pile in medium clay (cu = 39 kPa) at different
depths for both the cases with (V = Vult) and without (V = 0) vertical loads. For all considered depths, the soil
element in the case of (V = Vult) reached the surface failure earlier than the soil element in the case of (V = 0).
The variation of stresses (major, minor, vertical, horizontal) of a soil element adjacent to the pile, installed in
clay medium (cu = 39 kPa) at a depth of 3 m, with lateral deflections are plotted for both the cases with (V = Vult)
and without (V = 0) vertical loads in Fig. 6b. Figure 6b indicates that the inclusion of vertical load decreases the
soil stresses compared to the corresponding stresses in the case of pure lateral loading. Figure 6b shows also that
the orientation of the minor principle stress 3 and the major principle stress 1 in both analyses with and without
vertical loads are similar up to the maximum lateral deflection of 100 mm. Figure 6b shows also a principle
stress rotation after 8 mm lateral deflection since the horizontal stress x becomes larger than the vertical stress
z.

Figure 5. Mohr circles of a soil element adjacent to the pile and at a depth of 3 m in a clayey soil for (a) cu = 20 kPa, (b) cu =
30 kPa, (c) cu = 39 kPa and (d) cu = 64 kPa.

294
Figure 6. Analyses of pile installed in a medium clay (cu = 39 kPa) with and without vertical loads: (a) stress paths of soil
elements attached to the pile, and (b) stresses of a soil element adjacent to the pile at a depth of 3 m.

6 CONCLUSION

The influence of vertical loads on the behavior of laterally loaded pile in clay soil was investigated by means of
numerical modeling. The numerical models were conducted using the computer program FLAC3D and the models
were verified using laboratory model testing data. The verified numerical model was then used to perform a
parametric study considering different shear strength of clay to evaluate the lateral capacities of concrete piles
subjected to both lateral and vertical loads. The results of the lateral capacities of piles were determined and
compared for piles subjected to pure lateral loads and to combined vertical and lateral loads for several values of
vertical loads corresponding to 25 %, 50 %, 75 % and 100 % of the pile ultimate vertical capacity. Based on this
comparison, the presence of vertical loads decreases the lateral capacity by as much as 20 % of piles in clayey
soil depending on the level of vertical load, the values of the clay shear strength, cu and the lateral deflection.

REFERENCES

Achmus M., Thieken K. (2010) On the behavior of piles in non-cohesive soil under combined horizontal and vertical
loading. Acta Geotechnica, 5(3): 199–210.
Anagnostopoulos C., Georgiadis M. (1993) Interaction of axial and lateral pile responses. Journal of Geotechnical
Engineering, 119(4): 793–798.
Brown D.A., Reese L.C., O’Neill M.W. (1987) Cyclic lateral loading of a large scale pile group. Journal of Geotechnical
Engineering Division, ASCE, 113(11): 1326-1343.
Canadian Geotechnical Society. (2013) Canadian Foundation Engineering Manual (4thEd.), Richmond, B.C., Canadian
Geotechnical Society.
Comodromos E.M. (2013) Response prediction for horizontally loaded pile groups. Journal Geotechnical Engineering
Southeast Asian Geotechnical Society, pp123–33.
Georgiadis M., Saflekou S. (1990) Piles under axial and torsional loads. Computers and Geotechnics Journal, 9: 291-305.
Hazzar L., Karray M., Hussien M.N., Bouassida M. (2013) Three dimensional modeling of a pile group under static lateral
loading using finite differences method. GeoMontreal2013, Montréal, Québec, September 29 - October 3, paper 201.
Hussien M.N., Tobita T., Iai S., Rollins K.M. (2012) Vertical load effect on the lateral pile group resistance in sand
response. An International Journal of Geomechanics and Geoengineering, 7(4): 263–282.
Hussien M.N., Tobita T., Iai S., Karray M. (2014a) Influence of pullout loads on the lateral response of pile foundation.
GeoRegina2014, Regina, September 28 - October 1, Accepted.
Hussien M.N., Tobita T., Iai S., Karray, M. (2014b) On the influence of vertical loads on the lateral response of pile
foundation. Computers and Geotechnics, 55: 392s–403.
Hussien M.N., Tobita T., Iai S., Rollins K.M. (2010) Soil–pile separation effect on the performance of a pile group under
static and dynamic lateral load. Canadian Geotechnical Journal, 47(11): 1234–1246.
Itasca. (2009) User’s and theory manuals of FLAC3D: Fast Lagrangian analysis of continua in 3D, version 4. Minneapolis,
Itasca Consulting Group Inc 2009.
Karthigeyan S., Ramakrishna V.V.G.S.T., Rajagopal K. (2007) Numerical investigation of the effect of vertical load on the
lateral response of piles. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 133(5): 512–521.
Matlock H., Reese L.C. (1960) Generalized solutions for laterally loaded piles. Journal of Soil Mechanics Foundation
Division, ASCE; 86(SM5): 63–91.
McVay M., Casper R., Shang T.-I. (1995) Lateral response of three-row groups in loose to dense sands at 3D and 5D pile
spacing. Journal of Geotechnical Engineering, 121(5): 436–441. doi 10.1061/(ASCE)0733-9410(1995)121:5(436).
Ottaviani M. (1975) Three-dimensional finite element analysis of vertically loaded pile groups. Géotechnique, 25(2): 159–
174.

295
Poulos H.G., Davis E.H. (1980) Pile foundation analysis and design. Wiley: Singapore.
Randolph M.F., Wroth C.P. (1978) Analysis of deformation of vertically loaded piles. Journal of Geotechnical Engineering,
ASCE 1978, 104(12): 1465–1488.
Rollins K.M., Lane J.D., Gerber T.M. (2005) Measured and computed lateral response of a pile group in sand. Journal of
Geotechnical and Geoenvironmental Engineering, 131(1):103–114. doi 10.1061/(ASCE)1090-0241(2005)131:1(103).
Rollins K.M., Peterson K.T., Weaver T.J. (1998) Lateral load behavior of full-scale pile group in clay. Journal of
Geotechnical and Geoenvironmental Engineering, 124(6): 468–478. doi 10.1061/(ASCE)1090-0241(1998)124:6(468).
Roscoe K.H., Burland J.B. (1968) On the Generalized Stress-Strain Behavior of ‘Wet Clay’. Engineering Plasticity,
Cambridge University Press, New York, 535-609.
Tobita T., Iai S., Rollins K.M. (2004) Group pile behavior under lateral loading in centrifuge model tests. International
Journal of Physical Modelling in Geotechnics, 4(4): 1–11.
Zhang L.M., McVay M.C., Han S.J., Lai P.W., Gardner R. (2002) Effects of dead loads on the lateral response of battered
pile groups. Canadian Geotechnical Journal, 39(3): 561-575.
Zhu H., Chang M.F. (2002) Load transfer curves along bored piles considering modulus degradation. Journal of
Geotechnical Geoenvironnement and Engineering, ASCE, 128(9): 764–774.

296
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Particle image velocimetry analysis on the sinking of shallow foundation in 2D

Analyse d’images sur l’enfoncement de fondation superficielle en 2D

*1 2
P. Pizette and N.E. Abriak
1
Univ Lille Nord, F-59000 Lille, France & Mines-Douai, LGCgE GCE, F-59508 Douai
*
Corresponding Author : [Link]@[Link]

ABSTRACT – Punching experiment is developed to study the failure of 2D analogue soil. In order to follow the
kinematic behaviors of soil, Particle Image Velocimetry (PIV) analysis has been developed and tested in the case
of the shallow foundations. The results show that the field of the soil displacement under the foundation can be
followed via the PIV method. In particular, the image analysis results are qualitatively in good agreement with
the Prandlt scheme.

RÉSUMÉ - Un dispositif expérimental a été développé pour suivre le poinçonnement en 2D d’un sol analogique
dans le cas du chargement d’une fondation superficielle .Pour suivre par imagerie numérique la cinématique de
déformation du sol sous la fondation, une analyse par vélocimétrie par images de particules (Particle Image
Velocimetry, PIV) a été testée dans le cadre de ce poinçonnement. En particulier, le champ des déplacements du
sol analogique peut être récupéré. Les résultats montrent que les différentes zones de déformation sont
qualitativement en bonne adéquation avec le mécanisme de Prandlt

1 INTRODUCTION

The settlement, the bearing capacity and the soil failure behavior are main data to design a building structure as
for example the foundations. In parallel with the analytical and numerical approaches, the experimental
bidimentional approaches have been developed since the1950s to study the typical behavior of soil in interaction
with geotechnical structure. In particular, scale models with Schneebeli’s material have been widely used for
example to reproduce the active and passive earth pressures, the soil nailings, the foundation loadings, the tunnel
structures, the kinematics of silo discharge, etc..
This Schneebeli’s material, name usually given as from the name of the researcher who first used, is
composed of rods stacked up following to their axis (named also Taylor-Schneebeli model medium). This layout
of the rods enables only plane strains thereby reproducing the typical behavior of soil in 2D. This 2D approach
also allows making links between the global behavior of the granular media and the local parameters at the grain
scale (Abriak, 1998) (Misra and Poorsolhjouy, 2013). With the major development of the image analysis
technics since 2000s, the behavior of soils was became accessible in 3D via for example the microtomography
applied on the laboratory devices (Hall et al,2012). However, the Schneebeli’s approach remains used in
particular for the structure (Dolzhenko, 2002) (Jenck et al., 2009) and theory studies (Misra and Poorsolhjouy,
2013) and also thanks to the news development in image analysis methods carried out on the Schneebeli’s
materials (Sibille and Froiio, 2007), (Richefeu et al, 2012).
Particle Image Velocimetry is a photogrammetry method of flow visualization used in the fields of research
and education. PIV allows the instantaneous velocity of fluid to be measured by following tracer particles seeded
therein. In this paper, PIV analysis tracking has been carried out on the Schneebeli’s material in the case of
superficial foundation loading. The aims of this study are to show the potential uses of PIV method for the
granular media in 2D in the case of spread foundation and to set up a base for image analysis with the
perspective of a further more detailed experimentation program.

2 MATERIAL AND EXPERIMENTAL DEVICE

2.1 Schneebeli’s material

First introduced by G. Schneebeli, rods stacked up following to their axis reproduce a bidimentional analogue
granular medium by complying with the Mohr-Coulomb's criteria (Schneebeli, 1956).
In this study, the rods are cylindrical with 60 mm in length. Three various diameters have been used with a
fairly distribution of each diameter (3, 4 and 5 mm). The matter of the cylinders is duralumin. The friction local
angle (grain-grain friction) is approximately 18° and the intrinsic friction angle is approximately 26°-28°
(Abriak, 2006).

297
2.2 Punching apparatus
A dedicated punching apparatus has been developed in order to simulate the loading of a foundation until the
punching shear failure. As shown in figure 1, four steal extruded bars are used to form the rectangular frame
allowing the analogue granular matter to be contained.
A loading system is simply composed of a vertical steal cylindrical shank with 30 mm in diameter guided by
a Teflon bushing and of a load plate fixed on the top of the cylindrical shank allowing weights to be set down. At
each stage of the loading, the settlement and the applied stress are measured. Note that a more sophisticated
loading system is under development with a pneumatic piston. This system would allow the axial displacement of
the loading cylindrical shank to be controlled during a punching.
The foundation is positioned between the granular matter and the load cylindrical shank with either a knuckle or
3
a clamping. As the granular medium has a low density (the density of the duralumin is equal to 2900 kg/m ), the
foundation is wooden to simulate the entire problem via a scale model, i.e. the soil, the foundation and the load
effect of the structure building. We consider that the distance between the frame and foundation is sufficient to
have a semi-infinite condition.

Loadplate

Foundation
Single scale model
acting

0.6m
location of the
frame (rigid
Schneebeli’s materials
boundaries)

Figure 1. 3D drawing view of the punching apparatus

2.3 Image acquisition

The images of the punching of the analogue soil by the foundation were acquired by a digital camera. The
location of the camera has been fixed in order to avoid an image distortion (minimal distance from the device)
and to align the camera horizontally and vertically on the interesting part of the granular matter, i.e. under the
foundation. A colored square grid is drawing on the analogue granular matter at the initial stage to help the
2
following of the displacement patterns (each square measure approximately 25 × 25 mm
The figure 2 shows an example of image acquisitions in the case of foundation with 160 mm in width at four
stages of the load. In particular, the stage (t3) corresponds at the moment after the punching shear failure. Note
that for each initial stage (t0), the analogue granular matter was systematically expanded by hand to have an
initial packing as loose as possible.

Figure 2. image acquisitions of the soil punching (initial depth D of foundation is 100 mm)CRAMSG-2015 3
f

298
3 IMAGE ANALYSIS

3.1 Image preprocessing

Several preliminary stages are required in order to treat the image sequences before the PIV analysis. In this
work, the FIJI open source software was used to both the preprocessing and the PIV analysis. The following
stages were applied in order to obtain binary images:
- The interested area should be delimited by removing the not required data on the images as for example the
frame of the device.
- The background of the images could be treated by removing smooth continuous backgrounds via a rolling
ball algorithm (the radius of curvature should be at least as large as the radius of the diameter of rods)
- Each image is converted to 8-bit grayscale.
- One additional operation may be applied in odre to obtain a binary image. By make a zoom on one of
the faces of rods, white and black threshold is carried out in the zoom selection to distinguish the center of
each grain. Dilate or erode operations may be carried out to ameliorate the threshold. Note that this operation
is not turned out to be necessary: the PIV may be applied directly on a8-bit grayscale image. Some authors
used to improve the image acquisition the drawing of colored points located at the faces of the rods (Sibille
and Friio, 2007).

3.2 Particle Image Velocimetry analyses

The using of digital image analyzing is widely developed in particular to extract information from the images. A
manually analysis done manually on our rods would be time consuming and influenced by a personal bias. To go
to an automatized processing, we have developed here the use of Particule Image Velocimetry (PIV) plugin on
your image acquisitions.
Both for a first approach and to evaluate the feasibility for our rods in the case of the punching, we use the
PIV ImageJ plugin developed originally for the Traction Force Microscopy (TFM) applications (Qinqzong,
2011). Two images following one another were extracted from the punching test, thus the preprocessing
described above were applied to obtain a binary image which allow the distinction between the faces of the rods
to be drawn . The preprocessing is a delicate stage because some information could be loses during the threshold
operations. The iterative PIV operations which were performed in this study use the template-matching cross-
correlation type. Compare to a basic cross-correlation basic PIV, this PIV iterative method consists to decrease
several time the size of the interrogation window during an iteration to increase the accurate of the PIV.
A typical result of the PIV plugin has been shown in figure 3 in the case of the foundation described in 2.3 (the
foundation is 160 mm in length and 100 mm in-depth). The vectors represent the displacement field coloured as
a function of the displacement magnitude. In particular, the results show that the PIV analysis clearly highlights
the field of the displacement in several areas of soil. Thus, the results of the PIV are in good agreement with the
image observations of the shear localization observed on the untreated images (see figure 2).

Figure 3. PIV analysis exhibits the slip – line field in the case of spread foundation punching.

3.3 Punching mechanism

As shown in figure 4-a, the Prandtl mechanism consists to describe the flow of the soil under a foundation in
three distinct zones. This zones are represented in the case of our PIV analysis in the figure 4-b.
The figures show that the experimental tendencies via the PIV have been qualitatively in agreement with the
Prandtl mechanism. The area I under the foundation is a triangular form with a vertical translation. Note that the
value of angle depends of intrinsic friction angle of the analogous granular matter but also of the roughness of
the footing of the foundation.

299
The area II exhibits a lateral movement of the analogue soil corresponding to the radial shear zone. At last, the
area III corresponds to the backflow of the soil leaded by the punching. Note that the last area IV does not show
any sizeable influence on the soil (quasi motionless).

Figure 4. a) Prandtl system (from Venkatramaiah, C. (1993)) - b) Flow areas and displacement magnitude field obtained
by the PIV analysis
4 CONCLUSION

We have presented the potential to use of particle image velocimetry applied to bidimentional analogue materials
in the particular case of shallow foundation bearing capacity. A dedicated apparatus has been developed to punch
the analogue soil. The results showed that the PIV analysis provided valuable information of failure mechanism
under foundation. These PIV results can be exploited in order to compare the image data with other approaches
like as example FEM simulations, kinematics models (Baars, 2014), etc.
The future work should be carried out both to experimental and numerical ways. For the experimental way,
the load mechanism could be improved to control the displacement during the loading and the image acquisition
could be developed in order to catch the rotation of the rods (Richefeu et al., 2012). For the numerical way, the
main idea should be to model the foundation and the soil by discrete element method via a parallel in-house 3D
DEM code (Pizette and Rémond, 2014) or another DEM tool in order to compare the global and local behaviors
of soils and the interaction with the footing given both by numerical and experimental approaches.

REFERENCES
Abriak N.-E. (1998) Local friction effect of the global behaviour of granular media Mathematical and Computer Modelling
28, 121–133.
Abriak N.-.E and Caron J.F. (2006) Experimental study of shear in granular media Advanced Powder Technology 17, 297-
318.
Dolzhenko N. (2002) Etude expérimentale et numérique de modèle réduit bidimentionnel du creusement d’un tunnel.
Développement d’une loi de comportement spécifique, PhD, INSA Lyon.
Hall S.A., Desrues J., Viggiani G., Bésuelle P. and Ando E. (2012) Experimental characterisation of (localised) Deformation
Phenomena in Granular Geomaterials from Sample Down to Inter-and Intra-grain Scales Procedia IUTAM 4, 54-65.
Jenck O., Dias D. and Kastner R. (2009) Discrete element modelling of a granular platform supported by piles in soft soil : “
Validation on a small scale model test and comparison to a numerical analysis in a continuum” Computers and
Geotechnics 36, 917 – 92.
Misra A. and Jiang H.(1993) Measured kinematic fields in the biaxial shear of granular materials. Computers and
Geotechnics 20, 267–285.
Misra A. and Poorsolhjouy P. (2013) Micro-macro scale instability in 2D regular granular assemblies Continuum Mechanics
and Thermodynamics, 1-20.
Qinqzong T. (2011) Particle image velocimetry plugin (ImageJ) [Link]
Richefeu V., Combe G.,Viggiani G. (2012) An experimental assessment of displacement fluctuations in a 2D granular
material subjected to shear Géotechnique Letters 2, 113-118.
Pizette P. and Rémond S. (2014) Experimental and numerical studies of the effect of rolling friction for sandpile formation
WCCM XI conference, 20-25 july 2014, Barcelona, Spain.
Schneebeli G. (1956) Une analogie mécanique pour les terres sans cohésion. C.R. Acad. Sc. Paris, 125-126.
Sibille L. and Froiio F. (2007) A numerical photogrammetry technique for measuring microscale kinematics and fabric in
Schneebeli materials. Granular Matter 9, 183-193.
Van Baars S. (2014) The inclination and shape factors for the bearing capacity of footings, Soils and Foundations 54, 985–
992.
Venkatramaiah C. (1993) Geotechnical Engineering New Age International publishers. ISBN: 81.224.1793.0.

300
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Numerical Modeling of Piles in Expansive Soil

Modélisation numérique des Pieux dans un sol gonflant

N. ElSakhawy1, T. Nageeb2, R. ElBadawy3

1,2
Zagazig Univ., Zagazig, Egypt
3
Eng. Institute, Zagazig, Egypt

ABSTRACT - Expansive soil causes considerable damages to structures all over the world. It is necessary that
the expected heave of foundations should be calculated and consequently accounted for when designing
foundations resting over expansive soils. Pile foundations are an alternative for the cases of highly expansive
soils extending to deep layers under the foundations, especially in light weight structures. This paper presents
numerical modeling of piles in expansive soils using the finite element software (ADINA) in which the expansive
soil is modeled using the Cam-Clay soil model. As a verification of the proposed numerical model, a well
documented case study of piles constructed in expansive soil is considered. Results showed that ADINA can
effectively model heave in saturated expansive soil around pile foundation. Soil is exposed to ground water
pressure, and/or wetting due to inundation that can last for years.

RÉSUMÉ – Le Sol expansif cause des dommages considérables à des structures partout dans le monde. Il est
nécessaire que la houle attendue des fondations doit être calculée et, par conséquent représenté lors de la
conception des fondations reposant sur des sols expansifs. Les fondations sur pieux sont une alternative pour les
cas de sols très vastes s’étendant aux couches profondes sous les fondations, en particulier dans les structures
légères. Cet article présente la modélisation numérique des pieux dans les sols gonflants utilisant le logiciel
d'éléments finis (ADINA) dans lequel le sol gonflant est modélisée en utilisant le modèle du sol Cam-Clay.
Comme la vérification du modèle numérique proposé, une étude de cas bien documenté de pieux construits dans
le sol gonflant est considérée. Les résultats ont montré que ADINA est mieux pour le cas des pieux entourés par
un sol gonflant saturé. Le sol est exposée à la pression de l'eau, et / ou à l’eau sous terraine en raison de
l'inondation qui peut durer des années.

1- INTRODUCTION

Expansive soil swells when subjected to moisture; conversely shrink when dry. Soils with smectite clay,
(montrmorillonite minerals) exhibit the most profound swelling properties. Soils with a high percentage of
swelling clay have a very high affinity for water partly because of their small size and partly because of their
positive ions (Day, 1999).
Swell potential presents a guide for selecting the foundation type in expansive soils. For sites with moderate
to very high swell potential, straight or belled piles may offer a good foundation (Chao, 2007).
The main objective of this research is to present a model of piles in expansive soil. In order to validate the
proposed model, it was primitively aimed to reproduce the TRACON Case Study. TRACON building is located
at Denver, Colorado. It is located on a site having highly expansive soils, (Chao 2007).

2- The TRACON CASE STUDY

A typical soil profile is shown in Figure (1). Coal seams were encountered during drilling underneath the
claystone. Pneumatic piezometers were installed at the bottom of the uppermost major coal seam to monitor the
ground water conditions in the coal seams. The values of the pressure head in the coal seam ranged from
approximately 0 to 5.8 meters (Chao, 2007). Subsurface nuclear gauge tubes installed within the TRACON
facility indicated that the coal seam is a water source for the overlying claystone bedrock (Chao, 2007).
Average diameter and length of constructed piles in TRACON building were selected for numerical solution.
A pile with length of 8 meters, and 0.6 m diameter is modeled. That is the computed pile heave will be compared
to the average measured heave. The Lateral extent of soil domain was up to six times the pile diameter, and the
depth is limited to 18.3 m (end of pile boring of the case study). Pile monitoring for the TRACON building was
performed from September 2000 through June 2006. The maximum and average heave values of the pile
monitoring points since beginning of monitoring were approximately 3.9 and 1.5 cm, respectively.

301
0.0 Dry, Light, Yellowish, Brown Stiff to Medium Stiff Silty clay Fill with Minor
Fill
1.2
Clay Dry, to Slightly Moist Yellowish, Brown, Firm to Hard Silty Clay
2.6
Weathered Moist Very Dark Grayish Brown Firm to Hard Blocky Weathered claystone
Clay stone with Occasional Pockets of White Precipitate
5.5

Slightly Moist, Dark Yellowish, Brown and Dark Gray, Hard,

Claystone 1 Blocky, Claystone, with iron staining on Joints

10.5 Moist to wet, Black Dark Gray, and Dark Yellowish Brown, Coal
Coal Occasional Claystone and Sand Lenses
11.9
Clayston Same as Above for Claystone

End of Boring 18.3 meters

Not to Scale

Figure 1. Soil Profile of TRACON, Case Study (After Chao, 2007).

3- NUMERICAL MODEL OF TRACON CASE STUDY

Finite element analysis software ADINA with plane strain and axi-symmetric conditions is dapted in this
research. Single pile and the surrounding soil domain are modeled as axi-symmetric two dimensional elements,
with specifying porous media for modeling the soil. Roller supports were applied as boundary conditions on the
vertical sides of the model to allow for vertical movement (settlement). Hinged ones are used on the bottom of
the model to prevent any vertical or lateral movement at the bottom boundary.
Soil and pile model formulation were presented by 4-node element. Elastic Isotropic model is used to model
the pile material. Modulus of elasticity for the pile is set to 21000 MN/m2. Poisson's ratio and density are 0.25,
and 25 kN/m3, respectively.
Mohr Columb soil model was applied to represent the fill, sand, and non-expansive clay. Saturated expansive
soil is modeled with the Cam Clay Model (Schofield & Wroth, 1968). The main five parameters of the Cam Clay
model are shown in Figure (2). The four parameters: κ, λ, M, Γ, were chosen through four sets of expansive clay
types to identify the most appropriate set of parameters in modeling the pile heave in the TRACON case study.
Table (1) presents the applied sets of Cam Clay model parameters.

v Virgin Consolidation line

N (Normal compression line)
V Specific volume
Critical State Line
Γ λ p’ Effective mean stress
1 Swelling line  Slope of the normal compression Line (virgin
ĸ
(Loading Unloading Line) consolidation line, or critical state line in ν- ln p’ space
1 ĸ Slope of the swelling line in ν- ln p’ space
N Specific volume of normal compression line at unit
pressure, and it is dependent on units of measurements.
Γ specific volume of the CSL at unit pressure
M Slope of the CSL
1 p= p’p0 Ln p

Figure 2. Parameters of Cam Clay Soil Model.

The TRACON building is a steel frame structure supported on 146 bored piles and grade beam foundation
system with a slab-on-grade floor. The constructed piles range from 7.6 to 8.5 meters in length and from 46 to 76
cm in diameter.
The average diameter and length of constructed piles in TRACON building were selected for numerical
solution. A pile with length of 8 meters, and 0.6 m diameter is modeled. That is the computed pile heave will be
compared with the average measured heave. The lateral extent of soil domain was up to six times the pile
diameter, and the depth is limited to 18.3 m (end of pile boring of the case study). Typical numerical model of
TRACON case study is represented in Figure (3).

302
Table 1. Referenced Cam Clay Parameters Applied in the Case Study

Soil Cam clay Parameters Reference

M Λ Κ Γ
1- London clay 0.888 0.161 0.062 2.448 Parry *
2- Wiener Tegel V 1.01 0.122 0.026 2.130 Hvorslev*
3- Weald Clay 0.95 0.093 0.035 1.88 Parry*
4- Silty clay (CH) 1.02 0.049 0.018 2.01 Abdl Mofez, 2010

* Cited by Schofield and Wroth, 1968.

Studied Pile
Node
0.3 m 2m 4m

Fill

Clay

Pile
8m
Weathered Clay
Stone

Claystone

18.3 m

Coal

10m
Claystone

Figure 3. Numerical Model of TRACON Building.

3.1 Modeling of Pore Pressure, and Wetting in TRACON Case Study

The pneumatic piezometers in TRACON site show that there is a pressure head around the coal seam layer,
ranges from approximately 0 to 5.8 meters. This indicated that the coal seam is a water source for the overlying
claystone bedrock. Hence, water pressure is modeled with a head ranged from 6.4 m (thickness of coal seam
layer) at second stage till 12.8 m up in soil domain at the fifth stage of modeling as explained in the next section.
The wetting due to seasonal fluctuation and other run-off water sources is also modeled. The documented
water content profile, with TRACON case study, indicates that changes in water content are close to the coal
seam and are subjected to seasonal fluctuation (Chao, 2007). Water pressure profile due to inundation is
modeled as constant pore water pressure of 5.5 m, where ground water pressure affects the soil below this depth.

3.2 Phases of the Numerical Solution

Five phases are considered to fulfill the different loading for soil and pile. The pile is loaded with a pressure
intensity of 400 kN/m2 on the pile edge at the 4th phase. A schematic presentation of loading phases of the model
is shown in Figure (4).
These phases in sequence are:
1. Soil settlement due to own weight, and lasts for three years,

303
2. Soil heave due to ground water pressure till coal layer (6.4 m from bottom of soil domain), and wetting of
5.5 meter below ground surface,
3. Soil settlement due to pile installation,
4. Pile loading, and
5. Pile heave due to rising of ground water table above the coal layer, up to 5.8 m; 12.8 m from bottom of
soil domain, and full wetting till 5.5 m below ground surface. This phase extends for six years.
40 t/m2 40 t/m2

5.5 m 5.5 m
8m
8m

18.3 m
12.8 m

6.4 m

Phase 1 Phase 2 Phase 3

Phase 4 Phase 5

Figure 4. Schematic Sequence of Loading Phases for TRACON Case Study.

3 RESULTS AND DISCUSSION OF TRACON CASE STUDY

Pile heave was numerically calculated in the fifth phase (last phase in the analysis), which lasts for six years. In
This phase ground water table is raised to 12.8 m, and the depth of wetting is accounted in soil till 5.5 m depth.
Pile heave is calculated quarterly, at the edge point on the pile top, where most heave is expected to occur.
Figure (4) depicts the maximum, minimum, and average measured heave. The quartile computed pile heave
versus time, for the four Cam Clay sets is also shown. It is decided to use the set of soil parameters that mostly
represent the average measured heave values of TRACON building. From Figure (5), it can be noticed that
computed heave linearly increases with time for the four applied Cam Clay sets. Both Weald Clay and Wiener
Tegel show good agreement with the measured average heave, with Weald Clay much closer to the average
heave. Consequently the set of Weald Clay Cam Clay parameters is mostly suited for the soil deposits at the site
of TRACON building case. It has been adapted for further investigation of effect of pile configuration on the soil
heave.
Max Measured Min Measured Average measured Weald Clay
Wiener Tegel S. A. Mofez London Clay

40
Observed Pier Heave (mm)

0
0 9 18 27 36 45 54 63 72 81

Time ( Months)

Figure 5. Calculated Pile Heave at TRACON Building. Colorado.

5 CONCLUSIONS
Based on the current study, the finite element software, ADINA can effectively present the behavior of soil and
pile. ADINA can model heave in saturated expansive soil exposed to ground water pressure, and/or wetting due
to inundation that can last for years.
Weald Clay Cam Clay parameters are mostly suited for the soil deposits at the site of TRACON building. It
has been adapted for further investigation of effect of pile configuration on the soil heave.

304
REFERENCES

Abdul-Mofez, S. and Nurul-Islam, M. (2010) Modeling and Numerical Analysis of Expansive Soil in stress Path Tests.
ASCE, GeoFlrida, Advances in Analysis, Modeling & Design, (GSP 199), pp747-756.
ADINA, R and D Inc. (2011), ADINA, 8.7. 3.
Chao, K. C. (2007) Design Principles for Foundations on Expansive Soils. Ph.D. Colorado State University, Fort Collins,
Colorado.
Day, R. W. (1999) Geotechnical and Foundation Engineering Design and Construction, McGraw-hill, New York.
Schofield, A. N. and Wroth, C. P. (1968) Critical State Soil Mechanics, McGraw-hill, London

305
306
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Consequences of Geotechnical Uncertainties in Predicting Settlement of

Shallow Foundation on Sands

Conséquences des incertitudes géotechniques dans la prédiction de tassement de fondation

superficielle dans le sable

M.I. Lingwanda*1, S. Larsson2 and D.L. Nyaoro3

1
Mbeya University of Science and Technology, Mbeya, Tanzania
2
KTH - The Royal Institute of Technology, Stockholm, Sweden
3
University of Dar es Salaam, Dar es Salaam, Tanzania
*
Corresponding Author

ABSTRACT -Reliability based design, a rational approach to dealing with uncertainties demands quantification
of both load and resistance related uncertainties as inputs to design problems. In the process of obtaining design
values, soil’s natural inherent variability, errors in measurement process and errors in transforming measured
values to design parametersare the major contributors of the uncertainties in geotechnical designs. Consequently,
different soil characterization methods for determination of resistance parameters are associated with different
degrees of uncertainties. This paper demonstrates through simple procedures, the effect of different magnitudes
of soil modulus uncertainties to the predicted settlement of shallow foundation in sands through pad foundations
of three different configurations. Four different site characterization methods were involved in estimating
constrained modulus of the soil; the standard penetration test, the cone penetration test, the light dynamic
probing test and the laboratory oedometer test, all performed side by side on the same site in Dar es Salaam,
Tanzania. The case study indicates that the effect of uncertainty in a probabilistic design depends on the selected
factor of safety. At the limit state, a change in coefficient of variation of soil design parameter has no effect on
the calculated probability of failure. At the safe region, an increase in the uncertainty of soil design parameter
result to a higher magnitude of the probability of failure.

RÉSUMÉ – Les erreurs de mesures sur site ainsi que celles induites au cours de la conversion des données
expérimentales en paramètres sont les principaux facteurs, en termes d’incertitudes, pour le dimensionnement en
géotechnique. Par conséquent, chaque méthode de caractérisation, en laboratoire ou in situ, est associée à un
degré d’incertitude donné. Dans ce manuscrit, on démontre à travers des procédures simples, l’effet d’incertitude
dans la mesure des modules élastiques des sols sur le tassement des fondations superficielles reposant sur des
sols granulaires. Suivant le cas d’étude réalisé sur le site expérimental Dar Es-Salaam, Tanzanie, on trouve que
l’effet d’incertitude lors d’une étude probabiliste de dimensionnement dépend du facteur de sécurité choisi. A
l’état limite, une variation du paramètre de dimensionnement n’a pas d’effet sur la probabilité de rupture. Dans la
zone sécuritaire, une augmentation d’incertitude de ce paramètre induit des probabilités de ruptures élevées.

1 INTRODUCTION

Site investigation for foundation design typically involves in-situ testing of the soil and/or collection of samples
for laboratory testing. Several soil testing methods are usually available for applicationdepending on availability
of equipment, type data required, experience and cost of performing the tests. In Tanzania, where this study is
based, the standard penetration test (SPT) is the most commonly used method of sub-soil investigation for
conventional foundations. In most projects, the SPT is performed while simultaneously collecting samples for
further laboratory analysis. At present, it is generally agreed that this kind of sampling and laboratory testing is
expensive in both time and monetary terms as compared to in-situ testing.
This study is part of an ongoing research with the main objective ofcutting the cost of geotechnical
investigations by incorporating other faster and much cheaper solutions to data acquisition such as the cone
penetration test (CPT) and the light dynamic probing (DPL). The CPT, although is one of the oldest and popular
in-situtesting methods, it is still relatively new in Tanzania. The DPL has been in use for quite a long time
although the data have been used mainly for site stratification purpose [Link] the uncertainties related to
individual test methods and their consequences are understood, this information may be helpful in objectively
deciding on type of tests or a combination of tests to employ in geotechnical investigation of a particular project.

307
The first part of the study (Lingwanda et al., 2014a) involved determination of relationship between SPT,
CPT and DPL data in order to expand applicability of DPL data to design process. In the second part,
(Lingwanda et al., 2014b), relationships were established between the in-situ data from SPT, CPT and DPL and
data from laboratory testing by the oedometer method to determine constrained modulus of sandy soil ( ).
Geotechnical uncertainties related to the in-situ and laboratory method were evaluated and reported. In this
article, the objective is to study the effects of the uncertainties in relation to prediction of settlement of shallow
foundations on sand using field and laboratory data collected from a site in Dar es Salaam, Tanzania. This
comparison is important because as noted by Akbas and Kulhawy (2009), most settlement prediction in normal
foundation are based on information from in-situ tests. Monte Carlo simulation process is used to determine the
probability of failure at different levels of uncertainties. The effect of assuming normally distributed random
variables in a probabilistic design is also studied.

Case 1 Case 2 Case 3

Figure 1. Foundation sizes in relation to site stratification for the three design alternatives

2 STUDY DATA

Data for this study were collected from a research dedicated site located within the University of Dar es Salaam
main campus. Three in-situ and one laboratory testing method were employed in the process of data collection;
the SPT, the CPT, the DPL and the laboratory oedometer. For details on how the test were performed and the
geology of the area where the data were collected, reference should be made to Lingwanda et al. (2014a&b). For
the purpose of the current study, a summary of statistical parameters including the mean ( ), the coefficient of
variation of spatial variability ( ), the coefficient of variation of the transformation error from in-situ data
to ( ) and the vertical scale of fluctuation ( ) are presented in table [Link] useful studies on
quantification of geotechnical uncertainties include Phoon and Kulhawy (1999a&b),Baecher and Christian
(2003) and Foye et al.(2006). As sampling location have been found to have effect on the designed foundations
(Goldsworthy et al., 2007), the parameters presented in table 1 were determined from closely spaced side by side
tests performed using the four methods. Uncertainties in the applied load are not considered in this study.

Table 1. Statistical parameters for the soil characteristic values

Method Measured property (m)
2
Oedometer 5.7 MN/m 0.24 - 2.7
SPT 44 0.41 0.28 3.3
CPT 14 MN/m2 0.33 0.28 1.1
DPL 29 0.32 0.26 1.0

3 DESIGN PROBLEM

To demonstrate the effect of uncertainties when predicting settlements, three simple design problems were set.
Square pad footings of dimensions indicated in figure 1 were proposed for determination of design loads Q1, Q2
and Q3 and eventually predict settlement at different load magnitudes. The foundations base dimensions were

308
chosen to be 2 m by 2 m for case 1, 1 m by 2 m for case 2 and 1 m by 1 m for case 3. The foundations were
assumed to rest in a homogeneous layer of silty sand whose soil properties are listed in table 1. As indicated in
figure 1, a layer of hard stratum is assumed to exist just below the 10 m deep homogeneous layer.

4 SETTLEMENT PREDICTION METHOD

D´Appolonia et al. (1968) method for determination of immediate settlement was adopted in this study primarily
because it involves application of soil´s constrained modulus, a parameter which was statistically evaluated in
previous [Link] the method, the immediate settlement is given by;

(1)

where = settlement, = foundation pressure, = footing width, = modulus of compressibility of the soil
and = Poisson´s ratio, = influence factor. If oedometer results are used for the design, becomes
constrained modulus , the lateral strain is restricted, therefore and equation 1 should be adjusted
[Link] Das (2007, p.221), the value should be averaged over a depth from the ground
level. There are many forms of influence factors suggested in literature. In this study, influence factors applied
by D´Appolonia et al. (1970) were employed. The factor is a function offooting geometry and the distance from
the footing base to the location of the immediate incompressible layer, as indicated in figure 1.

(a) (b)

Figure [Link] of data (a) Oedometer (b) SPT (c) CPT and (d) DPL

5 DESIGN PARAMETERS

Probabilistic design requires a full description of the probabilistic variables in terms of the mean, coefficient of
variation (COV) and the type of distribution. In this study, only the parameter is treated as a probabilistic
variable while the rest are deterministic. The mean of directly measured is simply obtained from table 1.
However, in order to apply equation 1 to predict settlement with in-situ data, the , and given in table
1 should be converted to equivalent . The conversion may be done with application of equations 2 to 4 which
were established by Lingwanda et al. (2014b). By substituting data from table 1 the value of average is found
to be 5.2, 5.6 and 5.8 MN/m2 respectively for conversion from , and data. Since this range of mean
is narrow, the major concern is put into the variation of the parameters in terms of COV while the mean is

309
kept constant at 5.7 MN/m2in all [Link] transformation to introduces some degree of uncertainty which
was quantified and presented in table 1 as .
(2)

(3)

(4)
where = constrained modulus of the soil, = tip resistance from CPT, = number of blows per 10cm
penetration from DPL and = energy corrected number of plows per 30cm penetration from SPT.
Normal distributions are usually proposed to describe natural phenomenon, geological and geotechnical
phenomena are among them. A supporting argument is that such phenomena are determined by a contribution of
large number of small effects making the distribution of the overall effect to be normal (Baecher and Christian,
2003, p. 307). One practical argument which opposes the application of normal distributions for geotechnical
parameters is that they allow negative values which cannot be accepted for performance functions and lognormal
distributions are proposed instead. Baecher and Christian (2003) argues that the error introduced by allowing the
negative values is insignificant compared to the calculated probability of failure. In this study, Kolmogorov -
Smirnov test was used to test for normality of , and data collected from the study area. The test
statistic rejected the normality hypothesis for all four data sets at 95% confidence. The histograms for data
distributions are presented in figure 2 where it can be observed that all data except values may be
approximated as normal distributions.

6 PROBABILISTIC DESIGN

Monte Carlo simulation is a robust and straight forward method for performing probabilistic analysis. Through
repeated execution of a desired deterministic process, the method yields a set of output variables which may be
subdivided into two sets separated by a desired limit. In settlement prediction studies, this limit may be taken as
the allowable settlement and the probability of failure ( ) may be determined by the ratio of summation of
events that settlement exceeded the allowable value to the total number of events. The necessary number of
repetitions ( ) for the simulation process has been related to the desired such that smaller values require
large number of repetitions (Massih et al., 2008) to achieve a stable solution. According to Phoon (2008, p. 6),
the rule of thumb is to have repetitions to be able to estimate the accurately. But values cannot be
known before hand and therefore should be represented by some typical values, . For geotechnical design
problems, values are usually smaller than 0.001 which implies that should be in order of 10,000. Uzielli
and Mayne (2011), adopted a more objective method that utilizes confidence level ( ) to estimate the minimum
in relation to as;

(5)
By choosing at , would mean a target probability of failure of magnitude
around , which is classified by U.S Army Corps of Engineers(1997) as a case between high and good
performance design.
In geotechnical analyses, the average value of a property is usually of great concern. The local average over a
given soil volume is influenced by the size of the particularvolume which in turn depends on the type of
construction and the failure mechanism. The implication of averaging volume is that the lower the averaging
volume the higher the possibility that this average deviates from the global value. One way of accounting for the
effect of averaging volume is by the application of the variance reduction function over the length L, expressed
by Vanmarcke (1977) as . The deviation from global mean due to the size of averaging volume affects the
variance (or COV) and may expressed by equation 6.

(6)

for or for (7)

(8)
where is the variance of the property locally averaged over the length . For a settlement prediction
problem, is taken as the depth of influence of the footing. The variance function in one vertical dimension is

310
determined by equation 7 and the total variation after this reduction is then determined through equation 8 for
application in design.
After considering design variables from table 1 and applying variance reduction according to equations 6 and
7, the total COV was determined for each design case. The design was kept constant at 5.7 MN/m2 for each
design case but the COVs were varied in accordance to the field investigation methods. For design case 1, the
design COVs are 0.14 for oedometer method, 0.28 for DPL, 0.30 for CPT and 0.38 for SPT. For the design case
2, design COVs are 0.16 for oedometer, 0.29 for DPL, 0.31 for CPT and 0.41 for SPT. For design case 3, a COV
of 0.12 represents the oedometer method, 0.28 for DPL, 0.30 for CPT and 0.37 for SPT. A range of COVs
between 0.10 and 0.50 is studied in each design case to accommodate all the four investigation methods. In the
analysis, first the design load at an allowable settlement of 25 mm is deterministically obtained for each case.
This load represent the limit state condition. Then the load is factored and the effect of factoring is studied. The
calculated allowable loads are used to predict settlement in a probabilistic framework.

7 RESULTS AND DISCUSSION

Design variables were put into a Monte Carlo process assuming normal distribution of values. The events
where were counted as failures and used to calculate . The percentage of cases were the magnitudes
of settlement were found to be less than zero ( were also recorded for eachrealization. Three design states
differentiated by the magnitudes of factor of safety (FS) were studied; FS = 1 (the limit state), FS = 1.05 and FS
= 1.1. The results of the analysis were tabulated in terms of the COVs against the calculated and the
percentage of . Twoeffects were studied, the consequence of magnitudes of COVs in the calculated and
the effect of negative events caused by the assumption of normally distributed data.

0.6 0.6 0.6

0.5 0.5 0.5

Probability of failure (Pf )

0.4 0.4 0.4

0.3 0.3 0.3

0.2 0.2 0.2

0.1 0.1 0.1

0 0 0
0 0.1 0.2 COV
0.3 0.4 0.5 0.6 0 0.1 0.2 COV
0.3 0.4 0.5 0.6 0 0.1 0.2
COV0.3 0.4 0.5 0.6

Figure [Link] of magnitude of COV on calculated (a) design case 1, (b) design case 2, (c) design case 3

4.1 Effect of coefficient of variation on the calculated probability of failure

The effect of uncertainties in form of COVs to the calculated is studied through figure 3 where at low COVs
(for example application of oedometer data), the effect is significantly different for distinct factors of safety
whereas for higher COVs (such as those caused by applying SPT data), the difference between values at
different factors of safety is smaller. The effect of COV appears nearly constant at the limit state but increases
with increase in COV at a higher magnitudes of factor of safety. The trends of variation in with COVs at specific
factors of safety appears similar for all design cases prompting the suggestion that the trends are a function of the respective
design model applied to predict settlement.
Figure 4 shows the trend of COV against at different limit states when results from the three design cases
are put together. At FS = 1.05 and FS = 1.1, the trend indicates an increase in with increase in COVs where
the rate is higher for the latter than the former case. However, at the limit state, the appears decreasing with
increase in COV. This condition is strange and could be the effect of negative generated random variables caused
by the application of normal distributions. The decrease in with increasing COVs is also observed in the
trends representing FS = 1.05 and FS = 1.1 but appears smaller. When the events are added to the events
of and obtained by the ratio of this summation to the total number of simulations, new trends of

311
COVs against indicate that at the limit state, the effect of COVs is nearly constant regardless of their
magnitudes (figure 4 a). For FS = 1.05 and FS =1.1, the drop of is lifted up when the negative events are
taken into consideration as indicated in figure 4b and c. Mathematically, the negative settlements are obtained
when the values of used in calculations are very large. Due to the symmetry of normal distributions, the
amount of very high and very low values should be the same and the addition of events to the
events in calculating probabilities of failure is reasonable.
4.2 Effect of the normal distribution assumption on the calculated probability of failure
The effect of negative events due to normal distribution assumption is studied in figure 5. The general
observation is that at low COVs less than 30%, the effect may be considered negligible but may be very
significant for COVs greater than 30%. The effect appear to increase exponentially with increase in COVs such
that in designs where magnitudes of COVs are higher, such as with SPT data, normal distributions should be
avoided and lognormal distribution applied instead. An alternative to dealing with the effect of negative values is
to add them to failure events and the sum used to calculate the probability of failure as indicated in figure 4.

Figure [Link] of magnitude of COV on calculated for the three design cases together (a) FS = 1 (b) FS = 1.05 (c) FS = 1.1

Figure [Link] of magnitude of COV on the percentage of negative output events

8 CONCLUSIONS

From the results of this study, the following conclusions can be drawn;
i. The trend of relationship between coefficient of variation and probability of failure is a function of the
design model applied. At a given factor of safety, the trend is unchanged regardless of the design
problem provided that the same design model is applied in the determination of the probability of
failure. When such trends are made available for different design models and factors of safety, one can
readily determine the effect of different magnitudes of geotechnical uncertainties to the calculated
probability of failure.

312
ii. The effect of coefficient of variation on the calculated probability of failure depends on the design factor
of safety. A change of design coefficient of variation has no effect on calculated probability of failure at
the limit state but may have significant effects for designs at higher factors of safety.
iii. Because of the different magnitudes of uncertainties associated to different geotechnical investigation
methods, different magnitudes of probabilities of failure are to be expected accordingly for data from
different investigation methods even if the means of design parameters derived from the individual
methods is the same. Because they are associated with higher magnitudes of uncertainties, in-situ
methods are likely to result to higher probabilities of failures than laboratory methods and consequently
a higher possibility of a conservative design. This information is missed when one applies deterministic
design method.
iv. The effect of normally distributed parameters in probabilistic design affect the calculated probability of
failure by giving falsely lower values especially when the coefficient of variation of the random variable
is greater than 30%. Lognormal distributions should be applied instead to avoid this effect.

ACKNOWLEDGEMENT

The authors wish to acknowledge the support of Sida (Swedish international development cooperation agency)
through the project ``Continuous Penetration Test as Alternative to Intensive Borehole Investigation´´ under the
Rural and Urban Infrastructure Development Program (Tanzania).

REFERENCES

Akbas S. O., Kulhawy F. H. (2009) Reliability-based design approach for differential settlement of footings on cohesionless
soils. Journal of geotechnical and geoenvironmental engineering, 135(12), 1779-1788.
Baecher G. B., Christian J. T. (2003) Reliability and statistics in geotechnical engineering. John Wiley & Sons.
Das, B. M. 2007. Principles of Foundation Engineering. 6th ed. s.l.:Thomson.
DÁppolonia D. J., DÁppolonia E., Brissette R. F. (1968) Settlement of spread footings on sand. Journal of Soil Mechanics
& Foundations Div.
DÁppolonia D. J., DÁppolonia E., Brissette R. F. (1970) Closure to Settlement of spread footings on sand. Journal of Soil
Mechanics & Foundations Div.
Foye K. C., Salgado R., Scott B. (2006) Assessment of variable uncertainties for reliability-based design of foundations.
Journal of geotechnical and geoenvironmental engineering, 132(9), 1197-1207.
Goldsworthy J. S., Jaksa M. B., Fenton G. A., Kaggwa W. S., Griffiths V., Poulos H. G. (2007) Effect of sample location on
the reliability based design of pad foundations. Georisk, 1(3), 155-166.
Lingwanda M I., Larsson S., Nyaoro D. L. (2014a) Correlations of SPT, CPT and DPL data for sandy soil in Tanzania.
Submitted to Geotechnical and Geological Engineering
Lingwanda M I., Prästings A., Larsson S, Nyaoro D. L. (2014b) Comparison of Geotechnical Uncertainties Linked to
Different Soil Characterization Methods. Submitted to Geomechanics and Geoengineering: An International Journal
Phoon K. K., (Ed.). (2008) Reliability-based design in geotechnical engineering: computations and applications. CRC
Press.
Phoon K. K., KulhawyH.(1999a) Characterization of geotechnical variability, Canadian geotechnical Journal, 36, 612 –
624.
Phoon K. K., Kulhawy H.(1999b) Evaluation of geotechnical property variability, Canadian geotechnical Journal, 36, 625 –
639.
Vanmarcke, E. H. (1977) Probabilistic modeling of soil profiles, Journal of the Geotechnical engineering Division, ASCE,
103(GT11), 1227 – 1246.
Youssef Abdel Massih D. S., Soubra A. H., Low B. K. (2008) Reliability-based analysis and design of strip footings against
bearing capacity failure. Journal of geotechnical and geoenvironmental engineering, 134(7), 917-928
Uzielli, M. and Mayne, P. W.(2011). Serviceability Limit State CPT - Based Design for Vertically Loaded Shallow Footings
on Sand. Geomechanics and Geoengineering: An International Journal, 6(2), 91 – 107.
U.S Army Corps of Engineers, 1997. Engineering and Design: Introduction to Probability and Reliability Methods for Use
in Geotechnical Engineering, Washington DC.

313
314
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Rehabilitation and underpinning of a building classified as a national patrimony

Réhabilitation et reprise en sous-œuvre d’un bâtiment classé comme patrimoine national

Sabrime BOUBAKER
Terrasol Tunisie, Tunis, Tunisia, [Link]@[Link]

ABSTRACT: Located in the center of a dense urban area of the Tunisian capital, the building to be modified is a
construction that dates from the early twentieth century and classified as a national heritage. The purpose of this
project is to preserve intact the original building facade by underpinning the internal structure and foundation
system. The site on which is based the building is characterized by the presence of a highly compressible soil
over several tens of meters. As there is no information about the initial foundation system of the building, a
specific campaign of surveys was conducted to analyze the structural elements present in the basement. The
purpose of this study will be to create a new level of basement and to insure the stability of the building for the
next decades.
In this paper will be detailed the project elements and the results of all structural and geotechnical investigations.
Then will be exposed the different undertaken stages for the operation of underpinning of the building.

RESUME: En plein centre d'un tissu urbain dense, l'immeuble à reprendre est une construction classée
patrimoine national, composé en R+3 datant du début du XXe siècle. Le but de ce projet est de pouvoir
conserver intacte la façade initiale du bâtiment avec reprise en sous œuvre de la structure interne et du système
de fondation en vue de la création d'un nouveau R+4 avec un niveau de sous-sol. Le site sur lequel est fondé le
bâtiment est caractérisé par la présence d'un sol très compressible s'étendant sur plusieurs dizaines de mètres. En
absence d'information sur le système de fondation initial du bâtiment, une campagne de reconnaissance
spécifique a été menée afin d'analyser les éléments de structures présents en sous-sol et de comprendre le
fonctionnement du système initial qui a permis d'assurer la stabilité du bâtiment sur plus d'un siècle.

1 INTRODUCTION
The building to be rehabilitated is composed by a first floor and three levels built on a 580 m² corner lot. The
construction is situated in a dense urban area at well known avenue of Tunis: the Habib Bourguiba Str
(geographic coordinates: 36°47’58’’N, 10°11’1’’E). It is delimited on the North and West sides by two roads.
The East side is adjacent to a similar old building and in the south side is next to a lot where has been
constructed another similar building been demolished since several years.
The project consists in removing and changing all the internal existing structure of the building to create a new
hotel compound of a first floor and four levels with an additional level of basement. The creation of such a new
basement will require an excavation of about 3 m depth.
All this work will be undertaken while keeping intact the initial facades considered as a national heritage.

2 PROJECT CONTEXT.

2.1. Building structure and fondation prospection

The construction is established on external bearing walls of big thicknesses built in freestones. The central
core, including the stairwell, is constituted by big supporting walls in masonry rubble stones and inserted bricks
(Figure 2).
Floors are shaped in full brick-built jack arches and metallic profiles. The whole is bond by a complex system
of floors – metallic stringers which insures the equilibrium of the structure by conferring it a monolithic
behavior.

The building dates from the early XXth century; no documentation on the foundation details was preserved
since the time. So, and to make a clearer idea about the existing system of foundation, we proceeded to
prospecting works in various areas of the construction.
Therefore, a 2.1 m depth excavation was opened up against a bearing wall of area 1.35 x 1.37 m² (Figure 2).
The auscultation of the underground structure let take the following hypotheses on the existing foundation mode:

315
Bearing walls rest on stone bases of a trapezoid shape until the depth of 2.1 m;
This pad lets the transfer of loads to a cyclopean raft foundation (which thickness is about 1.6 m
according to a borehole). This raft would be generalized on all the surface of the building.
The whole parts constitute a tight system (flow rate not exceeding 2.10-5 l/day and the initial water level not
restored even after a week).

Figure 2. Bearing walls constitution

Figure 2. Foundation system prospection

Besides, according to nearby excavation works on a similar old building, a similar foundation system was made
of trapezoid shaped stone bases laying on a cyclopean raft foundation. The subgrade was improved by a regular
mesh of rigid inclusions. These elements are made of eucalyptus trunks of 120 to 150 mm diameter.

Furthermore, the topographic surveys have shown that the building would have swing to the South-East side
by a maximal amplitude of 46 cm. This tipping would be due to the location of the building next to the nearby
constructions. Indeed, the building had been, according to its history, bounded on two sides by roads and on
other sides (South and East) by constructions. Besides, the foundation system such as it was explored supposes
that all the adjoining constructions rest on a structure equivalent to a common raft foundation whose settlement is
more important in the center (Figure 3).
On the other hand, and according to excavation works inside the building, it was noticed the presence of an
old flooring at 1 m under the ground level, buried on all the surface of the building. This give evidence of a long-
term generalized settlement that the construction would have sudden: it is about soil consolidation.

316
Figure 3. Settlement of the group of buildings

Therefore, and according to these diagnostics, we had concluded that the foundation system layout would be
as shown in the Figure 4.

Figure 4. Principle of the existing foundation system

2.2. Soil investigation results

Results of geotechnical investigations made on the site, allow highlighting the following geological levels:
- Level 1: (from 0 to 3.7 m) sandy fill put on a layer of calcareous blocks with joints of sand and gravel (the
existing raft foundation);
- Level 2: (from 3.7 to 18 m) gray muddy clay with shell debris the first ten meters;
- Level 3: (from 18 to 20 m) sandy greenish-yellow with shell debris;
- Level 4: (from 20 to 29 m) grayish brown clay with traces of silt and shell debris;
- Level 5: (from 29 to 32 m) yellowish gray laminated clay with traces of sand;
- Level 6: (from 32 to 45 m) grayish brown clay with silty crossings and cemented concretions;
- Level 7: (from 45 to 48 m) silty fine sand with traces of cemented concretions on the first meter;
- Level 8: (from 48 m) compact brownish clay with traces of silt.
The following table summarizes the average pressuremeter values measured.
Table 2. Soil mechanical properties
Level Z (m) Em (MPa) Pl* (MPa) Em/pl*
1 0 – 3.7 1.9 0.21 9
2 3.7 – 18 3.4 0.41 8
3 18 – 20 16.1 1.6 10
4 20 – 29 6.2 0.7 9
5 29 – 32 11.3 1.2 9
6 32 – 45 9.3 0.97 10
7 45 – 48 30 2.7 11
8 > 48 26.7 2.2 12

The water level is almost situated at 0,8 m depth below ground surface.

317
Several laboratory tests were made on some samples of level 1 to 6, it consisted of oedometer and triaxial
tests with Atterberg limit determination.
However, in level 4 and 5 no eodometer tests were done. These parameters are very important to estimate the
consolidation settlement of the new structure. Therefore, and while having an idea about the initial project loads
and its long term settlement, we have tried to make a model calibration of the soil based on the following
hypothesis:
- Project loads are composed of a general stress of 50 kPa and a linear load due to the walls weight estimated
at 300 kN/m.
- The similar neighboring buildings would have the same load distribution.
- The building settlement consists on a general settlement of about 0,5 m with a differential settlement of 0,4 m
to the South-East side;
- All soil layers are supposed to be normally consolidated;
- The ratio between the compression and the recompression index is considered equal to 10 (Cc/Cs = 10) ;
These entire hypotheses, coupled with the laboratory tests, let clayey soil oedometric calibration as following.

Table 3. Soil oedometric parameters

γh
Level OCR Cc/(1+e0)
(kN/m3)
1 17 0.18
3 19 0.12
1
4 20 0.05
5 20 0.07

3 UNDERPINNING SOLUTION DESIGN

3.1. Principal of the underpinning solution

Given the diagnostic results of the building, previously presented, and to ensure a new redevelopment of the
building on a basement level, the foundation system will be inspired from the existing system and will be
directed to a mixed foundation that consists of a generalized slab resting on a quasi-regular mesh of micropiles.
Underpinning works of the current system will begin by creating an internal rigid portal frame (main framing
structure of the future building) provisionally based groups of 30 m long micropiles. This structure will retain the
original structure of the building (including the front walls) before demolition of interior walls and floors.
Once all elements of internal structure demolished, earthworks to create the basement will be started gradually
and in parallel with the works of final raft construction. This raft will be sealed to the foundation members newly
created in previous step (micropiles groups) and will ensure “rigid box” behavior in the basement.
The principal of this underpinning system is shown in the Figure 5 below.

3.2. Design of the temporary retaining system

In order to begin earthworks without affecting the behavior of the original facade structure, it will be made at
the future columns positions, behind the outside bearing walls, a retaining system of reinforced concrete walls,
based on micropiles.

Figure 5. Principle of the solution to be undertaken

318
Foundation micropiles will be 30 m long equipped with tubular steel frames of 114 mm diameter and 7.5 mm
thick. They will be sealed by injection into boreholes of 250 mm diameter.

2.3. Design of the temporary foundation system

The temporarily foundation system will let maintaining the portal frame structure during demolition phases of
interior walls and roofs before placing permanent foundation system. This system is compound of groups of four
micropiles at each future column, realized from pre-excavation of 0.8 to 1 m depth. These micropiles will be
sized according loads corresponding to the self weight of the beams and columns of the portal frame.
Bearing capacity of micropiles is normally counted from 4 m depth. Moreover, and in order to add additional
security to the foundation system, we have considered only 30% of the effect of lateral friction throughout the
thickness of soft muddy clay (Level 2).
Thus, as these micropiles will be included in the final foundation system, we have considered elements of 30 m
in length, with a bearing capacity of isolated micropile given by the following table.
Table 4. Bearing capacity of micropiles
L (m) QSLS(*) (kN) QULS(*) (kN)
30 495.7 578.4
(*)
SLS: Serviceability limit states
ULS: Ultimate limit states
These results are supposed to be confirmed by at least one static axial load test (control test) on a micropile
foundation.

2.4. Design of the final foundation system

The final system will be a mixed type. It will consist of a foundation raft 40 cm thick resting on a semi-regular
mesh of micropiles of 250 mm in diameter and 25 to 30 m depth. Micropiles used for temporary foundation will
be an integral part of the final mesh.

The settlement under the foundation raft will have two components:
- Elastic settlements: immediate settlement consumed during the construction work progress;
- Oedometer settlement: consolidation settlement consumed very long term after dissipation of
excess pore pressures generated in the clay layers.
Elastic settlements are estimated using a 3-dimensional simulation of the raft foundation.

The soil is modeled as an elastic multilayer. Horizontal layers characterized by Young's modulus and Poisson's
ratio defined as following:

Where Em is the pressurmeter modulus and α is the soil rheological factor.

Micropiles are introduced by their equivalent stiffness. The maximum elastic settlement at the foundation raft is
about 1.7 cm.

The calculation of oedometric settlement take into consideration the effect of over-consolidation of the soil due
to the load of the initial embankment of 0.5 m and the existing slab 1.6 m thick, placed over the entire area of the
building for several years. The impact of unloading will be modeled by OCR values based on a model initially
normally consolidated as follows.

Table 5. OCR values fixed for oedometer settlement calculation

Level OCR
I 1.63
III 1.25
IV 1.19
V 1.14
The maximum calculated settlements are about 3.5 cm.

319
On the other hand, as the foundation system is floating (micropiles arrested at -25 and -30 m), it will be
necessary to estimate the subsidence that may occur under micropiles. Thus, it was considered the load transfer
of the project to the deep compressible layers by the foundation system. This led to an additional settlement of
about 1.3 cm. The total long term settlements would then not exceed the 5 cm.

4 PHASES OF WORK

In order to reduce the impact of earthworks on the elements of the original structure, including the foundation
elements, following phasing was recommended (Figure 6):
1. Generalized excavation of 0.8 to 1 m depth;
2. Achievement of temporary foundation micropiles (groups of 2 to 3 under each column) and final
micropiles (regular mesh under final raft);
3. Portal frame structure construction;
4. Demolition of interior walls and roofs;
5. Slabs construction for different levels

All this phasing has to be undertaken after securing the exterior by a peripherally confinement system to
avoid any disturbance of the existing structure

External side

Embankement

External side

Embankement

External side

Embankement

External side

Embankement

320
External side

Embankement

Figure 6. Underpining works phases

5 CONCLUSION AND PERSPECTIVES

When editing this paper, underpinning works were not yet undertaken. Several additional investigations have
to be made to fix most soil’s parameters and to confirm the study hypothesis.
Moreover, micropiles’ loading tests will be conducted to estimate the real bearing capacity of these
foundation elements.
Another alternative of the solution was proposed providing more space in the basement. The main idea was to
partially demolish the existing foundation pads so that basement’s peripheral structure can be executed just under
the first floor’s columns. However, the excavation works could not be undertaken before performing trial tests in
small areas of the existing building (not more than 2 m x 2 m surfaces).
REFERENCES

Fascicule 62 – Titre V : « Règles techniques de conception et de calcul des fondations des ouvrages de génie civil », LCPC-
SETRA.
Combarieu O. « L’usage des modules de déformation en géotechnique ». Revue Française de Géotechnique, N°114, 1er
trimestre 2006
NF P94-262 « Justification des ouvrages géotechniques », national application of the standard Eurocode7, July 2012.

321
322
Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Retaining Structures and Special Foundations at the Platinum Tower, in Maputo

Solutions de Soutènement et des Fondations Spéciaux à la Tour Platinum, à Maputo

[Link]*1, R. Tomásio2, R. Coelho3 and R. Nicolas4

1
JETsj Geotecnia Lda., Lisbon, Portugal, apinto@[Link]
2
JETsj Geotecnia Lda., Lisbon, Portugal, rtomasio@[Link]
3
JETsj Geotecnia Lda., Lisbon, Portugal, rcoelho@[Link]
4
ROSOND, Mozambique, [Link]@[Link]
*
Corresponding Author
ABSTRACT - Regarding the ongoing urban reconversion of the Maputo city, some high rise buildings are being
built with several underground floors, demanding the execution of peripheral retaining walls and deep foundation
solutions, to face local conditions, such as geological, geotechnical and vicinity. The aim of this paper is to
present the main adopted design and execution criteria regarding the retaining walls and deep foundation works
for the Platinum Tower, with 3 underground and 26 above ground floors, located in Maputo city center,
Mozambique.

RÉSUMÉ - En ce qui concerne la reconversion urbaine en cours de la ville de Maputo, quelques bâtiments de
grande hauteur sont construits avec plusieurs sous-sols qui déterminent l'exécution des parois de soutènement et
des solutions de fondations profondes, pour faire face aux conditions locales comme les conditions géologiques
et les conditions de voisinage. Le but de cet article c’est de présenter les principaux critères de conception et
d'exécution adoptées pour la solution d'excavation et parois de soutènement, bien aussi comme des fondations
profondes pour la Tour Platinum, avec 3 sous-sols et 26 étages au-dessus du sol, en construction dans le centre
de la ville de Maputo, au Mozambique.

1 INTRODUCTION

The excavation works were carried out through a partial top-down method. The peripheral retaining wall solution
was conditioned by the complex vicinity constraints. The adopted solution included the execution of concrete
bored piles, braced with part of the level -1 slab, avoiding more flexible and more invasive solutions, such as
temporary ground anchors (Pinto & Pita, 2011). The special foundations solution was conditioned by the high
loads, as well as by the complex geotechnical and geological conditions. Therefore, it was adopted a solution
with combined piled raft foundations - CPRF (Duc Long, 2011), associated with pile toe injections. This
methodology was initially used on bridges (Manai, 2011) to increase bearing capacity of the piles. For long and
large diameter piles this method is being proved as economic and time efficient.

2 MAIN CONSTRAINTS

2.1 Vicinity constraints

This building is located in a central area in Maputo City manly surrounded by small structures and important
roads, like Av. Julius Nyerere. One of the most important objectives for proposed solution is deformations
limitation for surrounding structures and infrastructures (Figure 1).

323
Figure 1. Site location and main vicinity conditions.

2.2 Geological and Geotechnical Constraints

The site is located in central/east zone of the Maputo City. The local ground is characterized by fine silty sand,
intercalated with moderately compact to compact sand. This information was validated during the initial site
investigation, allowing the definition of three geotechnical zones. The main geotechnical parameters, as well as
the geotechnical profile, used in design are presented in Figure 2. The ground water table was not detected at the
drilled depths (39m below the surface).

Figure 2. Geological and geotechnical parameters and geotechnical profile.

3 ADOPTED SOLUTIONS

3.1 Peripheral Retaining Walls

Taking into account the above constrains, the adopted solution for the peripheral retaining walls was a reinforced
concrete bored piles curtain. The bored piles Ø600mm, were executed with kelly bar and temporary head casing,
and were spaced 1m between axes. Above the final excavation level, the peripheral retaining wall bored piles
were lined with reinforced shotcrete (Figure 3).

324
Figure 3. Plan of the peripheral retaining walls, including the horizontal bracing system, and cross section 1-1.

Although the overall depth of the excavation, about 10m, the retaining wall was braced only at one level, by a
slice of the level “-1” final reinforced concrete slab (cast directly over the ground), as well as with a
steel/concrete truss (at the inclined access ramps), allowing an easily process of disassembly. This bracing
solution was used instead of the traditional temporary ground anchors both with economic advantages
(permanent elements used to brace de retaining walls during the excavation works) and technical (bigger stiffness
and deformations control) advantages (Pinto & Pita, 2011).
The horizontal bracing slab had a variable width between 6.8m and 7.8m, working as beam for the horizontal
loads due to the horizontal earth pressures. At the internal side it was supported by micropiles, sealed inside the
foundation bored piles. At the outside side, the slab was supported by the peripheral retaining wall bored piles
(Figure 4).

Figure 4. View of the excavation works, including the execution of the bracing slab.

3.2 Foundations

Considering the type and high level of loads transmitted to the ground, as well as the geotechnical and geological
conditions, it was adopted a combined piled raft foundations - CPRF solution (Duc Long, 2011), with bored
concrete piles, 1000, 1200 and 1500mm diameter. The toe of the 1200 and 1500mm diameter bored piles was
injected in order to eliminate the ground perturbation due the piles drilling and consequently to increase its
bearing capacity. As main design criteria it was considered the minimization of the overall number of piles.
When possible just two piles bellow each column were considered (Figure 5).
This methodology was applied initially in bridges (Manai, 2011) to increase the bearing capacity of large
diameter piles, with both economic and schedule advantages. All piles were executed were executed with kelly

325
bar and temporary head casing, from the platform at level “-1”, with an approximate length of drilling without
concrete of 7m (Figure 5). The raft foundation was executed with 0,35m thickness in general and 0,55m near the
pile caps zones. This variation was created to ensure a smooth transition of stiffness and consequently of shear
and bending stresses.

Figure 5. Plan of the foundations solution and view of the foundations works, with working platform at the slab -1 level.

As already stated, the bored piles were executed with kelly bar technology. The piles were designed to
transmit loads mobilizing both shaft and base resistance, compatible with a proper functionality of the building,
by limiting the foundations displacements.
Combined piled raft solution allowed a uniform distribution of displacements and forces between the raft
foundation and the bored piles. Some stresses are transmitted to the shallow ground (ZG2) relieving the loads
transfer to the piles. In order to transmit the service stresses defined in project of 8MPa or 9MPa to the ground,
the piles were embedded, at least 3 diameters in ZG1, characterized by very compacted sands, densified with the
grout injection at the toe. For maximize the design shaft resistance (between 110kPa to 135kPa in sands), all
piles had a minimum length of 20m.

4 SOLUTIONS DESIGN
4.1 Peripheral Retaining Walls

The behavior of peripheral retaining wall, for all the construction phases, was modeled on an interactive basis,
using the Plaxis 2D (geotechnical) and the SAP2000 (structural) software. The construction phases and the
geotechnical behavior, considering the “Hardening Soil” constitutive model, was modeled using Plaxis 2D. On
the other hand, the structural behavior of the bracing slab and the bracing truss was modeled in SAP2000, based
on bracing slab reactions, initially obtained from the geotechnical model. On a second stage, using the structural
software, the reinforced concrete bracing slab was modeled through shell thick elements and the bracing truss
was modeled with frame elements (Figure 6).

Figure 6. Horizontal displacements and plastic points at the peripheral retaining wall model, using Plaxis 2D.

326
4.2 Foundations

The behavior of the combined large diameter piled raft foundations was modeled through the SAP2000 software.
Three dimensional elements were used considering elastic springs, to simulate soil-structure interaction under the
slab and along piles shaft (Figure 7). The piles vertical deformability was previously calibrated using the Plaxis
2D software.

Figure 7. 3D model of CPRF solution and vertical estimated displacements - SAP2000.

5 QUALITY CONTROL AND QUALITY ASSURANCE

Considering the importance and dimension of the performed works, several procedures for quality control and
quality assurance where adopted, mainly:
 Pile integrity tests performed at the foundation bored piles, as well as at the retaining wall bored piles.
Sonic tests were carried out in all bored piles and crosshole tests were performed only in some
Ø1500mm bored piles, in order to validate the piles integrity.
 Monitoring and observation plan with inclinometers and topographic targets, allowing the assessment on
real time of the peripheral retaining wall behavior, with the settlement of alert and alarm criteria, based
on the numerical values obtained at the design stage. As a risk management tool, the site monitoring
would allow the adoption, on time, of measures to reinforce the structure and to avoid damages on
nearby structures and infrastructures (Figure 8).

Figure 8. Plan of the instrumentation devices and view of the inclinometer readings.

327
It should be pointed out that for the last stage of the excavation works the estimated maximum horizontal
displacement obtained using Plaxis 2D software at the retaining wall was about 20mm, while the real maximum
horizontal displacement on the same section was only 12mm, confirming the overall excellent behavior of the
peripheral retaining wall solution (Figures 9 and 10).

Figure 9. Horizontal displacements of I2 inclinometer estimated and measured values for the last construction phase.

Figure 10. View of the final excavation works.

6 MAIN CONCLUSIONS

The presented case study allows to pointing out the potential and level of demand of integrated geotechnical
solutions, mainly using special foundations and retaining walls for high rise buildings. In general the following
main issues in this kind of works should be underlined:
 Difficulty to perform and control the ground anchors behavior for different reasons, as legal framework
or technical issues. The last ones are related with the time dependent behavior of some tropical soils,
mainly due to changes on its water content, as well as the decrease of horizontal confinement during the
excavation works, leading to the loss of resistance and, consequently, the overall increase of the
solution deformability.
 The replacement of the traditional ground anchor bracing solution for slices of the final underground
slabs, allows a more effective deformations control of the retaining wall. On the other hand, the majority
of the bracing slabs can be integrated at the basement structural solution.
 Importance of the quality control and quality assurance procedures, mainly the monitoring and
observation plan, pile integrity tests and, whenever possible, load tests at the foundation piles.

328
 The large diameter piles toe injection has a positive influence on the piles bearing capacity and,
consequently, economic advantages, leading to the reduction of the overall execution schedule, mainly
for long piles.
 The advantages of the adoption of large diameter pile raft foundation (CPRF), mainly as high rise
buildings foundation solution, combined with the large diameter piles toe injection.

ACKNOWLEDGEMENT

The authors thank to the Owner the authorization for the writing and submission of this paper. The main
geotechnical works were well performed by the company “Dura - Soletanche Bachy”. The authors also point out
the important intervention of the supervision company, “Afaplan – Planeamento e Gestão de Projectos”, as well
as the important contribution of the company responsible for the remaining engineering projects, “Betar -
Estudos e Projectos”.

REFERENCES

Duc Long, P. (2010) Piled Raft – A Cost-Effective Foundation Method for High. Geotechnical Engineering Journal of the
SEAGS & AGSSEA, Vol. 41, No.3.
Manai, R. (2010) Enhancement of Pile Capacity by Shaft Grouting Technique in Rupsa Bridge Project, Geotechnical
Engineering Journal of the SEAGS & AGSSEA, Vol. 41, No.3.
Pinto, A. & Pita, X. (2011) Deep Excavations in Luanda City Centre, 15th African Regional Conference on Soil Mechanics
and Geotechnical Engineering, Maputo, Mozambique. Session 5 – Retaining Structures, pp. 269–274.

329
330
AMÉLIORATION DES SOLS

GROUND IMPROVEMENT

331
332
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Evaluation of Index and Strength Characteristics of Clay Soil Stabilized with Lime as
Subgrade Material in Jonglei, South Sudan
Evaluation des indices et des caractéristiques de résistance de sol argileux stabilisé à la chaux
comme matériau de la couche de base des chaussées à Jonglei, Sud- Soudan
S. Jjuuko*1, P.B.Matiop2 and D. Kalumba1
1
University of Cape Town, Cape Town, South Africa
2
Ndejje University, Kampala, Uganda
*
smljju001@[Link]/sjjuuko1@[Link]

ABSTRACT–Clay soil can be stabilized by the addition of small percentages, by weight, of lime, thereby enhancing
many of the engineering properties of the soil and producing an improved construction material. In this paper, the
performance of clay soil, in terms of index and strength properties, stabilized with lime as subgrade material in Kongor,
Jonglei state, in South Sudan was evaluated. The index and strength properties of unstabilized clay soil were used as a
control for comparison purposes. Fixed percentages of lime (2%, 4%, 6% and 8%), by weight, were mixed with clay.
The Atterberg limits, linear shrinkage, grading, Optimum Moisture Content (OMC), Maximum Dry Density (MDD)
and California Bearing Ratio (CBR) of the treated samples were determined. All the mixes experienced an increase in
their MDD and a decrease in their Plasticity Index (PI), Linear Shrinkage (LS), and OMC, as well as enhanced CBR, on
addition of lime. Lime percentage of 6% was determined as the optimum for stabilisation of clay soil from Kongor,
Jonglei state in South Sudan for use as subgrade material.

RÉSUMÉ–Un sol argileux peut être stabilisé par l’ajout en petites proportions, en masse, de chaux, pour améliorer
plusieurs propriétés d’ingénierie du sol; et ainsi produire un matériau amélioré de construction. Dans cet article, la
performance d’un sol argileux, en termes d’indices et de propriétés de résistance, stabilisé à la chaux comme matériau
de couche et sous-couche de base des chaussées, était évaluée à Kongor, Etat de Jonglei, au Sud-Soudan. Les indices et
les propriétés de résistance de sol argileux non stabilisé à la chaux étaient utilisés comme contrôle pour comparaison.
Des pourcentages définis de chaux (2, 4, 6 et 8%), en masse, étaient mélangés avec de l’argile. Les limites d’Atterberg,
le retrait linéaire, l’analyse granulométrique, la teneur en eau optimale, la densité sèche maximale et l’indice portant
californien (CBR) des échantillons traités étaient déterminés. Tous les mélanges ont eu une augmentation de leur
densité sèche maximale et du CBR, puis une réduction de leur indice de plasticité, de retrait linéaire et de la teneur en
eau optimale, suite à l’ajout de chaux. Le pourcentage optimal de chaux comme matériau de stabilisation du sol argileux
de Kingor, Etat de Jonglei au Sud-Soudan pour emploi dans la couche de base des chaussées était compris entre 4 et
6%.

1 INTRODUCTION

Road transport accounts for at least 98% of the passenger traffic in South Sudan hence a major contributor to the
economy, (MHPPE, 2011). Yet, most of the road network is in a poor state as a result of the two decades of war. Figure
1 shows the road network in South Sudan. According to the South Sudan’s ministry of Roads and Bridges (2012), the
World Bank and USAID intervened by increasing the funding to the road sector. Nevertheless, the current expansion
and construction of the road network is increasingly on problematic soils. Specifically in Jonglei state, particularly in
Kongor, road construction is mainly on expansive clayey soils hence early deterioration of the roads pavement
structure. This is due to the low shear strength, high swelling capacity, low bearing capacity and high levels of
instability of the founding soil.
According to Jjuuko et al. (2011), the re-use of materials found on or near site is more sustainable and
environmentally beneficial than traditional methods of “dig and dump”. In the case of problematic soils, such as
expansive clayey soils, Jirathanathaworn et al. (2002) suggest that some techniques are needed to improve their
mechanical properties in order to proceed with construction. In this paper, the performance of clayey soil stabilized with
lime as subgrade material in Kongor area, Jonglei state in South Sudan was evaluated in relation to index properties and
mechanical strength.

333
2 MATERIALS AND METHODS

2.1 Clayey soil

Clayey soil samples were collected from Kongor at depths between 0.3 m and 1.0 m from ground surface. These depths
were chosen for easy sampling and, they represent the most likely used depths in obtaining soils for road construction in
Kongor. Shallow trial pits, dug by means of a pick axe and shovel, were utilized to obtain both disturbed and
undisturbed soil samples. The soil profile was studied in detail and bulk samples collected in tightly sealed black
polythene bags to prevent the loss of natural moisture content.

Figure 1. Road network in South Sudan (Google, 2014)

The clayey soil was generally light grey in colour. It had liquid limit and plastic limit of 49.3% and 18.6%
respectively. The plasticity index was calculated as 30.7%. From the plasticity chart shown in Figure 2, the soil was
classified as clay of intermediate plasticity (CI) according to the Unified Soil Classification System.
The soil had maximum dry density (MDD) of 1649 kg/m3 and Optimum Moisture Content (OMC) of 14.8% as
shown in Table 1. Though its OMC meets the requirement of 15%, its MDD does not meet the requirement of 2000
kg/m3 as recommended by Jjuuko et al. (2011) hence the need to be stabilized.
Table 2 shows the guidelines for selecting a suitable method of stabilisation. According to these guidelines, the
clayey sample used in the study was suitable for lime stabilisation since it had more than 25 % of its particle size
passing the 75 µm sieve size and plasticity index between 10 and 20.

Figure 2. Plasticity chart

334
Table 1. Physical and mechanical properties of different blends
Lime LL PL PI LS MDD OMC CBR after 4-days soak
percentage (%) (%) (%) (%) (kg/m3) (%) (100 % MDD)
0 49.3 18.6 30.7 12.9 1731 12.8 9
2 50.4 26.3 24.1 12.1 1681 14.8 11
4 51.9 37.3 14.6 7.1 1649 16.3 15
6 55.7 39.5 16.2 9.3 1566 23.8 19
8 40.4 22.2 18.2 10.0 1478 26.8 18

Table 2. Guide to selecting a method of stabilisation (Austroads, 1998)

MORE THAN 25 % PASSING 75 µm LESS THAN 25 % PASSING 75 µm
Form of Stabilisation PI<10 10<PI<20 PI>20 PI<6 * PI * %Passing 75µm PI<10 PI>10
Cement and
Cementitious Blends
Lime
Bitumen
Bitumen/Cement Blends
Granular
Miscellaneous
Chemicals1
Usually
Usually
Key Doubtful not
suitable
suitable
1
Should be taken as a broad guideline only. Refer to trade literature for further information
Note: The above forms of stabilisation may be used in combination, e.g. lime stabilisation to dry out material and reduce plasticity,
making them suitable for other methods of stabilisation.
PI = Plasticity Index

2.2 Sample preparation

Sample preparation and all the laboratory investigations on the untreated and treated soil specimens were performed
according to the relevant American Standard of Testing and Measurement (ASTM) methods of testing as specified in
the Standard Technical Specification (2006) for Roads and Bridge Works for Ministry of Roads and Transport,
Government of Southern Sudan.
The bulk soil samples were air-dried for laboratory testing. In order to investigate the effect of lime stabilizer on the
properties of clayey soil, specimens with specified amounts of lime added to the soil samples were prepared by mixing
lime with soil in quantities of 0 %, 2 %, 4 %, 6 % and 8 % of weight. The percentages were chosen in relation to the
recommended 6 % percentage of lime by weight for road base stabilisation according to the standard technical
specifications for roads and bridges in South Sudan of 2006. The mixing was done mechanically on a metal tray. For
consistency, soil was mechanically blended before mixing. Tests of physical properties of the different soil/lime blends
were conducted. The tests included: Atterberg limits, Linear shrinkage, Compaction and California Bearing Ratio
(CBR) test.

2.3 Tests conducted

2.3.1 Atterberg limits test

The cone penetrometer method was used in the study to determine the liquid limit of the clayey soil/lime blends
according to ASTM D4318-05. As the moisture content of the soil sample was increased incrementally, the penetration
of the cone was noted and plotted against the respective moisture content.
From the same soil sample, a specimen was dried to near its plastic limit by air drying. It was then moulded into a
ball and rolled between the palms of the hand and glass plate to threads of nearly 3 mm in diameter. The soil was now
considered to be at the plastic limit and its moisture content determined.

2.3.2 Compaction test

The compaction test was performed in accordance with ASTM D1557 for modified proctor. The soil specimen was
mixed with suitable amount of water and compacted in five layers. Each layer was compacted with 4.5 kg rammer from
a dropping height of 450 mm. MDD and OMC was determined from the graph of dry density against moisture content.

2.3.3 CBR test

The test was carried out for all design mixes in accordance with ASTM D1883. Fresh sets of 7 kg air-dried soil were
mixed with suitable amount of water to their OMC. Each layer was compacted with 65 blows using a 4.5 kg hammer at
335
a drop of 450 mm. The compacted soil and mould were weighed and then soaked in water for four days. After the four
days of soaking, the samples were placed under the CBR machine following standard procedures. Load applied was
recorded at varying penetrations to give a stress-strain curve from which the CBR was computed.

3 RESULTS AND DISCUSSION

3.1 Atterberg limits

The cone penetrations for the different blends were plotted against the corresponding moisture contents. The resulting
liquid limits were obtained. The liquid limits, plastic limits, plastic indices and linear shrinkages of all the blends were
plotted against the concentrations of lime as shown in Figure 3.
The liquid limits and plastic limits generally increased with increased lime content up to 6% and then reduced. The
increased addition of lime, a cementitious product, increases the binding ability of the mixtures and its capacity to retain
moisture. Blends having liquid limit greater than 50% are indicative of high plasticity soils not suitable for road
construction, (Jjuuko et al., 2011).
The plasticity index generally reduced with increased lime content. After initial mixing, the calcium ions (Ca++)
from hydrated lime migrate to the surface of the clay particles and displace water and other ions. The soil becomes
friable and granular, making it easier to work and compact. This reduces the plasticity index of the soil and its tendency
to shrink and swell dramatically as shown in Figure 4. The process is called flocculation and agglomeration (Negi et al.,
2013).
Liquid limit Plastic limit Plasticity index Linear shrinkage

50
Moisture Content (%)

0
0 2 4 6 8
Lime Content (% Dry Mass)

Figure 3. Liquid limit versus lime content

3.2 Linear shrinkage

Linear shrinkage of the blends reduced from 12.9 % of the unstabilized sample to 7.1 % at 4 % lime addition and then
rose to 10.0 % at 8 % lime addition. Generally the soil linear shrinkage, swell potential and swell pressure are
significantly reduced by lime treatment. This is attributed to a decreased affinity for water of the calcium-saturated clay,
a reduction in the number of interconnected voids and the formation of a cementitious matrix that resists volumetric
shrinking and expansion, (Technical Note 39, 2000).

3.3 MDD and OMC

The dry densities for the different blends were plotted against the corresponding moisture contents. The resulting
MDDs and OMCs were obtained and plotted against lime concentration. As seen from Figure 4 and Figure 5, the MDD
and OMC of the resulting blends decreased and increased respectively with higher lime content.
The reaction between lime and soil is exothermic. The heat generated evaporates moisture in the soil hence
additional water needed to achieve maximum dry density. In addition, Figure 6 shows that the addition of lime to soil
increases its solid range and reduces its plastic range, hence more water needed for the soil to reach its plastic range.
Also finer lighter lime grains increasingly replace soil particles in a given volume. These smaller lime grains have a
larger surface area accounting for increased OMC.
The reduction in MDD of the clayey soil sample with increased lime content is an indication of improvement of the
compaction characteristics of the soil-lime blends. This is because the agglomerated and flocculated particles of soil
occupy larger spaces.

336
1800

MDD (kg/m3)
1700

1600

1500

1400
0 2 4 6 8
Lime Content (% Dry Mass)

Figure [Link] versus lime content

30
26
OMC (%)

22
18
14
10
0 2 4 6 8
Lime Content (% Dry Mass)

Figure [Link] versus lime content

Before Treatment After Treatment

(PI) (PI)’

Solid range Plastic range Liquid range Solid range Plastic range Liquid range

Figure 6. Effect of liming on the consistency of soil (Negi et al., 2013)

3.4 CBR

The CBR test results of samples blended with lime are shown in Figure 7. The 4-day soaked CBR generally increased
with increasing lime content.
The pH of the soil quickly increases to around 12 when lime is added. This enables the clay particles to breakdown.
In the process, silica and alumina are released and react with calcium from the lime to form calcium-silicate-hydrates
and calcium-aluminate-hydrates, (Negi et al., 2013). These products are essentially the same as the cemented products
responsible for strength and durability of conventional concrete, (Technical Note 39, 2000). They form a matrix that
contributes to the strength of lime-stabilized soil layers, transforming the soil into a hard, relatively impermeable layer
with significant load bearing capacity. The reduction in CBR at 10% might be due to the excess lime in the clay not
required for the early strength gain as a result of flocculation and agglomeration.

337
20
18
16

CBR (%)
14
12
10
8
0 2 4 6 8
Lime Content (% Dry Mass)

Figure 7. CBR versus lime content

4 CONCLUSION

The study indicated that the clay soil was not suitable for subgrade material in its natural state as its MDD and CBR of
1731 kg/m3 and 9 % respectively were less than MDD and CBR of 2000 kg/m3 and 30% respectively recommended for
subgrade material in South Sudan roads and bridges technical standard specifications. Hence the need to be stabilized. It
was found in this study, that the properties of clay can be improved by adding lime.
The addition of lime to clay reduced its plasticity index, subsequently, the workability of the clay soil was improved.
In addition, the CBR of clay increased with increasing lime content up to 6 %, whereas reduction was observed at 8 %.
The reduction at 8 % was attributed to the excessive lime in the blend not essential for cation exchange. Therefore, the
lime content of 6 % was best suited for stabilisation of clay in Kongor.
Though lime stabilisation improved the clay properties, the MDD and CBR of 6 % lime content clay sample did not
meet the specified MDD and CBR. The results from a study by Jjuuko et al. (2011) investigating the effect of sand
stabilisation on the material properties of clay obtained from Busega in Kampala district, showed that the stabilized clay
met the specifications of MDD and CBR. However, the stabilized clay lost its cohesiveness (plasticity), consequently,
its workability. Therefore, from this study and study by Jjuuko et al. (2011), it was concluded that a combined lime and
sand stabilisation was best suitable for clay improvement in Kongor, Jonglei sate, South Sudan.

REFERENCES

ASTM Standard D422-63 (1998) Standard Test Method for Particle-Size Analysis of [Link] International, West
Conshohocken, PA, USA.
ASTM D1557 - 12 Standard Test Methods for Laboratory Compaction Characteristics of Soil Using Modified [Link]
International, West Conshohocken, PA, USA.
ASTM D1883-05 Standard Test Method for CBR (California Bearing Ratio) of Laboratory-Compacted [Link] International,
West Conshohocken, PA, USA.
ASTM Standard D4318-05 Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils. ASTM International,
West Conshohocken, PA, USA.
Austroads (1998), Guide to Stabilisation in Roadworks.
Google (2014) Road network in South Sudan.
[Link]
D6F159E6. Accessed on 1st, October, 201
Jirathanathaworn T., Nontanandh S., Chantawarangul K. (2003) Stabilization of clayey sand using fly ash mixed with small amounts
of lime. In Proc. of the 9th National Conventional on Civil Engineering, Engineering Institute of Thailand and Thammasat
University, Petchaburi, GTE 93-98.
Jjuuko S., Kalumba D., Bagampadde U. (2011) Use of locally available sand in stabilization of clayey Ugandan [Link] of
the 16th National Technology Conference “A Developing Uganda, the Engineers’ Test”, Kampala, Uganda, May 25-27, 138-
151.
MHPPE (2011) Ministry of Housing, Physical Planning and Environment, South [Link] communication.
Negi A.S., Faizan M., Siddharth P.D., Singh. R. (2013) Soil stabilization using [Link] Journal of Innovative Research in
Science, Engineering and Technology. Vol. 2, Issue 2, 448-453.
Standard Technical Specifications (2006) Roads and Bridge Works, Ministry of Roads and Transport, Government of Southern
Sudan.
[Link]
%20Standard%20Technical%[Link]?AttachmentID=7e1524a8-176c-4933-8e8a-8bfc7b8b93a9. Accessed on 1st,
October, 2014.
Technical Note 39 (2000) Lime Treatment of Clay Subgrades.
[Link] Accessed on 1st,
October, 2014.

338
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Reinforcement of Base Course Consist of Black Cotton Soil using Do-nou

Technology
Renforcement de la Formation de Base Consiste de Noir Coton Sol Grâce à la Do-noh technologie

Y. Daito*1, Y.Fukubayashi2, M. Kimura3

1
Kyoto University, Graduate School of Engineering, Department of Civil and Earth Resources Engineering, Kyoto,
Japan
2
Community Road Empowerment (NGO), Kyoto, Japan
3
Kyoto University, Graduate School of Engineering, Department of Civil and Earth Resources Engineering, Kyoto,
Japan

ABSTRACT –Black cotton soils havea huge water-absorption capacity whichleads to many problems in infrastructures,
especially roads built on black cotton [Link] countermeasures, such as soil replacement, stabilization by
chemicalmaterials, or reinforcement bygeotextile, are required to repair the roads and protect them from water
absorption. Almost all the countries in Eastern Africa have difficulty in promoting road construction when above
mentioned measures due to the financial constraint in local areas. The main subject of this paper, Do-nou, a Japanese
term for soil bags, has been applied to improve rural roads with locally available material and labor intensive method.
However, black cotton soilsare normally difficult to apply toDo-nou technology. Therefore, this paper aims to find an
adaptation method through small-size Do-nou compression tests, and a method that laps 2 sacs into 1 Do-nou is foundas
the most effective mode of adaptation. The effect of road improvements using the selected method is verified through
real-scale driving tests.

RÉSUMÉ –Les Vertisols ont une capacité d’absorption d’eau énorme qui conduit à de nombreux problèmes dans les
infrastructures, en particulier les routes construites sur ces types de sols. Les contre-mesures sont coûteuses, comme le
remplacement du sol, la stabilisation par des matériaux chimiques, ou le renforcement par géotextile sont autant de
mesures recourues pour réparer les routes et les protéger de l’absorption d’eau. Presque tous les pays d’Afrique de l’Est
ont des difficultés à promouvoir la construction de la route dans les zones rurales en raison de la contrainte financière
occasionnée par les mesures palliatives ci-dessus mentionnées. Le sujet principal de cet article, « Do-nou », un terme
japonais pour désigner « les sacs de terre » a été appliqué pour améliorer les routes rurales avec des matériaux
disponibles localement et par la méthode de travail intensif. Cependant, il est difficile d’appliquer la technologie Do-
nou normalement sur les Vertisols. Par conséquent, ce document vise à trouver une méthode d’adaptation par la petite
taille, des tests de compression des Do-nou ou une méthode qui consiste à associer 2 sacs en un Do-nou se trouve être
le mode d’adaptation le plus efficace.

1 INTRODUCTION

Black cotton soils are the major problem soils in Eastern Africa because of their swelling and shrinking properties.
These soils absorb water and expand as they get wet. On the other hand, they shrink and leave large voids as they dry,
so that unpaved roadson these soils become muddy and impassable in the rainy season (Mohan and Bhandari, 1978;
Morin, 1971). Normally, soil replacement and chemical stabilization or physical reinforcement such as geotextileare
adopted for roads constructed with these soils. However, these methods are only employedfor highways or main roads
near urbanized areas due to the high cost of implementing them. Actually, local villages built on these soils suffer from
loss of the trafficability of roads in the rainy season, and villagers have to wait indefinitely for repair work funded by
government. Therefore, a road rehabilitation method,which can be carried out by the local citizens themselves, is
required in order to change the situation.
The community participatory road rehabilitation method using sacs, which is called“Do-nou technology”, is being
developedfor the local villages of several developing countries. The possibility of road repair only through the efforts of
local villagershas beenrecognized (Kimura et al., 2011). On the other hand, Do-nou technology ordinarily uses non-
expansive soil, such as laterites for the inner soil, but the use of clay materials,such as black cotton soils, is unsuitable.
Hence, this paper aims to find an adaptation method for expansive clay into the Do-nou technology. At first, 1/5-
scaleDo-nou compression tests are carried out to find an adaptation method for the laboratory. Secondly, driving tests,
on real-scale trial roads of black cotton soils, are performed, which are based on the method from the first experiment,
to verify the effect of road improvementson a real level.

339
2 1/5 SCALE DO-NOU COMPRESSION TEST
2.1 Test methods and cases
Three methods are suggested to enable the adaptation of expansive clays into Do-nou technology, namely, Method 1:
reduce the air voids in theDo-nou inner soil material, Method 2: mix mountain sand into the expansive clay, and
Method 3: lap 2 bags together to form 1 Do-nou. These methods were verified by compression tests in which 1/5-scale
Do-nou test pieces were used after curing in [Link] the Ca type of bentonite is used as the expansive clay for the
inner soil, and Table 1 shows the difference in properties between the Ca type of bentonite and black cotton soils.
After filling the 128 cm3 ofdry conditioninner materials which volume is determinedfrom their dry density, the test
pieces are compacted up to keep the dimension which length, width and height are 80 mm, 80 mm and 20 mm and then,
the test pieces are cured in water so the pieces will absorb water from all surfaces for a maximum period of 7 days to
simulatethe conditions during the rainy season.
The compression tests were conducted for no-water curing test pieces as well as for 1-day, 2-day, 3-day, 5-day, and
7-day water curing test pieces. The loading was at the speed of1 mm/min displacement and was continuedto a
maximum of 5 kN. The loading was performed for 1 Do-nou and, during that time, all surfaceswere being pressed
(Figure 1). After the relation between the load and the displacement was obtained from the tests, the difference in the
cases was compared for each displacement of 1.63 kN, which is the wheel load of the 2-ton truck in the 1/5 scale. At the
same time, the water absorption ratio WAR (%) wasobtained.
Three patterns were broadly prepared based on the processes applied to make the Do-nou, which are shown in Table
2. Method 1 uses powdered bentonite in a dry condition, as the inner soil material, in order to reduce the air voids. On
the other hand, aggregated bentonite is used for comparison and its particle size is adjusted to lie in the range of 2.00
mm to 4.75 mm after drying the water-mixed bentonite, with a water content of 150%, in the drying oven. Method 2
mixes in mountain sand to create a dry condition of the bentonite, as with Method 1, whose particle size is shown in
Figure 2 and whose dry density d is 1.64 g/cm3. The 10%, 20%, 30%, 50%, and 70% levels of mountain sand in the
volume ratio are mixed into the bentonite. Method 3 uses bentonite only in the dry condition, but the number of sacs is
doubled. In Method 3, two patterns of lapping methods are proposed, namely, Pattern A that makes the Do-nou the
same as the methods in the other cases, that lap 2 sacs to make one, and Pattern B that laps the second sac in the
opposite direction after making the Do-nou in the usual manner. As mentioned above, the test pieces were used in
compression tests after being prepared.
݉ୟ− ݉ୠ (1)
WAR = × 100
݉ୡ

Where ma, mb and mc are the mass of test piece after water curing, mass of test piece before water curing, and the mass
of inner soil failed into the test piece.

2.2 Results and discussion

Firstly, Figure 3 shows the relation between the load and the displacement in the Method 1 cases, which are the cases of
aggregated and powdered bentonite under the day 0 and day 1 water curing conditions, only to compare the clear
differences, but compression tests were carried out for all the test pieces. The strength of the Do-nou using the
aggregated material (Case_A) decreased after only 1 day of water curing, while no clear difference was found in the
process of the 1 day of water curing in the case of the powdered material (Case_P). The increase of the WAR by water
curing days for both cases is shown in Figure 4. The case using the powdered material (Case_P) converges with the
water absorption ratio at 55%. On the other hand, the case using the aggregated material (Case_A) shows a ratio of over
70% after only 1 day of water curing. This difference is caused by the difference in air void ratios, for which the
powdered bentonite is na = 62.6% and the aggregated bentonite is na = 50.0%, so that the permeability in Case_A
becomes smaller than that in Case_P. Therefore, the use of a reduced air void material makes it possible to bring about
a decline in the absorption speed.

340
Table 1. Soils-expansive properties
Free
Material Type w p [%] w L [%]
Swell [%]
Black Cotton
Ca 29.29 79.95 190
from Kenya
Bentonaite
Ca 34.80 190.51 599
from Japan

Figure [Link] of compression method

Figure 2. Particle size distribution for mountain sand

Table 2 Outline of test cases

Secondly, all the results from Method 2 are shown in Figure 5, which presents the displacement of 1.63 kN loading
for each day of water curing, but Case_S0 is the same as Case_P, as seen in Figure.3. The test pieces mixed with 20%
sand are destroyed at day 5 of water curing until a load of 1.63 kN. The test pieces cannot withstand the load of 1.63 kN
until they are mixed with 50% mountain sand at day 7 of the water curing. Mixing the pieces with 70% mountain sand
reduces the displacement. The progress for the WAR in both cases is shown in Figure 6. The WAR at the same days for
water curing decreases with the increase in the sand mixing ratio and it will converge in all cases. Therefore, WAR can
be reduced by mixing in sand, but it is necessary to mix in at least 50% mountain sand to attain the effect of
reinforcement.
Finally, the displacement of a 1.63 kN load in Method 3 is shown in Figure 7, which contains the results of Case_S0
and Case_S7 for comparison. Both cases use a double sac for 1 Do-nou which can withstand the load of 1.63 kN at all
water curing days; they also show smaller displacement than the case with 70% mountain sand. In comparing the
lapping patterns, Pattern B makes stronger Do-nou than Pattern A. Both water absorption ratios show the convergence
at less than 50% in the process of time (Figure 8).
The doubling sacs increase the tensile strength of the wrapping material so that the expansive pressure of the inner
material will be counterbalanced. Thus the expansion is constrained leading the convergence of water absorption ratio
at lower level. Then, the inner material keep the plastic condition without absorbing water until it becomes liquid even
under water curing condition. The using double sacs shows the best effect to bear the traffic load rather than the case
that the aggregated clay is filled and the case that the clay mixed with sand at the ratio 70 % in volume is filled into Do-
nou.
As mentioned above, the method using 2 sacs for 1 Do-nou with Pattern B is selected as the expansive clay
adaptation method for the Do-nou technology.

341
5 80
Case_P_0day Case_A Liquid Limit 190.51 [%]
70
Case_A_0day

Water Absorption Ratio [%]

4 Case_P_1day
60 Case_P (Case_S0)

50
Load [kN]

3
40
Plastic Limit 34.80 [%]
2 t Truck Wheel Load
2 30
1.63 kN
20
1
Case_A_1day 10
0
0 0 1 2 3 4 5 6 7 8
0 24 6 8 10 Day
Displacement [mm]
Figure [Link] between load and displacement for Case_A and _P Figure [Link] ofwater absorption ratio for Case_A and _P
80

70
Case_S0

Water Absorption Ratio [%]

60 Case_S1
50 Case_S2
40 Case_S3
Case_S5
30 Case_S7
20

10
0
0 1 2 3 4 5 6 7 8
Water Curing
DayDay [day]

Figure [Link] of 1.63 kN for all cases of Method 2 Figure [Link] of water absorption for all cases of Method 2

10 80
Liquid Limit 190.51 [%]
Only S0 was broken by
tearing sac until 1.63 kN 70
Water Absorption Ratio [%]

8 [ from day5 ]
60

50 Case_Da
6
day 7 40 Case_Db
4
day 5
30 Plastic Limit 34.80 [%]
day 3
2 20
day 1
day 0 10
0
Case Case Case Case
_S0 _S7 _Da _ Db 0
0 1 2 3 4 5 6 7 8
Day
Figure [Link] of 1.63 kN for Case_S0, _S7, _Da and _Db
Figure [Link] of water absorption ratio for all cases of Method 3

3 DRIVING TEST

3.1 Test methods

A series of driving tests were conducted in order to clarify the effect of the improvement of the trafficability with the
identified method using two sacs through the 1/5 scale Do-noucompression tests. The tests were conducted on the
ground originally consist of black cotton soils in Kenya.
In this series of the driving tests, the 4 patterns of base courses were constructed as shown in Table 4. The site for
each pattern of driving test was 3 m in length and width, respectively. The base courses of Road A, B and C consisted
of two layers of sacs with a height 0.2 m and surface layer of murrum whose thickness is 0.05 m. The surface layer was
to protect Do-nou from being torn through the friction caused with driving truck and exposure to sun light. Murrum is
the local term for laterite and generally utilized as qualified material for road base course in Eastern Africa. The
specification of the Do-nou bags is 0.4 m in length and width and 0.1 m height after filling the 20 L of inner soil

342
materials and the compaction and it is made by polyester. The properties of murrum are shown in Table 3 and Figure 9
together with those of black cotton soils.

Table [Link] properties Table [Link] of the driving test

Black Cotton
Soil Material Murrum
Soil The Number
3 Case Inner Soil Surface Soil
r s [g/cm ] 2.68 2.53 of Sacs
w p [%] 29.29 38.43 Road A Murrum Murrum 1
w L [%] 79.95 23.16 Road B Black Cotton Soil Murrum 2
IP [%] 50.66 15.29 Road C Black Cotton Soil Murrum 1
Road D None None None
Free Swell [%] 190 20
3
r dmax [g/cm ] 1.25 1.71
w opt [%] 35.0 17.5
modified CBR [%] 1.6 12.6

0.005 0.075 2
100

80
Passing Ratio [%]

20
Black Cotton Soil
Laterite
0
0.001 0.01 0.1 1 10 100
Particle Size [mm]
Figure 9. Particle size distribution curve for materials Figure [Link] view of the driving test

In the Road A, murrum was filled into Do-nou. Only one sac for each Do-nou was wrapping the inner material. This
was the conventional method to apply Do-nou technology to reinforce the base course. Black cotton soils which particle
size wasadjusted to be less than 50 mm in dry condition were put inside Do-nou in the case of Road B, and two sacs are
used for wrapping the inner materials. Black cotton soils were utilized as inner material of Do-nou in the Road C, but
the number of sacs is only one. Through the comparison of the result between Road B and Road C, the effectiveness of
the doubling sacs on the improvement of the trafficability could be found. No countermeasures were applied for Road
D, which was only cleared.
The amount of water equivalent to the maximum monthly rainfall in Nairobi Kenya at 2012, which was 208 mm,
was sprinkled for 1 hours on each road site to recreate the conditions during the rainy season. After curing the soils for
24 hours, the 2-ton truck drove at a speed of less than 20 km/h until the number of driving times on the trial roads was
to be 300, which truck is 2,970 kg in weight and 2-axis and 2-reat wheel type.
The relation between the road displacement and the driving times on the road of the 2-ton truck was observed. The
road displacement was measured when the number if driving reached to 0, 10, 20, 40, 60, 100, 200 and 300 times. The
displacement was measured at the mid-section of the road in driving direction and at 300 mm forward and backward the
mid-section and then, the displacement data was represented by the average of the measured displacements at all the
three sections.
If it was impossible to continue, the driving was stopped. Moreover, two critical conditions to be passable by
vehicle were defined, namely, 1: When the average displacement where the tires pass reached to 150 mm, which was
the minimum ground clearance for normal vehicles, and 2: When the truck was stuck because of slippery surface after
the it was got wet and disturbed due to the driving cyclic load.

3.2 Results and discussion

Figures 11, 12, and 13 show the road surface profiles at the several times of drivingon Roads A, B, and C, respectively.
Road C ended up with displacement of more than 150 mm at the 60times of driving, which met to one of the critical
conditions for driving. However, driving on Roads A and B was able to be continuedeven after 300 times without
reaching to the critical conditions. On the other hand, it was impossible to continue the driving on Road D after only 7
times because of the slippery surface, while no displacement occurred. This was the critical condition defined as the

343
second. Figure 14 shows the relation between the number of driving times and the road displacement for each road
which was represented by the averaged deepest displacement at the right and left rut underneath the tires during driving.
The rapid progress to make the road displacement was recorded for Road C. On the other hand, the remarkable
displacements of Road A and B converge into about 110 mm.

15 15

10 10

Displacement [cm]
Displacement [cm]

5 5
Ground Level Ground Level
0 0
-5 -5
-10 -10
-15 -15
Minimum Road Clearance 15 [cm] Minimum Road Clearance 15 [cm]
-20 -20
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Distance [cm] Distance [cm]

Figure 11Road displacement on Road A Figure 12Road displacement on Road B

5
Road A
15 0 Road B
Road C
Displacement [cm]
10
Displacement [cm]

5 -5
Ground Level
0
-10
-5
-10 Minimum Road Clearance 15 [cm]
-15
-15 Minimum Road Clearance 15 [cm]

-20 -20
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Distance [cm] Distance [cm]

Figure 13Road displacement on Road C Figure 14Progress of road displacement in each cases

From the test results for Road D, if black cotton soils are utilized or exist as the road surface materials, they will make
the roads impassable in the rainy season, regardless of countermeasures performed, unless the soil materials themselves
are replaced. Therefore, as far as the road surface materials are not black cotton soils, like for Road C, the road can be
protected from worsening to be critical condition for driving. However, it was found that even Road C didn’t keep
sufficient trafficability during the driving test. It was not effective to adopt black cotton soils directly into the Do-nou
technology. Road B, which base course consist of Do-nou filled black cotton souk wrapped with two sacs the same road
improvement effect on trafficability as Road A, which use only murrum.

4 CONCLUSIONS
The method to utilize the black cotton soil as the material for Do-nou technology to improve rural roads was proposed
through 1/5 scale Do-nou compression test and driving test. It was found that two sacs wrapping black cotton soil
enabled to balance with the expansive pressure of inner material, hence made it possible to confine the water absorption
of the inner material of Do-nou. As a result, the Do-nou filled with black cotton soil wrapped with two sacs owned the
larger bearing capacity than the Do-nou filled with the mixture of the black cotton soil and mountain sand at the ratio
70% in volume wrapped with single sac.
It was confirmed that the improvement effectiveness on trafficability of the base course consist of Do-nou filled
with black cotton soil with the proposed method obtained the equivalent effectiveness to the one with Do-nou filled
with murrum.
The applicability, unit cost and productivity of the proposed method will be examined for dissemination as a cheap
and practical method to improve rural roads on black cotton soil in East Africa.

REFERENCE
D. Mohan and R. K. Bhandari (1978) Black Cotton Soil of India. The Japanese Geotechnical Society, 47.
W. J. Morin (1971) Properties of African Tropical Black Clay [Link], vol.1, 51.
Kimura, M. and Fukubayashi, Y. (2011) Applications of Participatory Road Maintenance using “Do-nou” Technology in Kenya,
Proc. 15th African Reg. Conf. on Soil Mech. and Geotech. 476-481.

344
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Analyse statique par éléments finis des fondations superficiellessur sol renforcé par
inclusions rigides
Static analysis by finite elements of the shallow foundations on ground reinforced by rigid
inclusions
DJERAFNora1, NECHNECHAmmar1
1
LEEGO, Faculté de Génie Civil, Université des Sciences et de la Technologie Houari Boumediene, Algérie

RESUME : La technique de renforcement des sols par inclusions rigides qui se rapproche théoriquement de la fondation
mixte puisqu’elle associe des éléments verticaux rigides et une semelle ou radier, se distingue de celle-ci par la présence
d’une couche intercalaire entre les éléments appelée « matelas de transfert» qui a pour rôle d’assurer une répartition des
charges apportées entre le sol lui-même et les [Link] système ainsi composé comporte différents modes
d’interactions entre les inclusions, le matelas de transfert de charge et le [Link] présent travail tente de mieux
comprendre le mécanisme de fonctionnement d’un tel système quand il est soumis à un chargement statique. Une
analyse globale est faite, en se basant sur un modèle tridimensionnel en éléments finis moyennant l’utilisation ducode
Plaxis 3D Foundations, L’accent est mis aussi sur l’influence des principaux paramètres sur le comportement du
système en termes de déformations et du report de charge.

ABSTRACT:The soil reinforcement technique by rigid inclusions associates a vertical rigid elements and a sole or a
slab, by introducing a spacer layer in between, the role of this layer called « Transfer mattress» is to ensure the charge
distribution between the soil and the [Link] so made system has different interaction modes, between the
inclusions, the transfer mattress and the [Link] worktries to better understand the functioning mechanism of such
system while it is subject to static load. An overall analysis of the system is made using plaxis 3D Foundations. Also in
this thesis work, a stress is put on the influence of the main parameters on system behavior in terms of deformation and
charge transmissions.

1INTRODUCTION

Les sols à faible portance, posent de réels problèmes pour la construction des ouvrages de génie civil (bâtiments
industriel, ouvrages routiers ou ferroviaires…) compte tenu des tassements importants qui peuvent survenir pendant et
après la [Link] nos jours, la raréfaction des terrains présentant des caractéristiques géotechniques satisfaisantes
ont contraint le développement de nouvelles techniques pour fonder les ouvrages sur ces formations géologiques
constituées le plus souvent de dépôts récents caractérisés par une forte compressibilité et/ou faible portance. On compte
parmi ces techniques le renforcement de sol par inclusions rigides, qui constitue une alternative des plus intéressantes
aux fondations sur sol compressible de type traditionnel.
Si l’utilisation de cette technique ne fait que croître, et si l’élaboration et la mise en œuvre sont assez bien maîtrisées
du point de vue technologique, il n’en est pas de même pour ce qui est des méthodes de calcul et de dimensionnement
qui se diversifient selon les cas et restent encore du domaine de la recherche. Ainsi La méthode analytique de
(Combarieu, 1990) qui considère que la force portante et le tassement pour une semelle rigide de faibles dimensions
constitue un cas intermédiaire entre celui d’une fondation superficielle sur sol stratifié et d’une fondation mixte, se base
sur un fondement tout à fait différent de celui de (Frossard et Glandy, 2003). Ces auteurs proposent une approche où le
système est décomposé en deux domaines, la force portante admissible de l’ensemble est supposée alors égale à la
somme des contributions séparées de la semelle et desinclusions.
Les études bibliographiques réalisées par plusieurs auteurs (Jenck, 2005, Le Hello, 2007et Antoine, 2010) et(les
recommandations ASIRI, 2012), notent un manque considérable des méthodes spécifiques dédiées à l’analyse complète
du comportement de fondations superficielles sur sols renforcés par inclusions [Link] majorité des études traitantde
ce sujet ont eu une approche locale du problème en se penchant sur l’analyse d’une partie du système.
Par une modélisation numérique en éléments finis, le travail présenté dans ce papier a pour objectif d’évaluer
l’apport du renforcement du sol par inclusions rigides sur le comportement des fondations par rapport aux fondations
superficielles classiques. Une analyse comparative du point de vu déformations et efforts a été ainsi faite. L’étude
paramétrique réalisée dans ce cas d’étude a permis aussi d’apprécier l’influence de quelques paramètres clés sur le
comportement du système vis-à-vis aux tassements et au report de charge.

345
2PRINCIPE DU RENFORCEMENT PAR INCLUSION RIGIDES

La technique des inclusions rigides verticales consiste à transférer la charge due à l'ouvrage vers une couche de sol plus
dure. Ce transfert se fait par la combinaison d'inclusions rigides traversant le sol compressible et assurent le
renforcement, et une plate-forme de transfert de charge disposée entre le réseau d'inclusions et l'ouvrage et qui a pour
fonction la réduction et l’homogénéisation des tassements sous l’ouvrage en assurant la répartition des charges entre les
inclusions et le sol compressible (Figure 1).
Fondation
Plateforme de
transfert de charge Géosynthétique éventuellement
Tête d'inclusion

Sol compressible
Inclusions rigides

Couche rigide

[Link] d’un système de renforcement par inclusions rigides

Contrairement aux systèmes classiques de fondation mis en œuvre en présence de sol compressibles (radiers sur
pieux par exemple) qui permettent la transmission à un substratum de meilleure qualité (sous le sol compressible) de
l’intégralité du chargement due à l’ouvrage à fonder, l’effet recherché dans le cas du renforcement de sols par
inclusions rigides est la transmission de la charge partiellement aux têtes d’inclusions, partiellement au sol en place.

3PRESENTATION DU MODELE D’ETUDE

Le cas d’étude porte sur l’analyse du comportement statique d’une semelle filante soumise à un chargement
uniformément réparti reposant sur un sol renforcé par un groupe d’inclusions rigides. Dans un premier temps une
configuration de base (référence) caractérisée par une fondation superficielle sur un sol sans renforcement est étudiée.
Par la suite une analyse du comportement de la fondation est faite après renforcement du sol.
Le site d’étude est constitué principalement de trois couches dont une formation compressible, avec présence d’une
nappe phréatique à une profondeur de 12.9m. Lesépaisseurs ; la nature et les caractéristiques mécaniques des différentes
couchessont donnéesdans le tableau 1.
La semelle filante rigide est en béton armé de dimensions (2x12x0.5) m. Le renforcement du sol est assuré par un
groupe de 24 inclusions de type pieux en béton d’un diamètre de 0.3m recouvrant une surface de 1.7m², soit un taux de
recouvrement de 4%, les inclusions traversent la couche compressible et vient se reposer sur la couche rigide. Pour les
deux éléments (fondation et les inclusions rigides) on suppose un comportement élastique linéaire avec les
caractéristiques données dans le tableau 2.
D’une hauteur de 0.5 m, le matelas de transfert de charge qui déborde de 0.6 m de part et d’autre de la semelle est
constitué d’un matériau granulaire (grave non traitée), ayant les caractéristiques données dans le tableau 2. Une loi
élastique linéaire avec critère de plasticité de Mohr-Coulomb est adoptée pour ce matériau.

Table 1. Caractéristiques du sol in situ

Couche Limon marneux Sable grossier

Argile limoneuse
Paramètres état vaseux graveleux-caillouteux
Profondeur de la couche (m) 0-3 3-9 9 - 18
Poids volumique h (kN/m3) 19 17 20
Cohésion C (kN/m²) 40 30 0
Angle de frottement  (°) 16 6 32
Angle de dilatance  (°) 0 0 2
Coefficient de Poisson  0.3 0.3 0.3
Module de déformation (kN/m²) 15600 5200 84000

Table 2. Caractéristiques des différents éléments du système

Eléments (kN/m3)(K E (MPa)  (°) C (kPa)
3
Plate-forme de transfert N/m
22 ) 150 0.3 35 0
Fondation 25 25000 0.2 - -
Inclusions 24 20000 0.2 - -

346
4. RESULTATS ET ANALYSES

4.1Analyse des déformations

Les courbe tassement-chargement

chargement illustrées sur la figure 2 montrent le résultat évident de la mise en place des
inclusions comme éléments de renforcement de sol, une réduction appréciable des tassements atteignant un facteur
moyen de 4. De même une réduction des déformations horizontales très significative a été observée, ave
avec un taux
moyen de 70 % pour les déplacements suivant l’axe z et atteint 78% suivant l’axe x.

Chargement (KN/m²) Chargement (KN/m²)

Déformation horizontale
0.00
0.00
Tassement (m)

0.04 0.01

(m)
0.08 0.02
0.12 0.03
0.16
0.04 Ux-solrenforcé
solrenforcé
0.20 sans inclusions
0.05 Uz-sol
sol renforcé
0.24 avec inclusion Ux-sol
sol non renforcé
0.06 Uz-sol
sol non renforcé

Figure 2. Courbe tassement-chargement

tassement chargement pour le cas du sol renforcé et non renforcé

L’un des résultats le plus intéressant de cette analyse concerne la répartition des déformations le long des fûts des
inclusions. Les courbes de la figure3, pour un même chargement, montrent que les tassements induits en sol non
renforcé sont concentrés au niveau des deux premières couches (fig.3a), la valeur maximale est enregistrée sous la
semelle (point M). Les tassements diminuent avec la profondeur jusqu’à un minimum à la limite de la zone
compressible (point B). Le rapport de tassements entre ces deux points est de l’ordre de 8%.

a. sol non renforcé [Link] droit des inclusions c. entre les inclusions

Figure 3.Répartition
3. des déplacements en profondeur

Pour le sol renforcé par contre (fig.3b et 3c), les tassements s’étendent jusqu’à la couche rigide, comme pour le sol non
renforcé le tassement maximal est observé au niveau de la surface (point N). Le point de tassement le plus faible se
localise au milieu
u de la couche du sable grossier (point c), avec une valeur représentant 15% du tassement global. Cette
constatation relate parfaitement le rôle des inclusions dans la reprise d’une partie importante des contraintes du sol
compressible et sa transmission vers
ers le sol porteur, uniformisant par cela les tassements induits tout le long du massif.

4.2Analyse des contraintes et des efforts

La figure 4 donne la répartition des contraintes totales au milieu de la semelle dans le sol non renforcé, et dans le sol
situé entre les inclusions pour le cas renforcé, on note des contraintes plus importantes dans le sol non renforcé entre 0
et 9m. Par contre au-delà
delà on remarque que les contraintes engendrées dans le cas renforcé dépassent légèrement celles
en sol non renforcé.
orcé. Cette différence apparaît beaucoup plus à la surface limite entre la couche compressible et la
couche rigide (à 9m). Cette évolution des contraintes paraît logique et peut s’expliquer par le fait que dans les couches
supérieures du massif jusqu’à 9m, une partie des contraintes générées initialement dans le sol non renforcé sont reprises
par les inclusions dans le cas renforcé.
Contrairement au sol situé entre les inclusions, la distribution des contraintes au droit des inclusions est tout à fait
inversée
sée (figure 5), on constate des contraintes en tête d’inclusions relativement faibles. Les contraintes les plus élevées
sont observées dans la couche compressible, avec un maximum à la surface de cette dernière (à 3m).

347
Contrainte totale P-sol
P non renforcé 2 400
Contrainte totale P entre inclusions-sol
inclusions renforcé

Contrainte totale (KN/m²)

240

Contraintes totales (KN/m²)

2 000
200
1 600
160
1 200
120
800
80
40 400
0 0

Profondeur (m) Profondeur le long de l'inclusion (m)

Figure 4.Répartition
Répartition des contraintes en profondeurFigure
profondeur 5.Répartition
Répartition des contraintes le long de l’inclusion

Le graphe 6 donne l’évolution de l’efficacité « E »(définit comme étant le rapport entre la charge transférée aux têtes
d’inclusions rigides à la charge verticale totale appliquée) en fonction du chargement, on remarque que même pour le
poids propre de la semelle (chargement extérieur nul), un report de charge est obtenu. Ceconstat peut rejoindre celui
indiqué par les expérimentations de (Berthelot &al., 2003)et (Van Eekelen& al., 2003) qu’une fondation superficielle
posée sur un remblai mince surmontant un terrain compressible renforcé par inclusions, modifie dès sa mise en œuvre le
mécanisme de report de charge.
Le graphe mis aussi en relief que l’amplitude
l’amplitude du chargement ne semble pas avoir une grande influence sur
l’efficacité, une augmentation de l’efficacité en pic est observée pour le premier chargement, puis l’évolution se fait très
légèrement.
80 Pointe
Tête
Efficacité en efforts (E %)

Efficacité en efforts (E %)
70 70

60 60
50 q= 00 KN/m²
50
q=40 KN/m²
40 q=120 KN/m²
40
30 q= 200 KN/M²
q= 280 KN/m²
30 Profondeur (m) q=320 KN/m²
Chargement (KN/m²) dans le sol compressible
en tête d'inclusion

Figure [Link]é
Efficacité en fonction du chargement Figure [Link]é
Efficacité en fonction de laprofondeur
Le graphe donnant la proportion de charge reportée aux inclusions en fonction de la profondeur et pour différents
chargements représenté à la figure 7 montre que quel que
que soit la valeur du chargement extérieur appliqué le report se
fait de la même manière le long du fût de l’[Link] proportion maximale de charge encaissée par les inclusions se
trouve au niveau de la couche compressible entre 3 et 6m, avec un taux moyen
moyen de 75%.Cet état va dans le même sens
que celui de (Plumelle, 1985) qui a remarqué que l’effort en tête reste relativement faible et que le maximum du report
est situé à quelques mètres au-dessous
dessous de la semelle.

4.3Etude paramétrique
4.3.1 Epaisseur du matelas
Les graphes de la figure8montrent une très faible influence de la variation de la hauteur du matelas sur les tassements
notammententre 0.4 et 0.6m, où les tassements obtenus pour des chargements inférieurs à 200 kPa n’enregistrent
qu’une
’une évolution de l’ordre de 2à 3%.On note cependant une augmentation plus importante entre 0.6 et 0.8 m, qui
varient de 5 à 10% en fonction du chargement.

40 50
Tassement (10-3 m)

Efficacité E (%)

30 45
20 40
H matelas 0.4 m
10 H matelas 0.5 m
H matelas 0.6 m 35
0 H matelas 0.8 m
30

Chargement (KN/m²)
Hauteur du matelas (m)
Figure 8. Influence de la hauteur du matelas sur les tassements et l’efficacité en eff
efforts

Contrairement aux tassements, on remarque que l’augmentation de la hauteur du matelas de répartition fait diminuer
le report des charges. Une régression de 17% est observée quand l’épaisseur du matelas augmente de 30 cm. Cet état
peut traduire que le matelas lui-même
même supporte une certaine quantité de charges, donc plus l’épaisseur est grande plus
l’effort transférée dans le matelas croit, ce qui diminue celui transmis aux inclusions. Ce constat a d’ailleurs été rapporté

348
dans les recommandations (ASIRI,, 2012) pour le cas du dallage, où une réduction de l’épaisseur du matelas augmente
l’efficacité des inclusions rigides.
4.3.2Module de déformation du matelas
Les résultats des tassements et des efforts induits en fonction de différents modules de déformation du matelas de
transfert de charge montrent une sensibilité à ce paramètre, une réduction moyenne des tassements de 12.7% est
enregistrée pour une augmentation de 100 kPa. Le taux de diminution est plus élevé pour des chargements plus
importants,, un taux de 14% est atteint pour un chargement de 300 kPa.
Inversement aux tassements, l’évolution du report de charges est proportionnelle à l’augmentation du module de
déformation du matelas. Cette sensibilité est d’autant plus importante que le module est faible. Ainsi on relève une
progression de l’efficacité de 17% pour le premier palier (entre 50 et 100 MPa), par la suite le taux diminue pour varier
entre 5 et 8% pour le même pas de variation du module de déformation, figure 9.
40
50

Efficacité E ( %)
Tassement (10-3 m)

30 45

20 40

10 E matelas=50 MPa
E matelas=100 MPa 35
E matelas=150 MPa
0 E matelas=200 MPa 30

Chargement (KN/m²) Module de déformation du matelas

matelas…

Figure [Link]
Influence du module de déformation du matelas sur les tassements et l’efficacité en efforts

4.3.3 Angle de frottement du matelas

Les simulations faites pour trois valeurs d’angle de frottement du matelas montrent une influence assez appréciable de
ce paramètre sur les tassements et les effortsComme montré sur la figure 10, les tassements sont ainsi inversement
proportionnelles à l’angle de frottement du matériau du matelas, plus le matériau est frottant moins sont les tassements
engendrés. Cette influence
nce est aussi fonction de l’intensité du chargement. Du point de vu report de charge, on note une
évolution de 16% pour une élévation de l’angle de fortement de 10°.
50
40
Tassement (10-3 m)

Efficacité E (%)

45
30
40
20
phi matelas=30° 35
10 phi matelas=35°
phi matelas=40° 30
0

Chargement (KN/m²) Angle du frottement du matelas (°)

(

Figure [Link]
Influence de l’angle de frottement du matelas sur les tassementset l’efficacité en efforts

4.3.4Taux de couverture
Les courbes tassement-chargement
chargement de la figure 11 pour trois valeurs du taux de couverture mettent en évidence une
réduction très significative des tassements à l’accroissement du taux de substitution. La réduction est d’autant plus
importante que le chargement est plus élevé.Pour une différence de couverture de 4%, les tassements diminuent de plus
de 50%.
Comme pour les tassements, le report des charges est aussi conditionné par le taux de couverture, une augmentation
de 25% est relevée pour une variation de 4% du taux de couverture, Elle atteint 40% quand «  »passe de 2 à 8%.
Fait évident qu’une augmentation de la surface d’inclusion favorise la réduction des tassements puisque ce
paramètre définie la proportion
portion du sol renforcé, donc plus le taux est élevé, plus le report de charge vers les inclusions
est augmenté, et le sol sera moins sujet à des déformations
50 Apha=2%
Tassement (10-3 m)

Alpha=4%
40 60
Efficacité E (%)

Alpha=8%
30
50
20
10 40

0 30

Chargement (KN/m²) Taux de couverture  (%)

Figure 11. Influence du taux de couverture sur les tassements et l’efficacité en efforts

349
4.3.5Module de déformation des inclusions
Les tassements en fonction de chargements donnés sur la figure 12 pour différents modules de déformations donnent
des courbes très proches révélant une réduction presque insignifiante d’un maximum de 4% pour un pas d’augmentation
du module de 5x106 KPa.
Comme pour les tassements, la variation du module de déformation des inclusions affecte très peu le taux de
transfert des charges, le taux d’augmentation de l’efficacité est seulement de 2.8% quand le module de déformations
croît de 10 à 25 GPa. Cette faible influence sur les tassements et l’efficacité peut s’expliquer par le fait qu’on se trouve
toujours dans la plage de l’inclusion rigide,
gide, et qu’on pouvait mesurer mieux la sensibilité du comportement en faisant
varier le module entre les inclusions rigides et semi rigides par exemple (colonnes à module contrôlé)

40 50
Tassement (10-3 m)

Efficacité E (%)
30 45

20 40
E inclusions =10E6 Kpa 35
10 E inclusions=15E6 Kpa
E inclusions=20E6 Kpa 30
0 E inclusions=25E6 Kpa
Module de déformation des inclusions (106…
Chargement (KN/m²)

Figure 12. Influence du module de déformation de l’inclusion sur les tassements et l’efficacité en efforts

5CONCLUSION

L’analyse statique de fondations superficielles reposant sur sol renfoncé par inclusions rigides moyennant le code de
calcul Plaxis 3D foundations a permis une étude globale du comportement d’un tel système.
Outre l’efficacité importante en termes de tassements obtenue par le système de fondation renforcé par des inclusions
rigides, des éléments assez appréciablesse rapportant au mécanisme de transfert des efforts à travers les différentes
composantes,
antes, et son effet sur la génération des tassements et leurs diffusions au sol support ont été aussi établis.
Ainsi en termes de déformations, les constatations les plus notables, concernent surtout la diffusion des tassements
en profondeur du massif par le biais du matelas de répartition, et des inclusions, et même de la fondation qui a un rôle
essentieldans l’uniformisation des tassements.L’étude a aussi montré que la proportion maximale de charge encaissée
par les inclusions se trouve au niveau de la couche
co compressible.
L’analyse de l’effet de quelques paramètres sur le comportement du systèmevis-à-vis
systèmevis vis aux déformations et au report
de charge a montré en premier l’influence considérable du taux de couverture, la hauteur et les caractéristiques
mécaniques du matelas de répartition (module de déformations et l’angle de frottement) semblent aussi avoir une
influence non négligeable.

REFERENCES
Antoine P.C. (2010) Renforcement par géosynthétiques des remblais sur inclusions rigides- rigides Etude des dalles sur sols renforcés au
moyen d’inclusions rigides ou non, Thèse de Doctorat, Université Libre de Bruxelles, Belgique.
ASIRI. (2012) Recommandations pour la conception, le dimensionnement, l’exécution et le contrôle de l’amélioration des sols de
fondations par inclusions rigides,, IREX (Institut pour la Recherche appliquée et l’Expérimentation en génie civil), France,
édition presses des Ponts.
Berthelot P, et al. (2003) Recommandations
ecommandations sur la conception, le calcul, l’exécution et le contrôle des colonnes ballastées ssous
bâtiments et ouvrages sensibles au tassement.
tassement
Combarieu O. (1990) Fondations superficielles sur sol amélioré par des inclusions rigides verticales,
verticales, Rapport des Laboratoires, GT
n° 41.
Frossard A, Glandy M. (2003) Dimensionnement d’une fondation superficielle
superficielle sur un sol renforcé d’inclusions,Fondsup (Symposium
International sur les Fondations Superficielles),
Superficielles Paris, France.
Jenck O.(2005) Le renforcement des sols compressibles par inclusions rigides verticales. Modélisation physique et numérique
numérique, thèse
dee Doctorat, Institut National des Sciences Appliquées de Lyon, Unité de Recherche en Génie Civil, France.
Le Hello B. (2007) Renforcement par géosynthétiques des remblais sur inclusions rigides-
rigides Etude expérimentale en vraie grandeur et
analyse numérique, Thèse
hèse de Doctorat, Université Grenoble 1, Joseph Fourier., France.
Plumelle C. (1985) Renforcement d’un sol lâche par inclusion de micropieux », Revue française de géotechnique, n° 30, pp 47 47-57.
Van Eekelens S.J.M, Bezuijen.A, Oung.O. (2003) Arching in piledembankments - experiments and design calculationscalculations, Foundations
Innovations,observations, design and practice. Thomas Telford, London.

350
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Etude de la résistance au cisaillement d’une argile surconsolidée expansive traitée

aux liants hydrauliques
Shear strength of an expansive overconsolidated clay treated with hydraulic binders
A. Mahamedi1, M. Khemissa*1
1
Laboratoire de Développement des Géomatériaux, Université Mohamed Boudiaf, 28000 M’sila, Algérie
*
Auteur correspondant (khemissa@[Link])

RÉSUMÉ– Cet article présente et analyse les résultats d’une série d’essais d’identification, de compactage et de
cisaillement direct effectués conformément aux normes algériennes sur une argile surconsolidée expansive traitée aux
liants hydrauliques (ciment Portland composé et chaux éteinte). Cette argile a été prélevée sur le site urbain de la ville
de Sidi-Hadjrès (wilaya de M’sila, Algérie), où d’importants désordres apparaissent fréquemment dans les
infrastructures routières, voiries et réseaux divers et dans les constructions civiles et industrielles de faible hauteur. Les
résultats d’essais obtenus montrent que les valeurs des paramètres géotechniques déduits de ces essais sont
concordantes et confirment l’amélioration de la résistance au cisaillement de cette argile naturelle traitée au ciment ou à
la chaux et compactée à l’optimum Proctor. Cependant, contrairement à ses caractéristiques minéralogiques qui ne
semblent pas être affectées par le traitement, cette argile naturelle expansive est caractérisée par une résistance au
cisaillement aussi bien drainée que non drainée sensible au dosage en stabilisant ; les meilleures performances sont
obtenues pour un traitement correspondant à 8% de teneur en ciment ou en chaux.

ABSTRACT – This paper presents and analyzes the results of a series of identification, compaction and direct shear
tests performed in accordance with the Algerian standards on an expansive overconsolidated clay treated with hydraulic
binders (composed Portland cement and extinct lime). This clay was obtained from the urban site of Sidi-Hadjrès city
(wilaya of M'sila, Algeria), where significant damages frequently appear in the road infrastructures, roadway systems
and various networks and in civil and industrial light structures. Tests results show that the geotechnical parameters
values deduced from these tests are concordant and confirm the shear strength improvement of this natural clay treated
with cement or lime and compacted at the Proctor optimum. However, contrary to its mineralogical characteristics
which do not seem to be affected by the treatment, this expansive natural clay is characterized by as well drained as
undrained shear strength sensitive to stabilizer content; the best performances are obtained for a treatment
corresponding to 8% cement or lime content.

1 INTRODUCTION

Le tissu urbain de la wilaya de M’sila en Algérie connaît de nos jours un développement considérable en raison d’une
démographie sans cesse croissante, d’où son extension vers des terrains vierges souvent moins favorables que ceux déjà
urbanisés. Cette wilaya est située dans une zone classée comme semi-aride, caractérisée par de faibles précipitations et
des écarts de température importants entre l’hiver et l’été. La géologie de cette zone comporte des formations argileuses
caractérisées par une forte variation de volume lorsque les conditions de leur équilibre sont modifiées. Une étude
antérieure effectuée par Khemissa et al. (2008) montre que ces formations argileuses sont très surconsolidées, peu
perméables et très peu sensibles au fluage ; leur surconsolidation est due au phénomène de retrait résultant d’une
dessiccation plus ou moins poussée. L’emploi en remblai ou en couche de forme en l’état naturel de ces argiles n’est
normalement pas envisagé. A sec, elles sont très difficiles à compacter puisque leur consistance varie de dure à très dure
et, humide, elles sont très collantes. Cependant, leur emploi peut éventuellement être décidé sur la base d’un traitement
spécifique aux liants hydrauliques (LCPC-SETRA, 2000a). Cet article présente les résultats d’une étude effectuée sur
une argile surconsolidée expansive prélevée dans la commune de Sidi-Hadjrès (wilaya de M’sila, Algérie), où
d’importants désordres apparaissent fréquemment dans les infrastructures routières, voiries et réseaux divers et dans les
constructions civiles et industrielles de faible hauteur. L’étude effectuée vise à déterminer les paramètres physiques et
mécaniques de cette argile naturelle traitée au ciment Portland composé et à la chaux éteinte, puis à analyser l’effet du
traitement sur ses caractéristiques de cisaillement et de rupture.

2 DESCRIPTION DE L’ARGILE ETUDIEE

Les échantillons utilisés ont été prélevés entre 1,3 et 1,7 m de profondeur dans une couche d’argile marno-gypseuse,
jaunâtre à brunâtre atteignant 1,5 à 4,5 m de profondeur selon les endroits. La figure 1 présente le fuseau
granulométriquede cette argile naturelle et la figure 2 son diffractogramme aux rayons X. Les essais d’identification
effectués sur les échantillons testés conformément aux normes algériennes en vigueur (IANOR, 2010) ont permis
d’obtenir les plages de variation et les valeurs moyennes des paramètres géotechniques présentées dans le tableau 1 et
leur composition chimique donnée dans le tableau 2.
351
100

Pourcentage de passant (en poids) (%) 75

- % < 2 mm=95,5%
25
- % < 80 mm=73,2%
- % < 2 mm=25,7%

0
10 1 0,1 0,01 0,001
Diamètre des grains (mm)
Figure 4. Fuseau granulométrique de l’argile de Sidi-Hadjrès (wilaya de M’sila, Algérie).

Figure 2. Diffractogramme aux rayons X de l’argile de Sidi-Hadjrès (wilaya de M’sila, Algérie).

Tableau 1. Paramètres géotechniques de l’argile de Sidi-Hadjrès (wilaya de M’sila, Algérie).

Paramètres Symboles Plages de variation Moyennes Normes suivies
Profondeur z (m) 1,30-1,70 1,50 -
Teneur en eau naturelle wnat(%) 13,21-13,46 13,34 NA 5209
Poids volumique humide h(kN/m3) 20,4-24,2 22,3 NA 5210
Poids volumique sec d(kN/m3) 18,0-21,4 19,7 NA 5210
Limite de liquidité wL (%) 81,5-86,7 83,7 NA 16213
Limite de plasticité wP (%) 30,6-36,6 32,8 NA 16213
Indice de plasticité IP (%) 50,1-51,9 51,0 NA 16213
Indice de consistance Ic (%) 1,33-1,47 1,38 NA 16213
Valeur de bleu du sol VBS 7,40-9,77 8,31 NA 5288
Refus à 2 mm %<2 mm 95,0-96,0 95,5 NA 5232
Refus à 0,08 mm %<0,08 mm 64,6-81,9 73,2 NA 5232
Teneur en argile C2m (%) 20,5-30,9 25,7 NA 5251
Activité de l’argile Ac 1,95-2,02 1,98 NA 5251
Teneur en eau optimale wopt (%) 19,2-19,6 19,51 NA 5262
Densité sèche maximale d-max 1,59-1,61 1,60 NA 5262
Coefficient de dégradabilité DG 2,68-3,50 2,97 NA 5281
Coefficient de fragmentabilité FR 2,93-3,51 3,25 NA 5282

Tableau 2. Composition chimique de l’argile de Sidi-Hadjrès (wilaya de M’sila, Algérie).

Constituants SiO2 CaO MgO Fe2O3 Al2O3 SO3 K2O Na2O Pertes au feu
% 43,38 14,66 2,55 4,02 11,36 11,55 1,51 1,12 10,03

352
Ces valeurs peu dispersées pour le prélèvement effectué semblent indiquer que le massif est homogène. Selon la
classification des Laboratoires des Ponts et Chaussées en vigueur en France (Magnan, 1988), il s’agit d’une argile très
plastique (At), très consistante avec une activité importante de sa fraction argileuse (présence de montmorillonite
calcique).L’analyse chimique effectuée sur cette argile montre que les éléments prépondérants sont la silice, les
carbonates et l’alumine. L’analyse diffractométrique aux rayons X montre que la silice est cristallisée sous forme de
quartz (21%) et les carbonates sous forme de calcites (79%). Selon la classification des sols fins et des matériaux
rocheux évolutifs en usage en France (LCPC-SETRA, 2000b), cette argile appartient à la sous-classe A4 (Ip>40 ou
VBS>8 : argiles et argiles marneuses très plastiques) et est considérée comme peu fragmentable (FR<7) et peu
dégradable (DG<5). Les modifications de sa teneur en eau sont, par contre, accompagnées de retrait ou de gonflement.
L’abaque de Casagrande adapté aux sols gonflants par Chen (1988) et Dakshanamurthy et Raman (1973) montre que
cette argile est caractérisée par un potentiel de gonflement élevé à très élevé (figure 3). Les classifications de
Ranganatam et Santyanarayana (1965), de Seed et al. (1962) et de Williams et Donaldson (1980) indiquent, elles aussi,
un potentiel de gonflement élevé à très élevé. La classification du BRE (1980) conduit, par ailleurs, à un potentiel de
retrait très élevé.

Potentiel de gonflement (Dakshanamurthy et Raman, 1973)

Potentiel de gonflement (Chen, 1988)

nul faible moyen élevé très élevé

60
très élevé
Indice de plasticité, Ip (%)

Argiles très plastiques (At)

45
40
élevé

Argiles peu plastiques (Ap)

30
moyen

Présente étude
10
faible

Khemissa et al. (2008)

0
0 20 30 40 50 60 70 80 90 100 120
Limite de liquidité, w L (%)

Figure 3. Classification de l’argile de Sidi-Hadjrès (wilaya de M’sila, Algérie).

3 PROGRAMME EXPERIMENTAL ET PROCEDURES D’ESSAI

Outre les essais d’identification, le présent programme expérimental comporte des essais de cisaillement direct à la
boite de Casagrande de type consolidé non drainé (essais CU) et de type consolidé drainé (essais CD) exécutés sur
échantillons saturés ou non saturés après leur compactage aux conditions de l’optimum Proctor. Ces essais ont été
effectués sur sol non traité (échantillon témoin) et sur sol traité à différentes teneurs en ciment ou en chaux ; le ciment
utilisé est de type Portland composé (CEM-II/B 42.5 N NA 442 – MATINE) fabriqué localement dans la cimenterie
Lafarge de Hammam Dalâa (wilaya de M’sila, Algérie) et la chaux éteinte (NHL) provient des usines ERCO de
Hassasna (wilaya de Saïda, Algérie).
Les teneurs pondérales en ciment ou en chaux considérées sont 0% pour l’échantillon témoin, 2%, 4%, 6%, 8%,
10% et 12% pour les échantillons traités. Les échantillons soumis aux essais ont été confectionnés à partir du mélange
avec le stabilisant de la quantité nécessaire du sol séché puis finement broyé ; l’ensemble étant malaxé vigoureusement
à la main à sec puis humidifié à la teneur en eau optimale wopt (i.e. densité sèche maximale correspondant à l’optimum
Proctor d-max). Tous les essais ont été conduits à la température ambiante comprise entre 20 et 25°C au mois de mai.
Les procédures expérimentales suivies dans chaque type d’essai ont été conformes autant que possible aux méthodes
d’essais en vigueur en Algérie (IANOR, 2010) et les méthodes d’interprétation des résultats d’essais s’inspirent
beaucoup des connaissances acquises sur ce type de sol à travers le monde (Al-Rawas et Goosen, 2006 ; LCPC-SETRA,
2001 ; CRR, 2004). La figure 4 présente les résultats des essais de compactage effectués à l’optimum Proctor sur
l’argile traitée à différentes teneurs en ciment ou en chaux. Ces résultats constituent un gage de bonne répétabilité de
l’essai de compactage et indiquent une bonne reconstitution du sol en place comme en laboratoire.

353
0,2
0,2

gd-max/wopt
gd-max/wopt=0,078
gd-max/wopt

gd-max/wopt=0,079
0,1 0,1

0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Dosage en ciment (%) Dosage en chaux (%)
Figure 4. Résultats des essais de compactage à l’optimum Proctor de l’argile traitée au ciment ou à la chaux.

4 RESULTATS DES ESSAIS ET DISCUSSION

On se restreint à ne présenter ci-après que les principaux résultats de cette recherche, c’est-à-dire l’influence du
traitement au ciment ou à la chaux sur les caractéristiques minéralogiques et de plasticité de l’argile étudiée, d’une part,
et sur ses caractéristiques de cisaillement et de rupture, d’autre part.

4.1 Effet du traitement sur la minéralogie de l’argile

La figure 5 présente les diagrammes d’évolution des principaux minéraux constitutifs de l’argile déterminés par
diffractométrie aux rayons X avant et après traitement aux différents dosages en ciment ou en chaux utilisés et
compactage à l’optimum Proctor. On peut noter que le traitement aux liants hydrauliques de cette argile naturelle
expansive n’affecte pas ou peu ses caractéristiques minéralogiques, de sorte que les teneurs en calcite et en quartz ne
semblent pas évoluer de façon significative. Autrement dit, la forme et l’assemblage de ces deux types de minéraux
prépondérants ne sont que très peu affectés par l’ajout de ciment ou de chaux à l’argile avant d’être compactée aux
conditions de l’optimum Proctor.
Calcite Quartz
100
Minéraux (%)

21 28 26
39 39 35 39

50
79 72 74
61 61 65 61

0
0 2 4 6 8 10 12
Dosage en ciment (%)
Calcite Quartz
100
Minéraux (%)

21 29 25 19 26
32 37

50
79 71 68 75 81 74
63

0
0 2 4 6 8 10 12
Dosage en chaux (%)
Figure 5. Effet du traitement au ciment ou à la chaux sur la minéralogie de l’argile compactée à l’optimum Proctor.

4.2 Effet du traitement sur la plasticité de l’argile

La figure 6 montre que l’indice de plasticité et la limite de liquidité diminuent avec les dosages en ciment ou en chaux
considérés, mais davantage avec la chaux qu’avec le ciment. Cette diminution se traduit par une réduction de la
plasticité de l’argile qui devient moins sensible à l’eau, donc mieux compactable, avec en prime un potentiel de
gonflement plus faible. La réduction du gonflement de l’argile traitée lui procure une certaine stabilité vis-à-vis des
déformations dues aux variations saisonnières de teneur en eau et, par conséquent, un comportement durable (vis-à-vis
de l’usure des particules génératrices de particules fines plastiques).

354
Potentiel de gonflement (Dakshanamurthy et Raman, 1973)

Potentiel de gonflement (Chen, 1988)

nul faible moyen élevé très élevé
60

très élevé
Indice de plasticité, Ip (%)
Argiles très plastiques (At) (Echantillon témoin)
45
40

élevé
Argiles peu plastiques (Ap) Ciment Chaux
30 2% 2%

moyen
4% 4%
20 6% 6%
8% 8%
10 10% 10%
faible

12% 12%
0
0 20 30 40 50 60 70 80 90 100 120
Limite de liquidité, w L (%)

Figure 6. Effet du traitement au ciment ou à la chaux sur la plasticité de l’argile.

4.3 Effet du traitement sur la résistance au cisaillement de l’argile

Les figures 7 et 8 présentent les courbes d’évolution de la cohésion non drainée ccu et de l’angle de frottement non
drainé cu, d’une part, et de la cohésion effective c' et de l’angle de frottement effectif ', d’autre part, de l’argile traitée
aux différents dosages en ciment ou en chaux utilisés ; lesquels paramètres ont été déterminés au pic des courbes de
cisaillement [Link] peut noter que les valeurs de la cohésion et de l’angle de frottement (i.e. la résistance au
cisaillement) du sol saturé après compactage à l’optimum Proctor sont moins élevées que celles du sol non saturé, aussi
bien traité que non traité. Toutefois, l’évolution de ces paramètres avec le dosage en ciment comme en chaux n’est pas
régulière : elle est tantôt croissante tantôt décroissante.

250 50
Ciment Chaux
CU non saturé CU non saturé 45
Angle de frottement non drainé, jcu (°)

CU saturé CU saturé
200 40
Cohésion non drainée, ccu (kPa)

150 30

100 20

50 10
Ciment Chaux
5 CU non saturé CU non saturé
CU saturé CU saturé
0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Dosage en stabilisant (%) Dosage en stabilisant (%)
Figure 7. Effet du traitement sur la résistance au cisaillement non drainée de l’argile compactée à l’optimum Proctor.

355
250
Ciment Chaux
CD non saturé CD non saturé
CD saturé CD saturé
200

Cohésion drainée, c' (kPa)

150

100

0
0 2 4 6 8 10 12
Dosage en stabilisant (%)
50
Ciment Chaux
45 CD non saturé CD non saturé
CD saturé CD saturé
Angle de frottement effectif, j' (°)

0
0 2 4 6 8 10 12
Dosage en stabilisant (%)
Figure 8. Effet du traitement sur la résistance au cisaillement drainée de l’argile compactée à l’optimum Proctor.

L’examen des courbes permet de constater que l’effet du traitement de cette argile naturelle expansive ne devient
efficace qu’à partir d’un dosage de 4% de ciment ou de chaux, mais davantage pour le sol non saturé que saturé. Dans
tous les cas, les valeurs optimales des paramètres de cisaillement drainés et non drainés sont obtenues pour un
traitement correspondant à 8% de ciment ou de chaux. Cependant, il faut amasser beaucoup de données sur le même
type de sol pour comprendre pourquoi on obtient une plus grande résistance à 8% qu’en utilisant un plus grand dosage
en ciment ou en chaux. Par ailleurs, une étude antérieure effectuée sur cette même argile traitée aux mêmes stabilisants
(Khemissa et Mahamedi, 2014 ; Mahamedi et Khemissa, 2014) montre que la cohésion non drainée cu (obtenue à partir
des essais de cisaillement direct à la boite de Casagrande de type non consolidé non drainé : essais UU sur échantillons
non saturés après leur compactage aux conditions de l’optimum Proctor) augmente avec le dosage en stabilisant, tant
pour le traitement simple que pour le traitement mixte (figure 9). Ce résultat se traduit, dans les deux cas, par une nette
amélioration de la capacité portante de l’argile. Cette amélioration est davantage obtenue pour le traitement simple avec
la chaux qu’avec le ciment et le meilleur résultat est obtenu pour un traitement mixte correspondant à un dosage de 8%
de ciment et 4% de chaux.

356
250 250
Cohésion non drainée, cu (kPa) Ciment

Cohésion non drainée, cu (kPa)

Chaux 200
200

150
150
100
100
50
50
0
0
0 2 4 6 8 10 12
Dosage en stabilisant (%) Dosage Ciment/Chaux (%)
a/ Traitement simple b/ Traitement mixte
Figure 9. Effet des traitements simple et mixte sur la cohésion non drainée de l’argile compactée à l’optimum Proctor.

5 CONCLUSION

Cet article avait pour finalité de caractériser le comportement en laboratoire d’une argile naturelle expansive traitée au
ciment Portland composé et à la chaux éteinte en vue de son utilisation en construction routière. Le choix du site urbain
de Sidi-Hadjrès (wilaya de M’sila, Algérie) a été motivé en raison de son extension vers des zones à risque, où
d’importants désordres apparaissent fréquemment dans les infrastructures routières, voiries et réseaux divers et dans les
constructions civiles et industrielles de faible hauteur. Les échantillons de sol testés ont été identifiés comme une argile
marneuse très plastique. Les différentes classifications basées sur les caractéristiques d’identification montrent que cette
argile naturelle est caractérisée par un potentiel de gonflement élevé à très élevé ; le gonflement étant pour partie dû à la
structure minéralogique des sols (teneur élevée en montmorillonite) et aux variations de leur teneur en eau (cycles de
dessiccation-réhumidification des sols). Les résultats obtenus permettent de conclure que le traitement simple aux liants
hydrauliques a pour effet d’améliorer la résistance au cisaillement de l’argile, donc sa capacité portante, mais davantage
avec la chaux qu’avec le ciment et les meilleures performances sont obtenues pour un traitement mixte correspondant à
un dosage de 8% de ciment et 4% de chaux. Cependant, ces résultats ne constituent qu’une première approximation qui
nécessite la réalisation de nombreux essais pour être mieux élucidés.

REFERENCES

Al-Rawas A.A., Goosen M.F.A. (2006). Expansive Soils: Recent advances in characterization and treatment. Taylor & Francis
Group, London, United Kingdom.
BRE-UK (1980).Building Research Establishment, United Kingdom.
Chen F.H. (1988). Foundations on expansive [Link] in Geotechnical Engineering, Elsevier Publishing Co., vol. 54, 464
p.
CRR (2004). Code de bonne pratique pour la stabilisation du sol au ciment. Recommandations du Centre de Recherches Routières,
R 22/60, Bruxelles, Belgique.
Dakshanamurthy V., Raman V. (1973). A simple method of identifying an expansive [Link] and Foundations, 13, 97-104.
IANOR (2010). Recueil des normes algériennes : secteur des travaux publics. Institut Algérien de Normalisation, Algérie.
Khemissa M., Mekki L., Bakir N. (2008). Comportement oedométrique des argiles expansives de M’sila (Algérie). In: Actes du
Symposium International Sécheresse et Constructions, Presses du Laboratoire Central des Ponts et Chaussées, France, 2008, 229-
234.
Khemissa M., Mahamedi A. (2014). Cement and lime mixture stabilization of an expansive overconsolidated clay. Applied Clay
Science, Elsevier, 95, 104-110, [Link]
LCPC-SETRA (2000a). Traitement des sols à la chaux et/ou aux liants hydrauliques : Application à la réalisation des remblais et
des couches de forme. Guide technique.
LCPC-SETRA (2000b). Guide des terrassements routiers : Réalisation des remblais et des couches de forme. Guide technique.
Magnan J.P. (1980). Classification géotechnique des sols. A propos de la classification LPC. Bulletin de liaison des Laboratoires des
Ponts et Chaussées, France, 105, 49-52.
Mahamedi A., Khemissa M. (2014). Stabilization of an expansive overconsolidated clay using hydraulic binders. HBRC Journal,
Elsevier, in-press, [Link]
Ranganatam B.V., Santyanarayana B. (1965). A rational methods of predicting swelling potential for compacted expansive clays. In:
Proc. 6th ICSMFE, Montreal, 1965, vol. 1, pp. 92-96.
Seed H.B., Woodward R.J., Lundgreen R. (1962).Prediction of swelling potential for compacted clays. Journal of the Soil Mechanics
and Foundations Division, ASCE, 88(SM 4), 107-131.
Williams A.B., Donaldson G.W. (1980).Developments related to building on expansive soils in South Africa: 1973-1980. In: Proc.
4th ICSMFE, Denver, 1980, vol. 2, pp. 834-844.

357
358
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

On the behaviour of columnar-reinforced foundations

M. Bouassida*1, M. Klai1,S. Tabchouche2and M. Mellas2
1
Université Tunis El Manar, Ecole Nationale d’Ingénieurs de Tunis., LR14ES03-Ingénierie Géotechnique, BP 37 Le
Belvédère 1002 Tunis, Tunisia
2
Université de Biskra, Faculté des sciences et technologies, Biskra, Algeria
*
Corresponding author

RÉSUMÉ –Ce papier étudie la prédiction du comportement de fondations reposant sur sol renforcé par colonnes de
sable. Ledimensionnement est basé sur un taux de substitution optimiséd’après la méthodologie proposée par Bouassida
& Carter (2014). La prédiction du comportement du sol renforcé est analysée par des calculs numériques effectués par
les codes Plaxis V9.2D et FLAC3D en déformation plane et en axisymétrie. Deux cas d’études tunisiens de
deuxréservoirs pétrolierssont examiné[Link] adoptant le comportement élastique linéaire pour les constituants du sol
renforcé, on remarque que les prédictions du tassement en utilisant le logiciel Columns 1.01 sont en bon accord avec les
résultats numériques obtenus par les codesPlaxis, en déformation plane et en axisymétrie et par FLAC 3D. La
comparaison entre les prédictions avec plusieurs approches constitue une validation des résultats obtenus avec le
logiciel Columns 1.01. En adoptant un critère de rupture de type Mohr-Coulomb pour le matériau constitutif des
colonnes et d’écrouissage pour le sol mou, l’accélération de la consolidation a été mise en évidence d’après l’évolution
du tassement prédite par le code Plaxis.

ABSTRACT–The paper studies the prediction of behaviour of foundations resting on a soil reinforced by sand and
stone [Link] preliminary design is based on optimized improvement area ratio predicted by the methodology
proposed by Bouassida& Carter (2014). The prediction of thebehaviour of thereinforced soil is analyzed by numerical
computations carried out by Plaxis V9.2D and FLAC3D codes in plane strain and axisymmetric conditions. Tunisian
case historyof an oil tank is investigated. When oneconsiders the linear elastic behaviour for constituents of reinforced
soil, it is noted thatthe predictions of settlements from Columns 1.01 software are in good agreement with numerical
results obtained by Plaxis and FLAC3D codes. Comparison between those approaches shows the validity of results
using the Columns 1.01 software. By adopting the Mohr-Coulomb failure criterion for columns material and the
hardening soil model for soft clay, the evolution of long term settlement predicted by Plaxis code showed the
acceleration of the consolidation of thecompressible soft clay due to the enhanced drainage property of column
material.

KEYWORDS - behaviour, case history, follow-up, numerical, sand column, sand, soft clay.

1 INTRODUCTION

The study of the behaviour of engineering structure built on a soil reinforced by columns increasingly knew, since the
last decade, several contributions. As for other geotechnical applications, the development of finite element and finite
difference codes in conjunction with advanced constitutive laws for soils enhanced the accuracy of numerical
computations using either 2D or 3D modelling of reinforced soil by a group of columns (Six et al, 2013).
When dealing with columnar reinforcement of soft soil, the evolution of settlement resulting from the acceleration of
consolidation is with high importance in regard to long stability of the studied structure.
In this paper, tow main goals are targeted. First objective is to discuss the usefulness of the linear elastic behaviour in
describing the behaviour of constituents of reinforced soil by columns. Accordingly, the predictions by Columns 1.0
software (Bouassida & Hazzar, 2012) are compared with results predicted by the group of columns modelling in linear
[Link] comparison between settlement predictions, obtained by the linear elasticity model and Mohr Coulomb
failure criterion,is made regardless the consolidation aspect. Finally, the influence of columns material is foreseen when
the consolidation phenomena are taken into account in the case where reinforcement of soft soil is considered.

2. STUDY OF REINFORCED SOIL BEHAVIOUR

2.1 Foundations on compressible clays reinforced by end-bearing columns

Tunisian study case is considered to carry out numerical computations using the FLAC3Dand Plaxis codes: the case
study is the ‘’Oil Tank at La Goulette‘’. The reinforcement by end-bearing sand columns of soft clay layer of 10 m
thickness has been considered for an applied surcharge load of 80 kPa by a circular tank of 30 m diameter. Consider the
allowable settlement of reinforced soil equals 10 cm, the optimized Improvement Area Ratio (IAR), as determined by
the Columns 1.01 software, is 29.95% that was executed by 729 sand columns of 0.6 m in diameter with axis to axis
spacing of 1.06 m installed in triangular pattern.

359
Figure 1depicts the FLAC 3D numerical model of reinforced soil which consists of a central row crossing the tank
center with dimensions: 30.74 m in length along x direction; 2.12 m in width along y direction and 10 m in depth as
length of end-bearing sand columns. This reinforced soil model comprising 29 columns is modeled by generated mesh
constituted by 18320 finite element zones and 16947 grid-points. The reinforced soil model subjected to the tank load
of 80 kPa has the following boundary conditions: horizontal displacement along x and y directions is assumed zero;
horizontal and vertical displacements are zero at the stratum depth level.

Figure 1. The FLAC 3D numerical modelling of reinforced soil (Oil tank at Zarzis, Tunisia)

Computations are carried out by adoptingelastoplastic behaviourand the Mohr-Coulomb failure criterion without
taking into account the primary consolidation option. The geotechnical parameters of the numerical model are given in
Table 1.

Table 1Geotechnical parameters of the numerical model generated by the FLAC 3D software
Thickness or Young’s modulus Poisson’s Friction Total unit weight
Parameters Cohesion (KPa)
length (m) (MPa) ratio angle (°) (KN/m3)
Columns material 10 20 0.33 38 0 18
Soft soil 10 2.0 0.33 0 24 17

The evolution of the vertical displacement at the surface of the unreinforced soil vs the number of iterations, at
centerline of tank, is plotted in Figure 2 that indicates a maximum vertical displacement of 20.4 [Link] evolution of
vertical displacement of reinforced soil at centerline of tank versus number of iterations is plotted in Figure 3.

Figure 2. Evolution of vertical displacement of unreinforced soilFigure [Link] of vertical displacement of reinforced soil

Table 2 summarizes the predictions of settlement as obtained by the FLAC3D code and by the methods incorporated
in Columns 1.01 software (Bouassida & Hazzar, 2012) that considers the linear elastic behaviour both for the initial soil
and column material. It is noticed that the settlement prediction by FLAC 3D code is close to the settlement
360
predictionobtained by the analytical method proposed by Bouassida et al (2003) which consider the behaviour of a
group of columns similarly as that modeledin the FLAC3D program in this generated models. The relative difference
between the predicted settlements is less than 1.2% that is quite negligible.

Table 2Settlement predictions obtained by the FLAC3D and Columns 1.01 software
Bouassida et al French method
Prediction method FLAC 3D code Balaam & Booker (1981)
(2003) (2011)
Unreinforced soil 20.4 38.37 20.8 20.8
Reinforced soil 9.88 10.0 8.6 7.48

Considering the elastoplastic behaviour (without the consolidation option) of the constituents of the reinforced soil by
columns, the numerical results obtained by the FLAC3D code have confirmed the results provided by the optimized
methodology of design of column-reinforced foundations embodied in the Columns 1.01 software. Although the
elastoplastic behaviour for constituents of reinforced soil by columns was assumed by the numerical FLAC 3D
[Link] results were found identical to those predicted by the new methodology of design. In witch,a linear elastic
behaviour is considered for constituents of reinforced soil (Bouassida & Carter, 2014). Such result can be explained by
the fact that settlement prediction is essentially foreseen under admissible loads that do not induce large deformation
within the reinforced soil. It follows the predictions both by linear elastic and elastoplastic constitutive laws can be
quite similar.

2.2 Comparison between numerical predictions in plane strain and axisymmetric analyses

The plane strain condition can be adopted for structures having a current cross section that extends on a quite length as
compared to the dimensions of the cross section of studied structure. The linear loading condition of such structures
shows that the displacement in the perpendicular direction to the cross section of the considered structure is negligible.
In turn, axisymmetric analysis applies for structures having geometrical and loading symmetries with respect to a
revolution axis. The common property for plane strain and axisymmetric analyses relies on their inadequate modelling
of a truly 3D structure.
The Tunisian case study of oil tank (La Goulette) is now considered to compare the predicted behaviour of unreinforced
and reinforced soft clay of Tunis by sand columns both in plane strain and axisymmetric conditions. This numerical
study has been conducted by the Plaxis V9.2D finite element code adopting the hardening soil model (HSM) as
constitutive law for Tunis soft clay. The simulation of the behaviour of the Tunis soft clay specimens subjected to
oedometer and triaxial tests showed that the HSM predictions are in agreement with the measured data rather than
predicted results obtained by the Cam-Clay modified model (Klai, 2014).
Performing numerical computation in plane strain condition passes by the determination of the equivalent width, B, of
squared tank that has the same area as the circular tank of radius R = 15 m, B   R  26.587m .
Predictions by the HSM (plane strain condition)
The evolution of the settlement of the unreinforced soil at the boundaries of the oil tank,illustratedin Figure 4 shows a
maximum consolidation settlement of 4.2 cm. The predicted maximum consolidation settlement is 21.8 cm at the
centerline of the tank, the consolidation time approaches ten years. It is clearly seen that the long term settlement due to
the consolidation of soft clay is slightly larger than the settlement of 20.8 cm predicted by different methods that adopt
the linear elastic behaviour (see Table 2).
As illustrated in Figure 4, the differential settlementof the tank could be problematic for the stability of the tank. The
normalized differential settlement is of about 0.3% that exceeds notably the allowable limit of 0.1%. Therefore, to avoid
excessive settlements which can compromise the stability of the tank, the need of columnar-reinforcement of Tunis soft
clay is strongly recommended to reduce the settlement and to accelerate the consolidationby choosing a drained column
material like sand with enhanced hydraulic conductivity.

The behaviour of soft clay reinforced by sand columns in plane strain condition has been analysed using the perfect
elastoplastic law and the Mohr-Coulomb failure criterionhavebeen adopted for the sand as constitutive column material
with geotechnical properties displayed in Table 4.

Table [Link] properties of sand and constitutive column material

Thickness or Young’s Poisson’s Friction Cohesion Unit weight

Parameters
length (m) modulus (MPa) ratio angle (°) (kPa) (kN/m3)
Sand (columns) 10 20 0.33 38 0 18
Tunis soft clay 10 2 0.4 0 24 17

361
Figure 4. Evolution of the settlement of the unreinforced soil using plane strain modelling

The evolution of the consolidation settlement, in plane strain condition, is plotted in Figure 5 both at the centerline
and at the boundariesof the oil tank where predicted values are 7 cm and 3.8 cm respectively.
The quasi constant distribution of settlement at the surface ofthe reinforcedsoilis partially owed to the presence of
blanket drainage layer. One can consider that the differential settlement equals the difference between predicted
settlements at the centerline and the boundariesof the oil tank. The comparison between Figures 4 and 5well shows that
the reinforcement by sand columns greatly reduces the differential settlement from 17.7 to 3.2 cm that plays an
important benefit for the stability of the oil tank.

Figure [Link] evolution of the consolidation settlement of the reinforced soil (plane strain condition)

Predictions by the HSM (axisymmetric study)

The axisymmetric study is carried out comparing the predicted 2D behaviour of unreinforced soil and reinforced
soil by sand columns both in plane strain and axisymmetric conditions by implementing numerical computations
with Plaxis software. The adopted parameters of the hardening soil model for Tunis soft clay are given in Table 5.

Table 5 Parameters of the hardening soil model for Tunis soft clay

c [kN/m3] e E’ [kPa] G [kPa] c Eoed,c[kPa] c [°] kh= kv [m/day]

18 0.5 20,000 7519 0.33 29,630 38,0 10
362
The evolution of the consolidation settlement, versus time, at the surface of the unreinforced soil is illustrated in
Figure 6 showing that the primary consolidation is expected to end within 12.5 years. The maximum settlement at the
centerline of tank equals 27 cm that exceeds the predicted settlement in plane strain condition. The same remark holds
for the predicted settlements at boundariesof the tank that are 8.3 cm and 7 cmrespectively in axisymmetric and plane
strain conditions. However, when adopting the consolidation procedure, it was found that the end of the primary
consolidation settlement remains unchanged (12.5 years) both for the plane strain and axisymmetric analyses.
The evolution of the consolidation settlement at the surface of the reinforced soil by sand columns (axisymmetric
study) is plotted in Figure 7. From this figure, it is noticed that the settlement at the centerline of the tank of 9.4 cm is
greater than the predicted value of 7 cm in plane strain condition (Figure 5). The same remark holds for the predicted
settlement at the boundariesof the oil tank: 4.5 and 3.8 cm respectively in axisymmetric and plane strain conditions.

Figure 6. Evolution of the consolidation settlement of the unreinforced soil (axisymmetric condition)

Figure [Link] of the consolidation settlement at the surface of the reinforced soil

363
Table6 summarizes the predicted settlements of the consolidation at the centerline of the oil tank of La Goulette both
in plane strain and axisymmetric studies (performed by the Plaxis V2.9 and the 3D model generated by the FLAC 3D
code). It is noticed that the predicted settlements of the unreinforced soil are quite similar withthe plane strain
computation using Plaxis and FLAC3D programs. In turn, axisymmetric study withPlaxis leads to anoverestimated
prediction of the consolidation settlement of the unreinforced soil. For the case of the reinforced soil, it is found that the
plane strain condition slightly underestimates the settlement prediction compared to that obtained by Plaxis software in
axisymmetric condition and the one obtained by FLAC3D model. It isconcluded that comparable results are obtained
using any computational tools (Plaxis or FLAC) independently of the adopted model either 2D (plane strain and
axisymmetric conditions) or 3D.

Table [Link] predictions at centerline of the La Goulette oil tank by Plaxis V2.9 and FLAC 3D codes.

Numerical tool Option computation Unreinforced soil (cm) Reinforced soil (cm)
Plaxis V9.2D Plane strain 21.8 7.0
Plaxis V9.2D Axisymmetric study 24.8 9.4
FLAC 3D 3D model 20.4 9.88

3 CONCLUSIONS

This paper focused on the study of the behaviour of foundations installed on compressible soils reinforced by sand
columns. A Tunisian case study has been investigated by performing numerical computations withthe Plaxis V9.2 and
the FLAC 3D codes.
All those methods consider the linear elastic behaviour both for the initial soil and column material. It is noticed that the
settlement predictions obtained with theFLAC 3D code are close to those obtained with the analytical method proposed
by Bouassida et al (2003). The latter uses the group of columns model as for the FLAC3D code. The relative difference
between these predictions, of 0.59 % and 3.74% respectively for the unreinforced soil and the reinforced soil, are
negligible. This first example confirms the usefulness of analytical methods of settlement predictions programmed
within the Columns 1.01 software that provides the optimized reinforcement.
It is noticed that the difference between predicted settlements of unreinforced soil is not significant; therefore the use of
any computational tool (Plaxis or FLAC) or any study hypothesis (plane strain or axisymmetric study) leads to
comparable results from which the need of reinforcement by columns can be justified. In turn, when dealing with the
modelling of reinforced soil, it is revealed that the plane strain condition slightly overestimates the settlement
prediction. Indeed, such hypothesis is not realistic for cylindrical [Link] axisymmetric condition can be adopted
since its settlement prediction is close to that obtained from a 3D modelling.

REFERENCES

Balaam, N.P. and Booker, J.R. Analysis of rigid rafts supported by granular piles. Int. J. of Num. & Analyt. Meth. in Geomech.,
5(4), 1981, 379-403.
Bouassida,M., Guetif, Z., de Buhan P. and Dormieux L., “Estimation par une approche variationnelle du tassement d'un sol renforcé
par colonnes.”Revue Française de Géotechnique. 102(1), 2003, 21-29.
Bouassida M. and Hazzar L. Novel tool for optimized design of reinforced soils by columns. Ground Improvement: Proc. ICE, 2012,
London 165, Issue 1, 31 –40.
Bouassida, M. and Carter, J. P. (2014).Optimization of Design of Column-reinforced Foundations. Int. J. Geomech., Volume 14,
Issue 6 (December 2014)10, 1061/(ASCE)GM.1943-5622.0000384.
French document (2011). Recommandations sur la conception, le calcul, l’exécution et le contrôle des colonnes ballastées sous
bâtiments et sous ouvrage sensible au tassement (in French). Comité Français de Mécanique des Sols, Version n°2, March 16, 32 p.
Klai, M. On the behaviour of Tunis soft clay – Application to the study of foundations stability (in French).PhD Defended 16 Oct.
2014, National Engineering School of Tunis. Tunisia.
Six, V., Mroueh, H. Shahrour, I. and Bouassida, M. Numerical Analysis of Elastoplastic Behaviour of Stone Column Foundation.
Geotech Geol Eng. Jrn 2012, 30:813–825.

364
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Reduction of clay soil collapsibility using cement-polymeric fibre admixture

Réduction de la tendance à l'effondrement de d’un sol argileux à l'aide de mélanges de fibres

ciment-polymère
M.M. El-Emam*1, M.F. Attom2, and Z.H. Khan3
1
Associate Professor, American University of Sharjah, Sharjah, United Arab Emirates
2
Professor, American University of Sharjah, Sharjah, United Arab Emirates
3
Assistant Professor, American University of Sharjah, Sharjah, United Arab Emirates
*
Corresponding Author

ABSTRACT - This study presents evaluation of collapse potential of clay soil treated with cement-polymeric fibre
admixture. Different cement/fibre ratio were mixed together to form the cement-fibre admixture used in this study. In
order to investigate the effect of admixture content percent on collapsibility potential, admixture was added at 0% to 5%
by dry weight of the soil, in 1% increment. Three different soils with different clay contents and plasticity indices were
selected to conduct the current investigation. According to classification tests, the three soils can be considered with
moderate to high collapsibility potential. Double oedometer tests were conducted to measure pure and stabilized soil
collapsibility index. Specimens were prepared inside the oedometer rings using static (kneading) compaction. In
addition, samples have been compacted at different initial water contents in order to isolate the effect of water content
in the soil collapsibility. Results indicate that, increasing cement/fibre ratio and/or the admixture content percent
significantly decreases the collapse strain at inundation. Moreover, the collapsibility index decreased with increasing
the admixture content beyond 3% for soil I and 2% for soil II and III. For admixture contents less than 2% to 3%, the
reduction on the collapsibility index is negligible. For all the soil used in this study, with 5% admixture content and a
cement to fibre ratio of 6 by weight, the soil moves from very high collapsible range to almost no collapsibility range.
Finally, results indicated that the collapsibility index decreased as the initial water content increased. This result is valid
for all admixture content percent and all cement to fibre ratio.

RÉSUMÉ-Cette étude présente une évaluation du potentiel d'effondrement d’un sol argileux traité avec différents taux
de mélange fibres de ciment et d'un polymère. Ratio ciment/fibres différentes ont été mélangés pour former le mélange
de ciment-fibre utilisé dans cette étude. Afin d'étudier l'effet pour cent contenu mélange sur l'effondrement potentiel,
adjuvant a été ajoutéà 0 % à 5% du poids sec du sol, en incrément de 1%. Trois types de sols différents avec différentes
teneurs en argile et indices de plasticité ont été choisis pour mener l'enquête en cours. Selon les tests de classification,
les trois types de sols peuvent être considéré comme modéréàélevé effondrement potentiels. Essais oedométriques
double ont étéréalisées pour mesurer l'indice d'effondrement du sol pur et stabilisé. Spécimens ont été préparés à
l'intérieur de l'oedomètre anneaux à l'aide de compactage (pétrissage) statique. En outre, des échantillons ont été
compactés à teneur en eau initiale différente afin d'isoler l'effet de la teneur en eau dans l'effondrement du sol. Résultats
indiquent que, augmentant le ratio ciment/fibres et/ou le pourcentage de contenu mélange sensiblement diminue la
souche d'effondrement à l'inondation. En outre, l'indice d'effondrement diminue avec l'augmentation de la teneur du
mélange au-delà de 3% pour le sol j'ai et 2% pour le sol II et III. Pour mélange contenu moins que 2% à 3%, la
réduction sur l'index de l'effondrement est négligeable. Pour tous les sols utilisés dans cette étude, à teneur en mélange
de 5% et un ciment ratio fibre de 6 en poids, le sol se déplace de très haute gamme pliable à presque aucune tranche
d'effondrement. Les résultats obtenus, pour tous les pourcentages du mélange fibres de ciment et polymère, indiquent
que l'indice d'effondrement diminue lorsque la teneur en eau initiale augmente.

1 INTRODUCTION

Many geotechnical engineering problems occur due to presence or addition of water to soil. Soils that decrease in
volume due to wetting under sustained load are termed as collapsible soils. Collapsible soils can sustain large loads
under unsaturated conditions but exhibit significant settlement (or collapse) when its degree of saturation increases.
Saturation of soils can increase due to many reasons such as fluctuations in ground water table, flooding, impoundment
of dams, poor drainage, and irrigation. The behavior of collapsible soils has extensively been investigated (Tadepalli et
al. 1992; Vu and Fredlund 2004) and the collapse potential of soils has been observed by many researchers to be
affected by the initial water content, the initial dry density, and the method of compaction (Lawton et al. 1989; Fredlund
and Rahardjo 1993; Jotisankasa et al. 2007). Increase in wetting affects the matric suction and diagenesis of soils
resulting in the relocation of soil particles and consequently the collapse (Casagrande 1932; Barden et al. 1973; Mitchell
1976).

365
Collapse potential is generally evaluated by testing soil samples in one dimensional oedometer (ASTM D5333) or
with the help of triaxial test equipment. Single or double point oedometer test could be used to evaluate the collapse
potential. Previous studies have found insignificant differences between the two testing methodologies. Different
criteria of evaluating collapse potential can be applied to estimate the collapse (Abelev 1984; Clevenger 1985; Denisov
1963; Feda 1998; Lin and Wang 1998; Handy 1973). Collapse potential of a soil can be reduced by many methods as
indicated by many researchers and implemented by practitioners (e.g. Anayev and Volyanick 1986; Houston and
Houston 1997; Hansen et al. 1989; Rollins and Rogers 1994; Clemence and Finbarr 1981). These methods include (but
are not limited) to soil replacement, pre-wetting, controlled wetting, compaction control, chemical stabilization and
cement or lime treatment.
This study presents a full testing program using one dimensional oedometer test to evaluate, reduce or eliminate the
collapsibility potential of soil using cement-fibre admixture. Different percentages of cement-fibre admixture were
mixed with the soil at different initial physical conditions. Enough numbers of specimens from each soil were tested in
one dimensional oedometer at three initial dry densities and initial water contents. The samples were prepared in such a
way to have a potential for collapse. All samples were prepared in the odometer ring using kneading compaction to
ensure the uniform low density of the soil. The standard collapsible test was conducted on the prepared samples in
agreement with ASTM (D5333).

2 EVALUATION OF COLLAPSE POTENTIAL

The behavior of collapsible soils can be attributed to certain conditions that affect the settlement. Barden et al. (1973)
and Mitchell (1993) identified that a collapse would initiate if the soil fabric is partially saturated, a stress that creates a
metastable soil matrix is applied, and addition of sufficient water. Soils generally exhibit interparticle cohesion at the
contact level because of digenesis due to iron oxide or carbonates and clay particles and because of the matric suction in
shallow soils (Collins 1978). Negative pore water pressure is developed within the soil due to capillary action because
of low water content in unsaturated collapsible soils. The interparticle bond due to capillary action (matric suction)
exists due to negative pore water pressure. Matric suction decreases with increasing saturation because of increasing
pore water pressure and approach zero at full saturation. Collapsible soils dependent on matric suction to maintain the
metastable structure at high porosity. In the absence of this bond, collapsible soil experience sudden loss of volume and
the loss of volume increases with increasing applied pressure.
The type of bonding significantly affects the collapse due to different rates of reduction in matric suction during
saturation (Tadepalli et al. 1992). It is therefore intuitive to use cementing agents to decrease the collapse potential of
the soils. Although the total amount collapse could be similar; the rate of collapse also depends on the interparticle
contact. The rate of collapse decreases with cementation and smallest rate of collapse is observed in clayey bonding of
contacts (Clemence and Finbarr 1981). The behavior of collapsible soils suggests that collapse, due to increase in
saturation, will occur when the degree of saturation is below a critical value (Jennings and Knight 1975a&b). They
suggested that the critical degree of saturation (Sr) varies from Sr = 6% to 95%. The critical degree of saturation
increases with decrease in particle diameter (D50).
The critical conditions that affect the collapse of most soils are degree of saturation, dry density of the soil, and the
effective confining stress (Lawton et al. 1989). These factors along with other stabilization techniques are generally
evaluated in one dimensional oedometer test. The oedometer test is performed using one of the two most common
variations. The first method is based on testing two identical specimens under different wetting conditions (Jennings
and Knight 1975a); whereas, in the other method,a single specimen is tested according to the testing methodology
presented by Houston et al. (1988) and ASTM (D5333). Typically, a specimen is placed in the oedometer cell with
initial dry density, water content, and stabilizers if any, and consolidated until the normal stress of 200 kPa. At this load,
the specimen is saturated and the consolidation test is completed at other normal stresses that is expected to be applied
in the site. The settlement recorded at the normal stress of 200 kPa is used to assess the collapse potential. The
schematic of collapse behavior during a typical oedometer test is presented in Figure 1. The collapse potential according
to ASTM (D5333) is assessedconsidering:

e200 H 200
Cp  100  100 (1)
1  eo Ho

where Cp = Collapse Potential (in %), Δe200 = Change in void ratio upon wetting under 200 kPa of normal pressure, eo =
Initial void ratio, ΔH200 = Change in sample height on wetting under 200 kPa of normal pressure, and Ho = Initial
height.
The collapse potential can be used to assess the severity of problem in the field (Jennings and Knight 1975b).
Similar criterion is suggested by Lin and Wang (1988). They presented the collapsibility potential as the collapsibility
Index with essentially similar ranges as presented in Table 1. They; however, measured the changed in height (H) due
to saturation of the specimen. The collapsibility index (IC) is given by Equation 2.

366
H1  H 2
IC  100 (2)
Ho

Table [Link] of collapse potential and severity of problem (Lin and Wang, 1988).
Collapse Potential, IC ( %) Severity of Problem
0-1 No Collapsibility
1-5 Medium Collapsibility
5-10 High Collapsibility
10-20 Very High Collapsibility
>20 Extremely Collapsible

dry specimen
200 kPa
Void ratio, e

collapse strain at
inundation (e)

saturated specimen
Pressure, n (kPa)
Figure [Link] oedometer test to evaluate collapse potential.

whereH1 and H2 are the heights of the specimen before and after saturation,at stress 200 kPa.
Abelev (1984) presented the pioneering work of relating change in void ratio (due to saturation) to the collapsibility
of soils. The collapsibility coefficient (S) related the change in void ratio (e) during saturation to the initial void ratio
(eo) as: S = e/(1+eo). According to Abelev (1984), the soil’s potential to collapse increases as the collapsibility
coefficient becomes larger than s = 2%.
Denisov (1964) evaluated the collapse potential by comparing the void ratios at water contents corresponding to
natural and liquid limits. The study indicated that soils are susceptible to collapse if the void ratio at any state is larger
than the void ratio at liquid limit. Feda (1960) proposed the relationship between the water content, the degree of
saturation and the plasticity of the soil for the evaluation of the collapse potential. The collapse potential (iC) is given by
Equation3.
m
 PL
Sr
IC  (3)
PI
Where;m is the natural water content, Sr the degree of saturation, PL the plastic limit, and PI the plasticity index. The
soil is considered collapsible if the collapsibility index (Equation 3) becomes less than 1. Feda (1960) also correlated
the porosity of the soil to the collapse potential. Soils having porosity (n) greater than 40 % are susceptible to collapse.
The main objective of this work is to evaluate the effect of cement-fibre admixture on the behavior of collapsibility
of the clayey soil. To achieve the research objective, three types of clay soil were selected and the initial physical
properties such as plastic limit, liquid limit, maximum dry density and optimum moisture content, and soil gradations
were determined in agreement with ASTM. The soil samples were prepared at initial low water content and initial low
dry density to ensure that they all samples are susceptible to collapse. Different percentages of cement-fibre admixture
were added to the soils by dry weight of the soil. The prepared specimens were tested in the one dimensional oedometer
test, and test results were analyzed to evaluate the effect of admixture on soil collapsibility.

3 EXPERIMENTAL PROGRAM AND MATERIALS

3.1 Soil and Fibres

Three different types of clayey soils were selected with different clay content and different indices. The initial physical
properties of the soils were obtained such as Atterberg’s limits,gradation, maximum dry density and optimum moisture
content in agreement with ASTM procedures. Initial physical properties of the tested soils areshown in Table 2.
367
The three types of soils were prepared at different initial dry density in order to satisfy Lutenegger and Saber (1988)
criteria (see Figure 2) and to ensure the collapsibility of the soil. The soils were then mixed with five different cement-
fibre percentages; 1%, 2%, 3%, 4% and 5%. Ratios of cement to fibre are 2:1, 4:1 and 6:1. A total numbers of 78 soil
specimens were prepared from the three types of soil with different initial dry density and different cement-fibre content
and percentages.
Identification of collapsible soil is best accomplished by testing specimens in single or double odometer tests.
However, dry density of soil together with either liquid limit or percentage finer than No. 200 sieve can be used to
evaluate the collapsibility of soils. Figure 2, shows the relation between dry density-liquid limit and collapsibility of
soils according to Lutenegger and Saber (1988).
After determining the mechanical and plasticity properties of soil samples, liquid limit of samples are plotted versus
dry density as shown in Figure 2. Each soil type is represented in two states; the maximum dry density state and the
remolded dry density state. At the maximum dry density, the three soil types are considered non collapsible according
to Lutenegger and Saber (1988) (Figure 2). The three soil types are remolded at a dry density less than the maximum
dry density, shown in Table 2. Figure 2 shows that all soil samples, at the remolded dry density, can be considered as
collapsible soil with moderate to high collapsibility potentials. The only exception is the remolded sample of soil_3.
This soil type showed a collapsible nature, according to Lutenegger and Saber (1988) (Figure 2). Therefore, this sample
is considered collapsible in lower degree.
50
Lutenegger and Saber (1988)
60
Initial dry density (Ib/ft )
3

Collapsible
70

soil_III Non Collapsible

soil_I
90 at d(max)
soil_II at d(remolded)
100
20 30 40 50 60 70 80
Liquid limit (%)
Figure [Link] potential of the soil according to the commonly used criteria for determining
soil collapsibility potential.
3.2 Soil-Fibre mixture preparation

Three types of clayey soils were selected and used to achieve the objective of this research. Specimen from each type of
clay were prepared in a predetermined initial water content and dry density and mixed with cement-fibre admixture at
different percentages. The cement-fibre admixture content is defined as: Wad/Ws where Wad is the dry weight of the
cement-fibres admixture, and Ws is the dry weight of soil. The quantities of cement, fibres and soil needed to prepare
each specimen were weighed at high accuracy. The initial water content of the specimen which represents its natural
water content is determined as Ww/Ws, where Ww is the weight of water added to the specimen. All solid components
(i.e. soil, cement, and fibre) are mixed together in dry state, and then water was added to the mixture and mixed
manually, to prevent fibre segregation and to obtain to obtain a homogeneous mixture. Water content for all soil
specimens ranged from 10% to 20% by weight of dry soil.
After the soil-cement-fibre admixture was mixed thoroughly with water, one-third of the specimen was poured
inside the oedometer ring and tamped statically with a steel rod of 10 mm diameter. This process is called kneading
compaction, which is suitable for fine grain soil (Holtz and Kovacs, 1981). In this process, weights of soil, cement,
fibre, and water were pre-estimated such that when it inserted into the oedometer ring volume and compacted with pre-
calibrated kneading effort, it gives the required dry density of the specimens. The specimen is compacted in three layers
with almost equal height (i.e. one-third of the oedometer ring height). The initial water content and the dry density were
calculated to ensure that the soil behaves as a collapsible soil according to Lutenegger and Saber (1988).

368
Table 2. Different soil composition, indices, and grain size distribution

Soil No. Soil_I Soil_II Soil_III

1. Atterberg Limits and Compaction properties
Liquid Limit (%) 52 42 31
Plastic Limit 27 24 17
Plasticity Index 25 20 14
Maximum dry density (kN/m3) 13.1 13.5 14.3
Optimum moisture content (%) 32 26 21
2. Grain Size Distribution
Sand (%) 14 19 20
Silt (%) 28 41 45
Clay (%) 58 40 35
Activity 42 50 37
Specific Gravity Gs 2.66 2.65 2.65
3. Mineralogical composition
Kaolinite (%) 17.1 25.5 31.3
Ellite (%) 31.5 26.3 36.7
Montmorillonite (%) 13.9 12.3 8.9
Chlorite (%) 17.9 14.1 9.1
Classification (USCS) CH CL CL

3.3 Single Oedometer Test

Single oedometer test which use only a single soil specimen were conducted according to the procedure recommended
by ASTM (D 5333-92) (1993), and Jennings and Knight (1975a) suggested procedures. In these tests, a remolded
sample is placed in the oedometer at specific compaction characteristics (i.e. water content and dry density). Then a
small load is applied to the specimen in order to ensure that the soil sample is in full contact with the pores plate at the
bottom of the oedometer and it is fully seated. After that, the specimen is gradually loaded up to the anticipated field
loading conditions. In this test series, a stress of 200 kPa is assumed to represent the expected load that the collapsible
soil can support. At this stress level, water is added and the specimen is allowed to be saturated. Then the deformation,
caused by inundation,is recorded for 24 hour. Afterwards, load increments proceeded up to 800 kPa.

4 RESULTS AND DISCUSSION

The (e-log) characteristic curves generated from single oedometer test for soil_I are plotted in Figure 3. It can be noted
that e is the largest for the soil without admixture (solid curve in Figure 3a). As the admixture-soil percent increased,
the reduction in void ratio, e decreased. The soil stabilized with 5% fibre-cement admixture showed the smallest
change in void ratio, e. In Figure 3b, it can also be seen that the consolidation behavior after and before inundation are
affected by the fibre-cement content. While the soil without admixture showed the smallest void ratio after inundation,
it did not show the same trend before inundation(smallest void ratio). The soil with 1% fibre-cement admixture showed
the largest decrease in void ratio, before and after inundation. Finally, the soil with 5% fibre-cement admixture content
showed the smallest decrease in void ratio, before and after inundation.
Change in void ratio at inundation versus (cement-fibre) admixture content is shown in Figure 4for soil_1. Three
different cement/fibre ratios are considered in Figure 4. Figure 4 indicated that the collapse strain at inundation, e,
reduced significantly with increasing admixture/soil content percent, for all cement/fibre ratio. For admixtures with 2
and 4 cement/fibre ratios, the addition of fibre-cement admixture slightly reduced the collapse strain. Once the
admixture/soil content exceeded a threshold value of 3%, the collapse strain significantly reduced with the
admixture/soil content. This threshold value was found to be reduced (2%) in the case where the cement-fibre ratio was
6. It could be concluded that the threshold value, at which the (fibre-cement) admixture/soil content starts to
significantly reduce the collapse strain, decreased with increasing cement/fibre ratio. Finally, note that the effect of
cement addition on the collapse strain becomes significant beyond a cement/fibre ratio of 4.

369
1.8 2.0
Cement/Fiber ratio = 2 Mixture/Soil percent Mixture/Soil percent
Cement/Fiber ratio = 2
0.0% 1.8 0.0%
1.0% 1.0%
1.7 2.0% 2.0%
3.0%
1.6
3.0%
4.0% 4.0%
1.6 5.0% 1.4 5.0%
Void ratio

Void ratio
1.2
1.5
1.0

0.8
1.4
0.6

1.3 0.4
10 100 10 100 1000
2 2
Pressure (kN/m ) Pressure (kN/m )
a) (e-log) full test path b) (e-log) path before inundation
b)
Figure 3. Single oedometer tests for remolded samples with different admixture ratios for the soil_I (cement/fibre ratio = 2).

0.5
Threshold value Cement/Fiber ratio
2.0
Collapse strain at inundation (e)

0.4 4.0
6.0

0.3

Threshold value
0.2

0.1

0.0
0 1 2 3 4 5 6
Admixture/Soil (%)

Figure [Link] of collapse strain at inundation (e) with (cement/fibre) admixture/soil content at different cement/fibre ratio, for
soil_I.

4.1 Collapsibility Index

Collapsibility index, calculated according to Equation 2, versus admixture content is respectively shown in Figures 5, 6
and 7 for soils I, II and III. For each soil, different cement-fibre ratios have been used to study the effect of cement
content on the collapsibility index. Figure 5 indicated that, before treatment, soil_I has the highest collapsibility index
among the 3 soil types, which is close to 20%. This may be related to the higher clay content of soil_I compared to the
other 2 soil types. Soil_III, with the smallest clay content, showed the smallest collapsibility index, as illustrated in
Figure 7. However, all three soil types, tested in this investigation showed collapsibility indices located in the very high
collapsibility range, as showed in Figures 5, 6 and 7. In all three figures, there are admixture content values where
significant reductions of collapsibility index started to be clear. These values are defined as the threshold admixture
contents.
For soil_I, it can be observed that there are two threshold values; 3% for the specimens stabilized with cement/fibre
ratios of 2 and 4, and 2% for specimen stabilized with cement/fibre ratio of 6 (see Figure 5). For soil_II, the threshold
admixture contents are 2% for the specimens stabilized with cement/fibre ratios of 2 and 4, and 1% for the specimen
stabilized with cement/fibre ratio of 6 (see Figure 6). For soil_III, it can be observed that there is one threshold
admixture content of 2% for all cement/fibre ratios.

370
20
Threshold values Soil_I with:
w = 10%
3
d = 10 kN/m
15 Cement/Fiber ratio = 2

Collapsibility index, Ic (%)

Cement/Fiber ratio = 4
Very High collapsibility range Cement/Fiber ratio = 6

High collapsibility range

Medium collapsibility range

0 No collapsibility
0 1 2 3 4 5 6
Admixture (%)

Figure [Link] of the collapsibility index with the (cement/fibre) admixture contentfor different cement/fibre ratios (soil_I).

20
Soil II with:
w = 10%
Very high collapsibility range 3
d = 12 kN/m
15 Threshold value
Collapsibility index, Ic (%)

Cement/Fiber ratio = 2
Cement/Fiber ratio = 4
Cement/Fiber ratio = 6

High collapsibility range

Medium collapsibility range

No collapsibility
0
0 1 2 3 4 5 6
Admixture (%)

Figure [Link] of the collapsibility index with the (cement/fibre) admixture contentfor different cement/fibre ratios (soil_II).

Beyond the threshold values, it can be observed that, for soil_I, an increase of the admixture content from 2.5% to
3.5% is needed to reduce the collapsibility indices from the very high to high collapsibility range, depending on the
cement/fibre ratio, as shown in Figure 5. To further improve the soil collapsibility index, a 1% increase in the admixture
content is enough to reduce the soil collapsibility from high to medium. Finally, to completely remove the collapsibility
potential of soil_1 (no collapsibility zone), an admixture content of 5% or larger is needed, regardless of cement/fibre
ratio.
For soil_II, admixture content from 2% to 2.5% is needed to reduce the collapsibility indices from the very high to
the high collapsibility zone (see Figure 6). However, for further reduction of the soil collapsibility index, an additional
1% increase of the admixture content reduces the soil collapsibility indices to the low collapsibility range. To move the
soil specimens to the zone of no collapsibility, the soil requires 5% admixture content with cement/fibre ratio equal to 6.
For cement/fibre ratios of 2 and 4, the soil requires admixture percent content larger than 5% to be in the no
collapsibility range.
For soil_III, Figure 7, admixture content from 1.5% to 2.5%, depending on the cement/fibre ratio, is required to
improve the soil collapsibility from very high to high collapsible zone. To further reduce the soil collapsibility index, a
1% increase of the admixture content is needed to reduce the soil collapsibility from high to medium. To further
improve the soil and reduce the collapsibility index to the no collapsible zone, 5% admixture was enough for this soil,
regardless of the cement/fibre ratio. It can be concluded that the admixture content required to improve the soil
collapsibility from the higher zone to the lower zone decreased with the reduction of the clay content in the soil. In
addition, the admixture content required to improve the soil collapsibility is decreased with increasing cement/fibre
ratio.

371
20
Soil III with:
w = 10%
3
Very high collapsibility range d = 13 kN/m
15 Cement/Fiber ratio = 2

Collapsibility index, Ic (%)

Threshold value Cement/Fiber ratio = 4
Cement/Fiber ratio = 6

High collapsibility range

Mediume collapsibility range

0 No collapsibility
0 1 2 3 4 5 6
Admixture (%)

Figure 7. Variation of the collapsibility index with the (cement/fibre) admixture contentfor different cement/fibre ratios (soil III).

16
Soil_II with cement/fiber ratio = 6
14
Water content (%)
Very high collapsibility range
12 w = 10%
Collapsibility index Ic (%)

w = 12%
10 w = 14%
w = 16%
8 w = 18%
w = 20%
6 High collapsibility range

2 Mediume collapsibility range

0
0 1 2 3 4 5
Admixture content (%)

Figure 8. Variation of the collapsibility index in function of the admixture contentfor different initial water contents (soil_II with
cement/fibre ratio of 6).

4.2 Effect of Initial Water Content

Figure 8 shows the effect of the admixture content on the collapsibility index for different initial water contents for
soil_II. The admixture used in this case has a cement/fibre ratio of 6. From Figure 8, it is clear that the effect of
(cement/fibre) admixture is more efficient in reducing the collapsibility index when the soil has lower initial water
contents. However, as the initial water content of the soil increased, its collapsibility index decreased and the effect of
(cement/fibre) admixture in reducing the collapsibility index is also decreased. It can be observed that the largest
admixture effect in reducing the collapsibility index was for soil with 10% initial water content, while the smallest
effect was for the soil with 20% initial water content. Finally, 5% admixture content appeared to be an optimum value
as it fully eliminates the soil collapsibility potential, regardless of the soil initial water content. This conclusion is clear
in Figure 8 as the collapsibility index, for all soil with different initial water content, is located in the range of no
collapsibility at admixture content equal to 5%.

5 CONCLUSION

Based on tests on improved and non-improved soil conducted in this study and the results presented in this paper, the
following concluded remarks can be summarized:
1. Collapsibility index and potential increased as the clay content of the soil increased. Adding cement/fibre admixture
to collapsible soils leads to a reduction in the void ratio change resulting from an inundation. In addition, the
consolidation behavior after and before inundation are influenced by the fibre-cement content.

372
2. The addition of fibre-cement admixture to collapsible soil slightly reduces the collapse strain up to a threshold
value. Once the admixture/soil content exceeds this threshold value, the collapse strain, significantly reduces with
increasing the admixture/soil content. The threshold value decreased as the cement/fibre ratio increased.
3. For each soil, there is a threshold admixture value where significant reductions of soil collapsibility index are
observed with increasing the admixture content. The threshold value increased as the clay content of the soil
increased. This threshold value decreased with the cement/fibre ratio of the admixture.
4. As the clay content of the soil increases, the admixture content needed to stabilize the soil from higher to lower
collapsibility range increases. In addition, the admixture content required to improve the soil collapsibility is
decreased with increasing cement/fibre ratio. 5% of admixture content is suitable in order to improvethe soilsstudied
in this research (no collapsible zone).
5. For the water content range used in the current study (i.e. 10% to 20%), the collapsibility index decreases as the soil
water content increases. Therefore, flooding the collapsible soil with water to increase the water content could
significantly reduce the collapsibility index.
6. Effect of cement/fibre admixture is more efficient (in reducing the collapsibility index) with soils presenting low
initial water contents. However, as the initial water content of the soil increased, the effect of cement/fibre
admixture in reducing the collapsibility index is decreased.

REFERENCES

Abelev, Y.M. (1984) The essential of designing and building on microporous soils. Stroitel’naya Promyshelmast No. 10
Anayev, V.P., Volyanick, N.V. (1986) Engineering geologic peculiarities of construction work on loessial soils, Proceedings of the
5th International Congress of the Association of Engineering Geologists, 2: Buenos Aires, 659-665.
ASTM D5333–92 (1993) Standard Test Method for measurement of collapse potential of soils, Annual Book of ASTM Standards, 4:
343-345.
Barden, L., McGown, A. and Collins, K. (1973) The collapse mechanism in partly saturated soil. Engr. Geology, Amsterdam, 7: 49-
60.
Casagrande, A. (1932) The structure of clay and its importance in foundation engineering, Journal of Boston Society of Civil
Engineers, 19: 168-209.
Clemence, S.P., Finbarr, A.O. (1981) Design consideration for collapsible soils, Journal of the Geotechnical Engineering Division,
ASCE 135 (11), 305-317.
Clevenger, W.A. (1985) Experiences with less as foundation [Link] American Society for Civil Engineers, 123, pp:
51-80
Collins, K. (1978) A scanning electron microscopy study of natural engineering soil,.Ph.D. thesis, University of Strathclyde,
Glasgow, Scotland, U.K.
Denisov, N.Y. (1963) About the nature of high sensitivity of Quick clay. [Link]. Grunt, 5 pp: 5-8
Feda, j. (1988) Collapsible loess upon wetting. Engineering Geology, 25 pp: 263-269
Fredlund D.G., Rahardjo H. (1993) Soil Mechanics for Unsaturated Soils, New York, John Wiley and Sons, inc.
Handy, R.L. (1973) Collapse loess in Iowa, Proceedings of the Soil Science Society of America, 37: 281-284.
Hansen, L.A., Booth, R.B. and Beckwith, G.H. (1989) Characterization of a site on deep collapsing soils. Foundation Engineering,
Current Principles and Practices, ASCE, New York, 191-208.
Holtz, R.D. and Kovacs, W.D. (1981) An Introduction to Geotechnical Engineering. Prentice Hall.
Houston, S.L., Houston, W.N. (1997) Collapsible soils engineering. Unsaturated Soil Engineering Practice, Geotechnical Special
Publication, ASCE, Proceedings of the 1st Geo Institute Conference, Logan UT, USA, Part 68, 199-232.
Houston, S.L., Houston, W.N. and Spadola, D.J. (1988) Prediction of field collapse of soils due to wetting, Journal of Geotechnical
Engineering, American Society of Civil Engineers, ASCE, 114: 40-58.
Jennings J.E., Knight K. (1975a) The additional settlement of foundation due to collapse of sandy soils on wetting. Proc. 4th Int.
Conf. on soil Mech. and [Link]. 1, 316-319.
Jennings J.E., Knight K. (1975b) A guide to construction on or with materials exhibiting additional settlement due to collapse of
grain structure, Proceedings of the 6th Regional Conf. for Africa on Soil Mech. and Foundation Eng., Durban, South Africa, 1:
99-105.
Jotisankasa, A., Ridely, A., Coop, M. (2007) Collapse behavior of compacted silty clay in suction-mentioned oedometer apparatus.
Journal of Geotechnical and Geoenvironmental Engineering, ASCE 133 (7), 867-877.
Lawton, E.C., Fragaszy, R.J., James, H.H. (1989) collapse of compacted clayey sand. Journal of Geotechnical Engineering, ASCE
155 (9), 1252-1267.
Lin, Z.G., Wang, S.J. (1988) Collapsibility and deformation characteristics of deep seated loess in [Link]. Geology, 25 pp: 271-
282.
Lutenegger, A.J., Saber, R.T. (1988) Determination of collapse potential of soils. Geotechnical Testing Journal 11 (3), 173-178.
Mitchell, J.k. (1976) Fundamentals of Soil Behavior, Wiley. New York.
Rollins, K.M. and Rogers, G.W. (1994) Mitigation measures for small structures on collapsible alluvial soils, Journal of
Geotechnical Engineering, 120: 1533-1553.
Tadepalli, R., Rahardjo, H. and Fredlumd, D.G. (1992) Measurement of Matric suction and volume changes during inundation of
collapsible [Link] Testing Journal, 15: 115-122.
Vu, H. Q., Fredlund, D. G. (2004) The prediction of one, two- and three dimensional heave in expansive soils, Canadian
Geotechnical Journal, 41:1-25.

373
374
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Effect of Salinity on Hydrated Lime Stabilized Swelling Clays from Sudan4

Influence de la salinité sur le comportement d’une argile gonflante du Soudan traitée par la chaux
hydratée
A.M. Elhassan (1,*)& A.M. Elsharief (2)
1
Assistant prof. Sudan University of Science & Technology, Khartoum, Sudan, adelabdallah@[Link],
adilabdallah01@[Link]
2
Associate prof. Building and Road Research Institute, University of Khartoum, Khartoum Sudan,
aelsharief@[Link]
*
Corresponding author

ABSTRACT- Saline and sodic soils are found in the central and eastern clay plains of Sudan which are dominated by
highly plastic fertile swelling clays. Lime has been found to be an effective stabilizer for swelling clays. This paper
presents the outcome of a laboratory testing program intended to study the effect of salinity on engineering properties of
the lime stabilized swelling clays. Typical highly plastic swelling clay from Alfao in eastern Sudan was tested by
applying artificial sodicity. The natural salinity was ECe 0.5 dS m-1 and artificial samples were prepared at 2.5, 7.0 and
12 dS m-1 related to the volume of Na-CL concentration used. The increase in sodicity for the swelling clay soil resulted
in decrease of its plasticity and linear shrinkage and increase in strength for the untreated and lime treated sodic soil
tested.

RESUME- Les sols salins et sodiques se trouvent dans les plaines argileuses centrales et orientales du Soudan qui sont
dominés par des argiles gonflantes fertiles très plastique. Lime s’est révéléêtre un stabilisant efficace pour le gonflement
des argiles. Ce document présente les résultats d'un programme de tests de laboratoire destinéàétudier l'effet de la
salinité sur les propriétés mécaniques de la chaux stabilisée argiles gonflantes. Les argilesgonflantes très de Alfao dans
l'est du Soudan ontété testées en appliquant une sodicité artificielle. La salinitéde l’état naturel ECe 0,5 dS m-1 et des
échantillons ont été préparés artificiellementà 2,5, 7,0 et 12 dS m-1 en relation avec le volume de la concentration de Na-
CL utilisé. L'augmentation de la teneur en sodium pour le sol d'argile gonflante entraîne une diminution de sa plasticité
et du retrait linéaire et augmentation de la force pour le sol sodique non traitée et de la chaux traitée testé.

Keywords: Stabilization, salinity, hydrated lime, plasticity, linear shrinkage, free swell, unconfined compressive
strength

1. INTRODUCTION

Salinity and sodicity are separate and unique descriptions of the impact of soluble salts in soil and water. Sodicity
represents the relative predominance of exchangeable sodium compared to other exchangeable cations, mainly calcium,
magnesium, potassium, hydrogen and aluminium and is expressed as ESP (exchangeable sodium percentage). The
sodium adsorption ratio, SAR, is another expression of sodicity that refers to the ratio of adsorbed sodium and the sum
of calcium and magnesium [Van, 2001]. Soil salinity is a characteristic of soils relating to their content of water-soluble
salts. Salinity and sodicity are both measured by the electrical conductivity ECe expressed in decisiemens per meter (dS
m-1) of the extract of the saturated soil paste [Charman, 2000]. A strong association between saline and sodic soils has
been reported for soils from Sudan. Typical values for Sudanese soils are less than 20 dS m-1 [Elhagwa, 1998; Mustafa,
1986].
Saline and sodic soils are dominant in Sudan especially within the arid lands in the central clay plain (the area
between the White and the Blue Niles) and the terraces of the Nile north of Khartoum [Mustafa, 1986; Elhagwa, 1998].
Most of these areas are known to be covered by Vertisols or potentially expansive clay soils. Salinity seems to increase
with aridity and highly saline soils occur where the average rainfall is less than 200 mm which is the case for the
northern areas of Sudan and of Khartoum in particular [Mustafa, 1986]. According to Musfafa most of the salt
accumulation occurs at soil depth 0.3 to 0.6 m. In most soils average salinity values tends to increase with depth as does
its coefficient of variation. The dominant cation for soils from Sudan was found to be Na+ [Mustafa, 1986].
Almost 40% of the area of Sudan is covered with highly plastic potentially expansive clay plains (Vertisols). These
plains are characterized by poor traffic-ability during the rainy season. Many villages and small towns are linked with
un-paved roads therefore are totally isolated during the rainy season. Experience and research have shown that addition
of hydrated lime substantially reduces plasticity and subsequently improvetraffic-ability of the highly plastic expansive
soils.

375
Previous researchersstudied the improvement of saline clay soils using different chemical stabilizers. Ziaie (2011)
studied the volume change behavior of saline soils and also the effect of lime on the improvement of these soils using
laboratory tests, namely, consolidation, swelling pressure and free swell test. The results showed that lime reduced the
liquid limit and free swell and increased the coefficient of consolidation of stabilized saline clay soils.
Given the fact that the clay plains of the central and northern Sudan are characterized by the presence of saline and
sodic soils and their improvement with lime could be technically and economically viable [Elsharief and Mohamed,
2000] the influence of soluble salts on lime stabilization of these soils need to be investigated.

2. MATERIAL, TEST PROCEDURE AND RESULTS

The experimental testing program was aimed at studying the effects of salinity on lime stabilization of a typical highly
plastic potentially expansive clay soil from the clay plains of Sudan. The clay sample studied here was obtained from
Alfao in eastern Sudan and is characterized by its low salinity. It was intended to prepare clay samples with electrical
conductivity levels up to about 12dSm-1 to cover the salinity/sodicity levels normally found in Sudanese clay
[Link] the good association between salinity and sodicity, the dominance of Na+ in the pore water of Sudanese
soils and the ease of preparing and adding sodium chloride salt to the soil, this investigation mainly assessed the effect
of salinity on the lime-soil reaction.
A bulk sample was obtained from 0.5 m depth in Alfao. Its electrical conductivity was measured and was found to be
0.5dS m-1. The bulk sample was divided into four samples. The first one was considered as natural sample whereas the
other three samples were mixed with sodium chloride solutions with different concentration levels to prepare the saline
samples. About 1.0 kg of dry soil from each of the three samples was mixed with sodium chloride (Na CL) solution.
The concentrations were 100 mmol, 150 mmol and 200mmol respectively. The samples were then left for two days to
dry in the air after which their salinity levels were measured by electrical conductively. The values obtained were 2.5,
7.0 and 12 dS m-1, related to the volume of (Na CL) concentration used. The sample preparation procedure ended up
with four samples, a natural one with ECe value of 0.5 dS m-1, and samples with ECe values 2.5, 7.0 and 12.0 dS m-1.
The next step was to treat the samples with calcium hydroxide (lime). The optimum lime content (OLC) of the tested
soil was found from previous studies to be 6.5% (Elhassan 2013). It was decided to add 3.0% andthe OLC (6.5%) to the
prepared soil samples. Each of the four samples was further divided to three subsamples; two of them were mixed with
lime, one with 3.0% and the other with 6.5% whereas the third was not mixed with lime (0%). The following tests were
then performed on each of the subsamples: Atterberg limits, linear shrinkage (LS), free swell (FS) and unconfined
compressive strength (UCS). The tests were performed followingthe guides of BS 1377-1990 procedure. The results are
presented in Table 1. The zero percent lime and 0.5 dS m-1represent the natural un-stabilized soil properties.

Table 1. Atterberg limits, LS, FS and UCS for different salinity/sodicity levels
Sample Lime% Atterberg Limits LS% FS% UCS (kPa)
Ec (dSm-1)
LL PL PI
Alfao 0 66 29 37 19 168 275
0.50 3 61 40 21 9.9 83 406
6.5 55 45 10 5.0 32 1565
Alfao 0 66 29 37 18 167 281
2.5 3 61 40 21 9.1 78 521
6.5 55 45 10 4.8 30 1654
Alfao 0 64 30 34 17 158 290
7.0 3 60 40 20 8.1 69 580
6.5 51 41 10 4.7 28 1681
Alfao 0 61 32 29 16 151 297
12 3 58 40 18 6.2 59 610.6
6.5 50 42 8 4.5 27 1705.7
Note: LS for linear shrinkage; FS for free swell; LL for liquid limit; PL for plastic limit; PI for plasticity index

3. DISCUSSION OF THE RESULTS

Salinity/sodicity effects on plasticity and strength had been studied by imposing artificial sodicity to a typical swelling
clay sample from Alfao in eastern Sudan. The natural salinity of Alfao soil was very low, i.e. ECe 0.5 dS m-1. Soil
samples with various levels of sodicity were prepared. The Atterberg limits, linear shrinkage, free swell and unconfined
compression strength were obtained for the natural,un-treated sodic and lime-treated sodic samples (Table 1). The
Figures 1 through 5 are plotted in this format to clearly observe the changes in the engineering properties for samples
mixed with different amounts of sodium chloride, untreated or treated with the same amount of lime.
This investigation has shown that liquid limit, plasticity index, linear shrinkage and free swell of the untreated sodic soil
from Alfao generally decreased with increase in sodicity. The decrease is more effective for the untreated sodic samples
(0% lime) with ECe equals 7.0 and 12.0 dS m-1(Table 1 and Figures 1, 2, 3 and 4). The unconfined compression
strength showed steady increase with increase in sodicity for the untreated sodic soils (Figure 5). The results clearly
show that sodicity reduces the activity of the tested soil and tends to improve its engineering properties.
376
As for the lime-treated soils small increase in sodicity(2.5 dSm-1) did not affect the liquid limit and plastic limit of the
sodic samples (Figures 1 and 2). However, there is small effect, noticed for the samples with ECe 7.0 and 12.0dS m-1.
The sodicity effect on plasticity decreases with increase in lime content. The linear shrinkage and free swell decreased
with increased sodicity for the samples treated with the same amount of lime. The unconfined strength increased with
increase in ECefor the lime treated sodic soils with the same amount of lime (Figure 5). Therefore, the increase in
sodicity caused an increase in strength for the untreated and lime-treated soils (Table 1 and Figure 5). Sodicity did not
inhibit or adversely affect the reaction between lime and the swelling soil.

Figure 1. Effect of salinity and lime content on liquid limit

Figure 2. Effect of salinity and lime content on plasticity index

Figure 3. Effect of salinity and limecontent on linear shrinkage

377
Figure 4 Effect of salinity and lime on free swell

Figure 5. Effect of salinity and lime content on unconfined compressive strength

4. CONCLUSION

Clay soils (Vertisols) in northern and the central clay plains of Sudan are known to be saline. Salinity increases with
aridity. A strong association between saline and sodic soils has been reported for soils from Sudan. This paper studied
sodicity effects on the engineering properties of lime stabilized tropical swelling soil from Sudan or the effect of
sodicity on lime-treated vertisols. Sodium chloride solution was added in the laboratory to a typical swelling clay from
Alfao in eastern Sudan which has very low content of dissolved salts to prepare samples with different levels of salinity
(2.5, 5.0. 7.0 and 12.0dS m-1). The liquid limit, plasticity index, linear shrinkage, free swell and unconfined
compressive strength tests of the untreated sodic and lime treated sodic soil samples with 3.0 % and 6.5% (optimum
lime content) were carried out.
This study has shown that salinity decreases the plasticity, free swell and linear shrinkage of the untreated clay soil,
especially for the samples measuring 7.0 and 12.0 dS m-1. Salinity also increased the unconfined compression strength
of the untreated soils. The treated samples behaved in a similar manner for the samples treated with the same amount of
lime. Generally the increase of sodicity caused decrease in the activity (as reflected by the plasticity, shrinkage and
swelling) and increase in strength of the lime treated sodic soil tested. Therefore, sodicity tends to contribute positively
or to improve the engineering properties of swelling soils when stabilized with lime.

REFERENCES
Charman, P.E.V., “Soils. (2000): Their Properties and Management”, Oxford University Press, South Melbourne.
Elhagwa, A. and Buramha. (1998), “Review of saline and sodic soils of Sudan”, Proceedings of the 1st work-shop of the
land and Water Rresearch Center, Wad Madani, Sudan.
Elhassan.A. M. (2013), “Lime stabilization of tropical clay soils from Sudan, "PhD. Thesis, University of Khartoum, Sudan.
Elsharief, A.M. and Mohamed, S.A. (2000), “Technical and economical viability of lime stabilization of expansive soils for road
construction in Sudan”, Sudan Engineering Society Journal, Vol. 47 No. 38,p 5-16.
Mustafa. (1986), “Salt affected soils in the Sudan”, Reclamation and Revegetation Research, 5: p 115-124.
Van de Graaff, R. (2001), “Explaining the Mysteries of Salinity, Sodicity SAR, and ESP in On-site Practice”, in: Proceedings of
On-site ’01 Conference: Advancing On-site Wastewater Systems. 2001.
Ziaie, R. (2011), “Improvement of Volume Change Characteristics of Saline Clayey Soils”,Journal of Applied Sciences, 11(1):
p. 76-85.
378
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Geotextile-encased columns (GEC) and local available mineral aggregate (e.g. sand)
An innovative and economic foundation system for embankment on soft soil
Colonnes géotextiles encastrées (GEC) et agrégats minéraux disponibles (par exemple le sable)
Un système de fondation innovant et économique pour remblai sur sol mou

Laurent Sakou1, Johann Bruhier2, ARAB Rabah3 , Robert Medjo4, Peter Assinder5
1
Huesker Synthetic GmbH, Germany; Email: sakou@[Link]
2
Huesker SAS, France; Email: [Link]@[Link]
3
Huesker SAS, France; Email: [Link]@[Link]
4
Cameroonian committee of geotechnical engineering (CNGC), Cameroon; Email: medjoekorobert@[Link]
5
Huesker Synthetic GmbH, Germany; Email: assinder@[Link]

ABSTRACT -Road embankments, rail embankments or dikes over soft subsoil supported by piles or stone columns
have important advantages compared to “conventional” unsupported embankment foundation when compared in terms
of bearing capacity, serviceability and duration of construction. In extremely soft soil conditions, lateral support can be
problematic for stone columns. The “Geotextile-Encased sand/gravel Columns” (GECs) is an alternative system to
conventional piles of any kind and thereby eliminating at the same time the impossibility of constructing e.g. stone
columns in very soft soil due to insufficient lateral support. General principles, design concept, construction techniques,
optimization possibilitiesare shortly described. Two typical projects with monitoring systems, the High speed railway
Amsterdam-Paris and the extension of AIRBUS Hamburg-Finkenwerder in Germany, are presented.

Keywords: Geotextile Encased Columns, innovation, economic foundation system, embankment on soft soil

1. INRODUCTION: GENERAL PRINCIPLES AND DESIGN CONCEPT

Beginning in 1994, the German contractor Möbius, with the assistance of Huesker Synthetic and Kempfert & Partners,
developed a system for the foundation of embankments in soft and very soft soilareas. The general idea was to create an
alternative to conventional piles or columns of any kind; and atthe same time allow the possibility of constructing
compacted granular columns in very soft soils. Compactedgranular column techniques are usually limited to soft soils
with undrained cohesion (undrained,unconsolidated shear strength) cu or su =15 kN/m². The problem was solved by
confining thecompacted sand (or gravel) column in a high-modulus geosynthetic encasement. The general scheme of
the bearing system with Geotextile Encased Columns (GEC) is depicted in Figure 1.1 (left). Due to the higher
compression stiffness of the GECs a load concentration takes place on top of them thus reducing the stresses and
compression of the soft foundation soils in between. The vertical load on a GEC generates also horizontal radial normal
stresses outwards and radial widening of the column. This consequently provokes a counter-pressure from the
surrounding soft soils and a confining resistance from the encasement. With an appropriate encasement the system
reaches equilibrium. The design concept is based on the “single cell concept” as illustrated in Figure 1.1 (right).

Figure 1.1: General scheme of the GEC-system (left) [Huesker] and “single cell” design method (right) [EBGEO]

Assumptions, further explanations, detailed design recommendations and equations can be found in the German Code
EBGEO (2011).

379
2. CONSTRUCTION TECHNOLOGIES

Two different options are generally available with regards to the GEC construction technology. The first option is the
displacement method (Figure 2.1) where a closed-tip steel pipe is driven down into thesoft soil followed by the insertion
of the circular weave geotextile and sand or gravel backfill. The tip opens, the pipe is pulled upwards under optimized
vibration designed to compact the column fill. The displacement method is commonly used for extremely soft soils (e.g.
su< 15 kN/m²).The second construction option is the replacement method (Figure 2.2) with excavation of the softsoil
inside the pipe. This method uses an open pipe where special tools remove the soil during or afterdriving the pipe down
into the ground. The rest of the operation is identical to the displacement [Link] excavation method is likely to be
preferred with soils with high penetration resistance or whenvibration effects on nearby buildings and road installations
have to be minimised.

Figure 2.1: Displacement method of construction [Huesker] Figure 2.2: Replacement method of construction [Huesker]

The advantage of the displacement method compared to the excavation method is based on thefaster and more
economical column installation and the effects of pre-stressing the soft [Link] it is not necessary to excavate
and dispose soil. An overview of more than 30 executed projects around the world during the last about 15 years with
more than 2,000 km of installed columns shows that in more than 80% of the casethe displacement method was applied
(Assinder, 2013).

3. OPTIMIZATION EXAMPLE FOR A GEC FOUNDATION

Generally the vertical behavior (settlement) of a GEC-foundation can be controlled by the fill (e.g. sand) and by the
encasement. Additionally the percentage of columns (area ratio = total column area / total embankment basal area) can
be varied. The typical GEC area ratio valueis between 10% and 20%. In Figure 3.1 different possible options are
displayed for a typical simplified situation. The X-axis corresponds to GEC area ratios of 10 to 20%. The Y-axis
includes the predicted settlement. The four curves correspond to geotextile encasement typewith differentring tensile
strength. In the case herein an 8.0 m high embankment (ɣ= 19 kN/m², φ= 27.5°, c= 0 kN/m²) is founded on 15 m soft
soil (ɣ= 7 kN/m², φ= 18°, c= 17.5 kN/m², M = 2 MPa). Let us assume a max allowed settlement of ca. 0.6m (dotted
®
line in Figure 3.1). In this case numerous options are possible: For example: 20% area ratio with Ringtrac 200, 15%
® ® ®
area ratio with Ringtrac 300,13% area ratio with Ringtrac 400, 11% area ratio with Ringtrac 500or 10% area ratio
®
with Ringtrac 600. The difference between 20% and 10% area ratio does not look impressive, but it means in fact a
reduction by two times of the number of GECs (say half the fill, half the encasement, half the installation time, halt the
energy consumption, a strongly reduced carbon fingerprint).

4. TWO EXAMPLES OF ROAD AND RAILROADS ON GEC

4.1 High speed rail link Amsterdam-Paris, Netherlands, 2002

This new super-express (up to 300 km/h) rail link was built in 2000-2003. Due to the high speed the requirements in
terms of allowable deformations of any type were very stringent. The route had to cross at Westrick-Breda for some
hundreds of meters an old closed landfill from the 50ies. The waste had been never neither sorted, nor homogenized,
nor compacted, had significant degree of chemical contamination and was additionally mostly saturated. This resulted
in very low (varying) strength and deformation characteristics. The final optimized solution (Figure 4.1) comprised a
GEC foundation for the core embankment with a geogrid on top of the GECs. As material for the encasements PVA
(Polyvinylalcohol) was chosen due to two advantages: the high ring tensile modulus resulting in low settlements of the
system and the high chemical resistance in a wide range of environments. The project was completed within few
months by the German contractor Möbius in 2002.

380
Settlements
[m]

Area ratio [%]

®
Figure 3.1: Example of optimization potential of GEC´s [Draft, Ringtrac S version [Link]]

Figure 4.1: Typical cross-section at Westrick-Breda for the high-speed train Amsterdam-Paris

Due to its high category the project was accompanied with measurement programs. The settlements in the non-
supported area are about 0.5 m higher than in the GEC-supported area (Figure 4.2), although the load on the non-
supported area below the shoulder of embankment is lower (roughly the half). Due to the ductile behaviour of the
("softer") GEC-support there is a "mild" settlement transition from supported to non-supported area (Figure 4.2).
Despite the inhomogeneity of the waste fill the max settlement difference across the GEC-supported area is only ca.
0.05 m. The most settlements occur in within 1.5 months (beginning of July-mid August). No further increase of the
settlement was measured after the system was completely constructed (Figure 4.3).The system is since about ten years
under intensive train traffic and demonstrating good performance.

381
Figure 4.2:Westrick-Breda: Total settlementnon supported and GEC-supported area

Figure 4.3: Westrick-Breda: Increase of settlement non supported and GEC-supported area

4.2 Extension of AIRBUS Hamburg-Finkenwerder, Germany, 2001

Besides using the GEC foundation system in embankment construction there are meanwhile experiencesin major
hydraulic construction projects. Especially the area extension of the airplane dockyard in Hamburg-Finkenwerder with
approx. 140 hafor the production of the new Airbus A380 has to be mentioned. The area-extensionis located in the
“Mühlenberger Loch” adjacent to the west of the existing [Link] area-extension is carried out by enclosing the
polder with a 2.4 km longdike to fill up in the area under buoyancy, see Figure [Link] dike foundation was realized
with about 60,000 geotextile encased columns with a diameter of 80 cm, which were sunk to the bearing layers with
depth between 4 and 14 m below the base of the dike footing. This dike constitutes the new main water protectiondike
of the airplane dockyard. Furthermore another 10,000 columns were installed to relocate the existing “Finkenwerder
Vordeich” towards the river Elbe and to avoidsludge replacement, to increase the stability and to decrease the
settlements of thedike. Typical soil conditions are shown in Figure 4.4.

382
Figure 4.4:Concept to reclaim land by the construction of a polder and typical soil boundary conditions

Due to the different soil conditions along the dike length 7 measurement cross sections were necessary. In a typical
measurement cross section, 4 groups are placed, each containing one earth pressure gauge and one water pressure gauge
above the soft soil layer, and two piezometers within the soft soil. In each cross section, one horizontal and two vertical
inclinometers are used for the examination of the deformation behaviour. The measured settlements in dike section VI
are shown in Figure 4.5.
Due to the foundation system GEC the dike could be constructed to a flood water save height of 7 m in a construction
time of approx. 9 months. To complete the dike up to approx. 10 m, inclusive a cover of organic clay, a construction
time of approx. 15 month was necessary.
fill height [m]
max. Settlement [m]

Figure 4.5: Forecasted and measured Settlements, for example in section VI

The dike “Finkenwerder Vordeich Süd” is only partly founded on encased columns. In the part outside the main load
area vertical drainage is used to accelerate the settlements. Figure 4.6 shows typical measurement results pointing out
the different settlement reduction in the part with encased columns (thickness of soft soil about 7 m) and the part with
vertical drainage (thickness of soft soil about 4.5 m).

383
Settlement [m] Dike height [m]

Distance to dike axis dike [m]

Figure 4.6: Measured settlements ‘Finkenwerder Vordeich’ (for example)

The foundation system proved its value by flexibility during installation and by short time of consolidation. Therefore it
was possible to build up the dike almost continuously in separate layers. The dike stability for all construction stages
and the final stage were calculated with effective shear parameters φ' and c' considering the pore water overpressure due
to the particular stage-load. The stability was verified using circular sliding surfacesand failure mechanism with rigid
soil [Link] increaseof shear strength due to the concentration of stress above the columns was considered by the
implementation of alternative shear parameters by Raithel (1999).
As a result of the stability calculations, a geocomposite with a high tensile strength (maximum high tensile force 500-
1000 kN/m) in the dike base, perpendicular to the dike centerline was needed, to accelerate the construction of the dike
and to obtain a high degree of stability in the initial stage of construction. It was also necessary to increase the stability
since the area behind the dike was to be raised to a height of 5 to 8 m above sea level.

[Link]

The case studies show that theGEC-foundation system hasbecome an interesting technological and economical
alternativeto conventional foundation systems. By combining the geotextile encased columns with horizontal geotextile
reinforcement(load transfer mat) it is meanwhile possible to construct successfully foundations in extremely difficult
subsoil circumstances for Railway. The effectiveness respectively thesettlement reduction and the required properties of
the geosynthetic encasement can be forecasted with sufficient and high reliability.

REFERENCES

Alexiew. D. et al. 2010. Foundation of a coal coke stockyard on soft soil with geotextile encased columns and horizontal
reinforcement. Proceedings of the 9 th International Conference on Geosynthetics, Brazil, pp. 1095-1099
Alexiew, D., Raithel, M., Küster, V., Detert, O. (2012). 15 years of experience with geotextile encasedgranular columns as
foundation system, Proc. Int. Symposium on Ground Emprovement IS-GI, ISSMGE TC 211, Brussels.
Assinder, P.J; Alexiew, D. (2013).Road and rail embankments on geotextile encased columns:An overview, including suitability of
application for African Soils, GhIGS GeoAfrica 2013 Conference, Ghana
Brokemper. D. et al. 2006. Design and construction of geotextile encased columns supporting geogrid reinforced landscape
embankments. Proceedings of the 8th International Conference on Geosynthetics, Yokohama. pp. 889-892
Di Prisco, C., Galli, A., Cantarelli, E., Bongiorno, D. (2006). Geo-reinforced sand columns: small scaleexperimental tests and
theoretical modelling, Proc. 8th Int. Conf. on Geosynthetics, Yokohama, Millpress, Rotterdam, 1685-1688.
DGGT. 2010. Empfehlungen für den Entwurf und die Berechnung von Erdkörpern mit Bewehrungen aus Geokunststoffen-EBGEO.
Ernst & Sohn, Berlin

384
EBGEO (2011).Recommendations for Design and Analysis of Earth Structures using [Link]
Geotechnical Society (DGGT), Ernst & Sohn, Essen-Berlin.
Lhotzky, K.; Friedl, G.; Lhotzky, S. (2008) Hydrostatische Offshore – Setzungsmessungenbeim Bau des JadeWeserPorts,
Interdisziplinäre Messaufgabenim Bauwesen, Weimar 2008
Raithel. M. 1999. Zum Trag- und Verformungsverhalten von geokunststoffummantelten Sandsäulen, Schriftenreihe Geotechnik. Heft
6. Universität Kassel
Raithel, M., Kempfert, H.-G., Kirchner, A. (2002): Geotextile-Encased Columns (GEC) for Foundation ofa Dike on very Soft Soils.
Proc. 7th Intern. Conf. on Geosynthetics, Nizza, pp 1025–1028.
Raithel. M. 2008. Calculation techniques and dimensioning of encased columns. Proceedings of the 4th Asian Regional Conference
on Geosynthetics, Shanghai. pp. 718-723

385
386
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Use of Stone Columns for liquefaction mitigation

Utilisation des colonnes ballastées pour la réduction de la liquéfaction
* 1
Z. BENSALEM , W. FRIKHA, M. BOUASSIDA
1
LR14ESO3 Ingénierie Géotechnique. Ecole Nationale d'Ingénieurs de Tunis, URIG, ENIT, BP 37, LE BELVEDERE
1002 TUNIS, TUNISIA
*
[Link]@[Link]

RÉSUMÉ –Les colonnes ballastées représentent une technique de renforcement des sols qui permet d’améliorer la
capacité portante et de réduire le tassement. Cette technique est largement adoptée pour réduire le potentiel de
liquéfaction. Dans le présent travail, une quinzaine d’étude de cas ont été traitées dans le but d’évaluer l’efficacité des
colonnes ballastées vis-à-vis de la réduction du risque de liquéfaction. Les résultats montrent que ce risque est
considérablement réduit suite à la mise en place des colonnes ballastées. Certaines recommandations pour le
dimensionnement vis-à-vis de la liquéfaction sont présentées.

ABSTRACT– Stone columnsare a ground improvement technique employed to increase the bearing capacity and to
reduce settlements of weak soils. Installation of stone column is widely adopted as a liquefaction [Link] this
paper, fifteen case studies are considered to assess the effectiveness of the stone column technique to mitigate against
[Link] results show that the potential of liquefaction is considerably reduced after stone columns installation.
Design recommendations for liquefaction analysis of reinforced soils are presented.

1 INTRODUCTION

Liquefaction is the most hazardous damage during an earthquake. Liquefaction damage induces sand boils, excessive
settlement, lateral spread, landslides and slope failure, loss of bearing capacity, etc.
Seed and Idriss (1971) developed the basic "simplified procedure" for liquefaction evaluation which involves the
comparison between the Cyclic Shear Stress Ratio (CSR) caused by the design earthquake with the capacity of the soil
to liquefaction resistance expressed in terms of Cyclic Resistance Ratio (CRR). The simplified procedure was
developed from empirical evaluations of field observations and field and laboratory test data. This procedure has been
modified and improved based on in situ tests, such as those developed by Seed (1979), Seed and Idriss (1982), Seed et
al. (1985),Youd and Idriss (1997), Youd et al. (2001), Cetin et al. (2004), Idriss and Boulanger (2008).
To reduce the potential risk of liquefaction, various ground improvement methods are used including densification,
reinforcement, grouting/mixing and drainage. Installation of granular columns like gravel drains, granular piles and
stone columns is the most widely adopted method for liquefaction mitigation (Mitchell and Wentz, 1991). Adalier and
Elgamal (2004) reviewed the current state of the stone column method as a liquefaction countermeasure. In fact,
installation of stone column by vibro replacement provides drainage by reducing excess pore pressures generated by
cyclic loading and by accelerating its dissipation (Baez 1995,Mitchell et al. 1995) and increases the density of the
surrounding soil which increase the Cyclic Resistance Ratio CRR. The introduction of stiffer material (stone) can
potentially carry higher stress levels and thereby reduces stresses in the liquefiable soil.
The present paper examines the effectiveness of stone column installation in reducing the liquefaction potential. Fifteen
case studies based on stone column reinforcement project are considered, where SPT and CPT tests were performed
before and after stone column installation. The Cyclic Resistance Ratio CRR is then evaluated based on these in-situ
test results using the Idriss and Boulanger’s (2008) procedure.

2 SIMPLIFIED PROCEDURE FOR LIQUEFACTION EVALUATION

Seed and Idriss (1971) proposed a "simplified procedure" for liquefaction evaluation based on empirical evaluations of
field observations and field and laboratory test data. Seed and Idriss procedure involves comparing the Cyclic Shear
Stress Ratio (CSR) caused by the design earthquake with the capacity of the soil against liquefaction expressed in terms
of the Cyclic Resistance Ratio (CRR). The factor of safety (FS) against liquefaction is defined:

CRR
FS  (1)
CSR
It is argued that a FS less than 1.0 means that the soil is susceptible to the initiation of liquefaction.
Seed and Idriss (1971) formulated the following equation to estimate the Cyclic Shear Stress Ratio (CSR) induced by
earthquake ground motions, at a depth z below the ground surface:

387
 ave    a max 
CSR   0,[Link] . v .  (2)
 v' 0  '   g 
 v 
where amax is the peak horizontal acceleration at ground surface generated by the earthquake, g is the acceleration due to
gravity,  v and  'v are total and effective vertical overburden stresses respectively and rd is the shear stress reduction
factor evaluated using the Idriss and Boulanger (2008) recommendation. El Ouni et al. (2009) evaluated the variation of
CSR given by Eq (2) of vibrocompacted loose sand.
Empirical methods based on measurements of in situ soil strength and observations of field performance in previous
earthquakes have been developed to predict soil liquefaction resistance (CRR):

CRR  MSF.K  .CRRM 7.5,1atm (3)

where CRRM=7.5,1atm is the cyclic resistance ratio of the soil adjusted to 1 atmosphere of effective overburden pressure
for the earthquake magnitude M=7.5, MSF is the Magnitude Scaling Factor for earthquakes of magnitude other than 7.5
and Kσ is the overburden correction factor for the overburden stresses.
The present paper focuses on Idriss and Boulanger (2008) procedure for evaluating soil liquefaction resistance based on
SPT and CPT results. Idriss and Boulanger (2008) recommended a correlation between the CRRM=7.5,1 atm and the SPT
and CPT penetration resistance , respectively, via the following expressions:
N 2 3
 1,60,cs  N1,60,cs   N1,60,cs


 
 14.1   126    23.6  
     
CRR M 7.5,1atm  exp (4)
 4 
   N1,60,cs   2.8 
  25.4  
   
q 2 3
 c1Ncs   q c1Ncs    q c1Ncs  
 540    
 67   80  
CRR M 7.5,1atm  exp (5)
 q 
4 
   c1Ncs   3 
 
  114  
where N1,60,cs is the corrected blow count adjusted to a clean sand equivalent and qc1Ncsis the equivalent clean-sand value
of the corrected tip resistance.

3 ANALYSIS OF LIQUEFACTION POTENTIAL OF PRE- AND POST- STONE COLUMN INSTALLATION

Stone columns installation is commonly adopted for liquefaction mitigation and has proven its effectiveness in many
cases studies (Mitchell and Wentz, 1991).Stone columns help to mitigate liquefaction through one or more of these
ways (Baez 1995; Adalier and Elgamal,2004): i) the installation process of stone columns densifies the surrounding
soil, which increases the liquefaction resistance (CRR), ii) the stone columns act as drains due to higher permeability
than that of liquefiable soil and allow the rapid dissipation of excess pore pressure from the soil and iii) the stone
columns act as reinforcing inclusions due to their higher stiffness than that of surrounding soil.

3.1 Cases studies

In order to study the liquefaction potential of reinforced soil with stone column, several case studies using project data
are considered. Site investigation consisted of in situ testing including SPT and CPT tests performed before and after
stone column installation. Table 1 provides a summary of these case studies showing different design parameters of
stone column such as treatment depth, spacing, diameter and area replacement ratioη.An earthquake magnitude of 7 and
a maximum ground surface acceleration amax=0.20 g are used in liquefaction analyses and liquefiable layers soils
properties were given.

3.2 Analysis of Pre- and Post- Liquefaction Potential using SPT and CPT Correlations

Comparative results of N1,60,cs and qc1N,cs before and after stone column installation were reported in figures 2 and 3
respectively. A significant increase in penetration resistance was observed after stone column installation, which
confirmed the densification effect and reach in some cases more than five times the initial resistance. For these different
case studies, varied degrees of densification were observed. In fact, the degree of resulting densification is a function of
soil type, fines content, soil plasticity, pre-densification relative density, vibrator type, stone shape, stone column area,
and spacing between stone columns (Adalier and Elgamal 2004). Such measurable densification increased the
liquefaction resistance.

388
Table [Link] studies of stone column reinforcement

GWT Treatment Spacing Diameter

Project Soil Site Grid η (%) Source
depth (m) depth (m) (m) (m)
Loose to
medium
Christchurch, Mahoney and
dense sandy 1.5 to 1.8 10.5 Triangular 1.65 0.6 12
New Zealand Kupec(2014)
and silty-
sand
Hinckley Interbedded
12.2 to Rollins et
Drive ,Utah, silts and ~2.8 Triangular 2.44 0.91 14
13.7 al.(2012)
USA. sands
Debottlenecki
Silty Fine
ng project 1.8 to 0.8 to Basarkar et
Sand–Silty 2.8 7 to 8 Triangular 25
,New 2.5 0.85 al. (2009)
Clay deposits
Delhi,India
Shepard lane, Silty sand
Rollins et
Farmington, and sandy - 7.5 Triangular 2 1.1 27
al.(2006)
Utah silt
Marina Del Inter-bedded Shenthan
~2 - - 2.4 0.9 11
Rey, CA silty soil (2005)
Webster
Lee et al.
Street Tube, Loose fine
~0 15 Triangular 2.4 0.57 ~16 (2005)
California, sand
USA
Earth
Chen and
embankment
Silty sand ~2.25 5.1 to 5.7 Triangular 2.4 1 17 Bailey
project,wester
(2004)
n Washington
Full-scale
Weaver et al.
lateral load 0.9
Loose sand 1.5 Square 2.4 11 (2004)
tests,
California.
Steel oil
storage 12.5- Duzceer
Silty sand 1.5 15 Square 2.2 to2.5 1
tanks,Georgia 16.5 (2003)
.
East Anglia
1.5 and Slocombe et
trial Clayey sand ~0 - - 0.9 -
2.1 al. (2000)
Borcelik
Loose
Gemlik Cold Durgunoglu,
medium ~0 12 Triangular 1.5 0.5 10
Steel Rolling et al.,(1995)
dense sand
Mill,Turkey
Silty and Blanchard
Oxnard,
sandy 2.74 6 Square 2.74 0.9 ~10 and Clements
California
alluvium (1993)
Monterey
Baez and
Sports Center 1.8, 2.4
- - - - - - Martin
in Monterey, and 3
(1992)
CA
Port of zarzis, Very loose Gambin.
0.8 to 1 6.5 Triangular 2 1.2 33.3
Tunisia silty sand (1992)
Steel Creek
Dam Slightly Keller et al.
- - Triangular 1.8 0.9 22.6
California, clayey sand (1987)
USA

389
The Cyclic Resistance Ratio (CRR) profiles were then calculated based on N1,60,cs and qc1N,cs values deduced from measured
values before and after stone column installation. Figure 4 and 5 show Cyclic Resistance Ratio (CRR) plotted as a function of
N1,60,cs and qc1N,csrespectively. The obtained results indicate that stone column reinforcement significantly increases the Cyclic
Resistance Ratio (CRR). In fact, the increase in the degree of soil densification leads to an increase of the cyclic liquefaction
resistance.

Figure 2. Pre and Post reinforcement corrected Figure [Link] and Post reinforcement qc1N,csvalues
N1,60,csvalues

Figure [Link]- and Post reinforcement CRRvalues based Figure [Link]- and Post reinforcement CRRvalues based
on SPT-correlations on CPT-correlations

The Cyclic Shear Stress Ratio (CSR) of each site was calculated and then adjusted for the reference values of M=7.5 and  v'
=1atm as:

1 1
CSR*  CSR. . (12)
MSF K

The resulting CSR* values were plotted against the values of N1,60,cs and q c1N,cs in Figure 6 and Figure 7 respectively withthe
boundary curve (i.e the Cyclic Resistance Ratio (CRR) adjusted to M=7.5 and  v' =1atm). From these figures, it was observed
a significant increase of N1,60,cs and q C1N,cs values after stone column installation. These figures show that the majority of
plotted data recorded beforereinforcement were located to the left of the boundary curve and are classified as liquefiable soils.
After stone column installation, the data points were, mainly, found on the right of the boundary curve where no liquefaction
occurs. It should be noted that in some cases, before stone column, potential risk of liquefaction was not considerable,
390
however after reinforcement, it decreases. It is also noted, from figure 7, that the liquefaction risk from CPT data still remains
after reinforcement in some cases.

Figure [Link] between CSR* and N1,60,cs [Link] between CSR* and qc1N,cs

Figure [Link] before and after stone column reinforcement Figure [Link] before and after stone column reinforcement
based on SPT-correlations based on CPT

Figures 8 and 9 illustrate a comparison of the safety of factor FS calculated before and after stone column reinforcement
based on SPT and CPT results, respectively. After stone column installation, the most of FS values are greater than 1.0 which
indicates that liquefaction is prevented in the improved soil under the design seismic loading. In order to make comparison
between the estimation of FS based on SPT and CPT results, the field tests should be performed for the same project and at the
same location.

4 CONCLUSION

In the present analysis, the effectiveness of stone column in reducing liquefaction potential, as mentioned in literature, is
confirmed through the increase of liquefaction resistance (CRR). Based on fifteen case studies, comparison between SPT and
CPT values measured before and after stone column installation confirmed the densification effect of this ground improvement
technique. The corresponding Cyclic Resistance Ratio (CRR)was calculated and results show that after stone column
installation liquefaction resistance increases considerably. It can be concluded that field tests performed before and after stone
column installation are recommended for liquefaction analysis and improvement effect should be considered for the design of
stone column to mitigate liquefaction.

391
REFERENCES

Adalier, K., and Elgamal, A. (2004). Mitigation of liquefaction and associated ground deformations by stone columns. Engineering
Geology, 72(3-4), 275–291.
Baez, J. I. (1995). A Design Model for the Reduction of Soil Liquefaction by Vibro-Stone Columns. Ph.D. Dissertation,University of
Southern California,Los Angeles, CA.
Baez, J. I., and Martin, G. (1992). Liquefaction observation during installation of stone columns using the vibro-replacement technique.
Ground Modification, Geotechnical News, 41–44.
Basarkar, S. S., Panse, V., and Wankhade, R. . (2009). Ground strengthening by vibro-stone columns-a case study. Indian Geotechnical
Conference IGC, 446–450.
Blanchard, J. D., and Clements, K. . (1993). Site improvements with stone columns in stratified silty soils. Third International Conference
on Case Histories in Geotechnical Engineering,St. Louis, Missouri, 1027–1033.
Cetin, K. O., and Seed R.B., Der Kiureghian A., Tokimatsu K., Harder L.F. Jr, Kayen R.E., M. E. S. (2004). Standard penetration test-based
probabilistic and deterministic assessment of seismic soil liquefaction potential. Journal of Geotechnical and Geoenvironmental
Engineering, ASCE,130(12), 1314–1340.
Chen, B. S., and Bailey, M. J. (2004). Lessons learned from a stone column test program in glacial deposits. GeoSupport 2004: Drilled
Shafts, Micropiling, Deep Mixing, Remedial Methods, and Specialty Foundation systems,ASCE,Reston,Va, 508–519.
Durgunoglu, H. T., Kulac, H. ., Nur, O., Ikiz, S., Akbal, O., and Olgun, C. . (1995). A case study on determination of soil improvement
realization using CPT. Proceedings International Symposium on Cone Penetration Testinds,Vol.2,Swedish Geoechnical Society
Report 3:95, 2, 441–446.
Duzceer, R. (2003). Ground Improvement of Oil Storage Tanks Using Stone Columns. Proceedings of the 12th Pan American Conference in
Soil Mechanics and Foundation Engineering, Cambridge, MA, 1681–1686.
El Ouni, M. R., Bouassida, M., and Braja, M. . (2009). Vibrocompaction improvement of Tunisian liquefiable sands. Proceedings of the 17
Th International Conference on Soil Mechanics and Geotechnical engineering,Alexandria,Egypt, 1–4.
Gambin, M. P. (1992). Reducing Liquefaction Potential by Deep Soil Improvement. Chap. II-4 in “Recent Advances in Earthquake
Engineering and Structural Dynamic.”
Idriss, I. M., and Boulanger, R. W. (2008). Soil Liquefaction During Earthquakes. Monograph MNO-12, Earthquake Engineering Research
Institute, Oakland, CA, 261.
Keller, T. O., Castro, G., and Rogers, J. H. (1987). Steel Creek Dam foundation [Link] Improvement-A Ten Year Update.
ASCE,Geotechnical Special Publication,No12, 136–166.
Lee, T. S., Dash, U., and Anderson, R. (2005). Pipe Pile Stone Columns at Webster Street Tube, Oakland, California. Innovations in
Grouting and Soil Improvement, 1–15.
Mahoney, D. P., and Kupec, J. (2014). Stone column ground improvement field trial : A Christchurch case study. NZSEE Conference.
Mitchell, J. K., Baxter, C. D. P.,and Munson, T. C. (1995). Performance of improved ground during earthquakes. Proc., Soil Improvements
for Earthquake Hazard Mitigation, Geotechnical Special Publication No. 49,ASCE, 1–36.
Mitchell, J. K., and Wentz, F. K. (1991). Performance of improved ground during the Loma Prieta earthquake. Report No. EERC91/12,
Earthquake Engineering Research Center, University of California, Berkeley.
Rollins, K., Price, B. E., Dibb, E., and Higbee, J. B. (2006). Liquefaction Mitigation of Silty Sands in Utah Using Stone Columns with Wick
Drains. Ground Modification and Seismic Mitigation, 343–348.
Rollins, K., Wright, A., Sjoblom, D., White, N., and Lange, C. (2012). Evaluation of liquefaction mitigation with stone columns in
interbedded silts and sands. Proc. 4th Intl. Conf. on Geotechnical and Geophysical Site Characterization, Taylor and Francis Group,
London,2, 1469–1475.
Seed, H. B. (1979). Soil liquefaction and cyclic mobility evaluation for level ground during earthquakes. Journal of the Geotechnical
Engineering Division, 105(2), 201–255.
Seed, H. B., and Idriss, I. M. (1971). Simplified procedure for evaluating soil liquefaction potential. Soil Mechanics and Foundations,Div.,
ASCE, 97:SM9, 1249–1273.
Seed, H. B., and Idriss, I. M. (1982). Ground Motions and Soil Liquefaction during Earthquakes. Earthquake Engineering Research
Institute, Oakland, California, 134.
Seed, H. B., Tokimatsu, K., Harder, L. F., and Chung, R. (1985). Influence of SPT procedures in soil liquefaction resistance evaluations.
Geotechnical Engineering, 111(12), 1425–1445.
Shenthan, T. (2005). Liquefaction mitigation in silty soils using composite stone column. Ph.D. Dissertation, University at Buffalo, Buffalo,
NY.
Slocombe, B. C., Bell, A. L., and Baez, J. I. (2000). The densification of granular soil using Vibro methods. Geotechnique, 50(6), 715–725.
Weaver, T., Ashford, S., and Rollins, K. (2004). Performance and analysis of a laterally loaded pile in stone column improved ground. 13th
World Conf. on Earthquake Engineering, EERI, Vancouver,B.C,Canada.
Youd, T. L., Idriss I .M, Andrus R. D., Arango I., C. G., Christian J. T, Dobry R., Finn W. D. L., H. L. F. J., Hynes M. E., Ishihara K.,
Koester J. P., L. S. S. C., Marcuson W. F III, Martin G. R., Mitchell J. K., M., Y., Power M. S., Robertson P. K., Seed R. B., S. I. K.,
H. (2001). Liquefaction resistance of soils: Summary report from the 1996 NCEER and 1998 NCEER/NSF workshops on evaluation
of liquefaction resistance of soils. Journal of Geotechnical and Geoenvironmental Engineering, 127(10), 817–833.
Youd, T. L., and Idriss, I. M. (1997). NCEER Workshop Proc. on Evaluation of Liquefaction Resistance of Soils, Natl. Center for
Earthquake Engineering Research (NCEER), State Univ. of New York at Buffalo, NY.

392
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Study on effect of Sisal fiber-reinforcement on Tensile strength of clay soil using newly
developed tensile strength measuring apparatus.
Etude de l'effet du renforcement d’un sol argileux par les fibres de Sisal sur la résistance à la traction en
utilisant un appareil de traction directe nouvellement développé.
Chebbi Meriem1,Jamei Mehrez1,Guiras Houda1

Laboratory of Civil Engineering, National School of Engineers of Tunis, Tunisia

Corresponding author: Chebbimeriemenit2005@[Link].

ABSTRACT- The tensile strength of soil is an important mechanical factor in the design of slopes, dams, embankments and
hydraulic barriers, where tensile cracks are possible to occur. In fact, compacted soils used in geotechnical constructions such
as dams and clayey liners in waste containment facilities can suffer from cracking due to tensile failure. In order to increase
soil tensile strength, discrete fiber reinforcement technique was proposed. An innovative tensile apparatus was developed to
determine the tensile strength characteristics of fiber reinforced soil. The effects of fiber content, dry density and water content
on the tensile strength were studied. The results indicate that the developed test apparatus was applicable in determining
tensile strength of soils. Fiber inclusion can significantly increase soil tensile strength and soil tensile failure ductility. The
tensile strength basically increases with increasing fiber content. Also, the tensile strength of fiber reinforced soil increases
with increasing dry density and decreases with decreasing water content. For example, the tensile strength at a dry density of
1.7 g/cm3 is 3 times higher than that at 1.3 g/cm3. It decreases by 67 % as the water content increases from 18 % to 45%. Also,
it is observed that the tensile strength of fiber reinforced soil is dominated by fiber pull-out resistance, depending on the
interfacial mechanical interaction between fiber surface and soil matrix.

RESUME- La résistance à la traction du sol est un facteur mécanique important dans la conception des barrages, des digues et
des barrières hydrauliques, où des fissures de traction sont possibles à se produire. En effet, les sols compactés utilisés dans les
constructions géotechniques tels que les barrages et les revêtements argileux dans les installations de confinement des déchets
peuvent souffrir de fissuration due à la rupture en traction. Dans le but d'améliorer la résistance à la traction du sol, la
technique de renforcement par des fibres a été proposée. Un appareil de traction innovant a été développé pour déterminer les
caractéristiques de résistance à la traction du sol renforcé. Les effets de la teneur en fibres, la densité sèche et la teneur en eau
sur la résistance à la traction ont été étudiés. Les résultats indiquent que l'appareil développé est applicable à la détermination
de la résistance à la traction du sol. L’inclusion des fibres peut augmenter considérablement la résistance à la traction du sol
ainsi que sa ductilité. Aussi, la résistance à la traction du sol augmente avec l'augmentation de la densité sèche et diminue avec
la diminution de la teneur en eau. En outre, on observe que la résistance à la traction de sol renforcé par des fibres est dominée
par la résistance à l'arrachement des fibres, dépendant de l’interface entre la surface de la fibre et de la matrice du sol.

1 INTRODUCTION

Although it’s relatively low value compared with compressive or shear strength, the tensile strength of soil is considered as an
important mechanical parameter which is useful in many geotechnical engineering application. Indeed, the tensile strength of
soil is difficult to be precisely measured due to the lack of laboratory techniques. In recent years, increasing attention has been
paid to the investigation of soil tensile characteristics, and to the development of new methods of improving and measuring
soil tensile strength (Trabelsi et al., 2010, Chen et al., 2009; Hu et al., 2009).
Discrete fiber reinforcement is a recently developed technique to improve soil mechanical behavior (Yetimoglu Ahmad et al.,
2010; Consoli et al.,2011). In comparison with conventional geo-synthetics (strips,geotextile, geogrid, etc.), the advantages of
using discrete fiber are as follows: (1) The discrete fibers are simply added and mixed randomly with soil, like in mixing soil
with cement, lime, or other additives. (2) Randomly distributed fibers limit potential planes of weakness that can develop in
the direction parallel to the conventionally oriented reinforcement. (3) The inclusion of fiber only changes the physical
properties of soil and has no impact on the environment. For these reasons, researchers have shown an increasing interest in
mechanical behaviors of fiber reinforced soils. Although the mechanical behaviors such as shear strength, compressive
strength and bearing capacity of fiber reinforced soil have been extensively studied in the past few decades, the effect of fiber
reinforcement on soil tensile strength behavior has not been well understood yet. In this study, discrete fiber reinforcement was
proposed to improve soil tensile strength. An innovative tensile apparatus was developed for this purpose. Using this
apparatus, a series of direct tensile tests were conducted on fiber reinforced soil in order to determine the tensile strength
characteristics. The influences of fiber content, water content and dry density on tensile strength were analyzed and associated
mechanisms were discussed.

393
2 DIRECT TENSILE TESTS

2.1 Materials and sample preparation

2.1.1 Characteristics of the Fibers

The sisal fibers used are natural fibers. They are characterized by a density of 9.4 kN/m3, tensile strength of 510 MPa,
elastic modulus of 11.8 GPa, and failure strain of 5.6% (Morel et al. 2000). Observation under an optical microscope showed
that the diameters of the fibers ranged between 0.1 and 0.35 mm. However, most of the fibers tested had a diameter between
0.12 and 0.15 mm, which gave an average diameter, D, of the fibers equal to 0.15 mm. The length, L, of the short fibers was
fixed at 30 mm (Figure 1a).Electronic microscopic photography showed that the surface of the fiber was sufficiently smooth
(Figure 1b).

Figure 1: (a) Group of sisal fibers and (b) microscopic view (x100) of a single fiber (Mehrez J et al, 2013).

2.1.2 Clayey Soil Properties

The used soil was sampled from the region of Thibar (Beja), west of Tunisia. Physical properties of this soil such consistency
limits (liquid limit, WL, plastic limit, WP, and plasticity index, Ip) were obtained by following the guide lines provided by NF P 94-
056 andNF P 94 051, respectively. The specific gravity, G, was determined by employing an Ultra-pycnometer. The results obtained
from these investigations are presented in Table 1. The collected soil was first dried at 90°C, broken into pieces, and passed through a
2-mm sieve. The subsequent processing preparation consisted of the following steps:
 Crushing of the clumps of dry clay;
 Passage through a sieve with a 2mm mesh;
 Addition of water to reach the optimum moisture condition (OMC)+2% and
 Storage of the samples in hermetically sealed plastic bags for homogenization of the water content.

2.1.3 Preparation of the Fiber-Soil Mixtures

For the reinforced soil, the sisal fibers were hand mixed with the dried clay (Figure.2 (a)) and then water was added
(Figure. 2(b)). To homogenize the water content (Figure 2(c)), the wet soil was hermetically sealed (Figure 2(d)) for a
minimum of 2 days.

Table 1: Properties of soil sample

Specific gravity Consistency limit Compaction study

Liquid limit (%) Plastic limit (%) Plasticity index W (%) Dry density(g/cm3)

2.65 60 33 27 23 1.69

Figure 2: Steps for preparing the specimens: (a) dried soil1fibers; (b) adding water; (c) mixture; (d) time cure in hermetically sealed plastic
bag. (Jamei et al, 2013).

394
2.2 Test method and device
Soil tensile strength can be measured using many methods, which mainly can be divided into two categories: indirect and
direct methods. For the former one, the tensile strength is calculated according to some empirical correlations. The common
indirect tensile test methods include Brazilian tensile test, flexure beam test, hollow cylinder test, double punch test, etc.
(Ghosh and Subbarao, 2006; Kim et al., 2012). Since the determination of tensile strength by indirect method is based on a
series of theoretical assumptions, the calculated value usually cannot directly reflect the intrinsic mechanisms of soil tensile
behaviors.
Moreover, the measured tensile strength significantly depends on the employed test methods. Despite these limitations, many
researchers still favored it for its great testing convenience. The direct test method is generally considered as the most reliable
approach to measure tensile strength. An increasing tensile load is directly applied to the two ends of the specimens until
tensile failure occurs. The monitored maximum tensile load issued to determine the tensile strength. Generally, the obtained
tensile strength from direct tensile test is of high accuracy. However, it has been accepted that direct tensile tests are difficult
to perform due to problems related to specimen preparation and test procedures. Thereby, this method needs development of
new test setups and methodologies (Tang et al, 2010 a; Nahlawi et al., 2004).
In this context, direct test method was employed. In order to facilitate the test procedures, a special compaction and tensile
mould was designed to prepare dogbone specimens (Figure 4).The tensile mould consists of two halves of the split mould. A
neck at the centre of the specimen is formed to reduce the specimen width from 35 cm to 13cm, allowing the specimen failure
to occur at the centre section during tensile [Link] compaction, the required quantity of soil mixture was put in the
compaction mould and statically compacted to the target height at different dry densities γd=1.1g/cm3, 1.2g/cm3, 1.3g/cm3,
1.4g/cm3and 1.5g/cm3.

Figure 3:Newly developed tensile strength measuring apparatus Figure 4: Tensile mould
.
Direct tensile tests were then conducted on the prepared dogbone specimens, using a newly developed apparatus as shown in
Figure 3. A tensile load is applied on the movable part of the mould. This tensile load increased with the increase in
displacement of the movable part of tensile mould. The evolutions of tensile load and associated displacement were recorded
by displacement transducers (with the resolution of 0.01mm).
The tensile strength σt was then calculated by dividing the maximum tensile load Tmax by the cross sectional area A (Lxh) at
the specimen’s neck .The tensile strength σt can then be written as:

σt= Tmax/A

L
h

(a) (b) (c)

Figure 5:Compacted soil - (a) planeview of compacted soil, (b) side view of soil, (c): soil after test

3 RESULTS

Using the self-developed tensile test apparatus, direct tensile tests were performed on many prepared specimens. The
obtained tensile strengths of various tests were presented. In the following sections, the tensile characteristics of fiber
reinforced soil and the effects of fiber content, water content and dry density on the tensile strength will be discussed.

395
3.1. Tensile curves

Figure 6 shows the typical tensile curves (tensile strength versus displacement) of the specimens with different fiber
contents. It can be seen that the tensile strength increases monotonously with increasing displacement before reaching the peak
value, where tensile failure occurs. After that, for the specimen without fiber reinforcement, the tensile load drops to zero
suddenly, indicating brittle tensile failure of the specimen. For all the fiber reinforced specimens similar to the unreinforced
specimen, there is a reduction in the applied load once the peak value was attained. However, due to the presence of fibers, a
residual tensile load is maintained after failure, suggesting that the fiber inclusion is effective in improving soil failure
ductility. This is because, on formation of tension cracks, further overall displacement of the specimen is concentrated on the
elastic/plastic extension of the fibers across the opening and the bond slippage of the fiber anchorage in either side of the
opening.

Figure 6:The typical tensile curves of specimens with different fiber contents.

3.2. Effect of fiber content

Figure 7 shows changes in tensile strength with different fiber contents. For the specimen without fiber inclusion, the
measured tensile strength is 32kPa. Fiber inclusion can improve soil tensile strength, and the tensile strength increases with the
increase of fiber content. For instance, the tensile strength increased by 112 %, from 32 kPa to 68 kPa, when the fiber content
increased from 0% to 0.5%. This is mainly because when the specimen is subjected to tensile load, the slide of fibers in soil
matrix is restricted by the interfacial mechanical interactions between fiber surface and soil particles. Consequently, fibers are
capable of sharing some tensile load in soil matrix, and therefore increase the tensile strength. With increasing fiber content,
the number of fibers per unit volume increased. The acquired reinforcement benefit of fibers to the tensile strength was
therefore more pronounced.
3.3. Effect of dry density
In Figure 8, it can be observed that the tensile strength of fiber reinforced soil increases with increasing dry density. When
dry density increased from 1.1 g/cm3 to 1.7 g/cm3, the related tensile strength was increased from 9 kPa to 56 kPa. It is
believed that the dry density affects the contact conditions of soil particles. A higher dry density leads to more contacts
between soil particles. This, in turn, leads to a greater bonding force between particles and hence greater tensile strength.
Moreover, the increase of dry density also gives rise to an increase of fiber/soil interfacial contact area, which can improve the
interfacial mechanical interactions and increase the interfacial shear strength. As a result, the fiber reinforcement benefit might
be enhanced as compared to lose compacted specimen.

Figure 7: Changes of soil tensile strength with fiber [Link] 8: Changes of soil tensile strength with dry density.

396
3.4. Effect of water content

In order to understand the influence of water content on soil tensile strength, four groups of specimens with different water
contents, i.e. 18%, 23%, 33% and 40%, were prepared and subjected to direct tensile test. The obtained tensile strengths were
presented in Figures 09 (a) and 09 (b). It can be seen that the increase of water content resulted in a decrease of tensile
strength. When water content was increased from 18% to 40%, the tensile strength was decreased from 49 to 5 kPa. This
observation can be explained from the following two aspects:(1) Cohesion and suction decrease with increasing water content,
thus the bonds between soil particles are weakened; (2) The increase of water content may also weaken the interfacial
mechanical interactions between fiber and soil matrix, and decrease the capability of fiber to bear the tensile load.

Figure 9-a Figure 9-b

Figure 9-a, 9b: Changes of soil tensile strength with water content (Fiber content=0, 1% Dry density=1,7g/cm3).

4. CONCLUSIONS
Direct tensile tests were conducted on discrete fiber reinforced soil using the newly developed apparatus. The effects of fiber
content, water content, and dry density on soil tensile strength were considered and the obtained results were discussed. The
following conclusions can be drawn:
(1) The developed 8-shaped tensile mould constitutes a simple means of fixing soil specimen during tensile test. The proposed
method and the developed apparatus provide a convenient way for measuring the tensile strength of geo-materials.
(2) Very small dosage of fiber inclusion can significantly enhance soil tensile strength. The tensile strength basically increases
with increasing fiber content. As the fiber content increases from 0% to 0.2%, the tensile strength is increased by 65.7%.
Moreover, fiber reinforcement changes soil’s brittle tensile failure behavior to ductile tensile failure behavior. Due to the
presence of fibers, a residual tensile load was maintained after tensile failure.
It suggests that the fiber reinforcement is a favorable ground improvement technique, and has the potential to increase soil
cracking resistance and the stability of earth structures.
(3) Tensile strength of fiber reinforced soil increases with an increase in dry density. It is mainly because that a higher dry
density leads to more contacts between soil particles, and also gives rise to an increase of fiber/soil interfacial contact area.
Consequently, the fiber reinforcement benefit on tensile strength is improved.
(4) Tensile strength of fiber reinforced soil decreases with increase in water content. It is because the bonds between soil
particles and the interfacial mechanical interactions between fiber surface and soil matrix are weakened by adding water.

REFERENCES
Ahmad F, Bateni F. (2010) Performance evaluation of silty sand reinforced withfibers. Geotextiles and Geomembranes;28(1):93-9.
Chen YL,Wang M,(2009) Tensile and compressive strength tests on artificial frozen soft clay in Shanghai. Chinese Journal of Geotechnical
Engineering:31(7): 1046-51.
Consoli NC, Casagrande MDT.(2007) Performance of fiber-reinforced sand at large shears strains. Geotechnique; 57(9):751e6.
Ghosh A, Subbarao C.(2006) Tensile strength bearing ratio and slake durability of class F fly ash stabilized with lime and gypsum. Journal
of Materials in Civil Engineering; 18(1):18e27.
Houcem T,Mehrez J, Houda G, Hatem Z, Enrique R, Sebastia O.(2010) Some investigations about the tensile strength and the desiccation
process of unsaturated clay. "European physical [Link] topics", vol. 6, p. 12005-p1-12005-p8.
Hu HJ, Jiang MJ, Zhao T, Peng JB, Li H. (2010) Effects of specimen-preparing methods on tensile strength of remolded loess. Rock and
Soil Mechanics; 30(Suppl. 2):196e9.
Kim TH, Kim TH, Kang GC, Ge L.(2010) Factors influencing crack induced tensile strength of compacted soil. Journal of materials in civil
engineering; 24(3):315-20
Jamei M, Villard P, Guiras, H (2013) Shear failure criterion based on experimental and modeling results for fiber reinforced clay. American
society of civil engineers.
Morel JC,Ghavami K, Ali M. (2000). Theoretical and Experimental Analysis of Composite Soil Blocks Reinforced with Sisal Fibers
Subjected to [Link] international;13(2).
Nahlawi H, Chakrabarti S, (2004) A direct tensile strength testing method for unsaturated geomaterials. Geotechnical Testing Journal;
27(4):1e6.
397
398
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Experimental study of Tunis soft soil improved by deep mixing column

Etude expérimentale de la vase de Tunis renforcée par colonnes deep mixing

Wissem FRIKHA, Habib ZARGAYOUNA, Samia BOUSSETTA, Mounir BOUASSIDA

Université de Tunis El Manar/École Nationale d’Ingénieurs de Tunis/05-UR-11-06, BP 37 Le Belvédère, 1002 Tunis, Tunisia

ABSTRACT
Tunis soft clay belongs to the category of problematic soils because of its weak strength characteristics, high compressibility
and high content of organic matter (22%). Deep Mixing Column (DMC) presents a well-developed technique to improve the
properties of Tunis soft soils to increase the bearing capacity and to reduce settlement. This paper describes a laboratory
experimental program performed to assess the method effectiveness in improving the shear strength of Tunis soft soil. The two
main objectives of this research are to evaluate the effect of reinforcement by DMC on the behavior of the Tunis soft soil and
to study the influence of some parameters on its strength characteristics. Five parameters are studied from the performed tests:
i) curing time: CT=7, 14, 21 and 28 days; ii) injection pressure: IP=50,100 and 150 kPa; (iii) water-cement ratio w/C=0.7, 1
and 1.2; (iv) rotation rate V=15, 30 and 40 rpm and (v) consolidation pressure: 3=100,150 and 200 kPa.

1. INTRODUCTION

The development of infrastructure in the urban and metropolitan zones, in many countries, leads to the construction of
numerous projects on deposits of soft soil. The solution to overcome the difficulties of construction on problematic
compressible soils is to install deep foundations or to improve the soil properties. Multiple soil improvement techniques can be
used such as stone columns, grouting, thermal recompression and deep mixing columns. Nowadays, The technique of deep
mixing (DM) for soft clay stabilization has been widely used for its economic and timeliness efficiency. Deep mixing is a soil
modification technique in which the soil is mixed in-situ using rotating blades with bender (such as cement, lime, fly ash, etc.),
chemical or biological agents to improve the engineering and environmental properties of soft or contaminated ground
(Porbaha, 1998).
The essential feature of deep mixing column (DMC) technique is that columns of stabilized material are formed by mixing the
soft soil in place with a ‘binder’ and the interaction of the binder with the soft soil. There are two methods of deep stabilization
namely: dry deep mixing and wet deep mixing. The dry and wet methods of deep mixing correspond to the incorporation of
binder in dry form and wet deep mixing respectively. Deep mixing is applicable over a wide range of soil types including
clays, clayey silts, sands, sandy silts, organic clays and peat (Ahnberg et al. 1995; Porbaha, 1998; Japanese Geotechnical
Society, 2000; Bruce, 2001).
Several laboratory studies has been conducted to simulate the deep soil mixing (DSM) procedures using various types of
binder such as cement, lime, gypsum, bottom ash, fly ash, bentonite, furnace slag (Al-Taaba et al. 1999; EuroSoilStab, 2002;
Shen et al. 2003). The most widely used binders for soft and expansive soils include lime and cement. It is important to
emphasize that the increase in binder content results in increased strengths, (Ahnberg et al. 1995; Asano et al. 1996; Babasaki
et al. 1996; Chen et al. 1996; EuroSoilStab 1997; Taki 2003 and Huat 2006). The characteristics of material used (soil and
stabilizer) and conditions of installation (mixing and curing) are the main factors that affect the behavior of lime/cement
columns.
This paper details a laboratory experimental program performed to assess the method’s effectiveness in improving the shear
strength of Tunis soft soil. The two main objectives of this research are to evaluate the effect of reinforcement by DMC on the
behavior of the Tunis soft soil and to study the influence of selected parameters ( i) Curing Time: CT; ii) Injection Pressure:
IP; (iii) ¨water-Cement ratio: w/C; (iv) rotation rate: V and (v) consolidation pressure: б3) on its strength improvement.

2. TESTED MATERIALS

1.1. Soil Characterization

The soil of Tunis city plateau (Tunisia) is composed of three important formations: an embankment (from 1 to 8 meters of
thick), a soft muddy complex called hereafter Tunis soft soil (from 30 to 70 meters depth) and sandy-clay that can be assumed
as rigid substratum (Touiti et al. 2009). Tunis soft clay belongs to the category of problematic soils because of its weak
strength characteristics high compressibility, high content of organic matter (22%). The recorded sensitivity, defined as the
ratio of the undistributed shear strength to remold, for Tunis soft soil can reach 20 (Touiti et al. 2009). Tunis soft soil is
classified as highly plastic silt with very low consistency. Since 1994, several laboratory studies have been conducted at the
National Engineering School of Tunis on reinforced Tunis soft soil, as example by sand columns (Bouassida, 1996).

399
The studied Tunis soft soil is extracted at 7.5 meters depth in the Avenue Bourguiba at Tunis City centre. Its natural water
content is 163 %, the liquid and plastic limits are in the order of 109 % and 37.4 % respectively, and the specific gravity is
2.65. The groundwater level is located at about 1.0 meter below the ground surface. The sample is extracted during the drilling
work for cast pile installation. The soft soil muddy complex is grey, has a characteristic smell and contains shell debris.. It is
noticed that the Tunis soft soil is made out of 85 % of particles of dimension smaller than 80 μm, and 68% of particles smaller
than 2 μm. According to the XP P 94-011 French standard, the classification of tested soil is highly plastic clay.

2.2 Materials Used for Soil Stabilization

The material used for the fabrication of DM column in this experiment isan ordinary artificial Portland cement produced from
JbelOeust company (BirM'cherga, Zaghouan, Tunisia), type CEM – II / A –L Class 32.5 – Tunisian Standard NT 47-01 (1992)
and European Standard (EN 197-1:2000).

3. EXPERIMENTAL PROGRAM
3.1. Specimen preparation

To obtain a homogeneous soil mixture, the first step consists in sieving the natural soft soil by the wet method through a 100
µm sieve opening. The sieved soft soil is then dried in open air up to165% water content is obtained (equal to 1.5 times the
soft soil liquid limit). The slurry sample is then subjected to an initial K0 consolidation. The consolidometer cell is a hollow
Plexiglas cylinder of 280 mm internal diameter and 190 mm height. By using a spoon, the mould is carefully filled by the
grout which is gently tamped to remove any penetrated air. The consolidometer is then filled by the slurry and the Tunis soft
soil specimen is formed. The remolded specimen is drained vertically by two porous metallic disks placed at the top and
bottom of the consolidometer. Silicone grease is spread on the internal contact surface between the remolded soil and the tube
to allow for an easy unmolding of the specimen. Once the predetermined height of the specimen is reached, the consolidation
loading cap is placed and the loading starts slowly to provide the desired consolidation stress. The consolidation of the
remolded Tunis soft soil specimen ends when the measured settlement remains constant. This procedure is adopted for all tests
during which consolidation stress increased progressively from a stress level of 5 kPa until 60 kPa. Every level of stress is
maintained constant for about one week until settlement stabilization. After this consolidation procedure, the specimen height
is almost 140 mm.

3.2. Used Equipment

Figure 1 shows the mock-up of experimental equipment established and used for laboratory DM column installation. As in
field, the laboratory model ensures both drilling and injection of cement grout. The drilling tool is composed of a copper
hollow tube of 10mm inner diameter and 8mm outer diameter (8/10) which can move in vertical (as descent and ascent) and
rotational directions and a specific drilling head. The drilling head tool (Figure 2) consists of two cutting inclined blades
lightly helical to facilitate lowering and insertion of the mixing tool and the recovery without gluing the ground. The head of
the tool is equipped with a screw-nut for cleaning the tool after the formation of the column. Such a design of the drilling head
tool significantly reduces the possibility of transported all mixing into the sample's top surface. The size and shape of the head
tool are taken after the model presented by Larsson (2003).

Figure 1. Laboratory equipment of DM column installation Figure 2. Installation of DM columns

400
The grout cement, placed initially in a specific transparent tank is injected into the specimen through the drill pile and the
drilling tool under pump control pressure. The tank of grout cement is a graduated transparent circular Plexiglas tube (diameter
of 30 mm, length of 650 mm and volume of 115 cm3). The pressure into the tank is applied by means of a compressor using a
regulator valve that enables the control and adjustment of the pressure injection of the cement grout. The tank is filled by
cement grout using a vacuum pump.
A step motor is placed for applying the rotational forces. The drilling pipe is fixed in the hollow axis of the wheel and, then,
can rotate at the same rate as for the wheel. The control gear offers different rotation rates, depending on the ratio of the
gearing. The rotation rates taken are equal to 40, 30 and 15 rpm. The drilling pipe is connected with a fixed nozzle to the tank
of grout cement. The link assembly is designed to provide the rotation of drilling pipe without rotating the pipe connected to
the tank of grout. This method allows a sealed flow of grout under pressure from the tank to the injection part of the drilling
tool. The assembly (motor, gears, drill tools, etc.) is fixed to a hydraulic cylinder allowing vertical translation using second
electrical step motor. The translation speed taken is 440 mm / min. The grout cement is prepared using a mechanical stirrer by
introducing adequate ratio of cement and water (w/C), few seconds before the beginning of tests to avoid segregation.

3.3 Installation of DM columns

The installation of the DM column resulted from the translation speed and the rotation rate of the drilling pipe. The mixing
time is between 20 and 25 seconds for all performed test.

a. Penetration of drilling tool

The penetration of the drilling (mixing) tool is performed during the first phase. The movement of mixing tool across the depth
of preconsolidated Tunis soft soil specimen is at a vertical translation speed of 440 mm/min and rotation rates of 15, 30 and 40
rpm. This result in an initial disaggregation of the tested soil.

b. Mixing and installation of DM column

During mixing, the grout is injected continuously into the pre-consolidated Tunis soft soil specimen to form DM columns.
During this installation process, the recovery speed and the rotation rate of drilling tools are similar to the penetration phase
(rotation rates of 15, 30 and 40 rpm and a translation speed of 440 mm/min). The rotation rates correspond to the order of
magnitude used by AFNOR NF EN 14679 standard- December 2005 (20-60 rpm).

c. Extraction of a specimen and curing

For all the considered curing times (7, 14, 21 and 28 days), and immediately after DM column installation, the specimens of
70 mm diameter and around 140 mm height are immersed in water at 20°[Link] design of deep mixing column usually
considers the estimated strength 28 days in post installation of the DM columns.

d. Triaxial shear tests (CU + u)

After the cure phase the reinforced specimen are subjected to undrained consolidated triaxial tests performed according to
AFNOR standard NF P 94-074 (CU+u). The considered rate of axial displacement during the shear phase is of 0.03 mm/mn.
Table 1 summarizes all the details of the performed experimental program that comprised fifteen triaxial tests grouped into
three series.
The tested samples are subjected to different stress paths: i) isotropic consolidation is applied under three confining stress:
100, 150 and 200 kPa, ii) shear loading applied under undrained condition, at constant controlled strain rate, with recorded
excess pore pressure. Five parameters are analyzed :i) Curing Time: CT=7, 14, 21 and 28 days; ii) Injection Pressure:
IP=50,100 and 150 kPa; (iii) ¨water-Cement ratio w/C=0.7, 1 and 1.2; (iv) “Rotation rate” V=15, 30 and 40 rpm and (v)
consolidation pressure: 3=100,150 and 200 kPa (Table 1).
The first series comprises CU+utriaxial tests performed on unreinforced specimens of Tunis soft soil (Tests numbers 1, 2 and
3) under consolidation stress 3 of 100, 150 and 200 kPa.
The second series consists of three triaxial tests performed on the Tunis soft soil reinforced with DM column under
consolidation stress of 3= 100, 150 and 200 kPa. These tests are done using constant values of: i) Curing time: CT=14 days,
ii) Injection pressure: IP=100 kPa, (iii) ¨water-Cement w/C=1, (iv) rotation rate V=30 rpm.
The third series is a parametric study composed from nine tests performed under consolidation stress б3= 150 kPa. The
reference test is number 5 from series 2 with CT=15d, IP=100 kPa, w/C=1, V=30 rpm and б3= 150 kPa. For tests 7 and 8,
injection pressure IP equals s 50 and 150 kPa respectively. For tests 9 and 10, the ratio w/C is consider equal 0.7 and 1.2
respectively. In the tests 11, 12 and 13, the Curing time CT is hold 7, 21 and 28 days. In the last tests (14 and 15) rotation rates
is taken equal V=15 and 40 rpm.

401
Table 1. Summary and characteristics of tests

3

Num
CT PI V
Name w/C

t
(days) (kPa) (rpm) (kPa) Repetition
1 S_3100 100 4

I 2 S_3150 150 3

3 S_3200 200 3

4 S_C_3100 14 100 1 30 100 2

II 5 S_C_3150 14 100 1 30 150 4

6 S_C_3200 14 100 1 30 200 3

7 S_C_PI50 14 50 1 30 150 2

8 S_C_PI150 14 150 1 30 150 2

9 S_C_E/C 0,7 14 100 0.7 30 150 3

III 10 S_C_E/C 1,2 14 100 1.2 30 150 3

11 S_C_TC7 7 100 1 30 150 3

12 S_C_TC21 21 100 1 30 150 2

13 S_C_TC28 28 100 1 30 150 3

14 S_C_V15 14 100 1 15 150 2

15 S_C_V40 14 100 1 40 150 3

4. INTERPRETATION OF RESULTS

The results of the triaxial tests are interpreted focusing on the effect of the different parameters mentioned above on the
undrained shear strength. For consolidated undrained triaxial tests (CU+u), the undrained shear strength is defined as half of
the peak of deviatoric stress (Cu = qmax/2)

4.1 Effect of consolidation stress

Figure 3 presents the evolution of the undrained shear strength (Cu) and the peak pore pressure, as a function of the
consolidation stress, obtained from the series 1 and series 2. For all cases, one can observe the quasi-linear evolution of the
undrained shear strength (kPa) and pore pressure (kPa) in function of consolidation stress (kPa).

150

S: Tunis soft soil

S_C: Tunis soft soil reinforced by DMC
100
S_C
Cu (kPa)

S
50

w/C=100 %; IP=100 kPa; CT=14d; V=30 rpm

0
100 150 200
Consolidation stress  (kPa)
3

Figure [Link] versus consolidation stress 3

402
Indeed, a significant increase in the undrained shear strength is noticed, when Tunis soft soil is improved with DM column,
from 17.7 to 46 kPa for 3=100 kPa, 25.4 to 71.2 for 3= 200 kPa and 32.3 to 139.4 for 3= 300 kPa.

4.2 Effect of injection pressure

Figures 4a and 4b compare the undrained shear strength (Cu) and the peak pore pressure, as a function of the injection pressure
IP. An increase evolution, of the undrained shear strength (kPa) as a function of IP (kPa) is observed. Indeed, there is a
significant increase in the undrained shear strength (from 64 to 87 kPa) when the injection pressure increases from 50 to 150
kPa. The increase in injection pressure from 100 to 150 kPa (1.5 times) gives an improvement of undrained shear strength
from 71 to 87 kPa (1.25 times). Further, the value of maximum cement grout injection pressure equals 150 kPa, is almost six
times the undrained shear strength of the soft soil of Tunis. The recommended value of injection pressure IP =2-4 CU (Shen,
1988) gives a lower improvement result.
Undrained shear strenght, Cu (kPa)

100 140
3=150kPa; w/C=100%; CT=14d; V=30rpm 3 =150kPa; w/C=100 %; CT=14d; V=30rpm

120
80

u max (kPa)
100
Cu
80
60
umax
60

Tunis soft soil reinforced by DMC Tunis soft soil reinforced by DMC
40 40
50 100 150 50 100 150
Injection pressure IP (kPa) Injection Pressure IP (kPa)
Figure 4. Cu and Δu versus IP (kPa)

4.3 Effect of the water-cement ratio w / C

Figures 5a and 5b show the undrained shear strength (Cu) and the peak pore pressure, as a function of the water-cement ratio
w/C. It can be noticed the significant decrease in the undrained shear strength from 180 to 71 kPa (251%) when the ratio w/C
increases from 0.7 to 1. When the ratio w/C is superior to 1 (1.2), the undrained shear strength decrease slightly from 71 to 47
kPa (151%). It is also notice here, the importance of the influence of the parameter water-cement ratio w/C on the
improvement of undrained shear strength i.e. if the ratio w/C increases with 43 % (0.7 to 1), the undrained shear strength
decreases with 151 % (180 to 71 kPa).

200 120

110
150
u max (kPa)

Cu 100
umax
Cu (kPa)

100
90

50
80
3 =150kPa; IP=100kPa;CT=14d; V=30rpm 3=150kPa; IP=100kPa; CT=14d; V=30rpm
0 70
0,6 0,8 1,0 1,2 0,6 0,8 1,0 1,2
Ratio w/C Ratio w/C

Figure 5. Cu and Δu versus w/C

4.4 Effect of Curing time

It is noticed, from Figures 6, that the undrained shear strength and the peak pore pressure of reinforced soil increase quasi-
linearly with time, which is attributed mainly due to the pouzzolanic reaction. Obviously, the curing time of 28 days gives a
403
better performance on undrained strength compared to other curing times. The undrained shear strength increases more than
twice (from 58 to 128 kPa) when CT increases from 7 to 28 days.

150 200

125
150

u max (kPa)
100
Cu (kPa)

75 Cu
100
umax
50

3=150kPa;w/C=100%;IP=100kPa;V=30rpm 3=150kPa;w/C=100%;IP=100kPa;V=30rpm
25 50
0 7 14 21 28 0 7 14 21 28
Curing time (days ) Curing time (days)

Figure 6. Cu and Δu versus CT (days)

4.5 Effect of rotation rate of drilling tool

Figures 7a and b present the obtained undrained shear strength (Cu) and peak pore pressure, in function of different rotation
rates of drilling tools V (rpm) used in the installation of DM columns. It can be noticed the quasi-linear evolution, of the
undrained shear strength (kPa) in function of V (rpm). Indeed, there is a significant decrease in the undrained shear strength
(from 110 to 55 kPa) when the rate of installation increases from 15 to 40 rpm. When the rate of drilling tools is important, the
grout cement injected cannot give a well-formed column and the mixing is unsuitable. For this reason, it is necessary to ensure
the proper selection of the rate of rotation of drilling tools when installation of DM column.

125 200

100 Cu
150
umax
u max (kPa)

75
Cu (kPa)

50
100

3=150kPa;w/C=100%;IP=100kPa;CT=14d 3 =150kPa;w/C=100%;IP=100kPa;CT=14d
0 50
10 20 30 40 50 10 20 30 40
V(rpm) V (rpm)

Figure [Link] and Δu versus V (rpm).

5. CONCLUSION

The experimental work presented in this paper is to study in a first step, the performance of DM column stabilization using
cement grout on the improvement of Tunis soft soil. The performed tests give good results for both the undrained shear
strength and the drained angle of friction. In a second step, the influence of test parameters is studied on the mechanical
properties of the Tunis soft soil reinforced with DM column. These parameters are: i) consolidation stress 3 ii) Curing time:
CT, (iii) injection pressure: IP, (iv) water-cement ratio w/C, (v) rotation rate of drilling tools V.
The main findings of the experimental program are as follows:
1. The increase in undrained shear strength depends essentially on the ratio w/C. When the ratio w/C increases only to
43 % (0.7 to 1) the Cu decreases until 151 % (180 to 71 kPa).

404
2. The injection pressure IP used in DM column installation play also an important role. The increase in injection
pressure from 100 to 150 kPa (1.5 times) gives an improvement of undrained shear strength from 71 to 87 kPa (1.25
times).
3. The laboratory analysis showed that shear strength starts to increases in a short period after installation (7 days). The
undrained shear strength of Tunis soft soil reinforced with DM column increases quasi-linearly with Curing time CT.
The undrained shear strength increases more than twice (from 58 to 128 kPa) when CT increases from 7 to 28 days.
4. There is a significant decrease in the undrained shear strength (from 110 to 55 kPa) when rotation rate of drilling
tools V (rpm) increases from 15 to 40 rpm. When the rate of rotation of drilling tools is important, the grout cement
injected cannot give a well-formed column and the mixing unsuitable.
5. The dimensionless ratio r=Cu(reinforced) /Cu(initial) increases from 2.6 to 4.3 with the increase in consolidation
stress from 3=100 kPa to 3=200 kPa.

REFERENCES

Ahnberg H., Johansson S-E., Retelius A., Ljungkrantz C., Holmqvist L., (1995). Cement and lime for deep stabilization of soil –a chemical
physical investigation of improvement effects. Swedish Geotechnical Institute, Report 48, Linkoping, 170. 43 pp
Al-Tabba, A., Ayotamuno, M.J., and Martin, R. J. (1999) “Soil mixing of stratified contaminated sands.” Journal of Hazardous Materials,
Elsevier, 53–75
Asano J, Ban K. and Takahashi K. (1996).Deep mixing method of soil stabilization using coal [Link] the IS-Tokyo 96, 2nd Int. Conf.
on Ground Improvement Geosystems, Tokyo, pp. 393-398.
Babasaki, R. M., Terashi, T. Suzuki, A., Maekaea, M, Kawamura, E., and Fukazawa. (1996). JGS TC Report: Factors Influencing the
Strength ofImproved Soil.” Grouting and Deep Mixing, Proc. Of the IS-Tokyo 96, 2nd Int. Conf. on Ground Improvement
Geosystems, Tokyo,V2, pp. 913-918.
Bruce, D. (2001). ―An introduction to the deep mixing methods as used in geotechnical [Link] [Link] verification and
properties of treated ground. Report No. FHWA-RD-99-167, US Department of Transportation, FederalHighway Administration.
Bouassida M., (1996) Etude expérimentale du renforcement de la vase de Tunis par colonnes de sable- Application pour la validation de la
résistance en compression théorique d’une cellule composite confinée. Revue française de Géotechnique, N° 75, pp. 3-12.
Chen, X. L., Liu, Y.H., and Zhang, S. D. (1996). Design methods of the cement soil retaining wall Proc. Of the IS-Tokyo 96, 2nd Int. Conf. on
Ground Improvement Geosystems, Tokyo, 1, pp. 475-480.
EuroSoilStab 2002. Development of Design and Construction Methods to Stabilise Soft Organic Soils: Design Guide Soft Soil Stabilisation.
CT97-035 1. Project No. BE 96-3177, Industrial & Materials Technologies. Ministry of Transport Public Works and
[Link] (Brite- EuRam III), European Commission.
EuroSoilStab, Project [Link]-CT97-0351 (1997).Design and testing of mixtures of binder materials and organic [Link] 2.1 and
[Link] of design and construction methods to stabilize organic soils.
Huat, B. B. K. (2006). Effect of Cement Admixtures on the Engineering Properties of Tropical Peat [Link] Electronic Journal of
Geotechnical Engineering, 11(B).
Japanese Geotechnical Society Standard. (2000). Practice for Making and Curing Stabilized Soil Specimens Without Compaction. (JGS
0821-2000).
Larsson S. (2003). Mixing Processes for Ground Improvement by Deep [Link] Thesis, Royal Institute of Technology. ISSN 1650-
9501, Stockholm.
Porbaha A. (1998). State of the art in deep mixing technology. Part I: Basic concepts. Proceedings of the ICE - Ground Improvement. 2(2): 81–
92.
Shen, S. L., Miura, N., and Koga, H. (2003). “Interaction mechanismbetween deep mixing column and surrounding clay during
installation.”Canadian Geotechnical Journal, 40(2): 293-307.
Taki O. (2003). Strength properties of soil cement produced by [Link] the 3rd Int. Conf. On Grouting and Ground Treatment, New
Orleans. ASCE, Geotechnical special Publication N° 120, pp. 646-657
Touiti, L., Bouassida, MandVan Impe, W. (2009). Discussion on Tunis soft soil [Link] and Geological Engineering. 27(5):
631-643.

405
406
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Experimental study for the mechanical characterization of Tunis soft soil reinforced by a
group of sand columns
Etude expérimentale de la vase de Tunis renforcée par un groupe de colonnes de sable
Wissem FRIKHA, Wassim KAFFEL, Mounir BOUASSIDA
Université de Tunis El Manar – Ecole Nationale d’Ingénieurs de Tunis. LR14ES03-Ingénierie Géotechnique. BP 37 Le
Belvédère, 1002 Tunis. Tunisie

ABSTRACT - An experimental investigation has been conducted to study the mechanical properties of remoulded soft soil of
Tunis reinforced by a group of sand columns. The tested soft soil, extracted from Tunis centre city at a depth of 15m,
wasremoulded and consolidated along an initial K0 consolidation path up to a vertical stress of 140 kPa. All reinforced soil
specimens werethen subjected to consolidated undrained triaxial shear tests with measured excess pore-pressure (CU+u). In
addition to the unreinforced control specimens, three different types of reinforced specimens were used, namely reinforced
specimens with a single sand column, reinforced specimens with three sand columns, and reinforced specimens with four sand
columns. All reinforced specimens had the same area replacement ratio. Testing results have shown that the number of
reinforced columns has a significant effect on the mechanical characteristics of the reinforced soft soil. Keywords: group sand
columns, Tunis soft soil, triaxial test, undrained shear strength, and friction angle.

1. INTRODUCTION

The soil of Tunisia Tunis city plateau is composed of three main formations: a 7m fill layer (from a depth of 1 to 8 m), a soft
soil (from a depth of 8 to 70 m), and a sandy-clay layer that is assumed as a substratum. The first twenty meters of Tunis soft
soil layer is used as the foundation level for the majority of buildings in Tunis City. Tunis soft soil is considered as a
problematic soil because of its low strength and high compressibility. For this reason, foundation design on Tunis soft clay
requires a thorough study for the short-term as well as the long-term behaviors. According to classification test results
performed by Klai and Bouassida (2009), Tunis soft clay is a very plastic muddy soil with a high proportion of silt and varied
clay fractions. Saturated Tunis soft clay is classified as a highly plastic silt with very low consistency (Bouassida, 1996).
Building projects on such a problematic soil requires the use of deep piles with lengths that can reach up to 50 m. Thus, for
economic considerations, soil improvement techniques could present a solution for founding on Tunis soft soil. Among the
various current methods of improving in-situ soils, stone columns is considered as a cost effective soil improvement
technique especially for soft soils (Andreou et al. 2008, Frikha et al. 2014). Stone or sand columns reinforcing technique result
in an increase of bearing capacity as well as a reduction and acceleration of consolidation settlement. Furthermore, the rapid
installation process has made this technique quite competitive compared to other soil improvement techniques (Frikha et al.
2008, Frikha and Bouassida, 2014).
Several laboratory investigations were performed to estimate the performance of reinforced soft soil by granular columns
(Andreou et al. 2008, Frikha et al. 2013, Frikha et al. 2014). The main objectives of these studies were to validate some
theoretical results and to evaluate the effects of main design parameters on the overall behavior of the improved soil, such as:
applied load, improvement area ratio, boundary conditions, column installation techniques , grain size distribution of used
column material, column length, etc. (Frikha, 2010).
The group effect of stone columns ground improvement was studied using field tests (Goughnour and Bayuk, 1979, Munfakh
et al. 1984) and centrifuge tests to evaluate the general failure mechanisms (Terashi et al. 1991) and the extent width of
improvement (Kimura et al. 1983). Kaffezakis (1983) performed triaxial tests on clay specimens reinforced by a group of
stone columns. He reported a significant increase of lateral stress developed within the stone columns which increases as the
number of columns within the group increases. Bachus and Barksdale (1984) have concluded that there is only a slight
increase in the ultimate load bearing capacity per column when the number of columns increases. However, in their tests, the
effect of lateral confinement was more significant since the reinforcing columns were set close to the testing box borders. Hu
(1995) built a laboratory scaled model to examine the behavior of a cohesive soft soil reinforced by a group of stone columns
supporting a rigid footing. They found that reinforcement by a group of columns is more effective than that using an isolated
column. The interaction between columns and soft clay efficiently contributes to enhance the load bearing capacity and to
provide a wider transfer of loading. Wood et al. (2000) commented on the tests performed by Hu (1995) and confirmed that
the mode of failure of each column, in the case of group of columns reinforcement, depends on its location within the group,
its length, and the type of loading. Their results showed that the pre-failure mechanisms and failure modes of stone columns
group are different from those observed for a single column. They reported that the area replacement ratio affects the columns’
interaction extent and the load transferred to the soft clay in between columns. This research concluded that a significant
improvement of bearing capacity relies on a minimum area replacement ratio of 25%. Black et al. (2007) have found that clay
specimens reinforced by a single end-bearing column (fully penetrating column over the specimen length) showed a 33%
increase in strength. They have noted that the installation of a group of columns with the same area replacement ratio does not
provide particular difference in the load-carrying capacity. With regard to settlement in drained condition, a group of columns
makes a significant difference in stiffness of the composite material. The result, presented by Black et al. (2007), indicates that
407
a group of columns can lead to a possible reduction of stiffness when compared to a single column with similar area
replacement ratios. Ambilyand Ghandi (2007) reported that the stiffness of a single column and a group of six columns
(spaced more than 3 times the columns diameter) are almost comparable. It has been noted, using the unit cell concept, that the
behavior of a single column can simulate the field behavior of an interior column belonging to a group of columns.
In conclusion, some authors found that, for a given replacement area ratio, the increase of columns number improves the
strength characteristics of the reinforced soil, while other researchers confirmed that there is no difference between the
behavior of single column and thatof a group of columns. The main goals of this paper are to evaluate the effect of
reinforcement by a single column and a group of sand columns on the strength characteristics of Tunis soft soil and to examine
the effect of the contact surface between columns and soft soil on the shear strength of improved soil. To achieve these
objectives, an experimental study using consolidated undrained triaxial tests with excess pore pressure measurements (CU+u)
were carried out on remoulded Tunis soft soil specimens reinforced with sand columns.
Due to the enhanced drainage property of reinforcing sand columns, the consolidation of soft soil is highly accelerated, hence,
at this stage, the reinforced soil can be considered consolidated. As the applied load occurs rapidly, this second phase
simulates the undrained behavior of reinforced soil. Therefore, it is more appropriate to study the experimental behavior of
reinforced soil using the CU + u triaxial test since it reproduces the in-situ construction conditions during which non negligible
excess pore pressure are measured. Further, the CU + u test enables to determine the short-term strength characteristics of
reinforced soil, and to deduce the long-term shear strength properties.
These tests were performed by varying the number of columns (one, three and four columns) and the confining stress
(σ’3=100, 200 and 300 kPa) for a fixed area replacement ratio denoted as “”. The column diameters of specimens reinforced
by one, three and four columns are respectively 3.40, 2.00 and 1.60 cm. The results of those tests are discussed and compared
with the ones performed on unreinforced soft soil specimens.
The performance is quantified by the prediction of the strength properties to evaluate the safety factors of the improved soil
(after columns installation), and the homogeneous equivalent values of shear strength parameters (c and φ) replacing the real
composite material (soft soil - stone columns). As suggested by Christoulas et al. (1997), the equivalent strength characteristics
can be determined using the law of mixtures (Di Maggio, 1978):
c= (1-η) cs+ ηccc (1)
tan φ=(1-η) tan φs + η tan φc (2)

Where:cs and φs : shear strength parameters of soft soil (cohesion and friction angle respectively); cc and φc : shear strength
parameters of columns constitutive material; η: the area replacement ratio (AC/A); A: total cross-sectional loaded area; AC,
cross-sectional area of stone columns.
Note that Eqs (1) and (2) predict the equivalent shear strength characteristics of the reinforced soil only from the shear
strength characteristics of the soft soil and column material. Such a consideration relies on the expressions proposed by
Bouassida et al. (1995) who demonstrated, in the limit analysis framework that homogenized strength characteristics of the
reinforced soil only depend on the shear strength of the initial soil and column material and the area replacement ratio. It is
noticed that Frikha et al. (2014) found reasonable agreement between the predicted equivalent characteristics and experimental
results obtained on clayey soil reinforced by sand column. However, for example Bergado et al. (1996) suggested that the
homogenized strength characteristics also depend on the stress concentration ratio that describes the behavior of reinforced soil
in pre-failure phase.
2. STUDIED SOIL

The tested soft soil is grey in color, contains shell debris and has a characteristic smell. The material used to make the
reinforcing columns is siliceous sand classified as CEN in accordance to the European standards EN 196-1. The sand particles
are generally isometric and rounded in shape. The coefficient of uniformity of the CEN sand is Cu=6 and its dry unit weight is
γd=16 kN/[Link] friction angle of normalized CE sand is φ’=41.8° that also satisfies the Standard NF P 11-212. This standard
recommends minimum values for the friction angle of stone columns material equal to φ’= 40° for crushed material and φ’=
38° for rolled material. The friction angle of standard sand was chosen high enough to essentially provide a satisfactory
increase in the shear strength as specified in French Standard (Frikha et al. 2015)
The properties of Tunis soft soil are: liquid limit l=66.5 %, plastic limit, p=23.9%, plasticity index IP=42.6%,
natural water content =74%, compression index Cc= 0.64 and oedometer modulus E=1700 kPa. According to French
standard XP P 94-011 classification, the tested soft soil is classified as a highly plastic clay.
3. EXPERIMENTAL PROCEDURE

The first step consists in wet sieving the natural soft soil through the 100 µm sieve. Then, the sieved soft soil is air-dried until
its moisture content reached 100%, which represents approximately 1.5 l.
Second, the obtained slurry material is subjected to a vertical pre-consolidation pressure of K0 in a special mould
(consolidometer container). The consolidometer is a metallic hollow cylinder with 7cm internal diameter and 25 cm in height
(Figure 1). It is equipped with either 1, 3, or 4 vertical thin mould tubes covered by a filter paper to create the holes where the
normalized sand columns are subsequently installed. The diameters of inner mould tubes are deduced from the fixed value of

408
the area replacement ratio (η=22%) and the number of columns. In fact, to calculate the columns’ radius, the following
equation was used:

A ci
n. ACn  RC 
2

 i
  n.  n  (3)
A A  R 
Where: n= number of columns; R=radius of total cross-sectional reinforced area; RCn= radius of the cross-sectional
area of single column.
The radius of inner mould tubes, used for the reinforcement of clayey specimens with one, three and four columns,
are 1.70, 1.00, and 0.80 cm, respectively (Figure 1).
Rc = 1.7 cm Rc = 1.0 cm Rc = 0.8 cm

25 cm

7 cm 7 cm 7 cm

Figure1: Sectional view and plan view of Ko consolidation apparatus

The inner surface of the consolidometer was lubricated with silicone grease prior to slurry placement in order to reduce
roughness effects; thisallows an easy unmoulding of the specimens. The inner surface of the consolidometer is also lined with
filter paper to accelerate the consolidation. The slurry is then carefully placed using a spoon into the mould, where it is
frequently tamped with a plastic tamper, to avoid air formation during the filling phase. When the predetermined specimen
height is attained, the consolidation loading cap is placed to start the loading phase. The specimens were consolidated under a
vertical stress of 140 kPa for at least ten days. At the end of consolidation phase, the specimen height is 14 cm and the
diameter is 7 cm. During the pre-consolidation phase, drainage is allowed at the top and bottom of the specimen through the
installed upper and lower porous stones. A dial gauge on the loading cap is used to measure the axial deformation of the tested
specimen. The consolidation is considered complete when quasi-constant axial deformation is recorded. Thepressure is then
reduced to zero, and the loading cap is removed. The tubes are, then gently removed and the soft clay sample is placed in the
triaxial cell.
The cavities of the pre-consolidated specimen (after unmould and removal of the internal tubes) were filled by a dry sand in
small layers (20gr of sand per layer), in order to attain the final length of column (140 mm). Each layer was compacted by a
plastic tamper. The same weight (20 gr) of dense compacted sand was considered for column installation in all tests. The
reinforced soil specimen is finally reconsolidated and saturated in the triaxial cell under three isotropic confining pressures:
’3of 100, 200 and 300 kPa.

Figure [Link] of reinforced sample after preconsolidation and installation of columns

4. TRIAXIAL TESTS

The experimental program included 13 tests as summarized in Table 1. An axial displacement rate of 0.03 mm/min was used
in accordance with . These tests comprised the following steps:
Step 1: The specimen is subjected to full saturation so that the recorded Skempton coefficient defined as B  u  95% .
 3

409
Step 2: The specimen is subjected to an isotropic consolidation, under confining pressures σ’3= σ’1=100, 200 or 300 kPa until
quasi-constant volume variation is observed.
Step 3: The shear loading is applied, at constant confining pressure, in undrained condition with measured excess pore
pressure, up to specimen failure.
As observed in Table 1, most of the tests were performed twice to assure testing repeatability and, therefore, validate the
experimental procedure and testing equipment.

Table 1. Summary of test data

0 3 Substitution
Test name (kPa) (kPa) Repe Number ratio
tition of η
column
M_CU100_0C 140 100 2 0 0
M_CU200_0C 140 200 2 0 0
M_CU300_0C 140 300 2 0 0
M_CU100_1C 140 100 2 1 22%
M_CU200_1C 140 200 2 1 22%
M_CU300_1C 140 300 2 1 22%
M_CU100_3C 140 100 1 3 22%
M_CU200_3C 140 200 1 3 22%
M_CU300_3C 140 300 1 3 22%
M_CU100_4C 140 100 1 4 22%
M_CU200_4C 140 200 1 4 22%
M_CU300_4C 140 300 1 4 22%
0: vertical preconsolidation pressure applied before triaxial test

5. RESULTS AND INTERPRETATION

Figures 3 and 4 present typical results of the triaxial test carried out on unreinforced and reinforced specimens with single,
three and four columns subjected to different confining pressures. These figures show the variation of deviator stress, q=σ1-σ3=
σ’1-σ’3 calculated as the difference of total stress between the axial principal stress and the radial principal stress, and excess
pore pressure (Δu) respectively with respect to calculated axial strain in undrained conditions. The axial strain εa is calculated
as the measured displacement divided by specimen initial height.

150 160

140
125
120
Deviator q (kPa)

100
100
u (kPa)

75 80

60
50
40
25
M_CU100_0C M_CU200_0C
20 M_CU100_0C M_CU200_0C
M_CU300_0C
M_CU300_0C

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Axial strain a (%) Axial strain a(%)

Figure.3a Stress-axial strain curve during triaxial shear test Figure.3b. Pore pressure-axial strain curve during triaxial
for unreinforced specimen shear test for unreinforced specimen

410
350 160
M_CU100_1C M_CU200_1C M_CU300_1C M_CU100_1C M_CU200_1C M_CU300_1C
M_CU100_3C M_CU200_3C M_CU300_3C M_CU100_3C M_CU200_3C M_CU300_3C
M_CU100_4C M_CU200_4C M_CU300_4C M_CU100_4C M_CU200_4C M_CU300_4C
300 140

120
250
Deviator q (kPa)

100

u (kPa)
200
80
150
60

100
40

50 20

0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Axial strain a(%) Axial strain a(%)

Figure.4a Stress-axial strain curve during triaxial shear test Figure.4b. Pore pressure-axial strain curve during triaxial
for specimen reinforced with columns shear test for specimen reinforced with columns

The variations of the deviator stress and excess pore pressure peaks as a function of the number of column are drawn
in figures 5 and 6, respectively, for different confining stresses. It is noted that for all confining stresses, the values of peak
deviator stresses increase with the number of columns (n≥1) although the area replacement ratio is kept constant (η=22%).
When the number of columns increases from one to three, the peak deviator stress increases by 14%, 9% and 6% for the
respective confining stresses of 100, 200 and 300 [Link], when the number of columns increases from one to four, the
peak deviator stress increases successively by 31%, 24% and 16% for the same confining stresses. It can then deduced that as
the confining stress decreases, there is more effect of the number of column on the ultimate capacity of improved Tunis soft
soil. It can also deduced that for small confining stress, in in-situ practice of stone columns improvement, the effect of the
increase of the number of columns in shear strength (ultimate bearing capacity) is more pronounced at the soil surface than
deeper in the foundation.

350 160
M_CU100 (3'=100kPa) M_CU200(3'=200kPa)
300 M_CU300 (3'=300kPa) 140

120
250
umax(kPa)

100
qmax(kPa)

200
80
150
60
100
40
50 20 M_CU100 (3'=100kPa) M_CU200(3'=200kPa)
M_CU300 (3'=300kPa)
0 0
0 1 2 3 4 0 1 2 3 4
Number of columns
Number of columns

Figure.5. Peak values of deviatoric stress versus number of Figure.6. Peak values of pore pressure versus number of
columns for σ’3= 100 kPa, σ’3= 200 kPa and σ’3= 300 kPa columns for σ’3= 100 kPa, σ’3= 200 kPa and σ’3= 300 kPa

It can also be observed from Figure 7 that, for all specimens, the peak deviator stress (qmax) increases when the
confining stress increases from 100 to 300 kPa. For unreinforced specimens the peak deviator stress qmax increases by 246 %
(from 54.80 to 135.10 kPa) when the confining stress increases from 100 to 300 kPa. For reinforced specimens with one, three
and four columns, qmax increases by 267% (from 95.31 kPa to 254.10), 265% (from 107.99 to 270.30 kPa), and 237% (from
124.11 to 295.12 kPa), respectively, when confining stress increases from 100 to 300 kPa. Whereas, it can be noticed, for all
specimens, a slight decrease in excess pore pressure by increasing the number of columns and confining stress as well.

411
120 26
M_CU_0C
100 M_CU_1C
M_CU_3C 24
M_CU_4C
80

'(degrees)
22
60
qmax

40 20

20
drained frictional angle ' (°)
18
0
0 50 100 150 200 250 300 350 16
3 (kPa) 0 1 2 3 4 5
Number of columns

Figure 7. Peak values of deviatoric stress verus confining Figure 8. Drained frictional angel versus number of
stress columns

In addition, Figure 8 shows that the effective friction angle obtained for different samples increases with the number
of columns. In fact, when the number of column increases from one to three, the friction angle slightly increases by 2% and
when the number of column increases from one to four, the friction angle slightly increases by 3%.
Using the law of mixtures (equation (2)) with an area replacement ratio of 22% and a sand friction angle of 41.8°, the
equivalent friction angle is φ’=23.95°, that is slightly lower than the measured friction angle of reinforced specimen with
single column (φ’=24.1°). It can be observed that the law of mixtures provides an acceptable approximation of the effective
friction angle of Tunis soft soil reinforcedby a single sand column. However, correction should be used for the prediction of
friction angle of reinforced clayey soil with several columns. Indeed, as the number of reinforcing columns increases from
one, to four, with a fixed area replacement ratio, the friction angle slightly increases.
Moreover, For consolidated undrained triaxial tests (CU+u), the undrained shear strength is defined as half of the
q max  1   3
peak deviator stress ( cU   ).The law of mixtures (1), with an area replacement ratio of 22% leads to an
2 2
equivalent undrained shear strength that depends on the confining stress when varied from 100 kPa, 200 kPa to 300 kPa, as
shown by Figure 9. The predicted values of undrained shear strength are also comparable to those measured for a reinforced
specimen by single column. A correction should be also made when predicting the undrained shear strength of reinforced
Tunis soft soil with several columns. It is found that the undrained shear strength is also higher as the number of reinforcing
columns increases from one to four.

200
M_CU100 (3'=100kPa) mesured M_CU100 (3'=100kPa) calculated
M_CU200 ( 3'=200kPa) mesured M_CU200 (3'=200kPa) calculated
M_CU300 ( 3'=300kPa) mesured M_CU300 (3'=300kPa) calculated

150
cU(kPa)

100

0
0 1 2 3 4 5
 umber of columns

Figure 9 :Undrained shear stress versus number of columns for σ’3= 100 kPa, 200 kPa and 300 kPa
When the number of columns increases at a fixed area replacement ratio “η”, the contact surface between the reinforcing
columns and the clayey soil increases resulting in an increase in the soil-column interface friction, which enhances the strength
properties of the reinforced soft clayey soil.
6. Conclusions

. The experimental work presented constitutes a first attempt to see whether the group effect has some influence or not on the
bearing capacity of reinforced clayey soil by granular group of columns. The results of the various performed tests lead to the
following conclusions:
 For a fixed area replacement ratio value (η=22%), the increase in reinforcing columns, leads to the improvement of
undrained shear strength and effective friction angle both associated with the decrease in pore pressure.
412
 The interactions between group of columns and the soft clay tend to improve the distribution of load bearing capacity.
 The results exhibit that the effective friction angle for a single column specimen is 37% higher than the obtained value for
unreinforced Tunis soft soil specimen. This demonstrates the interest for reinforcing Tunis soft soil.
 The friction angle slightly increases with the number of columns.
 The effect of columns number seems to be more pronounced at small confining pressure However, the peak pore
pressure decrease is more pronounced at higher confining stress when the columns number increases.

References

AFNOR, NF P 94-074, (1994) "Sols: reconnaissance et essais. Essais à l’appareil triaxial de révolution. Appareillage, Préparation des
éprouvettes et Essais (UU), (CU+u) et (CD)." French Standard, October 1994.
AFNOR, XP P 94-011, (1999) "Sols : reconnaissance et essais. Description—Identification— Dénomination des sols. " French Standard,
Août 1999.
Ambily P., and Gandhi, S.R. (2007) "Behavior of Stone Columns Based on Experimental and FEM Analysis." J. Geotech. Geoenviron.
Eng.133, [4], 405–415. DOI :10.1061/(ASCE)1090-0241
Andreou P., Frikha W., Frank R., Canou J., Papadopoulos V., and Dupla J.-C., (2008) "Experimental study on sand and gravel columns in
clay." Proceedings of the ICE - Ground Improvement. 161, [4], 189-198. DOI: 10.1680/grim.2008.161.4.189.
Bachus, R.C., and Barksdale, R.D., (1984) "Vertical and lateral behavior of model stone columns."Proceedings of the International
Conference on In-situ Soil and Rock Reinforcement, Paris, 99–110.
Black J., Sivakumar V., and McKelvey D., (2007) "Performance of clay samples reinforced with vertical granular columns". Can. Geotech.
J. [44], 89–95. 10.1139/t06-081.
Bouassida, M. de Buhan, & Dormieux, L.,(1995) “Bearing capacity of a foundation resting on a soil reinforced by a group of columns.”
Géotechnique, Vol. 45, n° 1, 25-34.
Bouassida M., (1996) "Étude expérimentale du renforcement de la vase de Tunis par colonnes de sable. Application pour la validation de la
résistance en compression théorique d'une cellule composite confinée." Revue Française de Géotechnique.2ième semestre. N° 75: 3-12.
Christoulas, S.T., Giannaros, C.H., and Tsiambaos, G., (1997) “Stabilization of embankment foundations by using stone columns.”Geotech
Geol Eng, 15, 247-258. DOI: 10.1007/BF00880828
Frikha W., Bouassida M., and Canou J., (2008) "Calibration of an Elastoplastic Model for the Prediction of Stone Column Ultimate Bearing
Capacity."Proceeding of selected sessions of Geo Congress [Link] and Geohazard [Link] Publication ASCE
Institute (invited paper). Edit K.R. Rebby et al, 604-611 (USA). DOI: 10.1061/40971(310)75.
Frikha, W. (2010) “Approche expérimentale sur modèle pour l’évaluation d’une argile molle renforcée par colonnes ballastées“, PhD thesis,
ENIT, Tunisia. Publisher: Editions Universitaires Européennes, 276 pages. ISBN 978-613-1-55837-5.
Frikha W., Bouassida M., Canou J. (2013) “Observed Behavior of Laterally Expanded Stone Column in Soft Soil”. Geotechnical and
Geological Engineering. February 2013. DOI 10.1007/s10706-013-9624-8.
Frikha, W. & Bouassida, M. (2014)“Prediction of stone column ultimate bearing capacity using expansion cavity model.”Proceedings of the
ICE - Ground Improvement. DOI: 10.1680/grim.13.00045
Frikha W., Bouassida M., Canou J. (2014)“Parametric Study of a clayey Specimen Reinforced by a Granular Column.” Int. J.
Geomech. , 10.1061/(ASCE)GM.1943-5622.0000419 , 04014078
Frikha W., Tounekti F., Kaffel W., Bouassida M. (2015). “Experimental study for the mechanical characterization of Tunis soft soil
reinforced by a group of sand columns”.Soils and [Link] 55, Issue 1, February 2015, Pages 181–191.
Goughnour, R.R. and Bayuk, A.A., (1979) “A field study of long-term settlements of loads supported by stone column in soft ground.”
Proceedings of the International Conference on Soil Reinforcement, Paris, 279-285.
Hu, W. (1995) “Physical modelling of group behavior of stone column foundations” PhD thesis, University of Glasgow.
Kaffezakis, G.J., (1983) "A model test investigation of the behavior of stone columns."Special Research Problem, Georgia Institute of
Technology, [Link] April 1983.
Kimura, T., Nakase, A., Saitoh, K. and Takemura, J., (1983) “Centrifuge Tests on Sand Compaction Piles.” Proc. 7th Asian Regional
Conference on Soil Mechanics and Foundation Engineering, Vol 1.255-260.
Klai M., Bouassida M. (2009) “Comparison between behavior of undisturbed and reconstituted Tunis soft clay.” Proceeds.2nd Int. Conf.
New Develop. In SMGE. Edit. Atalar et al. 28-30 May NEU Nicosia. pp 414 –419
Munfakh, G.A., Sarkar, S.K. and CasteIli, R.C., (1984) “Performance of a test embankment founded on stone columns.” Piling and Ground
Treatment, Thomas Telford Ltd., London.
Terashi, M., Kitazume, M. and Minagawa, S., (1991) “Bearing capacity of improved ground by sand compaction construction and testing.”
ASTM - SIP 1089 47-61.
Wood M., Hu D.W., and Nash D.F., (2000) “Group effects in stone column foundations.”Géotechnique, [50], 689–698. DOI:
10.1680/geot.2000.50.6.689

413
414
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Numerical analysis of embankments on soft clay improved by PVDs

Analyse numérique des remblais sur argile molle améliorée par PVD
M.A. Mansour*1, A.M. Samieh2and M.H. Mahmoud 3
1,2
Civil Engineering Department, Faculty of Engineering, Helwan University, Cairo, Egypt
3
Advanced Construction Technology Services ACTS, Doha, Qatar

ABSTRACT – The approach of Preloading fills coupled with Prefabricated Vertical Drains (PVDs) is one of the soil
improvement techniques exhibiting successful use in stabilizing soft clay formations. This study numerically investigated the
performance of a monitored earth embankment that was constructed on a very soft to soft clay formation improved by PVDs.
Numerical analysis was conducted using the two-dimensional finite element program PLAXIS, considering the smear effect.
A good comparison was revealed between the monitored and the predicted responses. In addition, a sensitivity study was
carried out to highlight the influence of some modelling parameters and design considerations that may be of a major concern
on designing such embankments. The considered parameters were the PVD length, the PVD spacing and the width of the
improved ground. Moreover, the impacts of smear zone permeability and the shear strength of soft clay were investigated.

RÉSUMÉ–L'approche de préchargement remplit couplé avec les drains verticaux préfabriqués (PVD) est l'un des sols
techniques d'amélioration présentant une utilisation reussite à stabiliser les formations d'argile molle. Cette étude
numériquement étudiée la performance d'un surveillé remblai du terre qui a été construit sur un très doux à doux formations du
molle améliorée par PVD. L'analyse numérique a été réalisée en utilisant le bidimensionnel element limité programme
PLAXIS ,considérant l'effet de balayage. Un bonne comparaison a été révélé entre le surveillé et les réponses prédites. En
outre, une étude de sensibilité a été réalisée à mettre en évidence l'influence de certains paramètres de modélisation et des
considérations de conception qui peuvent être de les préoccupations majeure inclus dans la conception de ces remblais. Les
paramètres considérés sont la longueur de PVD, l'espacement des PVD et la largeur de la sol améliorè. En plus, les effets de
zone de souillure perméabilité et la résistance au cisaillement de l'argile molle ont été étudiés. De plus, les impacts de frottis
zone perméabilité et la résistance au cisaillement de l'argile molle ont été étudiés.

1 INTRODUCTION

Soft clay formations are extensively located in many coastal areas around the world as such the north coast of Africa. Design
and construction of earth embankments on soft formations encounters the problem of excessive total and differential
settlements, large lateral displacement and inadequate bearing capacity of such formations. The need to improve the undesired
ground conditions for construction purposes provides challenges to geotechnical engineers. Many ground improvement
techniques have been developed to overcome engineering problems facing the construction of embankments over soft alluvial
deposits. The approach of preloading fills coupled with vertical drains,with different configurations, is one of the frequently
adopted soil improvement techniques of soft clay formations. The installation of vertical drains reduces the length of the
drainage path that accelerates the consolidation process.
Currently PVDs are successfully adopted for soft soil stabilization, especially for roads, railways, ports, airports and
various other infrastructure projects, to reduce the total and differential settlements, reduce the lateral displacement and
increase the ground bearing capacity (Arulrajah, 2005).
Different approaches have been adopted to analyze the behavior of earth embankments constructed on soft clay formations
stabilized by PVDs. These methods can be grouped into three categories; analytical, observational and numerical techniques.
Barron (1948) first presented the analytical solution to the problem of soil consolidation due to horizontal flow towards
vertical drains. Hansbo (1981) recommended some modifications to Barron’s solution, which dealt mainly with the simplified
assumptions of the equation to take into account the physical dimensions, characteristics of prefabricated drains and the effect
of drain installation. The observational methods uses settlement monitoring records for back-calculating the degree of
consolidation and some soil parameters, such as the coefficient of consolidation in case of using Asaoka’s method (1978).
Numerical analysis of earth embankments constructed on soft ground improved by PVDs, requires adopting three
dimensional (3-D) models because the consolidation around vertical drains is neither truly axisymmetric nor plain strain.
Three dimensional idealization is very infrequent for routine analysis due to its complexity (Mestat, 2001). To employ a
realistic two dimensional (2-D) finite element analysis for vertical drains, the equivalence or matching between the 3-D
condition versus the plane strain or axisymmetric conditions needed to be established (Hird, 1992; Bergado & Long, 1994;
Indraratna & Redana, 2000; Lin et al. 2000 and Chai et al. 2001). Hird (1992) pointed out that the matching of axisymmetric
and plane strain conditions can be done in three ways: (a) geometric matching of the spacing between the drains while
keeping the permeability the same, (b) permeability coefficient matching while keeping the drains spacing to be the same, and
(c) combination of permeability and geometric matching approaches where a plane strain permeability is calculated for a
convenient drain spacing. Bergado and Long (1994) proposed another matching approach for 2-D modelling of PVD effects.

415
The PVDs are transformed to continuous walls with the same spacing and with the assumption that the coefficient of
permeability is independent of the state of seepage flow.
Arulrajah et al. (2004) has adopted the two-dimensional (2-D) finite element program PLAXIS (Brinkgreve, 2002) to study
numerically the behavior of an embankment constructed on marine clays strata at Singapore. The PVDs were modeled by
adopting an axisymmetric unit cell model and by using a full scale plane strain model. The equivalent horizontal permeability
of surrounding soils was estimated using the approach of Lin et al. (2000). The predicted embankment performance was
compared via the field measurement and Asoaka's method (1978). Good agreement was noticed between the finite element
results and the field settlements. Except that the axisymmetric unit cell provided a slightly smaller settlement values than that
of the full scale model.
Hien (2008) analyzed the behavior of the monitored Muar Bypass road embankment, Malaysia. The embankment was
constructed on a soft clay formation stabilized with PVD and a geotextile layer along the embankment base. The numerical
simulation was carried out using the 2-D finite element program PLAXIS. The numerical results exhibited acceptable
comparison with the field measurements.
This research presents the results of a detailed numerical study conducted to investigatethe response of a monitored earth
embankment constructed on a soft clay formation improved with PVDs. A comparison between the numerically predicted and
the monitored responses is presented to assess the adequacy of the adopted numerical modelling approach. Also, a sensitivity
study is presented to investigate the influence of some parameters that may be of a major concern on designing such
embankments such as the PVD length,the PVD spacing and the improved ground width. These three parameters have a major
economical impact for such projects. In addition, the sensitivity study investigates the influence of the soft clay shear strength
characteristics and the smear zone permeability.

2 EMBANKMENT CONFIGURATION, MATERIAL PROPERTIES AND NUMERICAL MODELLING

A full scale earth embankment TS1 was designed, constructed and monitored at the proposed Second Bangkok International
Airport at Nong Ngu Hao, East Bangkok, Thailand (Bergado et al., 2002).The soil profile at the test site is relatively uniform,
consisting of 2m thick weathered crust overlying very soft to soft clay approximately 10 m thick. Underlying the soft clay is a
medium stiff clay layer, about 4 m thick, followed by a stiff clay layer extending down to 24 m depth which is in turn
underlain by a layer of dense sand. The ground water was encountered between 0.5 and 1.0 m below the ground surface. The
embankment was 4.2 m in height with side slopes of 3H: 1V. Figure 1 presents the embankment configuration. The
prefabricated vertical drains were installed in a square pattern at spacing of 1.5m and were of depth 12 m. The embankment
was constructed by first excavating 0.3 m below the ground surface, constructing the sand blanket to a level of 1.2 m above the
ground surface. Clayey sand was subsequently used to raise the embankment to its final height.

Figure [Link] TS1 configuration

Numerical analysis, in the current study, was carried out using the two dimensional finite element program PLAXIS
(Brinkgreve, 2002). The different soil layers were modeled using fifteen-node triangular elements. The installed PVDs were
modeledin PLAXIS using five-node drainage line elements which areacting as seepage boundaries. The adopted finite element
mesh consisted of 11063 nodes and 1363 finite elements. The boundary conditions were adopted as follows: vertical
boundaries have zero lateral movements, i.e. roller support, and were considered to be impermeable. The bottom horizontal
boundary was restrained vertically and horizontally and was considered to be permeable, modeling a sand formation. A fine
mesh option was considered for the entire mesh. Moreover, the zone improved with the PVD was locally refined.
The elasto-plastic Mohr-Coulomb model was used to simulate the behavior of the embankment materials, both the sand
blanket and the clayey sand layer. The natural soils were modeled using the hardening soil model which is considered an
advanced model for simulating the behavior ofboth soft and stiff soils. In brief, Hardening Soil model consists of a fixed
416
ultimate surface and a moving cap surface to model the continuous hardening of the material (Brinkgreve, 2002). Table 1
presents a summary of the parameters that were adopted in the numerical analysis of the current study. The vertical and
horizontal permeability coefficients of the different layers were converted in the current study to their equivalent plain strain
permeabilities(Bergado et al., 1996 and Indraratna & Redana, 1999) using the approach proposed by Hird (1992). In this
approach, the spacing between the drains is kept as it is in the numerical model (permeability matching approach). A ratio of
ten was considered between the horizontal permeability of the natural clay to the horizontal permeability of the smear zone.
The diameter of the smear zone was considered to be twice the mandrel diameter (Bergado et al., 1993b).
Staged construction was performed to allow the strength of subsoil to increase as consolidation takes place under fill load
increments. Construction of subsequent stages commences when the strength of the soil is sufficient to maintain stability. In
the current study, the embankment construction was carried out in six stages of variable fill thicknesses (1.0m, 1.5m, 0.5m
,0.65m , 0.35m and 0.20 m consecutively). During stage (3) of the embankment construction, a 5m width and 1.5m high berm
was added. The berm width was increased to 7m during stage (4) of construction (Bergado et al., 2002).

Table 1. Materials parameters adopted in numerical analysis

Clayey Sand Very soft Medium Dense

Parameter/Layer Crust soft clay stiff clay
sand blanket clay stiff clay sand
Model Mohr-Coulomb Hardening soil model
Cohesion (Kpa) 1 1 2 2 2 2 2 1
Friction angle (°) 33 36 30 20 22 28 30 37
Dilatency angle (°) 5 3 - - - - - 5
Young's modulus (MPa) 40 60 - - - - - 50
Primary loading stiffness modulus (MPa) - - 10 0.45 0.65 5 12 60
Unloading/reloading stiffness - - 30 1.5 2.1 15 36 180
modulus(MPa)
Poisson's ratio 0.3 0.3 - - - - - -
Unloading/reloading Poisson's ratio - - 0.2 0.2 0.2 0.2 0.2 0.2
Horizontal permeability (m/day) 8 20 1.14 E -04 4.4 E -05 2.3 E -05 2.1 E -04 5.0 E -05 1
Permeability ratio** 1 1 1 1.7 2 2.1 1.67 1
** Permeability ratio=horizontal Permeability/ vertical Permeability

3 NUMERICAL MODELLING RESULTS

The settlement profile along the embankment base just after construction and long term settlement,are shown in figure 2-a.
Comparing the two settlement curves, the figure shows that,in the long term case the settlement underneath the embankment
increases while the heave outside the embankment toe does not significantly decrease. The settlement increase is due to the
consolidation of different clay formations that lie in-between and underneath the PVDs. The predicted and the monitored
horizontal displacement trends just after construction and the predicted long term horizontal displacement along a vertical line
passes through the crest edge, are shown in figure 2-b. The figure exhibits that the maximum lateral displacement occurred
below the ground surface, within the soft / very soft clay layer. In addition, the monitored and the predicted trends of the
horizontal displacement are identical. However, figure 2-b shows that the numerical analysis slightly overestimated the
horizontal displacementjust after construction, a deviation of about 18%.
400 Horizontal displacement (mm)
(a)
200 0 20 40 60 80 100 120 140 160 180 200 220 240
0
0 -2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 -4
-200
Depth below ground surface (m)

-6
(b)
Settlement (mm)

-400 -8
-10
-600
-12
-800 -14
-16
-1000
-18

-1200 -20
Predicted settl. at E.O.C -22 Monitored hz displacement at E.O.C
-1400 Predicted long term settle. -24 Predicted hz displacement at end of construction
-26 Long term predicted hz displacement
-1600
Distance from the embankment C.L (m) -28

a-Vertical displacement profile at ground surface b- Horizontal displacement profile below the embankment crest
Figure [Link] deformational response.

The time-settlement profiles at a point located at the intersection of the embankment centerline and the ground surface are
shown in figure 3. The figure presents the settlements estimated using numerical analysis, Asaoka’s method in addition to the
monitored settlements. The predicted settlements using both the numerical model and Asaoka’s method is slightly greater than
417
the monitored values especially during the construction period. However, the numerically predicted settlement exhibits a good
comparison with the monitored values until the end of the monitoring program, (around 400 days). Figure 3 shows, also, that
the numerically long term settlement is less than the value estimated by Asoka’s method, a difference of around 25%.
Variation of pore water pressure with time near the embankment centerline and at the toe, at different depths, is shown in
figure 4. Figure 4-a compares the dissipation of the excess pore water pressure near the embankment centerline and exhibits
that the maximum excess pore water pressure occurred at a depth of around 4m below the ground surface. At the toe,figure 4-
b, the maximum excess pore water pressure occurred at a depth of around 6m below ground surface. After the end of
construction, figure 4 shows that the rate of dissipation of the excess pore pressure near the embankment centerline is
greater than that at the toe. This can be explained as, near and outside the embankment toe there are no PVDs,while as, the
zone at the embankment centerline is surrounded by PVDs that conventionally speed the dissipation of the excess pore water
pressure due to shortening the drainage path.
Time (Days)
0 100 200 300 400 500 600 700 800 900 1000
0

-200 Observed settl.

E.O.C
Predicted settl., F.E.M (Plaxis)
-400 Predicted settl., Asaoka method

-600
Settlement (mm)

-800

-1000

-1200

-1400

-1600

-1800

Figure 3. Settlement accumulation at point located on embankment centerline and at ground Surface.

Time (Days) Time (Days)

0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
0 0

-4 (a) -4
Excess pore water pressure (kPa)
Excess pore water pressure (kPa)

-8 -8
(b)
-12 -12

-16 -16

-20 -20

-24 -24

-28 -28

-32 -32
at depth = 2m at depth = 2m
-36 at depth = 4m -36 at depth = 4m
at depth = 6m at depth = 6m
-40 -40
at depth = 8m at depth = 8m
-44 -44

a. Excess pore water pressure distribution with time at 0.75 m away b. Excess pore water pressure distribution with time at the
from the embankment centerline at various depths embankment toeat various depths

Figure 4. Profiles of excess pore water pressure distribution with time .

4 SENSITIVITY STUDY RESULTS

A sensitivity analysis was carried out for the considered embankment to investigate the effect of some parameters that may
have an influence on the embankment deformational response. The numerically investigated parameters were PVD length,
PVD spacing, improved ground width, smear zone permeability and frictional angle of the soft clay.

4.1 Effect of the PVD length

To explore the effect of the PVD length on the embankment deformational response, the embankment under consideration was
reanalyzed considering that the PVDs were extended to the centerline of the medium stiff clay layer, i.e. 14m in length, and in
another case, the PVDs were installed to reach the end of this stratum, i.e. PVD length of 16m. The settlement accumulation
with time at a point located at the intersection of the embankment centerline and the ground surface is shown in figure 5. The
settlement trend along the embankment base and the horizontal displacement trend along a line through the edge of the
embankment crestare shown in Figure 6. These figures show that extending the PVD to 14 and 16m below ground surface to
reach the underlying stiff clay layer did not significantly affect the deformational response of the embankment.

418
Time (Day)
0 100 200 300 400 500 600 700 800 900 1000
0

Observed settl.
-200

E.O.C
Predicted settl.,length (ref.)=12m
Predicted settl.,length =14m
-400
Predicted settl.,length =16m

Settlement (mm)
-600

-800

-1000

-1200

-1400

-1600

Figure 5. Effect of the PVD length on settlement accumulation at intersection

Horizontal displacement (mm)
Distance from embankment C.L (m)
400 0 20 40 60 80 100 120 140 160 180 200 220 240 260
(a) 0
200
-2
(b)
0 -4

Depth below ground surface (m )

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 -6
-200
-8
Settlement (mm)

-400 -10
-12
-600
-14
-800 -16
-18
-1000
-20
-1200 -22 Long term predicted hz displacement ,length(ref.)=12m
Long term settle.,length(ref.)=12m
Long term settle.,length=14m -24 Long term predicted hz displacement,length=14m
-1400
Long term settle.,length=16m -26 Long term predicted hz displacement,length=16m
-1600 -28

ofembankment centerline and ground surface.

a. Vertical displacement profile at ground surface. b. Horizontal displacement profile at the embankment crest

Figure [Link] of the PVD length on the long term deformational response of the embankment.

4.2 Effect of the PVD spacing

To evaluate the effect of the PVD spacing, the embankment under consideration was reanalyzed considering the PVD spacing
to be 1m and 2m, in addition to 1.5m at the reference case. Figure 7 shows settlement accumulation with time at a point
located at the intersection of the embankment centerline and the ground surface. The figure exhibits that the final settlement is
independent of the PVD spacing. However, the degree of consolidation and the corresponding settlement at any time increases
with the decrease of the PVD spacing within the investigated range (1 to 2m). Therefore, in any project, the PVD spacing
should be optimized to reach the final settlement on the basis of time and economic restrictions.
Time (Days)
0 100 200 300 400 500 600 700 800 900 1000
0

-200 Observed settl.

E.O.C

Predicted settl.,spacing (ref.)=1.5m

-400 Predicted settl.,spacing=2m
Predicted settl.,spacing =1m
Settlement (mm)

-600

-800

-1000

-1200

-1400

-1600

Figure 7. Effect of the PVD spacing on settlement accumulation at intersection

of embankment center line and ground surface.

4.3 Effect of the width of the improved ground

The influence of width of the improved ground beneath the embankment was also examined. Two values of a coverage ratio
Wr, which is defined as the ratio of the width of the zoneimproved by PVDs to the wholewidth of the embankment base,
were analyzed. These two cases are; thereference case where Wr =74% and the case where ground along the whole width of
the embankment was treated (Wr =100%). Figure 8 shows the accumulative settlement at point of intersection between the
embankment centerline and the ground surface with time. The figure indicates that the width of the improved ground has no
effect on the settlement at the considered point. Figures 9-a presents the effect of the improved width on the settlement
419
profilealong the embankment base.. The figure shows that, the increaseinthe coverage ratio, for the considered variation
range, has no significant effect on the embankment settlement,
settlement while the heave outside the embankment toe slightly
decreased with increasing the improved width. Figure 9-bshows thehorizontal
ntal displacement trend aalong a line passing
through the edge of the embankment crest. The figure exhibits that the increase of the improved width resulted in an
insignificant increase in the lateral displacement.
displacement

Figure8.. Effect of the coverage ratio(Wr) on settlement accumulation at

Intersectionof embankment center line and ground surface.
surface

a. Vertical displacement profile at ground surface. b. Horizontal displacement profile below the embankment crest
[Link] of the coverage ratio(Wr) on the long term deformational response of the embankment.

4.4 Effect of Smear Zone Permeability

The installation of vertical drains by means of mandrel causes some disturbance of the subsoil surrounding the mandrel. The
disturbed zone is known as the smear zone. This disturbance leads to a reduction of the permeability coefficient of this zone zone.
To investigate the effect of such permeability variation
variation on embankment behavior, various values of the ratio (Rs),defined as
the ratio of the coefficient of permeability of the original soil (kh) to that inside the smear zone (k'h), wereanalyzed. Figure 10
showss the settlement accumulation with time at a point located at the intersection of the embankment centerline and the
ground surface. The figure points out that when PVDs
PVD were considered as perfect drains (no smear, i.e. Rs = 1), the predicted
settlement profile was overestimated and the inclusion of smear effect significantly improved the prediction. A better match
between the predicted and the monitored data was noted when Rs was taken equal ten, which agrees with the results reported
by Bergado et al. (1993b) for soft
oft Bangkok clay. In addition, the figure
f shows that the final settlement of the system was not
affected by the value of the permeability ratio, in spite of a greater degree of consolidation
consolidation with time for the perfect drain.
Time (Day)
0 100 200 300 400 500 600 700 800 900 1000
0

Observed settl.
-200
Predicted settl.,Rs (ref.)=10
E.O.C

Predicted settl.,Rs =5
-400
Predicted settl.,perfect drain
Settlement (mm)

-600

-800

-1000

-1200

-1400

-1600

Figure 10. Effect of the smear zone permeability on settlement accumulation at

intersection
tersection of embankment centerline
center and ground surface.
420
4.5 Effect of the Friction Angle of the Soft Clay

To assess the effect of the soft clay effective friction angle, Ф', values varying from 16o to 24o were implemented in the
numerical analysis. Figure 11-a shows the settlement accumulation with time at a point located at the intersection of the
embankment centerline and the ground surface. It can be concluded that the soft clay friction angle has no significant effect on
the predicted settlement trend. Whereas, figure 11-b shows that, thelong term horizontal displacement profile along a line
passing through the edge of the embankment crest, nonlinearly increaseswith the decrease of the friction angle. Reducing the
friction angle from 24oto16o, led to about 18% increase in the maximum horizontal displacement below the embankment crest.

5 CONCLUSIONS

This study numerically investigated the performance of a monitored earth embankment that was constructed on a very soft to
soft clay formation improved by PVDs. Two-dimensional plain strain finite element models were adopted in the current
numerical analyses. A good comparison was revealed between the monitored and the predicted responses. For the considered
embankment, the maximum settlement along the embankment base was found to occur near the embankment centerline.
Whereas, the maximum lateral displacement was found to occur below ground surface, within the upper soft/ very soft clay
layer. In addition, the rate of dissipation of the excess pore pressure near the embankment centerline was found to be greater
than that at the toe due to the existence of PVDs at the center.
Horizontal displacement (mm)
Time (Days)
0 100 200 300 400 500 600 700 800 900 1000 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280
0 0

Observed settl. -2
E.O.C

-200 -4
Predicted settl.,phi (ref.)=20
Depth below ground surface(m)
Predicted settl.,phi =16 -6
-400 Predicted settl.,phi =24 -8
Settlement (mm)

-10
-600
-12
-800 -14
-16
-1000 -18
-20
-1200
-22 Long term predicted hz displacement ,phi (ref.)=20
-24 Long term predicted hz displacement ,phi =16
-1400
-26 Long term predicted hz displacement ,phi =24
-1600 -28
a. Settlement accumulation at intersection of embankment b. The long term horizontal displacement profile
centerline and ground surface. below the embankment crest
Figure [Link] of the variation of the friction angle of the soft clay on the embankment behavior.

A sensitivity study was carried out to highlight the influence of some design parameters that may have significant impactson
the performance and the cost of constructing such embankments. The conclusions obtained are as follows:
 Eextending the prefabricated vertical drains through the entire thickness of clay formations is not necessary as it does not
significantly affect the deformational response of the embankment.
 Decreasing the PVDs spacing, accelerates the consolidation process and the settlementrate at any time. But the final
settlement is independent of the PVDs spacing.
 Increasing the ground improvement width across the embankment base to cover 100% instead of 74% of the total width
has no significant effect on the deformational response of the embankment, except for a slight reduction of heave near
the toe.
 Neglecting the smear effect (considering perfect drain) results in overestimation of the predicted degree of consolidation,
while inclusion of smear effect significantly improves the predicted results
 The variation of soft clay frictional angle, for the range considered in this study, did not significantly affect the
embankment settlement response.
 The horizontal displacement below the embankment crest increases with the decrease of the friction angle in a nonlinear trend.

REFERENCES

Arulrajah A., Nikraz H., Bo M.W. (2004) Factors Affecting Field Instrumentation Assessment of Marine Clay Treated with Prefabricated
Vertical Drains. Geotextiles and Geomembranes, 22, 415-437
Arulrajah A. (2005) Field Measurements and Back-Analysis of Marine Clay Geotechnical Characteristics under [Link] Thesis,
Curtin University of Technology.
Barron R. A. (1948) Consolidation of fine-grained soil by drain wells. Transactions ASCE, 113, 718-742.
Asaoka A. (1978) Observational procedure of settlement [Link] and Foundation, 18, No. 4, 87-101.
Bergado D.T., Mukherjee K., Alfaro M.C., Balasubramaniam A.S. (1993b) Prediction of vertical band drain performance by finite element
method. Geotextiles and Geomembranes J., 12, No. 6, 567-586.
Bergado D.T., Long P.V. (1994) Numerical analysis of embankment on subsiding ground improved by vertical drains and granular piles.
Proc. 13th International Conference in Soil Mechanics and Foundation Engineering, New Delhi, 1361-1366.

421
Bergado D.T., Anderson L.R., Miura N., Balasubramaniam A.S. (1996) Soft Ground Improvement in Lowland and Other
[Link] Press, NewYork, 427.
Bergado D.T., Balasubramanian A.S., Fannin R.J., Holtz R.D. (2002) Prefabricated vertical drain (PVDs) in soft Bangkok clay: a case study
of the new Bangkok International Airport project. Canadian Geotechnical Journal, 39, 304-315.
Brinkgreve R.B.J. (2002) PLAXIS Finite Element Code for Soil and Rock Analysis, Version 8. Lisse: Balkema.
Chai J.C., Shen S.L., Miura N. Bergado D.T. (2001) Simple method of modeling PVD improved subsoil. Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, 127 (11), 965-972.
Hansbo S. (1981) Consolidation of fine-grained soils by prefabricated [Link]. 10th International Conference in Soil Mechanics and
FoundationEngineering, Stockholm, Vol. 3, 12-22.
Hien Y. (2008) Effects of prefabricated vertical drain on the performance of embankment on soft [Link] thesis, University of
Teknologi, Malysia, 44-64.
Hird C.C. (1992) Finite Element Modelling of Vertical Drains beneath Embankments on Soft Ground. Geotechnique, 42 (3), 499-511.
Holtz R.D., Jamiolkowski M., Lancellotta R., Pedroni S. (1991) Prefabricated vertical drains: design and performance. CIRIA ground
engineering report: ground improvement, Butterworth-Heinemann Ltd, UK.
Indraratna B. Redana I.W. (1999) Closure: Plane strain modeling of smear effects associated with vertical drains. J. Geotechnical and
Geoenvironmental Eng. ASCE, 125, No. 1, pp. 96-99
Indraratna B. Redana I.W. (2000) Numerical modeling of vertical drain with smear and well resistance installed in soft clay. Canadian
Geotechnical Journal, 37(1), 132-146.
Lin D.G., Kim H.K. Balasubramaniam A.S. (2000) Numerical modelling of prefabricated vertical drain. Geotechnical engineering Journal
of the Southeast Asian Geotechnical Society, 31, No. 2, 109-125.
Mestat P.D. (2001) An Overview of 25 Years of Numerical Modelling of Test Embankments and [Link] Methods and Advances
in Aeromechanics, Desai et al. (eds.), Balkema: 1521-1526.

422
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Analysis OfPerformance Of Preloading Vertical Drain System For Improvement Of Soft

Clay Soil Comparative Study Between 2D And 3D Numerical Techniques
Analyse des performances des système de drainage vertical préchargement pour l'amélioration des sols
souple Clay étude comparative entre 2D et 3D Techniques numériques

M. H. Rabie*1 and M. A. El Hamed 2

1,2
Civil Engineering Department, Faculty of Engineering, Helwan University, Cairo, Egypt

ABSTRACT - This paper describes a comparative study between 2D and 3D numerical techniques for the performance of
preloading using vertical drain in soft clay [Link] analysis using 2D numerical technique was performed by finite
element code (Plaxis v8), equivalent vertical permeability of the drain was used to simulate plain strain condition. Settlement
analysis using 3D numerical technique was performed by finite difference code (Settle 3D). Modelling was carried out and
results are compared.
In addition, Extensive parametric studies were performed to investigate certain parameters effect such as spacing ratio (n),
penetration ratio, the effect of smear disturbance as well as permeability ratio, the diameter of smear zone and the drain
pattern.

RÉSUMÉ - Cet article décrit une étude comparative entre les techniques numériques 2D et 3D pour l'exécution de
préchargement utilisant drain vertical dans le sol de l'argile molle. L'analyse de paiement en utilisant la technique numérique
2D a été réalisée par le code d'éléments finis (Plaxis v8), la perméabilité verticale équivalente de la fuite a été utilisé pour
simuler l'état brut de souche. L'analyse de paiement en utilisant des techniques numériques 3D a été réalisée par le code de
différence finie (Settle 3D). La modélisation a été effectuée et les résultats sont comparés.

1. INTRODUCTION

Soft soil deposits are widespread in the coastal area. The soil possess low strength and high compressibility and thus
having low bearing capacity, therefore it needs to be improved to avoid excessive settlement and prevent stability failure.
Many ground improvement methods have been proposed and used in soft soil to improve its bearing capacity and minimize
the anticipated settlement.
Preloading is one of the most effective and economical methods to reduce settlement and improve the bearing capacity of
soft soil. The application of preloading helps in applying stress to the existing soil, thus; reduces the post construction
settlement. With the elimination of water from the soil, the soil strength is increased. However, this method does not have the
advantages of speeding up the process of pore water pressure dissipation hence settlement might takes years to complete.
Therefore, the preloading method is usually combined with vertical drain to speed up the consolidation process.
Hansbo (1981) derived an approximate solution for vertical drain based on the 'equal strain hypotheses by Baron's (1948)
solution' by taking both smear and well resistance into consideration. Presently, this analytical approach is the widely used
method in the design of vertical drains. Further development to reflect the axisymemetric condition of vertical drain in plane-
strain condition to employ a realistic two-dimensional finite element analysis for vertical drains was established by Indraratna
and Redana (1997), Line et al. (2000) Chai et al. (2001) and others. This method enables a less restrictive and comprehensive
analysis to be carried out in comparison to the analytical method.

2. SIMULATION OF VERTICAL DRAINS FOR 2D ANALYSIS

The vertical drain function is modeled by application of the drain element built in the program. Drains can be activated and
deactivated in calculation phases making it possible to account for the delayed installation of vertical [Link] paper focus
only on the permeability matching procedure proposed by Line et al; (2000) and Chai et al; (2001) methods.

2.1 Equivalent horizontal permeability methodLine et al; (2000)

Line et al; (2000) method proposed that, an equivalent horizontal permeability of soil surrounding the vertical drains in
consideration of the smear effect can be applied for an axi-symmetric unit cell analysis. The equivalent horizontal permeability
of surrounding soils, ke (Line et al., 2000) is as defined;

 r 
K hln  e 
K   rw 
 r   kh   r 
e

Where: ln  e     ln  s 
 rw   ks   rw 
423
re: radius of influence zone, rw: equivalent radius of vertical drain, rs: radius of smear zone, kh: horizontal permeability of the
undisturbed soil and ks: horizontal permeability of soil within the smear zone.

The conversion from radial flow of an axisymmetric unit cell to 2D plane flow of continues drainage wall system of plain
strain unit cell can be carried out as proposed by Line et al., (2000) by the following equation;
K hax 
K hpl 
  ln S   0 . 75 
6  ln  i    hax
n k
  
Where:   S   k sax 

khpl: Horizontal permeability of undisturbed zone, khax: horizontal permeability of undisturbed zone, ksax:
horizontal permeability of smear zone, ni: Influence ratio given by re/rw , S: Smear ratio given by rs/rw.

Figure1: Finite element mesh with Line et al.; 2000 method

2.2 Equivalent vertical permeability methodChai et al; (2001)

Chai et al. (2001) proposed a simplified method for approximating the effect of vertical drains. Vertical drains increase the
mass permeability in the vertical direction. Consequently, it is possible to establish a value of permeability of the natural
subsoil and the radial permeability towards the vertical drain. This equivalent permeability (kve) can be determined as follows:

2 . 26 L 2 k
k  (1  h
)k
 d e2 k v
ve v

de k d 3 2 k
  ln( )  ( h  1) ln( s )    ( l 2 ) h
Where: dw ks dw 4 3 qw
de: Equivalent influence zone diameter, dw: Equivalent drain diameter, ds: Smeared zone diameter, ks and kh:
Smeared and undisturbed horizontal permeability, L: Length of one-way drainage and qw: Discharge capacity.

Figure2: Finite element mesh with Chai et al., 2001 method

424
3. FINITE DIFFERENCE METHOD (3D ANALYSIS)

FDM analysis can be carried out conveniently using Settle 3D numerical modeling software. Settle 3D has been developed to
analyses the settlement of foundations under the application of circular, rectangular, embankment or excavation loads
(Rocscience, 2009). Settle3D calculates stresses in three dimensions using Boussinesq's equations and the 2:1 method
(assumes load distribution varies with depth as 2H: 1V) and settlements in one dimension using Taylor's (1942) and Terzaghi's
(1925) theories. The software is convenient for settlement analysis with prefabricated vertical drains and surcharge
applications in which the input parameters are easily achievable and understandable. The governing equation in consolidation
has one additional independent variable i.e. time.

Figure 3: Finite difference 3D Model

4. CASE HISTORY

The test embankments at Saga Airport were reported by Bergado et al. (1996). Saga Airport is located 13 km south of the city
of Saga on reclaimed land close to Ariake BayCountry . The deposit mainly consists of soft and highly compressible Ariake
clay. For verifying the effect of vertical drain improvement for this deposit, three test embankments were constructed on
natural, PVD improved, and SD improved subsoil (Bergado et al. 1996).

Figure 4: Cross Section of Test Embankment in Saga Airport

The embankments had the same geometry, with a fill thickness of 3.5 m, base width of (71 X 71 m), and top width of (25 X 25
m). The filling speed was about 0.03 m/day. The PVD was installed to around 25 m deep with an area of 45 by 45 m. PVDs
were installed in a square pattern with a spacing S=1.5 m. The embankment construction time of about 200 days. Surcharge
load was 70 kPa (3.5 m fill thickness).

425
4.1 Subsoil condition

At the test site, the soft layer is about 25 m deep, consisting of three clay layers and two sand layers. The thickness of the soil
layers slightly vary at three embankment locations. The top weathered crust (B) is about 1.0 m thick, followed by the first soft
clay layer, Ac1, with a thickness of 2.5 to 3.0 m. A sand layer, As1, of 1.1 to 1.5 m thick, underlies Ac1. The main clay layer,
Ac2, below As1, is very soft, with a thickness of 13.5 to 15.5 m. Under Ac2 is a sand layer, As2, which has a thickness of 0.5
to 2.7 m. The third clay layer, Ac3, is soft to medium stiff and has a thickness of 1.3 to 3.6 m underlying a thick and dense
sand layer (DS). The soil profile for PVD section is indicated in Fig. 5.1. The groundwater level of 1.0 m below ground
surface.
The soil parameter for Saga Airport is summarized in Table 1.

Table 1: Soil parameters for Saga Airport test embankment

Kx (10-8 Ky (10-8
Layer E (kpa)  Cc Cr M e0 KN/m3
m/s) m/s)

B --- 0.25  0.028 1.3 2 15.0 11.45 7.6

AC1 --- 0.3  0.051 1.2 2 14.5 5.7 3.8
AS1 10000 0.2  --- --- --- 15.5 290 290
AC2 --- 0.3  0.10 1.2 2.5 14.5 2.64 1.76
AS2 15000 0.2  --- --- --- 16.0 290 290
AC3 --- 0.3  0.034 1.3 1.75 16.0 2.64 1.76
Ds 30000 0.2  --- --- --- 19.0 290 290

4.2 Settlement monitoring results

Figure 5:Monitoring results with the progress of backfilling.

4.3 Soil model

Plaxis (2D analysis), Mohr-Coulomb model was used for the Ariake clay (AC), sand layer (AS) and backfilling of
embankment. Mohr-Coulomb model is based on Mohr-Coulomb failure criterion. Settle 3D (3D analysis), the non-linear
model was used the Ariake clay (AC), sand layer (AS).

426
Table 2: Soil parameters adopted for 2D analysis and 3D analysis

Parameter B AC1 AS1 AC2 AS2 AC3 DS

model MC MC MC MC MC MC MC

behavior Un-drained Un-drained Drained Un-drained Drained Un-drained Drained

Khpl 2.69 1.34 6.82 6.21 6.82 6.21 6.82
(m/day) E-04 E-04 E-03 E-05 E-03 E-05 E-03
Kve 7.85 6.04 2.62 4.02 2.62 4.02 2.62
(m/day) E-02 E-02 E-01 E-02 E-01 E-02 E-01
E (kN/m2) 1200 600 10000 400 15000 800 30000
C (kN/m2) 1 1 1 1 1 1 1
Φ 28 28 35 28 35 28 40

4.4 Comparison of measured and computed time rate of settlement

Figure 6 displays the computed and measured normalized settlement versus time curve for each case.

Time (day)
0 100 200 300 400 500 600 700
0
Measured
Chaie Method
-0.5 Line Method
Settle 3D
Settlement (m)

-1

-1.5

-2

-2.5

-3

Figure 6: Measured and calculated settlement versus time

5. PARAMETRIC STUDIES

Many factors affecting the rate of settlement for case of vertical drains. There factor includes the effect of pattern type, drain
spacing and smear effect. The degree of consolidation for various mentioned parameter had been calculated. The degree of
consolidation had been calculated at two points. The first point (A) is at the centre line of an embankment while the second
point (B) is at the crest point of an embankment. Drains were assumed to be extended within the full width of embankment.

5.1 Effect of spacing ratio

As mentioned earlier, to evaluate the effect of the spacing ratio (n) on the degree of consolidation (U %). The study shall be
based on spacing ratio (n) of 17 and 51 for cases of fully and partially penetrating drain PR=100%, 73% & 50%. The models
names can be summarizes as follows:

427
(U%)

(U%)
100 100
Line Method
80 80 Chai Method
60 60 Settle 3D

40 40
Line Method
20 Chai Method 20
Settle 3D
0 0
5 15 25 (n) 35 45 55 5 15 25 (n) 35 45 55
Figure 7:: Degree of consolidation VS spacing ratio – point (A) – Figure 8: Degree of consolidation VS spacing ratio – point (A) –
(200 day – PR=100%) (200 day – PR=73%)

100
(U%)

80 Line Method
Chai Method
60 Settle 3D

0
5 15 25 (n) 35 45 55

Figure 9: Degree of consolidation VS spacing ratio – point (A) – Figure 10: Degree of consolidation VS spacing ratio – point (B) –
(200 day – PR=50%) (200 day – PR=100%)
(U%)
(U%)

100 100
Line Method Line Method
80 Chai Method 80 Chai Method
Settle 3D Settle 3D
60 60

40 40

20 20

0 0
5 15 25 (n) 35 45 55 5 15 25 (n) 35 45 55

Figure 11: Degree of consolidation VS spacing ratio – point (B) – Figure 12: Degree of consolidation VS spacing ratio – point (B) –
(200 day– PR=73%) (200 day – PR=50%)

5.2 Effect of smear zone

To
o evaluate the effect of the permeability ratio (kh/ks) on the degree of consolidation (U %). The study shall be based on
permeability ratio (kh/ks) of 1, 2 and 3 for cases of fully partially penetrating drain PR=100%, 73% & 50%.

428
Line Method Chai Method Settle 3D Line Method Chai Method Settle 3D
96 99 100 98 99 97 98
91 87 100
100
77 73 78 73 77 76
80 80 66 68
62
60 60

(U%)
(U%)

40 40

20 20

0 0
1 2 3 1 2 3
(kh/ks) (kh/ks)

Figure 13: Degree of consolidation VS kh/ks ratio – point (A) – (At Figure 14: Degree of consolidation VS kh/ks ratio – point (A) – (At
Certain time=200 day and n=17) for model (M5-a)
(M5 – (PR=100%) Certain time=200 day and n=17) for model (M7
(M7-a) – (PR=73%)

Line Method Chai Method Settle 3D

100

80
53 52 53 50 47 52 48 44 51
60
(U%)

0
1 2 3
kh/ks

Figure 15: Degree of consolidation VS kh/ks ratio – point (A) – (At Figure 16: Degree of consolidation VS kh/ks ratio – point (A) – (At
Certain time=200 day and n=17) for model (M8-a)
(M8 – (PR=50%) Certain time=200 day and n=51) for model (M5
(M5-a) – (PR=100%)

Line Method Chai Method Settle 3D Line Method Chai Method Settle 3D
100 100 99
91 96 87
94 93 100
100 84
80 80

60 46 50
(U%)

(U%)

60 44 45
40 36 40 40
35
40 40
20 20
0 0
1 2 3 1 2 3
(kh/ks) kh/ks

Figure 17: Degree of consolidation VS kh/ks ratio – point (B) – (At Figure 18: Degree of consolidation VS kh/ks ratio – point (B) – (At
Certain time=200 day and n=17) for model (M5-a)
(M5 – (PR=100%) Certain time=200 day and n=51) for model (M5
(M5-a) – (PR=100%)

5.3 Effect of drain pattern

The vertical drain are usually install in square patterns or triangular patterns. An extensive study had been carried out in order
to evaluate effect of drain patterns on the degree of consolidation (U%).

429
Line Method Chai Method Settle 3D Line Method Chai Method Settle 3D
96 99 100 97 99.5 100 77 78 77 76 78
100 73
80
80
60
60

(U%)
(U%)

40
40

20 20

0 0
Square patt. Traingular Patt. Square patt. Traingular Patt.
Type Of Patterns Type Of Patterns

Figure 19: Degree of consolidation VS pattern arrangement – point Figure 20: Degree of consolidation VS pattern arrangement – point
(A) – (Time=200 day and n=17) for model (M9) – (PR=100%) (A) – (Time=200 day and n=17) for model (M11) – (PR=73%)

Line Method Chai Method Settle 3D Line Method Chai Method Settle 3D

100 100

80 80 57
51 55 52
53 52 53 53 54 54 (U%)
60 39 40
(U%)

60
40 40
20 20
0 0
Square patt. Traingular Patt. Square patt. Traingular Patt.
Type Of Patterns Type Of Patterns

Figure 21: Degree of consolidation VS pattern arrangement – point Figure 22: Degree of consolidation VS pattern arrangement – point
(A) – (Time=200 day and n=17) for model (M12) – (PR=50%) (A) – (Time=200 day and n=51) for model (M9) – (PR=100%)

6. SUMMARY AND CONCLUSION

6.1 Observed from case study

 The results shows close agreement between measured values and that predicted by 2D and 3D analysis
 The results showed also that 3D analysis gives slightly higher settlement and in turn, slightly higher degree of
consolidation compared to 2D analysis.
 The difference between the 3D and 2D analysis vanishes for consolation time higher than 200 day.

6.2 Observed from parametric studies

 In general, the degree of consolidation concluded from 3D analysis is higher than 2D analysis.
 For 3D analysis, the degree of consolidation for two points (A) & (B) which are located at the center and crest of
embankment respectively, is almost the same for fully and partially penetrating drain.
 For 2-D
D analysis, the degree of consolidation
consolidation for two points (A) & (B) which are located at the center and crest of
embankment respectively, is almost the same for fully and partially penetrating drain.
 For case of fully penetrating drain, method proposed by Line et al; (2000) for 2D analysis gives
gives a degree of consolidation
lowest compared to that method proposed Chai et al; (2001) for 2D analysis and also 3D analysis. The percentage of
decrease between the degree of consolidation concluded by Line et al; (2000) and that concluded by Chai et al; (2001) is
ranging between 10% and 26%. The percentage of decrease between the degree of consolidation concluded by Line et al;
(2000) and that concluded 3D analysis is ranging between 10% and 30%. Finally, the percentage of decrease between the
degree of consolidation concluded by Chai et al; (2001) and the 3D analysis is ranging between 7% and 10%.
 For case of partially penetrating drain, method proposed by Chai et al; (2001) gives a degree of consolidation lowest
compared to that method proposedosed Line et al; (2000) for 2D analysis and also 3D analysis. The percentage of decrease
between the degree of consolidation concluded by Chai et al; (2001) and that concluded by Line et al; (2000) is ranging
between 10% and 20%. The percentage of between the degree of consolidation concluded by Line et al; (2000) and that
concluded 3D analysis is ranging between 10% and 20%. Finally, the percentage of between the degree of consolidation
concluded by Chai et al; (2001) and the 3D analysis is ranging between
betwe 15% and 30%.

430
 Decreasing the spacing ratio decreases the difference between the results of 2D and 3D analysis. This means that both 3D
analysis and 2D analysis using Line et al (2000) as well as Chai et al; (2001) show good agreement for closed spaced
drains while for wide spaced drains a difference in behaviour is noticed comparing the 3D analysis and the two used
methods for 2D analysis.
 Triangular pattern gives higher degree of consolidation compared to rectangular pattern for 3D analysis and 2D analysis
using Chai et al; (2001) method. Line et al (2000) method does not account for pattern.

7. REFERENCES

1. Adachi, K. and Todo, H. 1979. A case study on settlement of soft clay in penang. Proc. 6th Asian Regional conf. on Soil
Mech. And Found. Engineering, 1: 117-120.
2. Chai, J. C. and Miura N. (1999) Investigations of Factors Affecting Vertical Drain Behavior. Journal of Geotechnical and
Geoenvironment Engineering. ASCE, March: 216-226.
3. Chai, J. C., Shen, S. L., Miura, N. and Bergado, D. T. (2001). Simple method of modeling PVD improved subsoil. Journal
of Geotechnical and Geoenvironmental Engineering, ASCE, 127 (11): 965-972
4. Hird C. C. et al. (1995). Modelling the effect of vertical drains in two – dimensional finite element analyses of
embankments on soft ground. Canadian Geotechnical Journal (32): 795-807.
5. Rocscience (2009). “Settle 3D 2.0 Settlement and consolidation Analysis”, Website:
[Link]

431
432
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

High strength earth blocks reinforced with natural and polymeric fibers
Blocs de terre de haute résistance renforcés avec des fibres naturelles et polymères
M.M. El-Emam*1, A.K. Al-Tamimi2, A. Aljallad3, M. Al-Moukdad3, A. Al-Marei3, and M. Noaman3
1
Associate Professor, American University of Sharjah, Sharjah, United Arab Emirates
2
Professor, American University of Sharjah, Sharjah, United Arab Emirates
3
Student, American University of Sharjah, Sharjah, United Arab Emirates
*
Corresponding Author

ABSTRACT -This paper presents strength and deformation evaluation of earth blocks reinforced with different types and fiber
contents. Two different types of reinforcing fibres were used: natural Palm Tree Fiber and synthetic fiber. The soil used in
casting the blocks is composed of dune sand obtained locally from Sharjah deserts, and clayey silt acquired from Al-Ain city,
UAE. Several mixes were casted and tested in the compression machine available at the AUS material lab. The main
parameters investigated in this study were; clay content, fiber type and content, in addition to cement inclusion. Results
indicated that, the addition of fiber, natural or polymeric, to carefully casted earth blocks improved the compressive strength of
these blocks significantly. In addition, clay content played an important role in dictating the compressive strength of the fiber
reinforced earth blocks. The maximum compressive strength of fiber reinforced soil-blocks showed more than 33% increase
above the maximum compressive strength recorded for identical soil block without fiber. In addition, the post- and pre-peak
strength behaviour of the earth block changed with increasing fiber content.

RÉSUMÉ– La prédiction de la résistance et de la déformation de briques de terre renforcés par différents types et proportions
de fibre est analysée. Deux types de fibres de renforcement ont été utilisés : des fibres naturelles de palmiers et des fibres
synthétiques. Le matériau utilisé pour le coulage des blocs est composé de sable de dune du désert de Sharjah et de silt
argileux issu de la ville d'Al Ain, Émirats Arabes Unis. Plusieurs mélanges ont été testés à l’essai de compression simple du
laboratoire. Les paramètres examinés dans cette étude sont : la teneur en argile, le type et le pourcentage des fibres, et le
dosage de ciment. Les résultats montrent que l'ajout de fibres, naturelles ou de polymères, de blocs de terre améliorent de
façon significative la résistance à la compression des blocs. En outre, la teneur en argile joue un rôle important sur la
résistance à la compression des blocs de terre renforcés de fibres. La résistance à la compression maximale de fibres renforcée
sol-blocs a montré plus de 33 % augmentation supérieure à la résistance à la compression maximale enregistrée pour bloc de
sol identique sans fibre. En outre le comportement en post et prérupture des briques de terre varieen fonction de la teneur en
fibres.

1 INTRODUCTION

Adobe blocks have been used for construction of buildings since ancient times. Heathcote (1995) reported earth-wall buildings
that have survived over 100 years old of service in England and France with careful maintenance. It was also reported in drier
climates such as Middle East. Morton (2008) reported that one third of the world populations are living in houses built of
adobe and indicated that this simple technology competed well with the modern materials and advanced methods in
construction industry. Construction of building using adobe has many advantages such as; the local availability of materials,
low cost, recyclable, and environment friendly. In addition, growing concerns and interests about environmental hazards
increased the used of earth as a building material. In this context, Venny Riza et al. (2011) indicated that compressed earth
bricks produced one-tenth of CO2 quantity produced by fired clay bricks compared to almost one ton of CO2 releases with
production of one ton of cement.. This makes the compressed earth blocks considered environmental friendly construction
materials. In addition, houses constructed using adobe provides better thermal insulation compared to other construction
materials, especially in tropically climates region such as Africa, South Asia and Middle East, where heat is extremely high
and poses an additional measure and concern. Therefore, thermal insulation becomes an important factor that must be
considered into the design of suitable and affordable housing in such regions. Despite of all these advantages, walls
constructed with adobe blocks have shown major disadvantages that needed to be investigated and improved. Among these
disadvantages comes the relatively large wall thickness, loss of strength on saturation and erosion during rainy seasons, and
strength degradation during earthquakes and winds. Elemam (2014) concluded that, stabilized soil might lose significant
portion of its strength during earthquake. Several researchers have worked in order to minimize or eliminate many of these
disadvantages. However, there are still many possibilities to explore more in enhancing properties of earth bricks. The current
research is a first trial to produce high performance earth blocks using locally available-natural material in UAE. It represents
the first stage in a long term research program to produce earth blocks to satisfy strength, durability, and thermal insulation
acceptable for the local construction industry. The work presented in this paper is focused mainly on the strength of the
compressed earth blocks.

433
2 LITERATURE REVIEW

Walker (2004) casted earth blocks with different sizes, different cement contents, and different soil properties that are
compacted statically at 2 MPa in specific molds. Blocks were cured for 28 days, and tested under uniaxial compression load in
the concrete cube testing machine. In addition to axial compression tests, modulus of rupture and erosion tests were also
conducted to indicate the overall quality and erosion properties of the blocks. Results reported by Walker (2004) showed an
increase in block compressive strength with cement content, which were in general agreement with findings of Houben and
Guillaud (1994) and Walker and Stace (1997). In addition; compressive strength was also improved with increasing block
density which attained by increasing compaction effort and/or clay contents. Furthermore, modulus of rapture was typically
improved with cement addition, while slightly affected with clay content and block geometry. Finally, the improvement in
erosion resistance was significant at low cement content.
Venkatarama Reddy and Gupta (2006) investigated the properties of stabilized mud blocks using cement-soil mortars.
Compressive stress-strain relationship and elastic properties of the block were investigated using three different cement-soil
mortars. Sandy soil mix was obtained by mixing red loamy soil and sand such that the resulting mix contained 9% clay, 17.7%
silt, and 73.3% sand. Cement contents of 6, 8, and 12% by weight of the mix were used in the study. It was concluded that, the
wet compressive strength increased and the water absorption decreased with cement content. In addition, initial tangent
modulus increased by 2.5 times when the cement content of the block increased from 6 to 8%. However, there was only a
marginal increase in the modulus when cement content increased from 8 to 12%. Finally, the values of compressive strain at
ultimate compressive strength for all blocks were in a very close range of 0.32% to 0.35%. Similar results have been reported
by Chan and Liang (2010) who studied the cement content effect on the unconfined compressive strength of earth cylinders.
Venkatarama et al. (2007) examined the effects of soil grading and clay content in the soil-cement block compressive
strength, water absorption, and durability. Four different types of soils with different gradations and different clay contents
have been used. Each soil was mixed with two different percentages of cement, and subjected to different tests including; axial
compression, bending, saturated water content, initial rate of absorption, and linear and lateral expansion on saturation tests.
Venkatarama Reddy et al. (2007) concluded that the largest compressive and flexural strength are obtained at 16% clay content
regardless of the cement ratio. In addition, the rate of water absorption decreased while the saturated water content increased
when the clay content increased. Finally, the minimum weight loss obtained at 16% clay content, and linear expansion, initial
tangent modulus and Poisson ratio increased with clay content.
Murillo et al. (2005) used low cost, locally available henequen natural fibers together with clay to produce reinforced soil.
Different fiber contents were used mainly to improve bending strength, thermal insulation, and shrinkage limit. Tests indicated
that, the compressive strength decreased and strains at peak stress increased as the fiber content increased. The reduction in
compressive strength was attributed to the increase in porosity of the reinforced soil with fiber content. This increase in
porosity led to a reduction in the specimen dry density, and therefore a reduction in strength. This result was in good
agreement with Khalaf and DeVenny (2002) who reported a reduction in compressive strength due to increase in the porosity.
However, increase in the porosity was found to improve the thermal insulation of the soil. Despite the improvement in both
tensile and bending performance of the soil, Murillo et al. (2005) noticed that the miniature cracks developed due to shrinkage
reduced the compressive strength, and good orientation of fiber bundles could reduce size and number of these cracks.
The objectives of the research work presented in this paper are to study and compare the strength characteristics of
stabilized earth blocks. To accomplish these objectives, an experimental program was conducted using both stabilized and
unstabilized earth blocks casted manually at the American University of Sharjah (AUS) geotechnical lab. Earth blocks are
casted from local sand, silt and clay and stabilized with both natural and synthetic (Polymeric) fibers. The influence of soil
constituents, fiber type and content, and cement addition on the compressive strength and stiffness of the tested blocks was
studied. Stress-strain behavior of different blocks with different properties is investigated in details. Experimental results and
conclusions are presented, together with recommendations for strength future improvement.

3 EARTH BLOCK MATERIALS AND EXPERIMENTAL PROGRAM

3.1 Soil and Fibers

The natural soil used for casting stabilized and unstabilized earth blocks was collected locally from two different sourced in
United Arab Emirates (UAE). The sandy soil was collected from the dunes of Sharjah, while, clayey-silt soil was collected
from a farm close to Al’Ain city. Characteristics of the soil, including grading, Atterberg limits, and compaction test results,
are outlined in Table 1a.
Two types of polymeric (synthetic) fiber have been acquired for this investigation, namely; black fiber (BF) and white fiber
(WF). Both types are manufactured of Polyolefin produced from simple olefin (CnH2n). Yarns of both types of fiber have
similar geometrical dimensions (i.e. length and diameter). However, yarns of white fiber showed less stiffness compared to
black fiber, in addition to smoother surface. Mechanical, physical, and thermal characteristics of black fiber as provided by the
manufacturer are shown in Table 1b. However, Figure 1 shows different types of fiber used in this study.
A third type of fiber which is a natural Palm Trees Fiber (PTF) was used in this investigation. Palm trees fiber is locally
and excessively available in UAE. The tensile strength and stiffness of PTF have been investigated by Marandi et al. (2008)
through tensile strength tests. The results indicated that, maximum tensile strength of 63.32 MPa was achieved at 11% Strain.

434
In addition, the PTF specific gravity (of solids) was 0.92 with elastic modulus of 600 MPa. In the current study, the moisture
absorption characteristics of the palm tree fibers were examined by soaking the fiber samples and weighing after 24 hours. The
results indicate that maximum water absorption of 144% was achieved. Marandi et al. (2008) measured maximum water
absorption of 187% for a palm tree fiber threaded to 0.35 mm diameter yarns. It is expected that the threaded palm tree fiber
has greater water absorption capacity compared to the full palm tree strands. The effect of water absorption, volume change of
the PTF on the fiber-soil adhesion characteristics and on the stress-strain behavior of the earth block is still under
investigation. It should be noted that yarns of the three different types of fiber have been adjusted to 30 mm length before
mixed with soil.

Table 1. Properties of soil mixture and synthetic fiber

a) Soil Properties b) Fiber Properties (Black)
Characteristic Value Characteristic Value
Sand content 50% Length 40 mm
Silt content 41% Tensile Strength 533 MPa
Clay content 9% Young’s Modulus 7.1 GPa
Liquid limit 27% No. of fiber per Kg >70,000
Plastic limit 17% Specific Gravity 0.91
Optimum moisture content 18% Melting point 150-160 C

Figure 1. Different types of fiber: a) White synthetic fiber (WF), b) Black synthetic fiber (BF), and c) Natural palm tree fiber (NF).

3.2 Soil-Fiber mixture preparation

Many activities took place in the background prior to mixing soil and fiber to cast earth blocks. The natural soil constituents
were prepared by initially air drying and sieving to remove particles larger than 5 mm. Then, dried clay clods were broken
down manually using a hummer specified for this purpose. Pre-specified weights of dried sand, silt and clay were then
thoroughly mixed together in a plastic container, prior to addition of the fiber content. Once soil constituents were mixed
thoroughly, a pre-specified weight of fiber was added to the dry soil and mixed until a uniform fiber distribution throughout
the soil attained. Figure 2a shows the soil-fiber mixture during mixing process. An amount of water sufficient to attain the soil
optimum moisture content of 18% was gradually added to the dry soil-fiber mix. Despite of block static compaction, rather
than dynamic compaction, experience indicated close correlations between Proctor and static compaction optimum moisture
contents (Al-Jabri et al. 2009).
Prior to casting process, wooden molds were polished with minimum amount of oil to ease the extraction of the earth
blocks out of the molds at later stage. The wet soil-fiber mixture was then placed in the mold in 3 layers, each layer was
tamped manually using aluminum stick with 1”x1” cross section, to remove as much internal air voids as possible. Casting
process and manual tamping of the mixture inside the mold is shown in Figure 2b. Compaction procedure during block casting
is a major factor that can significantly affect the compressive strength of the earth blocks. Results reported by Guettala et al.
(2002) indicated 70% increase in the compressive strength as a result of 4 times increase in the compaction effort. This result
is in agreement with Bahar et al. (2004) who reported 50% increase in the compressive strength of earth blocks as a result of
using dynamic compaction instead of static compaction.
Unconfined compression machine, in the geotechnical Lab., was used to apply static pressure at the top of the block inside
the mold, as shown in Figure 2c. All earth blocks were compacted using the manually operated compression machine that
applied 2.0 MPa static-pressure for 30 minutes. The static compaction effort was carefully controlled throughout the
production of all blocks in order to avoid different compaction efforts effect. After static compaction, blocks were subjected to
constant consolidation pressure of 11.5 kPa for 48 hours. After consolidation, earth blocks were extruded from the mold and
carefully stacked to be air dried. Initially, the drying process required to leave the blocks to dry at room temperature for 4 days
prior to placing them under the sun for 10 days, for full drying. The logic behind allowing the specimen to dry at room
temperature initially is to prevent any sudden shrinkage due to the excessive heat and exposure to sunlight that the block may
experience outdoors. A total of 21 different blocks composed of 2 unstabilized blocks, 16 blocks stabilized with different fiber
435
contents and types, and 3 blocks with 0.5% fiber content and 0.5% cement content, were prepared for the experimental
program. Figure 3 shows 4 blocks ready for compression test.

(a) (c)

(b)

Figure 2. Earth block casting procedures.

Figure [Link] blocks fully dried and ready to be tested.

3.3 Compressive Strength Test

Compressive strength is the most universally accepted value for determining the quality of earth blocks. Block compressive
strength is intensely related to the soil types, stabilizer content, cement-content, types of soil (plasticity index), compaction
pressure and types of compaction. Since there is no standard testing for earth blocks, it is usual to determine the compressive
strength of these blocks using ASTM (1984) or BSI (1985) testing method, that mainly specified for fired clay bricks and
concrete masonry block (Walker, 1995; Walker, 2004; Oti et al. 2009). In case of non-availability of compression machine,
Walker (1995; 2004) suggested to use the three points bending test, and proposed that the unconfined compressive strength is
about five to six times modulus of rupture.
In this study, compressive strength of the blocks was measured in accordance with ASTM (1984). After each block was
fully dried and weighed, its dimensions are measured carefully. The block was then placed between the steel plates of the
concrete compression machine available at the AUS structure lab. The block was adjusted at the center of the steel plate and
the compression load was applied in small increments until the block failure. Load-deformation values of the block were
recorded by the computer attached to the testing machine. The stress at any deformation was calculated using the gross area of
436
the block normal to the load, and the associated strain was calculated using the gross height of the block parallel to the load
direction. Results of a series of compressive strength tests are analyzed, compared and discussed in the next section. These
results highlight the effects of clay content, fiber type and content, and cement addition in both ultimate compressive strength
and strain, and modulus of elasticity (stiffness) of the fiber reinforced earth blocks.

4 RESULTS AND DISCUSSION

4.1 Effect of Clay Content

Effects of clay content on the ultimate compressive strength of unstabilized earth blocks are shown in Figure 4. The figure
indicated that the block compressive strength increases as the clay content increases until reaches the peak value at 14.4 clay
content. The block compressive strength dropped thereafter when the clay content increased beyond the 14.4 clay content. This
result is in good agreement with finding of Venkatarama Reddy et al. (2007) which reported a maximum compressive strength at
clay content of 16.3%. The slightly higher optimum clay content reported by Venkatarama Reddy et al. (2007) compared to the
current study could be attributed to their coarser sand compared to the sand used in the current investigation. Coarser sand may
require large clay volume to fill the larger voids entrapped between the particles compared to relatively fine sand. Results of
the current study also indicated that, as the clay content increased, the block secant stiffness, E50 was significantly decreased.
This is expected as the sand matrix is stiffer compared to the clay matrix. In addition, the strain at peak strength and strain at
50% peak strength increased significantly with clay content. Based on this behavior, 9% clay content was selected to be used
in this study as it satisfied compressive strength, secant stiffness, E50, and compressive strain at both peak strength and 50%
peak.

4 60 16
Peak Strength Stiffness (E50) (a) Axial Strain at Peak Strength (b)
Axial Strain at 50% Peak Strength
3 45 12
Stiffness, E50 (MPa)

Axial Strain (%)

Strength (MPa)

2 30 8

1 15 4

0 0 0
9 10.8 14.4 18 9 10.8 14.4 18
Clay content (%) Clay content (%)

Figure [Link] compressive strength, secant stiffness, and strains versus clay contents for unstabilized earth blocks.

4.2 Effect of Fiber Content

Effects of synthetic (polymeric) fiber content on stress-strain behavior of earth blocks are shown in Figure 5. While the
unstabilized blocks attained their peak strength at 4% strain, the fiber-stabilized blocks reached the peak strength at strains
higher than 4%. For both polymeric fiber types shown in Figure 5, strain required to mobilize peak compressive strength
increased as the fiber content increased. This could be attributed to the strain required to mobilize the friction between the
fiber and the soil matrix. As the fiber content increased, the volume of soil matrix confining each fiber strand reduced and
therefore, more strain needed to mobilize the full frictional resistance at the soil-fiber interfaces. This soil-fiber interface
behavior is more pronounced when the effect of fiber type is compared, Figure 5a&b. At the same fiber content percent,
blocks stabilized with white fiber (Figure 5b) attained their peak strength at strain larger than blocks stabilized with black fiber
(Figure 5a). It was explained earlier that, black fiber is stiffer and has a rougher surface area compared to white fiber.
Therefore, the boned strength (soil-fiber interface) of white fiber is expected to be less compared to black fiber. This explained
the larger strain required to mobilize peak strength in case of blocks stabilized with white fiber, compared to black fiber.
Visual inspection of many failed earth blocks indicated that the fiber did not fail in tension, instead failure occurred due to
pullout of the fiber yarns from the soil matrix (i.e. interface failure).
For blocks reinforced with 0.0% and 0.25% fiber, there was clear strain softening (i.e. reduction on strength with strain)
once the compressive strength reached peak values. For blocks with fiber contents higher than 0.25%, there was strain
hardening (i.e. increase in strength with strain) after peak. In fact there are no clear peaks in stress-strain curves for block with
fiber content larger than 0.25%. Therefore, for all blocks, strength at 4% strain was designated as peak strength in the analysis.
This typical value (i.e. 4% strain) was selected as reference strain because it is corresponding to peak strength for unstabilized
earth block. Strength and stiffness for earth blocks at 6% strain is also presented for additional analysis of the block behavior.
Figure 5 also indicated that, the fiber inclusion did not improve the block behavior significantly at initial stage (i.e. up 2%
strain). Instead, fiber inclusion led to reduction of strength at initial stage of testing, compared to unstabilized block. This
could be attributed to the less stiffness of fiber compared to soil matrix and to the relatively small block strain that was
437
insufficient to mobilize the full interaction between the fiber and the soil matrix. Therefore, at this stage the fiber plays a
negative role in reducing the block strength and stiffness. The only exception of that are blocks stabilized with 0.75% fiber
content, both black and white fibers. At this fiber content, it is expected that both fiber matrix and soil matrix have continuity.
The continuation of fiber matrix allows it to behave as truss elements inside the block to resist the compression stress, which
increase the stiffness of the blocks. Increasing the fiber content beyond this percentage (i.e. 0.75%) resulted in discontinuity of
the soil matrix, and therefore less confinement to the fiber into the block. As a result, the block stiffness at this early stage
decreased.
6 6
Fiber content (%) Fiber content (%)
0.0% 0.0%
0.25% 0.25%
Uniaxial stress,  (MPa)

0.5% 0.5%
4 0.75% 4 0.75%
1.0% 1.0%

2 2

a) Black Synthetic Fiber b) White Synthetic Fiber

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Compressive strain (%) Compressive strain (%)

Figure [Link] of polymeric fiber on the stress-strain behavior of earth blocks.

6
Fiber content (%)
0.0%
0.25%
Uniaxial stress,  (MPa)

0.5%
4 0.75%
1.0%

c) Natural Palm trees Fiber

0
0 2 4 6 8 10
Compressive strain (%)

Figure [Link] of natural palm tree fiber on the stress-strain behavior of earth blocks.

Figure 6 shows the stress-strain relationship for earth blocks stabilized with different percentage of natural palm tree fiber.
Adding natural fiber to earth blocks improved the compressive stress-strain behavior of the block. This includes improving the
peak compressive strength and stiffness with the addition of natural fiber up to 0.5%. For fiber content larger than 0.5%, there
was significant improvement on the peak strength; however, stabilized blocks showed less stiffness compared to unstabilized
blocks. It could be clearly noticed that for blocks stabilized with 0.75% natural fiber, the increase in strength was negligible,
and the softening in the initial stage of the stress-strain curve was clear.
Figure 7 showed a block stabilized with 0.5% natural palm tree fiber during compression test, at different compressive
strain. The figure indicated that, up to 2% strain (Figure 7a), there is no cracks appeared in the block and the major
contribution in block stiffness is from soil matrix, with little contribution from the relatively stiff natural fiber. Therefore, the
initial stiffness of the block increased beyond the unstabilized block stiffness. As the compressive strain increased up to 4%,
the soil matrix reached a premature initial failure, which indicated by the first appearance of cracks, as shown in Figure 7b. At
this strain level, the peak soil-fiber friction mobilized, leading to the peak compressive strength of the block. Once the
compressive strain increased beyond 4%, the block was already in a state of progressive failure, which is indicated by the
larger cracks appears in Figure 7c. At this stage the earth block showed a state of stain softening as indicated in Figure 6.

4.2.1. Block Compressive Strength and Stiffness

The effect of fiber content on the maximum compressive strength and stiffness for earth blocks stabilized with both polymeric
and natural fibers are shown Figures 8 and 9, respectively. As stated earlier, the peak strength was selected to be at 4%
reference strain, which is coincident with the strain corresponding to peak strength of unstabilized blocks. In addition, strength
and stiffness at 6% reference strain are also plotted. It is clear from Figure 8 that the optimum range of polymeric fiber content

438
is 0.6% to 0.8%. This fiber content range satisfied both maximum compressive strength and maximum stiffness, at both 4%
and 6% strains. It can also be concluded that the black fiber is more effective in improving block stiffness compared to white
fiber. However, white fiber slightly more effective in improving compressive strength compared to black fiber.
For earth blocks stabilized with palm tree natural fiber, Figure 9, it could be noticed that the optimum fiber content for both
stiffness and compressive strength is ranged from 0.2% to 0.6%. There is a sudden increase in compressive strength at fiber
content of 1%, however, it accompanied by a significant reduction on block stiffness. This is attributed to the more ductility of
the block stabilized with 1% natural fiber, which resulted in higher strain at 50% maximum strength. Therefore, it could be
concluded that the optimum natural fiber content is from 0.2% to 0.6%. This range is less than the 0.7 to 0.8% optimum
values.

a) Block at 2% compressive strain b) Block at 4% compressive

strain

b) Block at 6% compressive strain

Figure 7. Earth block stabilized with 0.5% natural palm tree fiber at different compressive strain.

5 100
a) Maximum compressive strength b) Maximum compressive strength
Black Finer
4 80 at 4% strain
at 6% strain
Stiffness, E50 (MPa)

White Fiber
Strength (MPa)

3 60 at 4% strain
at 6% strain
Range of
2 Black Finer optimum 40
at 4% strain fiber
at 6% strain content
1 White Fiber 20
at 4% strain Range of optimum
at 6% strain fiber content
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Clay content (%) Clay content (%)

Figure [Link] of polymeric fiber content on peak strength and stiffness of earth blocks.

439
6 120
a) Maximum compressive strength b) Stiffness E50
5 100
Palm Tree Finer
at 4% strain

Stiffness, E50 (MPa)

4 80 at 6% strain
Strength (MPa)

3 60

2 40 Range of optimum
fiber content
Range of optimum Palm Tree Finer
1 20
fiber content at 4% strain
at 6% strain
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Clay content (%) Clay content (%)

Figure [Link] of natural Palm Tree fiber content on peak strength and stiffness of earth blocks.

For polymeric fiber content, shown in Figure 8. The smaller optimum fiber content values for natural fiber compared to
polymeric fiber could be attributed to the smaller unit weight of natural fiber. At the same fiber content, the volume of natural
fiber will be significantly larger than that for polymeric fiber. Therefore, at higher values of fiber content, the volume of
natural fiber is significantly large, and expected to cause soil matrix discontinuities, leading to softening of the earth block.
This softening is reflected by less stiffness values at the higher natural fiber content percent.

6 6
a) 0.25% Fiber content Fiber Type b) 0.5% Fiber content
5 Black Fiber 5
Compressive Stress,  (MPa)

Compressive stress,  (MPa)

White Fiber
Natural Fiber
4 Unstabilized Block 4

3 3

2 2 Fiber Type
Black Fiber
White Fiber
1 1
Natural Fiber
Unstabilized Block
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Axial strain (%) Axial strain (%)
6 6
Fiber Type
c) 0.75% Fiber content d) 1.0% Fiber content
Black Fiber
5 White Fiber 5
Compressive stress,  (MPa)
Compressive stress,  (MPa)

Natural Fiber
4 Unstabilized Block 4

3 3

2 2 Fiber Type
Black Fiber
1 White Fiber
1 Natural Fiber
Unstabilized Block
0
0 0 2 4 6 8 10
0 2 4 6 8 10
Axial strain (%) Axial strain (%)

Figure [Link] of fiber Type and content on stress-strain behavior of stabilized earth blocks.

4.3 Effect of Fiber Type

Effect of fiber type and content in stress-strain response is presented in Figure 10. For fiber content equal to 0.25% (Figure
10a), the polymeric fiber, both white and black, resulted in initial softening of the block with slight increase in the peak
compressive strength compared to the unstabilized block. In the contrary, the block stabilized with natural fiber showed more
initial stiffness with slight increase in the peak compressive strength. This behavior could be explained in terms of fiber
stiffness and surface area. The natural fiber has stiffness and surface area much larger compared to the polymeric fiber.
Therefore, the natural fiber-soil interface friction is expected to be larger compared to the polymeric fiber-soil interface
440
friction. In addition, natural fiber added more stiffness to the earth block compared to polymeric fiber. It could be noticed that
the earth block stabilized with white polymeric fiber showed the largest initial softening at 0.25% fiber content (Figure 10a).
For 0.5% fiber content (Figure 10b), earth blocks stabilized with polymeric fibers (both white and black) did not show any
improvement in the block initial stiffness compared to unstabilized block. However, it showed significant improvement on
peak compressive strength. In the contrary, the block stabilized with natural fiber showed significant improvement in both
initial stiffness and peak compressive strength. It should be noted that, all earth blocks stabilized with both natural and
polymeric fiber showed strain softening after reaching the peak compressive strength.
For 0.75% fiber content (Figure 10c), the earth blocks stabilized with polymeric fibers started to show significant
improvement on the block initial stiffness compared to unstabilized block. In addition, blocks stabilized with polymeric fiber
showed peak compressive strength significantly larger compared to both unstabilized block and block stabilized with natural
fiber. In fact, the earth block stabilized with 0.75% natural fiber indicated significant reduction of the initial stiffness and slight
improvement of the peak compressive strength. This could be attributed to large volume occupied by the natural fiber which
led to a discontinuity of the soil matrix and therefore a reduction on the stiffness. It should be noted that the natural fiber has
specific gravity which is smaller compared to polymeric fiber.
For 1.0% fiber content (Figure 10d), earth blocks stabilized with the three fiber types showed reduction of initial stiffness
up to 2% axial strain, compared to unstabilized earth block. For axial strains larger than 2%, blocks stabilized with polymeric
fiber showed improvement in the block stiffness, however, the block stabilized with natural fiber still showing reduction in the
stiffness up to 4% axial strain. For axial strain larger than 4% blocks stabilized with all types of fiber showed strain hardening
(i.e. increase of the stress with strain). Although the block stabilized with natural fiber showed softening at early stage of
strain, it picked up strength after 2% strain and showed the largest improvement at 1% fiber content. It could be noticed that, at
this fiber content percent, the block stabilized with natural fiber showed the highest peak strength, followed by black
polymeric fiber. The least improvement at this fiber content percent was showed by the earth block stabilized by white
polymeric fiber as shown in Figure 10c.

5 CONCLUSION

Based on the parametric study and compression tests of stabilized and unstabilized earth blocks conducted in this study and the
results presented in this paper, the following concluded remarks can be summarized:
1. Earth block compressive strength increases and the secant stiffness, E50 significantly decreases with the increase of clay
content until certain (optimum) clay content, and decreased thereafter. In addition, the coarser the soil grading, the higher
the optimum clay content is.
2. For both polymeric fiber types, strain required to mobilize peak compressive strength increased as the fiber content
increased. In addition, blocks stabilized with relatively softer and smoother (white) fiber attained their peak strength at
strain larger than blocks stabilized with stiffer and rougher (black) fiber.
3. Inclusion of polymeric fiber led to reduction in strength and stiffness at the initial stage of testing (strain < 2%). However,
the inclusion of natural palm tree fiber improved the peak compressive strength and stiffness with the addition of natural
fiber up to 0.5%.
4. The optimum range of polymeric fiber content is 0.6% to 0.8%, which satisfied both maximum compressive strength and
maximum stiffness, at both 4% and 6% strains. In addition, black fiber is more effective in improving block stiffness,
while white fiber is slightly more effective in improving compressive strength.
5. For earth blocks stabilized with palm tree natural fiber, the optimum fiber content for both stiffness and compressive
strength is ranged from 0.2% to 0.6%.

REFERENCES

Al-Jabri K., Hago A., Taha R., Alnuaimi A., Al-Saidy A. (2009) Strength and Insulation Properties of Building Blocks Made from Waste
Materials. J. Material Civ. Eng, 21(5), 191–197.
American Society for Testing and Materials (1984).Standard test method for compressive strength of molded soil-cement [Link]
D1633.
Bahar R., Benazzoug M., Kenai S. (2004) Performance of compacted cement-stabilised soil. Cement and Concrete Composites 26 (7), 811-
820.
British Standard Institution (1985) British Standard specification for clay [Link] 3921.
Chan C.M., Liang P.M. (2010). Development of a Prediction Model for “Green” Compressed Stabilized Earth Bricks. Journal of Sustainable
Development, 3(3).
El-Emam M. (2014) Modulus and Damping of Reinforced Soil walls from Shaking Table Tests. Geomechanics and Geoengineering: An
International Journal, 9(4), 279-293.
Guettala A., Houari H., Mezghiche B., Chebili R. (2002) Durability of lime stabilized earth blocks. Courrier du Savoir 2, 61-66.
Heathcote K. A. (1995) Durability of Earthwall [Link] of Construction and Building Materials 9 (3), 185-189.
Houben H., Guillaud H. (1994) Earth construction: a comprehensive guide. IT Publications, London.
Khalaf M., DeVenny A.S. (2002) New Tests for Porosity and Water Absorption of Fired Clay Bricks. Journal of Materials in Civil
Engineering 14 (4), 334 – 337.

441
Marandi S.M, Bagheripour M.H., Rahgozar R., Zare H. (2008) Strength and Ductility of Randomly Distributed Palm Fibers Reinforced
Silty-Sand Soils. American Journal of Applied Sciences 5 (3), 209-220.
Morton T. (2008) Earth Masonry Design and Construction Guidelines. Berkshire: Construction Research Communications Limited.
MurilloG., Walker P.J., Ansell M.P. (2005) Henequen fibres for reinforcement of unfired earth [Link]. 11th International Conference
for Renewable Resources and Plant Biotechnology (NAROSSA), 121-126.
Oti J.E., Kinuthia J.M., Bai J. (2009) Compressive strength and microstructural analysis of unfired clay masonry bricks. Engineering
Geology, 109(3-4), 230-240.
Reddy B., Gupta A. (2006) Strength and Elastic Properties of Stabilized Mud Block Masonry Using Cement-Soil Mortars.J. Mater. Civ.
Eng. 18(3), 472–476.
Venkatarama Reddy B.V., Richardson Lal, Nanjunda Rao K.S. (2007) Optimum soil grading for the soil-cement [Link] of Materials
in Civil Engineering, 19 (2), 139.
Venny Riza F., Abdul Rahman I., Ahmad Zaidi A.M. (2011) Preliminary Study of Compressed Stabilized Earth Brick (CSEB).Australian
Journal of Basic and Applied Sciences, 5(9): 6-12.
Walker P. (1995) Strength, durability, and shrinkage characteristics of cement stabilized soil blocks. [Link]., 17(4), 301–310.
Walker P. (2004) Strength and Erosion Characteristics of Earth Blocks and Earth Block Masonry. J. Mater. Civ. Eng. 16, 497-506.
Walker P., Stace T. (1997) Properties of some cement stabilized compressed earth blocks and mortars. [Link]. 30, 545–551.

442
GLISSEMENT DES TERRAINS ET
STABILITE DES PENTES

LANDSLIDES AND SLOPE STABILITY

443
444
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Analyse de stabilité et stabilisation du versant urbanisé de Tizi-N’Béchar

(wilaya de Sétif, Algérie)
Stability analysis and stabilization of urbanized Tizi-N’Béchar slope (wilaya of Sétif, Algeria)
A. Seddiki1, M. Khemissa*1
1
Laboratoire de Développement des Géomatériaux, Université Mohamed Boudiaf, 28000 M’sila, Algérie
*
Auteur correspondant (khemissa@[Link])

RÉSUMÉ– Cet article présente les résultats d'une série de calculs de stabilité effectués sur le versant urbanisé de
Tizi-N’Béchar (wilaya de Sétif, Algérie) et propose une solution pour son confortement. Les calculs de stabilité
montrent que le versant est en équilibre critique, donc potentiellement instable. Parmi les solutions de
confortement possibles, l’allègement en tête de ce versant et son remblaiement en pied apparaissent comme la
technique la mieux adaptée, mais insuffisante sous l’action d’une légère secousse sismique associée à une nappe
phréatique affleurant le terrain naturel. Aussi, le renforcement du versant par inclusions de type clouage en tête
et pieux en pied combinées éventuellement à un soutènement en contre-bas du versant s’impose.

ABSTRACT – This paper presents the results of a series of stability calculations carried out on urbanized Tizi-
N’Béchar slope (wilaya of Sétif, Algeria) and proposes a solution for its stabilization. Stability calculations
show that the slope is in critical equilibrium, therefore potentially unstable. Among possible stabilization
solutions, the lightening at the head of this slope and its filling in foot seem as the best adapted technique, but
insufficient under the action of a light earthquake associated with a groundwater table levelling the natural
ground. Also, the slope’s reinforcement by inclusions of the nailing type at the head and piles in foot possibly
combined with a retaining wall in against-low slope is essential.

1. INTRODUCTION

La commune de Tizi-N’Béchar (wilaya de Sétif, Algérie) est située sur la RN9 aux coordonnées
géographiques 36° 25' 52" Nord et 5° 21' 36" Est et s'étale sur une superficie de 71,65 km2 (figure 1).
Cette commune connaît de nos jours une urbanisation sans cesse croissante dictée par une
démographie galopante, d’où son extension vers des terrains moins favorables que ceux déjà
urbanisés. Des instabilités de terrains en pente sont à craindre dans cette zone en raison de son
contexte géologique et géomorphologique en plus des conditions climatiques et géophysiques
caractérisant l'Atlas tellien dans le nord de l’Algérie.

Figure 1. Plan de situation de la ville de Tizi-N’Béchar (wilaya de Sétif, Algérie).

445
Cet article a pour objet d’analyser la stabilité d’un versant destiné à la construction de locaux commerciaux
dans la commune de Tizi-N’Béchar à proximité de la RN9 et de proposer une solution pour son confortement.
On présente d’abord une description sommaire du site étudié, on expose ensuite les conditions retenues pour les
calculs et leur mise en œuvre à l’aide des logiciels TALREN et PLAXIS. On discute enfin quelques solutions de
confortement du versant permettant de parer à d’éventuelles instabilités liées aux propriétés mécaniques des
sols, aux conditions hydrogéologiques et aux effets sismiques probables.

2 DESCRIPTION SOMMAIRE DU SITE

2.1 Contexte géologique et géomorphologique

La commune de Tizi-N’Béchar est caractérisée par un relief accidenté constitué essentiellement de montagnes et
de quelques collines. Le site étudié est caractérisé par une déclivité raide en amont et moyennement raide en
aval à proximité d’un talweg. D’après la carte géologique du secteur étudié, les formations géologiques des
terrains sont constituées d’alluvions du quaternaire et de formations telliennes représentées par des schistes noirs
durs.
La campagne de reconnaissance effectuée révèle l’existence de trois couches de sols d’épaisseurs variables
d’un sondage à l’autre. A l’emplacement des trois sondages effectués, on rencontre des sols hétérogènes
comportant de haut en bas :
- une couche de limon entre 0 et 3 m de profondeur,
- une couche d’argile limoneuse entre 1,5 et 3,5 m de profondeur,
- une couche de marne compacte entre 2 et 8 m de profondeur.
Le niveau de la nappe phréatique varie selon la saison entre le niveau du terrain naturel en hiver et à
quelques dizaines de mètres en profondeur en été.

2.2 Caractéristiques géologiques des sols

Les valeurs des teneurs en eau déterminées sur les échantillons limono-argileux à débris schisteux
varient de 15 à 17% avec un degré de saturation variant de 87 à 95%. Par contre, les valeurs des
teneurs en eau déterminés sur les échantillons marneux varient de 4 à 5% avec des degrés de
saturation variant de 70 à 80%. L’analyse granulométrique fait ressortir des sols composés
essentiellement de fines. Le tableau 1 donne les valeurs des paramètres mécaniques retenues dans les
calculs de stabilité du versant.

Tableau 1. Valeurs considérées des paramètres de calcul.

Couches de sol d (kN/m3) c' (kPa) ' (°)

Limon 17 10 15
Argilelimoneuse 16 13 20
Marne compacte 18 30 7

2.3 Caractéristiques sismiques

La commune de Tizi-N’Béchar est classée par les règles parasismiques algériennes en vigueur (RPA
99 – version 2003) comme une zone de moyenne sismicité (zone IIa). Toutefois, la position
géographique de cette ville laisse craindre une activité sismique comparable à celle des villes du nord
de l’Algérie qui connaissent une activité sismique intense (AFPS, 2003). Le tableau 2 donne les
valeurs des coefficients sismiques horizontalkh et vertical kv pris en compte dans les calculs de
stabilité sismique du versant.

Tableau 2. Valeurs du coefficient d’accélération de zone A (RPA 99).

Zone Groupe Description A

IIa 1A Ouvrages d’importance vitale (sécurité, hôpitaux) 0,25
1B Ouvrages de grande importance (bâtiments scolaires, mosquées) 0,20
2 Ouvrages courants (logements, bureaux) 0,15
3 Ouvrages de faible importance (hangars) 0,10

446
3 CALCULS DE STABILITE

3.1 Profil géométrique considéré

Le profil du versant de Tizi-N’Béchar est composé de segments délimitant les couches des sols
constituant le massif. Le massif est constitué de trois couches de sols comportant une couche de limon
reposant sur une couche d’argile limoneuse qui elle-même repose sur une couche de marne compacte.
Les conditions aux limites sont choisies par défaut : déplacements libres sur les deux côtés verticaux
et bloqués à la base (substratum).

3.2 Outils numériques de calcul utilisés

Les calculs de stabilité du versant ont été effectués à l’aide des deux logiciels suivants :
- le programme TALREN 4 (version 2.0.3) de calcul de stabilité et de renforcement des talus en ruptures
circulaires et non circulaires par la méthode des tranches,
- le code PLAXIS (version 8.2) de calcul en éléments finis des ouvrages géotechniques par la méthode de
réduction des paramètres de cisaillement.
L’analyse de stabilité sismique du versant a été effectuée par l’approche pseudo-statique. Le principe de cette
approche consiste à remplacer l’action sismique par une action statique équivalente qui tient compte de la
réaction probable de la masse de sols en mouvement (figure 2).

–kvW
W
+kvW k hW
Forces statiques Forces pseudo-statiques

Figure 2. Forces statiques et pseudo-statiques appliquées.

3.3 Localisation de la zone de rupture potentielle du versant

Les calculs effectués en équilibre limite à l’aide du programme TALREN sous l’effet du chargement gravitaire
seul montre que la ligne de rupture du versant est localisée dans la couche d’argile limoneuse (figure 3). Le
coefficient de sécurité correspondant au cercle critique est égal à Fs=1,04. Ce résultat permet de conclure que le
versant est en équilibre critique, donc potentiellement instable.

Figure 3. Localisation de la ligne de rupture potentielle du versant.

Les calculs effectués en plasticité suivant un modèle de rupture de type Mohr-Coulomb à l’aide du code
PLAXIS sous l’effet du chargement gravitaire seul s’accordent avec l’hypothèse d’une rupture rotationnelle du
versant et confirment les résultats des calculs obtenus en équilibre limite (figure 4).

447
Figure 4. Localisation de la zone de plastification des sols en mouvement.

4 RECHERCHE D’UNE SOLUTION CONFORTATIVE DU VERSANT

Parmi les solutions de confortement envisageables du versant, on se restreint à ne présenter ci-après que les
résultats liés à la position de la ligne de rupture du versant et la valeur du coefficient de sécurité correspondant
déterminés à l’aide du programme TALREN ; ceux obtenus à l’aide du code PLAXIS étant semblables.

4.1 Confortement par allègement en tête du versant

La figure 5 présente le profil du versant allégé en tête et la position de la ligne de rupture correspondante sous
l’action gravitaire seule. On notera une légère amélioration de la stabilité du versant à laquelle correspond une
valeur critique du coefficient de sécurité Fs=1,23.

Figure 5. Profil du versant allégé en tête et position de la ligne de rupture correspondante.

4.2 Confortement par remblaiement en pied du versant

La figure 6 présente le profil du versant remblayé en pied et la position de la ligne de rupture correspondante
sous l’action gravitaire seule. On notera une légère amélioration de la stabilité du versant à laquelle correspond
une valeur critique du coefficient de sécurité Fs=1,21.

Figure 6. Profil du versant remblayé en pied et position de la ligne de rupture correspondante.

448
4.3 Solution combinée : allègement en tête et remblaiement en pied

La figure 7 présente le profil du versant allégé en tête et remblayé en pied et la position de la ligne de rupture
correspondante sous l’action gravitaire seule. On notera une amélioration très sensible de la stabilité du versant à
laquelle correspond une valeur critique du coefficient de sécurité Fs=1,45.

Figure 7. Profil du versant allégé en tête et remblayé en pied et position de la ligne de rupture correspondante.

Toutefois, les calculs de stabilité effectués sous l’effet d’une forte secousse sismique ou d’un séisme
correspondant aux ouvrages d’importance vitale de la zone IIa (kh=0,125g) montre que le versant s’effondre
(Fs=0,96) (figure 8).

Figure 8. Profil du versant conforté et position de la ligne de rupture correspondant à une secousse sismique de zone IIa.

Sous l’effet d’une nappe phréatique affleurant le terrain naturel, les calculs de stabilité montre que le versant
demeure en équilibre critique (Fs=1,03) (figure 9).

Figure 9. Profil du versant conforté et position de la ligne de rupture correspondant à une nappe phréatique affleurant le
terrain naturel.

449
Cependant, sous l’effet combiné d’une légère secousse sismique ou d’un séisme correspondant aux ouvrages
courants (kh=0,075g) associée à une nappe phréatique affleurant le terrain naturel, les calculs de stabilité
montrent que le versant s’effondre (Fs=0,95) (figure 10).

Figure 10. Profil du versant conforté et position de la ligne de rupture correspondant à un effet combiné : séisme + nappe
affleurant le terrain naturel.

5 CONCLUSION

Cet article avait pour finalité d’analyser la stabilité du versant de Tizi-N’Béchar (wilaya de Sétif, Algérie)
destiné à la construction de locaux commerciaux et de lui proposer une solution confortative. Le contexte
géologique et géomorphologique des sols constituant le massif laisse craindre une éventuelle instabilité du
versant, notamment lors d’évènements climatiques et/ou sismiques probables. Les calculs de stabilité effectués
sous l’action gravitaire seule justifie cette crainte puisqu’ils montrent que le versant est en équilibre critique,
donc potentiellement instable.
Cependant, les calculs de stabilité du versant allégé en tête ou remblayé en pied indiquent une légère
amélioration de sa stabilité sous l’action gravitaire seule. Tandis que la solution combinée d’un allègement en
tête du versant et de son remblaiement en pied en est la plus avantageuse, mais insuffisante lorsque celui-ci sera
éventuellement soumis à une légère secousse sismique associée à une nappe phréatique affleurant le terrain
naturel.
Aussi, doit-on chercher à conforter le versant au moyen d’inclusions de type clouage en tête et pieux en pied
avec éventuellement un soutènement en contre-bas du talus.

REFERENCES

AFPS (2003). Le séisme du 21 mai 2003 en Algérie. Rapport de mission de l’Association Française du Génie Parasismique.
PLAXIS (2002).Finite element code for soil and rock [Link], Rotterdam, Brookfield, version 8.2.
RPA 99 (1999). Règles Parasismiques Algériennes. Editions de l’Office des Publications Universitaires, Alger, version 2003.
TALREN 4 (2005). Programme d’étude de la stabilité et du renforcement des talus en ruptures circulaires et non
circulaires. TERRASOL, version 2.0.3.

450
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

The present-day landslide of the Moulares phosphates mines (west-central

Tunisia): Origin and evolution
Le glissement de terrain actuel de la carrière de phosphates de Moularès (centre-ouest de la
Tunisie) : Origine et évolution
FeridDhahri*1,3, Ramdhane Benassi1, Kamel Zeyeni2 , and Noureddine Boukadi3
1
Faculty of Sciences de Gafsa /University of Gafsa, Gafsa, Tunisia
2
Gafsa Phosphate Company /Research Center of Metlaoui, Tunisia
3
Faculty of Sciences of Tunis, 2092, UR11ES15 Sedimentary environments, Petroleum systems and reservoirs
characterization /University of Tunis El Manar, Tunis, Tunisia
*
Corresponding Author

ABSTRACT -Active landslides either natural or manmade constitutes serious geologic hazard that impact the
earth’s natural environment and poses a potential danger to life and territory. They occur frequently in both
bedrocks and engineering soils within wide variety of processes and scales under all climatic conditions and
terrains and engender economic, social and natural environment disruptions. Particularly in mining area, the
human-induced weakness and disturbance of geologic structures are a contributing factor to landslides
occurrence. In such fields, the interaction of lithology, rock mass structure, active tectonics/artificial vibrations
and climatic conditions can trigger straightforwardly the landslide.
In Gafsa phosphates mines (west-central Tunisia) some ground disturbance and present-day deformation related
to underground and surface front mining were highlighted. Particularly in the Moulares phosphates mine, deep
ground deformation occurs within both broken sterile rocks and in place bedrock subsequent to the accumulation
of enormous mass of sterile. The ground damage evolved continuously since 2006 until today and large part of
the "verses" moved northeastward according the general ground slope as a landslide bodies. The occurrence
causes and the deformation style of rocks bodies within the Moulares mine landslide are discussed based on a
field survey with emphasis of the geological and geotechnical characteristics of the ground. This case study
constitutes a thematic key for modeling and interpretation of landslides and deformations occurrence in relation
with geological and geotechnical conditions in mining area.

RÉSUMÉ-Les glissements de terrain actifs, que ce soit d'origine naturelle ou anthropique, constituent des
risques sérieux pour l'environnement et la vie humaine. Ils apparaissent communément au niveau des massifs
géologiques et des sols d’ouvrages divers, selon des modalités et des échelles variés, sous diverses conditions de
climat et de terrain. Ils engendrent des perturbations d’ordre économique, social et naturel. Particulièrement, au
niveau des zones sièges d'exploitation minière, les perturbations des structures géologiques induites par l'homme
constituent un facteur contribuant à l’émergence des glissements de terrain. Dans ces zones, l'interaction entre la
lithologie, la structure des massifs rocheux, la tectonique active et les vibrations artificielles ainsi que les
conditions climatiques peuvent déclencher facilement les glissements de terrain.
Au niveau des mines de phosphates de la région de Gafsa (centre-ouest de Tunisie) quelques perturbations et
déformations actuelles de terrain, liée à l'exploitation souterraine et de subsurface sont mises en évidence. En
particulier dans la carrière de phosphates de Moularès, des déformations importantes se sont apparues au niveau
des verses de stériles ainsi qu’au niveau de certaines séries lithologique en place, suite à l'accumulation
d’énormes masses de stériles. La déformation du terrain a évolué de façon continue depuis 2006 jusqu'à
aujourd'hui et une grande partie des verses est déplacée gravitairement vers le nord suivant la pente du terrain, et
ceci, sous forme de glissement de terrain. Les causes de l’apparition, et le style des déformations des roches
accompagnant le glissement de terrain de la carrière de Moularès sont discutés en se basant sur une étude de
terrain et en mettant l'accent sur ses caractéristiques géologiques et géotechniques. Cette étude de cas constitue
un exemple thématique pour la modélisation et l'interprétation des glissements de terrain et des déformations y
associés, en relation avec les conditions géologiques et géotechniques dans les zones minières.

1 INTRODUCTION

Geologists and geotechnical engineers provided different definitions of landslide because of the wide variety and
the complex nature of these field deformations. Commonly, this term means a geological phenomenon that
includes downslope movement of ground under the effects of gravity and landform. For more detail, a complete
classification of landslides is given by Highland and Bobrowsky (2008) in "The Landslide Handbook–A Guide
to Understanding Landslides (Section I, Part B).

451
The triggering mechanisms of landslide are both natural (earthquake, weathering, slope, gravity, debris
flows, plastic deformation at depth (salt, clays) ...) and manmade (excavation, artificial vibration, undercutting
...). Mining activities induces wide weakness and disturbance of geologic structures and seems to be main
contributing factors to landslide occurrence. In fact, the landslide occurrence depends on several factors as much
as lithology, structure and topography, seismicity and/or artificial vibrations and climatic conditions. The fact
that industrial mines uses explosive that propagates the vibration in the ground, creates weak zones, excavations
and artificial slope, they constitutes a favorable area for landslides occurrence. This is can be amplified
especially in active tectonic regions with the contribution of topographic and climatic factors.
In the Moulares mine (west-central Tunisia) (Fig. 1), the exploitation of the phosphates levels engenders
deep ground disturbances and left enormous stocks of sterile. The unstable sterile stocks left in a high damaged
area evolved to large landslide that slips northeastward according the general ground slope. This work aims to
discuss the origin and the evolution of this landslide and to detail the main factor and the deformation style
implicated in this ground deformation that can be considered as case study of landslides in mining area.

Figure 1. A) location map of the study area; B) simplified geological map of the study area and neighboring region showing
the main structural features and the location of the Northern Table mine of Moulares; C) Detailed geologic map of the study
area.
2 THE MOULARES PHOSPHATES MINE: AN OVERVIEW

The phosphate mining in the Moulares area was started since 1904 using underground mines. Since 1994 the
exploitation is developed using surface front mining methods. The surface front mining methods is simplistic but
induces deep ground deformation. In fact, the phosphatic strata are intercalated with marls in the

452
ChouabineFormation which is overlain by the limestones of Kef Eddour Formation. The Kef Eddourlimestones
are broken after detonation of explosive installed in a series of holes drilled in area of rock face. Together with
the marls, intercalated between phosphate strata in the Chouabine Formation, they constitute a mixture of sterile
rock which was stocked on-site banked up behind the mine working face and creating platforms that exceed 30
meters height with abrupt cliffs.
3. GEOLOGICAL SETTING
The study area comprises the "northern table mine" of Moulares located in the Moulares folds range, to the
southwest of the Gafsa fault in central-western Tunisia (Fig. 1) and make part of the Southern Tunisian Atlas
(Zargouni 1985, Boukadi 1989) (Fig. 1B). Around the Moulares ranges, folds are typical of a bundle fold pattern
with E–W oriented axial plane that diverge near the Gafsa fault to NE–SW direction.
Close to this mine, outcropping series ranges from Upper Cretaceous to Quaternary (Fig. 1C). The Upper
Cretaceous deposits are represented by the Berda Formation Campanian-Maastrichtian in age and made of two
carbonated units intercalated by a marly unit. This Formation is overlain by the argillaceous Haria Formation,
Paleocene in age. Paleocene to Lower Eocene series records favorable conditions for the accumulation of the
phosphate deposits in the Gafsa basin within the Metlaoui Formation (Zaier et al. 1998, Galfati 2010). This latter
is divided into three members respectively the Selja member Paleocene in age, and made of carbonates, clays
and evaporites, the Chouabine member Ypresian in age and is made of eight phosphate-rich strata with
carbonated, cherty and clayey levels, and the Kef Eddour member Lutetian in age, and made of dolomitic
bioclasticlimestones with an intercalation of one metric phosphatic level agreed as the "Upper Phosphates". The
Oligocene–Quaternary deposits are mainly of continental facies and made of detritic alluvial and fluvial
deposits.
The Northern Table Mine is located in the northern flank of the StahSouda anticline (Fig. 1C). This flank
shows a carbonated ridge consisting of Kef Eddour Formation 20-30° NNE-dipping and limited by the
Moulareswadi to the north (Fig.4B). It is also affected by two ENE-trending faults (Fig.2). One of them crosses
all along the mine with 45° NNW-dip and is mostly buried by the sterile.

4. CHARACTERIZATION OF THE LANDSLIDE DEFORMATION AND EVOLUTION OF ASSOCIATE’S

STRUCTURES
The study area show a kilometer sizes transitional landslide, comprised of both debris and earth slides that are
well exposed upstream and downstream of the main landslide (Fig. 2). The first ground deformation and damage
was observed since 2006 as a longitudinal detachment zone in the highest part of the sterile deposit. The ground
damage evolved continuously until today and large part of the sterile-stock moved northwestward according the
general slope of the studied site. For the last three years, the displacement exceeds 10 meters in some wedge
exemplifying a complex landslide leading to the discontinuation of phosphates mining.
The landslide front, extend about 1000 m downward the moving rock bodies, parallel to the mine working
face (Fig. 2). In this part, the sliding deformation occurs within the bedrock below the sterile bodies (Fig. 3 A).
The main sliding surface is parallel to strata and occurs along the marly level of Chouabine Formation. Field
deformations show that the landslide propagation is most obvious to the west and tends to be progressively
amortized to the east.
The surface of rupture (Fig 3A) occurs at ~500 m elevation in the western part of the Northern Table mine. It
is partially hidden to the east due to the post-detachment sterile discharge (Fig. 3B). At its western end, this
surface of rupture shows two differential sliding stage with displacements of 3 to 10 meters (Fig. 3 A–B). In
fact, the sliding plane downward the landslide corresponds to a polished and striated 45° NNW-dipping fault
plane (Fig. 3C). This fault is associated with the presence of a centimetric level of gypsum and striated mud
surfaces (Fig. 3 C–D) indicating which can amplify the role of meteoric water in the sliding process.
The southeastern part of sterile discharge is stable and seems to be in place with no displacement (Fig. 3).
However, the northwestern part seems to be unstable and slides to the northwest. The boundary between these
two parts matches with the trace of an ENE-oriented fault (Fig. 1 C). It seems that the sliding movement begins
initially with the detachment of the sterile masses in the upstream part along the 45° NNW-dipping plane of this
fault on which sliding displacement has initiated (Fig. 3). Downward, sterile are deposited on the lower part of
the Chouabine Formation made of NNE-25° dipping strata of marls and thin layers of limestones (Fig. 3A). Due
to the ground slope, the lithology of the strata underneath the sterile bodies and the sterile excessive loading, a
new gliding surface will occur parallel to stratification (Fig. 3A).
Several deformations occurred within the displaced strata located between the sterile and the sliding surface
(Figs 2 and 3A). Theses deformations are differential and evolve from simple folds to thrusts with intermediate
folding stages and associated structures. The front of the displaced bedrock shows a cylindrical fold axis with an
overturned forelimb (Fig. 2 F), within the most displaced terrain, second order folds occurs (Fig. 2 H). Folding
resulted in the opening of inherited fractures and new extrados fractures occurrence (Fig. 2 H). The displaced

453
bedrock and associated structures moving down slope to strike and thrust the canceled working face along a
striated contact (Fig 2E–F–G–G’).

Figure 2. Photographs of the landslide and its associates structures in the Moulares mine. A) the main surface of rupture
which coincide with the 45° NNW-dipping plane of an inherited fault shows two sliding stages; B) Photograph showing
sterile fall with meteoric water erosin; C) The inherited fault aspect with polished and striated plane and centimetric gypsum
level; D) Striated mud surface testifying the role of meteoric water in the triggering mechanism of landslide; E) Panoramic
view showing the sterile bodies over the deformed bedrocks; F) Bedrocks shows a folded front with overturned forelimb; G
and G’) The landslide front shows striations indicative of the movement according to the general ground slope: H) Two fold
generations occurs when the displacement amplitude is important, extrados fractures also occurs.

454
Fig. 3: A) NW-SE cross sections showing the present-day structure of the Northern Table mine of Moulares. The sterile
rocks and their substratum made of marls and thin limestones levels slide northwestward and thrust downstream; B)
Interpretation of panoramic 3D views (Google earth) showing the morphology of the study area and the location of the cross
section (A).

4. DISCUSSION AND CONCLUSIONS

Landslides are extremely complex and occur in particular ground conditions due to internal and external driving
forces that are caused by physical or human infraction. The most common triggers are earthquakes, heavy rains,
thawing of frozen ground and frequent development from humans (Abbott 2004). Wang et al. (2011)
demonstrates that the landslide occurrence in surface mine is a complicated process in which the slope
undergoes the generation, expansion and connection of the fractures until the displacement increases up to the
landslide; a small portion of the high rocks fails due to the tension and the lower portion fail due to the shear.
In the study area, the triggering mechanism seems to be complex, and consisting of several factors. The most
significant factors are the structure/topography, the gravity and the lithology. Other factor seems to be also
contributing seriously in the triggering and the evolution of the landslide body in the northern table mine of
Moulares as much as meteoric water and seismotectonic and artificial vibrations.
The sliding surface can be divided into two parts; the first part coincides with the fault plane and the second
part, which constitutes the main sliding surface, occurs within the NNE-25° dipping marly levels of the
Chouabine Formation (Fig. 3 A). The fault plane shows a polished steep angle slope associated with a
centimeters thick level of gypsum which can be easy dissolved by meteoric water, increasing the instability of

455
the sterile bodies which are made of a mixture of broken rocks with low cohesion. The marly levels of
Chouabine Formation display NNE-25° dip angle and can be considered a plastic level that gives a better chance
for gravity to take over and cause slope failure especially when weakened by interaction between water and clay
materials an stated by Abbott (2004).
Water is one of the most important instability factors because it weakens earth material, decreases the
cohesion of rocks, increases weight and pressure in granular media and weakens the slope (Abbott 2004, De
Blasio 2011). Its role is primordial in the emergence of the sliding surface in both sterile and bedrocks. In the
study area, the main sliding surface occurs between the clays and the thin layers of limestones of the Chouabine
Formation of about 2 m underneath the coquina level considered as the base of phosphatic series. The fact that
the limestone levels are fractured facilitates the water seepage along the fractures network. Together with the
lithostratigraphic pressure and the slope the water seepage contribute to the emergence of a gliding surface
parallel to the 25° dipping strata.
Topography is one of the fundamental triggering factors in landslide occurrence especially within weak
rocks and steep slopes. In the disturbed ground such mine environment, the discharge of sterile on sloped
terrains influences the distribution of weight along slope and can cause failure as much as in the studied mine.
In active tectonic regions landslides are very abundant and linked to historical earthquakes and seismic
factors (Marzorati et al. 2002). In fact, the study area is characterized by low to mean magnitude seismic activity
which can instantly change the local force equilibrium and anticipating rock deformations. Seismic activity is
due to the sustained tectonic activity of the Gafsa fault characterized by high seismic hazard subsequent to
historical earthquakes (Said et al. 2011, Bahrouni et al. 2013). In addition, the phosphate exploitation activity
uses explosives to break the limestone of Kef Eddour Formation. Explosive detonation propagates the vibration
in the ground.
The climate of Moulares region is arid with an average annual rainfall not exceeds 175 mm. This reduces the
impact of meteoric water. However, torrential rains can be critical in the triggering and the evolution of
landslides in such environment same that recorded in September 2009 (about 160 mm per few hours) which
seems responsible for the first stage of landslide described in figure 2A.
In conclusion, this survey constitutes a case study for understanding and interpretation of the occurrence of
structures and deformation in mine and quarry environment. It shows that the occurrence of present-day
deformations within sterile and bedrock in the Northern Table mine of Moulares, west-central Tunisia, is due to
the interaction between several factors such as inherited structures, topography, slope, gravity and
seismotectonic and artificial vibrations. The marl dominated lithology and the steeply dipping sliding surface,
together with meteoric water, are favorable for rock failures and landslide occurrence. The sliding rocks bodies
move down along a combined surface: the upper part of the sterile waste slip, especially due to gravity and the
fault plane and the lower part is displaced after the onset of the slip plane parallel to the bedding planes in the
Ypresian marls sequence, base of Chouabine Formation.

REFERENCES BIBLIOGRAPHIQUES
Abbott P.L. (2004) Natural Disasters, New York, McGraw-Hill Companies, Inc., 460 p.
Bahrouni N., Bouaziz S., Soumaya A., Ben Ayed N., Attafi K., Houla Y., El Ghali A. andRebai N. (2013) Neotectonic and
seismotectonic investigation of seismically active regions in Tunisia: a multidisciplinary approach. Journal of
Seismology, April 01, Vol. 18, Issue 2, 235-256.
Boukadi N. (1989) Sur le plissement disharmonique et la dispersion des axes de plis dans les couloirs de décrochement :
l’exemple du faisceau de Moularès de Tunisie. C. R. Acad. Sci. Paris, t. 309, Série II, 2105-2110.
De Blasio F.V. (2011) Introduction to the Physics of Landslides, Lecture notes on the dynamics of mass wasting. Springer
Edit., 2011, XV, 408p. 202 illus. ISBN 978-94-007-1122-8.
Galfati I., BejiSassi A., Zaier A., Bouchardon J.L., Bilal E., Jordon J.L., Sassi S. (2010) Geochemistry and mineralogy of
Paleocene Oum El Khechebphosphorites (Gafsa-Metlaoui) Tunisia, Geochemical Journal, vol.44,189-210.
Highland L.M., Bobrowsky P. (2008) The landslide handbook–A guide to understanding landslides. Reston, Virginia, U.S.
Geological Survey Circular 1325, 129 p.
Marzorati S., Luzia L. and De Amicisb M. (2002) Rock falls induced by earthquakes: a statistical approach. Soil Dynamics
and Earthquake Engineering 22 (2002), 565–577.
Saïd A., Chardon D., Baby P., Ouali J. (2011) Active oblique ramp faulting in the Southern Tunisian Atlas. Tectonophysics,
499(1–4),178–189.
Wang D., Cao L-Z., Piao C-D., Xue Y-D., Wang M. (2011) Study on high and steep slope stability of surface mine based on
RFPA-SRM. Journal of Coal Science and Engineering (China).
Zaïer A., Beji-Sassi A., Sassi S., Moody, R.T.J. (1998) Basin evolution and deposition during the Early Palaeogene in
Tunisia. Petroleum Geology of North Africa. Geol. Soc. Lond., Spec. Publ., 132, 375-393.
Zargouni F. (1985) Tectonique de l’Atlas méridional de Tunisie : évolution géométrique et cinématique des structures en
zone de cisaillement. Thèse, Univ. Louis Pasteur, Strasbourg, 292 p.

456
Innovative Geotechnics for Africa - Bouassida, Khemakhem &Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Etude à la rupture des murs de soutènements ancrés par la méthode cinématique

des solides rigides
The failure Study of theanchoredretainingwallswithkinematic method of rigidsolid
[Link]*1,2,L.Belabed1,2
1
Département de Génie Civil et Hydraulique, Université 08 Mai 45 Guelma, Algérie
2
Laboratoire de Génie Civil et d’Hydraulique (LGCH), Université 08 Mai 45 Guelma, Algérie

RÉSUMÉ - Le renforcement des murs de soutènements ancrés peut être effectuée au moyen des tirants
d’ancrage capables de transmettre des forces de traction qui leurs sont appliquées à une couche de terrain plus
résistante. Ils sont mis en charge préalablement à l'application des actions afin de limiter les dé[Link]
détermination des paramètres influents sur le comportement mécanique des talus renforcés au moyen de tirants
précontraints à été réalisée par l’étude d’un nouveau modèle mécanique basé sur la méthode cinématique de
rupture des solides rigides. L’analyse de l'ensemble (tirant-sol) nous a permis de déterminer les longueurs
d'ancrage nécessaires et économiques pour différents types de sols.

ABSTRACT–The reinforcement ofanchored retaining walls can be performedbyground anchors. Able to

transmit tensile forces applied to them alayerofmoreresilientsoil; they areloaded before application actionsto
limit the deformations. The determination of parameters influencing behavior mechanical reinforced slope
byprestressedhasbeenperformed by thestudy of a newmechanicalmodel based on kinematic method of rigid
solids. Theanalysis of the overall (tied-soil)was used to determinethenecessaryandeconomiclengthsanchor for
differentsoil types.

1. INTRODUCTION

La vérification de la stabilité des ouvrages de soutènement (Murs cloutés, Murs ancrés, ….) se fait dans la
pratique généralement en étudiant le modèle mécanique classique « rupture circulaire » (Corfidir.A,2008). Dans
ce travail on va montrer qu’il existe d’autres modèles mécaniques prouvés par des essais en modèles réduits et
en vraie grandeurs, qui sont plus instables (défavorables) que ce modèle classique. L’objet principal de ce
travail est d’étudier la stabilité des murs ancrés vis-à-vis de la rupture d’ensemble parement-sol-ancrage. Dans
ce but on a étudié les différents mécanismes de rupture possibles, en particulier ceux basés sur la méthode
cinématique des solides rigides, afin de trouver le mécanisme ou le modèle mécanique le plus instable.

2. METHODE CINEMATIQUE DES SOLIDES RIGIDES ET PRINCIPE DE LA SECURITE MINIMALE

L’expérience a montréque les murs de soutènement ancrés dans la plus part des cas préféraient se rompre le
long des surfaces de glissement (Pellet F,1995). Le calcul de cet état limite de rupture peut être effectué donc
d’après la méthode cinématique (Salençon.J,1983). A travers la prise en compte d’une résistance au cisaillement
et éventuel les forces de cohésion le long des surfaces de glissement secondaire les réserves de sécurité du
système sont correctement exploitées. On suppose qu’à l’état de rupture des murs ancrés, les déformations de
cisaillement plastiques se concentrent dans les surfaces de glissement pendant que le reste du massif de sol ne se
déforme pas (Franke.E; Heibaum.M,1988). C’est seulement dans ces conditions qu’on peut appliquer la
méthode cinématique des solides de rupture rigides sur les murs de soutènement (Gäβler.G,1987).

3. MODELES MECANIQUES

En se basant sur les travaux (Gäβler ; Gudehus,1989) sur la formation des solides de rupture rigides en état de
rupture d’ensemble devant et en arrière des murs. Ces solides sont séparés entre eux par des surfaces de
glissement inté[Link] recherche de l’inclinaison critique de la surface de glissement profonde avec
l’horizontale (1) se fait itérativement jusqu'à ce que la longueur d’ancrage nécessaire est obtenue qui vérifie
juste l’équilibre du système. On procède de la même manière pour les trois modèles étudiés dans ce travail qui
sont :
Modèle Kranz : c’est un modèle classique.
Modèle I : à deux solides rigides : la ligne de rupture coupe le scellement au milieu.
Modèle II : à deux solides rigides : la ligne de rupture coupe le scellement à une longueur résiduelle de
scellement derrière la ligne de rupture.

457
q=10kN/m2

H/4

H

 =20kN/m3
f H =10 m
q=10kN/m2
 = 10°

Figure1. Modèle type

Figure.2Modèle KranzFigure.3 Modèle I Figure.4 Modèle II

3.1. Mécanismes à deux solides rigides : la ligne de rupture coupe le scellement au milieu.

Tableau 1. Liste des symboles (modèle I)

SYMBOLE DESIGNATION
a angle de frottement actif mur - sol
 Inclinaison de l’ancrage avec l’horizontale
f angle de frottement actif mur fictif- sol
Inclinaison de la ligne de glissement secondaire

avec la verticale
 Angle de frottement interne
Inclinaison de la surface de glissement principale
1
avec l’horizontale
G Poids propre du solide de glissement 1
P Surcharge
La composante horizontale de la pression active du
Eah
sol sur le mur de soutènement
La composante horizontale de la pression active du
E1h
sol du solde 2 sur le mur fictif
La composante horizontale de la force de cohésion
C1h
sur la ligne de glissement principale
La composante verticale de la force de cohésion sur
C2v
la ligne de glissement secondaire
Q Force de frottement
Ah La composante horizontale de l’effort d’ancrage

Pour ce modèle cinématique, l’étude de la stabilité de l’ensemble mur-sol-ancrage se ramène alors à l’étude
d’un seul corps rigide (1) sur lequel s’exercent la réaction A de l’ancrage sur le corps rigide (1), la réaction de la
paroi sur ce même massif Ea, le frottement du sol sur la ligne de rupture Q, les cohésions C1 et C2 , la surcharge
Pet le poids propre G du massif (1) (Pellet. F,1995). En outre aucune force d’ancrage n’agit sur le solide actif (2)
situé à l’arrière Figure 5,6, car la partie libre de l’ancrage n’est pas coupée par une surface de glissement
(Belabed,2000). C’est pour cela dans ce cas, on a remplacé le solide (2) par la pression des terres active E1.
L’inclinaison de la surface de glissement extérieure avec l’horizontale (11) est prise égale à celle de la surface
de glissement active (/4+φ/2) (FHWA - circulaire n°4,1999).

458
[Link] forces agissantes sur les corps rigidesModèle I.

L’étude de l’équilibre des forces verticales et horizontales donne après transformation algébrique l’équation
d’état limite à la rupture correspondante au niveau de l’ancrage.

[(G  P  E 1htan  δ f  ρ   C 2v )tan(θ1  φ) + A h  E1h] - (1)

[(E ah tanδ a + A h tanε  C 1htanθ1)tan(θ1  φ) + E ah + C1h + C2v tanρ ]  0
3.2. Mécanismes à deux solides rigides : la ligne de rupture coupe le scellement à une longueur résiduelle de scellement
derrière la ligne de rupture

Pour ce modèle en plus des efforts et des réactions illustrés sur la Figure 6 on a ajouté l’effort d’arrachement
du scellement A1 développé dans ce cas à une distance de la longueur de scellement coupéeLc derrière la ligne
de rupture principale.

Tableau 2. Liste des symboles (modèle II)

SYMBOLE DESIGNATION
Arup L’effort de rupture d’ancrage
Aadm L’effort admissible d’ancrage.
 Coefficient de sécurité d’ancrage
A L’effort d’ancrage déterminé par la méthode statique.
As L’effort de service calculé
a Espacement horizontal des ancrages
 Indice de transfert de l’effort par mètre de longueur
fa La valeur limite de frottement de l’ancrage.
L0 Longueur de scellement
Lc Longueur de scellement coupé derrière la ligne de rupture

L’équation d’état limite à la rupture au niveau de l’ancrage de mur de soutènement est déterminée selon le
modèle-II Figure 6.

[(G  P  E 1htan  δ f  ρ   C2v + A 1h tanε)tan(θ1  φ) + A h  E1h ] -

[(E ah tanδ a + A h tanε  C 1htanθ1)tan(θ1  φ) + Eah + (2)
C1h + C 2v tanρ + A 1h ]  0

A 1  Lc avec   
fa A  A rup
,a  s  η   1.5 (3)
 a A  A adm

459
Figure.6Les forces agissantes sur les corps rigidesModèle II.

D’après (DIN 4125)=1.5 et As=AadmAprèsTransformation algébrique on a trouvé :

1.5 A
κ= (4)
L0
4. COMPARAISON DES DEUX MODELES CINEMATIQUES

L’étude des deux modèles cinématiques I et II a été effectuée sur la base des équations d’équilibres
précédemment déterminées. Les études paramétriques sont faites avec les combinaisons des paramètres de sol φ
et c tableau 3.
Tableau .3 Les paramètres de sol

[°] 40 35 30 25 20
c[kN/m²] 0 5 10 15 20

On a développé un programme de calcul en turbo pascal basé sur cette méthode, en faisant varier
l’inclinaison de la surface de glissement (1) jusqu'à ce que la longueur d’ancrage nécessaire soit obtenue qui
vérifie juste l’équilibre du système. Ce calcul s’est effectué itérativement. Les résultats des études comparatives
effectuées pour = [0°, -5°, +5°]pour le cas de chargement le plus défavorable (q=10kN/m2, β=10°)tableau .4.

Tableau .4 Comparaison des longueurs d’ancrages Modèle I pour ρ[°]=(0,-5,5)

φ[°]/c[kN/m²]
L(m) 40/0 35/5 30/10 25/15 20/20
ρ[°]=0 10,56 12,09 13,64 14,09 16,96
ρ[°]=-5 11,46 13,06 13,58 14,83 18,4
ρ[°]=5 10,37 11,19 12,59 14,17 16,26

Tableau 5. Comparaison des pressions des terres E1 Modèle I pour ρ[°]=(0,-5,5)

φ[°]/c[kN/m²]
40/0 35/5 30/10 25/15 20/20
E1 [kN]
ρ[°]=0 125,17 162,4 213,12 267,12 397,1
ρ[°]=-5 269,9 307,31 356,66 427,02 539,44
ρ[°]=5 107,46 125,13 158,86 212,03 312,56

Tableau .6 Comparaison des longueurs d’ancrages Modèle II pour ρ[°]=(0,-5,5)

φ[°]/c[kN/m²]
40/0 35/5 30/10 25/15 20/20
L [m]
ρ[°]=0 10,02 11,38 12,75 13,92 16,27
ρ[°]=-5 10,77 12,18 12,86 14,21 17,59
ρ[°]=5 9,84 10,63 11,87 13,28 15,4

Tableau.7 Comparaison des pressions des terres E1 Modèle II pour ρ[°]=(0,-5,5)

φ[°]/c[kN/m²]
40/0 35/5 30/10 25/15 20/20
E1 [kN]
ρ[°]=0 117,04 148,7 190,85 242,38 366,13
ρ[°]=-5 143,08 185,75 221,03 288,9 483,37
ρ[°]=5 100,53 116,12 143,88 187,3 279,17

460
On remarque que la variation des inclinaisons de la surface de glissement secondaire () entre [-5,5] influe
sur les longueurs d’ancrages et les pressions actives derrière le mur fictif. Les longueurs d’ancrages sont
d’autant plus longues que la pression E1et la cohésion du sol augmentent. L’angle ρ[°]=5 donne les longueurs
d’ancrages les plus courtes

Figure 7 Comparaison des longueurs d’ancrage Figure 8 Comparaison des pressions des terres
ModèlesKranz , I et II derrière le mur fictif ModèlesKranz , I et II

Dans le cas où la ligne de rupture coupe le scellement à une longueur résiduelle derrière la ligne de rupture
Lc, le modèle II donne des tirants d’ancrages plus courts que les deux modèles (I,Kranz).

5. CONCLUSION

Dans ce travail nous avons étudié trois modèles mécaniques basés sur la méthode cinématique de rupture des
solides rigides. Après développement des équations d’état limite décrivant les mécanismes de rupture,on a pu
déterminer les longueurs d’ancrages en faisant varier l’inclinaison de la surface de glissement principale 1pour
les différents types de sols. On peut conclure que l’inclinaison de la ligne de glissement secondaire «mur
fictif »‘’ ainsi que lapression des terres E1 derrière le mur fictifinfluent considérablement sur l’estimation des
longueurs d’ancragescar ces derniers sont plus courtespour=+5°et E1faiblela variation dans les pressions des
terres derrière le mur fictif est due à laprise en compte de la force de cohésion C2 sur cette même ligne de
rupture.

REFERENCES

Belabed L.(2000)Murs de soutènement ancrés : Modélisation de la surface de glissement profonde,Revue française de

Géotechnique, n°92.
Corfidir A.(2008)la méthode de Kranz d’hier à aujourd’hui : une revue [Link]çaisede Géotechnique.
FHWA - Geotechnecal engineering circulair n°4(1999)Ground anchor and anchored system, Publication FHWA-F-99-017-
015, US. Department of transportation.
Franke E., Heibaum M.(1988) Ein betrag zum nachweis der standsicherheitand der tiefen gleitfuge, Bauingenieur63.
Gäβler G (1987)VernagelteGeländesprügeTragverhalten und standsichereitVeröffentlli- chungen, Universitätridericiana,
Karsruhe.
Gäβler, G. Gudehus G.(1989) Anchored Wall. Model tests and statiscaldesign,Proceeding of the 12th international
conference of soil mechanics and foundation engineering, Rio de janeiro.
Pellet F.(1995)Comportement mécanique d’une paroi expérimentale renforcée par clouage , Revue française de
géotechnique, (n°71).
Salençon J(1983)Calcul à la rupture et Analyse [Link] des Ponts et Chaussées, Paris.

Contribution to study and mapping of the risks induced by Tigzirt landslides

(Algeria)
Contribution à l’évaluation et à la cartographie du risque induit par les glissements de
terrain à Tigzirt (Algérie)

Lynda DJERBAL*1,IbtissamKHOUDI2, Nassima ALIMRINA1, Ramdane BAHAR1&Bachir MELBOUCI2

1
University of Science and Technology HouariBoumedienne, BP. 32 El Alia, 16111 Algers (LEEGO), Algeria.
2
University MouloudMammeri of Tizi-Ouzou, Laboratory of Geomaterials, Environment and Installations
(LGEA), Tizi-Ouzou, Algeria.

ABSTRACT - Management of the risks resulting from landslides becomes increasingly heavier in
Algeria. These phenomena cause very expensive equipments disorders, every year. The region of Tizi-Ouzou is
oneof the most affected regions in Algeria. Indeed, major landslides affect more than half of its departments.
Landslides in this region are very extensive (with areas that can reach more than 100hectares) and affect
sites with economic and cultural importance for the region and for the country. Landslides affecting the city
of Tigzirt are one of the most difficult cases to manage, given the economic and cultural importance of
the area and the extent of major landslides. The movement thataffects the centre of the town is the oldest and
largest one; it affects an area greater than 136 ha.
This paper focuses on the study and assessment of the risk levels of the primer and the reactivation oflandslides
in the city of Tigzirt together with the hazards caused by these instabilities for the population and the
built environment. The latter is evaluated by performing a cross between the risk of instability, evaluated using
Tritsh and Didier (1996) method, and the vulnerability of the site, urbanization. A mapping of the
landslides and the risks involved in these movements are also conducted as part of this work. In addition, these
studies allowed classifying a large area of the city in hazardous area.
Keywords: Tigzirt; Landslide; Risk; Vulnerability; Mapping.

Résumé – La gestion du risque dû aux glissements de terrain devient de plus en plus difficile en Algérie. Ces
phénomènes engendrent, chaque année, des coûts et des désordres qui pèsent très lourds sur le développement
des villes affectées. La wilaya de Tizi-Ouzou est l’une des régions les plus affectées en Algérie, vu que des
glissements majeurs affectent plus de la moitie de ces départements. Les mouvements de terrain qui caractérisent
cette région sont très étendus (avec des superficies qui peuvent dépasser 100 ha dans certains cas) et affectent
des sites d’importance culturelle et économique pour la région et le pays. Les glissements de Tigzirt constituent
l’un des cas les plus difficiles à gérer, vu l’importance économique et culturelle des zones affectées. Le
glissement de terrain du centre-ville est le plus ancien et le plus étendu ; il affecte une superficie supérieure à
136 ha.
Cette recherche a pour objectif l’étude et l’évaluation des niveaux d’aléa mouvement de terrain et du risque
induit par cet aléa dans la ville de Tigzirt. Le risque dû aux mouvements de terrain est évalué en effectuant un
croisement entre l’aléa, évalué avec la méthode de Tritsh et Didier 1996, et la vulnérabilité du site, urbanisation.
Une cartographie du risque est également réalisée dans ce travail. En effet, cette étude a permis de classer une
importante superficie de la ville de Tigzirt en zone dangereuse.
Mots-clés : Tigzirt ; Glissement de terrain ; Aléa ; Vulnérabilité ; Cartographie.

1 INTRODUCTION

Landslides are one of the most catastrophic geo-hazards that cause severe injuries, property destruction and
deaths all over the world. The landslide susceptibility and hazard assessment approaches constitutes the
fundamental tool to predicting, planning, assessing and mapping the risk. These studies are based essentially on
previous observed phenomena and includes collecting data, building a spatial database, extracting landslide-
related factors, calculating the frequency ratios, overlaying factors, making instabilities susceptibility maps and
verifying the maps(Lee and Dan 2005).
NorthernAlgeria, because of its morphology,hasasignificant sensitivity [Link] of
themostaffectedregionsinAlgeria the department ofTizi-Ouzou, locatedin the centerof this
[Link] mountainous [Link] of
themajorlandslidesthataffectingthe coastal city [Link] city is located at about 39 Km in the North of Tizi-
Ouzou (Fig. 1). It is a small coastal town known for its cultural and historical richeswith a population

463
ofabout11962 inhabitants. The instable slope, characterized by an inclination of about 15°, is locatedinthe
Cretaceous marl flyschbelongingtothetopSebaou-Isser. The centre city is affected by an important landslide
affecting an area of 136 ha. This study consists to mapthelandslidehazardatthecenter town of TigzirtusingTritsch
et al.(1996) method, based on hydrological and geomorphological interpretation of the landslides. This method
consists on a qualitative approach, which is the most used in Algeria, and use parameters easily evaluated in situ.
New combination is added to this method to assess the landslide hazard considering the vulnerability of the site,
represented by the densityof urbanization.

39 Km

Figure [Link] ofTigzirt city.

2. CAUSES OF THE ACTIVITY OF LANDSLIDES IN TIGZIRT CITY

The landslide is the result of a multitude of factors, which have natural and anthropogenic origin.

2.1 Natural factors

Regionalgeological andgeotechnical parameters effect

Tigzirt city centerislocatedinformationsbelongingtotheSebaou-IsserbasinFlysch(Guirous et al.

2014).Thegeological and geotechnical investigationscarried out by thelaboratoryLNHC2in 2002revealed the
presenceofalteredclayformation, locatedat the interface betweenthesubstrateandsurfacemarlclay layers(Guirouset
al., 2013).In addition,the site mechanical characteristics are,generally, low andsubstantiallyreducedin the
presenceof water.

Site morphology effect

The morphology of natural slopes is used to study the evolution and to delimit the sliding areas and constitute
one of the causes of their initiation. The studied site is characterized by a low inclination, which is between 13°
and 15°. It is composed by a multilayer of an inclined stratigraphy oriented in the direction of the slope
inclination.
The analysis of Ordnance Survey mapofTigzirt region has identified some old instability signs mentioned on
these maps, such as the main detachment niche of the landslide located at an altitude of about 270 m. This
observation shows that Tigzirt landslide is very old and has gone through very active periods in the past. The
visit, which we have performed on the site on January 2014, permitted to detect and locate several signs of
instability; these signs shown on the map of the site morphology (Figures 2 and 3).

464
Figure [Link] signs ofmovementobserved during January 18, 2014site visit.

Figure [Link] of landslide morphologicalsigns.

Effect of the hydrology and the climate

The hydrology of the studied site is characterised by the presence of two small rivers and some secondary
ravines(Guirous et al. 2014)(Fig.1). The lackofhydrography networks, allowing the drainage of surface water,
and the absence of deep drainage considerably increase the activity of landslides by promoting infiltration and
erosion of slopes, seenthat the surfacecrackformationsarepermeableand bedrock(Cretaceous marl) verysensitive
to water. These phenomenon are favoured by the climate of this region, which is characterized by a period
ofheavy rains that lasts about six months, with an average annual rainfall greater than 900 mm (LNHC, 2002).

2.2. Anthropogenic actions effect

The man greatly affects its environment; its actions may be beneficial like they can harm the environment and
cause or exacerbate natural hazards. The negative effects of human actions on a natural slope can be grouped
into two categories: actions on the geometry and weight of the ground and action on site hydrology (natural
streams deviation). Several human destabilizing activities were identified at Tigzirt City (Fig. 4):

465
2.2.1Human actionson the geometryand weight of theSite

- The significant overload of the foot and the centre of the unstable slopes and the uncontrolled cities
development;
- Erection of important works that require excessive earthworks which cause deletions of abutments and
the creation of embankments (changing the geometry of the slope).

2.2.2Human actionson the slope hydrology

- The lack of drainage of superficial water and deep water in the unstable site and its surroundings
(except of some natural gullies);
- Deviation of several natural streams and clogging of water sources by development of urban spaces;
- Abandoning the agricultural activities and deforestation of several slopes.

Figure [Link] pictures showing humans activities in Tigzirt instability.

3. ASSESSMENT AND MAPPING OF HAZARD GENERATED BY LANDSLIDES AT TIGZIRT CITY

Management of natural hazards is an onerous problem, especially when important issues are exposed to the
[Link] Landslide is among the most sensitive cases; as it affects a highly urbanized city (Fig.2)
characterized by a rich history and culture (ruins dating back to Roman times). It is then necessary to assess the
risk posed by the instability of land in this region.
The risk (or probability of occurrence), of reactivation and activation of landslides at the Centertown of Tigzirt,
is studied usingTritsch and Didier(1996)method.

3.1. Evaluation and mapping of hazard by TritschandDidier(1996) method

TritschandDidier (1996) quantified the probability of reactivation of instabilities by performing a combination

between the sensitivity of the site and the movement activity. Five parameters were considered by these authors
to define the sensitivity of the site ‘S’(Table 1) and four classes were distinguished to evaluate the activity of
instability (dormant, inactive, fresh, active). Site sensitivity, obtained by the equation (1) for the site of Tigzirt,
is placed in a scale of values of four classes (Table 2) and is combined with the activity of the movement to
evaluate the occurrence probability of reactivation of instabilities(the Risk of sliding).
S= (4[PF or LF] + 2ER + HF + DIR + 2PEN) × (10/3) (1)

466
Table 1: Parameters of Tritsch and Didier(1996)method.

Parameter Propriety Index Map

Pitch of PF < 0.2 m or LF> 10 m I=3
fracturing (PF) PF= 0.2~0.6 m or LF= 3~10 m I=2
or fracturing PF= 0.6~2 m or LF= 1~3 m I=1
network (LF) PF > 2 m or LF< 1 m I=0

Roughness  5mm, with clay filling. I=3

and spacing of <5mm, potential failure surface. I=2
the joints (ER) <1mm, altered surface. I=1
Closed, rough surface. I=0

Cracks Came running water I=3

moisture (HF) Drips I=2
Humidity I=1
Dry I=0

Structural < 5° I=3

direction with 5°~15° I=2
respect to 15°~30° I=1
front (DIR) >30° I=0

Inclination of -15° ~ -5° I=3

the crack -5° ~ 5° I=2
report to the 5°~ 15° I=1
front (PEN) >15° I=0

Table 2: Quantification of site sensitivity (S).

Sensitivity qualification Calculated index Sensitivity Map

Very favorable (dark green) < 25

Favorable (clear green) 25~49
Unfavorable (yellow) 50~75
Very unfavorable (red) > 75

The very low level of sensitivity is not observed for Tigzirt landslide; since the front direction (DIR) is
favourable to landslide activity (oriented in the slope direction) in the entire studied site.

3.2. Map of the risk

The risk is quantified by combining the site sensitivity level with the activity of this instability (Table 3) using
GIS1system. The obtainedresults allowed to classify an area of about 55ha in high risk and an area greater 84 ha
in medium (Fig. 5).

1
GIS: Geographic Information System. It is a way of database management designed to introduce, store,
manipulate, analyze, combine and display spatial data to solve complex problems of management and planning.
2
LNHC: National LaboratoryofHousingand Construction (Laboratoire National de l’Habitat et de la
Construction).

467
Tableau 3: Classification of risk usingTritsch and Didier (1996) Method.

Sensitivity Very Favourable Unfavourable Very

Activity favourable unfavourable
Dormant Negligible Low Low Medium
Inactive Low Low Medium Medium
Fresh Medium High High High
Active High High High High

600 700 m 602 000 m

4083800 m

4082 300 m

Figure 5. Map of the Risk of instability defined using Tritsch and Didier(1996) Method.

TritschandDidier(1996)evaluated the hazard at the site of Pontoise (France) by performing a combination

between the probability of occurrence (risk) and the intensity(activity of the phenomena) of the phenomenon.
However, these combinations are not valid for all instabilities (Tritsch and Didier, 1996). A new parameter
combinationwas then proposed to study the instabilities of Tigzirt. It is combination between the probability of
occurrence of instability (risk of instability) and the exposed issues(Djerbal. 2013; Djerbal and al. 2014) (Table
4).
Table4: Combination of parameters to assess the hazard at Tigzirt city.

Class 1 Class 2 Class 3

Stakes (Low) (Medium) (High)
Risk
Negligible Negligible Negligible Low hazard
Low Negligible Low hazard Medium hazard
Medium Low hazard Medium hazard High hazard
High Low hazard High hazard Very high hazard

The results show that a significant area of the city of Tigzirt is affected by a very high-to-high hazard (Fig. 6).
The affected area, the very high hazard extends over an area of 30 ha and the high hazard spans an area greater
than 42 ha,concern the city centre and the port area.

468
600 200 m 602 000 m
4083800 m

4082 300 m

Figure 6: Map of the Hazard levels evaluated using Tritsch and Didier(1996)Method.

4. CONCLUSION

A large landslide of great magnitude characterizes Tigzirt city; the instability affects the centre town of Tigzirt.
This landslide affects an urban area and spreadover an area of 136 ha. This study led to observe that the
important activity of the instabilities in this city is mainly due to the action of water and man.
The method of Tritsch and Didier (1996), used in this work, allows the evaluation of the occurrence probability
of the phenomenon (Risk) using site conditions and landslide parameters. The obtained results led to classify an
area of about 140 ha on danger zone. The hazard evaluated by Tiritch and Didier (1996) method is then crossed
with site vulnerability, represented by urbanisation density, to assess the risk induced by instabilities. This
method shows that an area of 72 ha is affected by a strong landslide hazard in [Link] work constitutes a
first approach to the landslide risk study in Tigzirt. However, quantitative and deterministic studies based on
Tigzirt landslide propertiesareneeded to better assess and map these phenomena.

REFERENCES

Guirous L, Djerbal L, Alimrina N, Melbouci B, and Bahar R, Caractérisation des glissements de terrain de la région de Tizi-
Ouzou (Algérie). In 1st International Conference on LandslidesRisk, AinDrahem – Tabarka, Tunisia, 14, 15 et 16 Mars
2013. pp. 117-128, 2013.
Guirous L, Dubois L, Melbouci M, Contribution à l’étude du mouvement de terrain de la ville de Tigzirt (Algérie), Bulletin
of Engineering Geology and the Environment, DOI : 10.1007/s10064-014-0624-6, 2014.
Djerbal, L., Analyse des mécanismes de déformation et de la rupture progressive du versant instable d’Ain El Hammam
(Algérie), These of Doctoratedegree of the University of Tizi-Ouzou, Algeria, july 1st, 2013.
Djerbal L, and Melbouci B, Contribution to the mapping of the landslide of Ain El Hammam (Algeria)’. Advanced
Materials Research, Vol. 601: 332-336, 2013.
Djerbal, L., Alimrina, N., Melbouci, B., &Bahar, R. Mapping and Management of Landslide Risk in the City of Azazga
(Algeria).In Landslide Science for a Safer Geoenvironment (pp. 463-468).Springer International Publishing, 2014.
Lee S, Dan NT. Probabilistic landslide susceptibility mapping in the Lai Chau province of vietnam: focus on the relationship
between tectonic fractures and landslides. Environ Geol 48:778–787, 2005
LNHC Laboratory, Étude géotechnique de la zone de glissement 1ière tranche-Tigzirt, Internal report, file N° 6.0.120.2002,
2002.
TRITSCH, J.J., DIDIER, C., Méthodologie pour la connaissance et l'identification des risques de mouvements de terrain».
Final research project report, INERIS, France, 1996.

Comparative study of analytical and numerical methods in slope stability. A

case study of an open slope mining in DRC.
Etude comparative des méthodes analytiques et numériques en stabilité des pentes. Cas d’une
mine à ciel ouvert en République Démocratique du Congo.

J. Shimatu Mbuyi1,2,3, H. TshimangaNzokona1 and V. Fariala Muchanga1

1
Université de Mbujimayi, Mbujimayi, RDC
2
Université Officielle de Mbujimayi, Mbujimayi, RDC
3
Institut National du Bâtiment et Travaux Publics, Kinshasa, RDC

ABSTRACT -Slope stability in a mining shall be monitored and controlled to ensure the smooth running of the
mining activity. This paper presents an analysis of the slope stability of a massive kimberlite located at
Mbujimayi, a diamond city in DRC. The study is made from the results of a structural geology and those
geomechanical testing of rock. It is conducted in order to prevent failure. Two methods were used:
- The analytical method or limit equilibrium method of determining the limits of the stability conditions. In this
case, the safety factor is a ratio between the resistant forces and the motion forces.
- The numerical method or analysis of stresses and strains. Here, Plaxis, a finite element software was used.
The comparative analysis of these two methods has shown that the values of the safety factor calculated by
numerical methods are low compared to those of standard analytical methods with between 15 and 18%.
Numerical methods, in addition to determining the safety factor, provide additional information on the state of
stability of rock mass that allow us to know the overall behavior of the slope and to manage it well putting us
further risks.
Key Words : Slope stability, stereogram, safety factor, finite element, discontinuity

RESUME : La stabilité des talus dans une mine doit être suivie et contrôlée en vue d’assurer la bonne marche de
l’activité minière.L’article présente l’analyse de la stabilité des talus d’un massif kimberlitiquesitué à
Mbujimayi, une ville diamantifère de la RDC. L’étude est faite à partir des résultats d’un levé structural et ceux
des essais géo mécaniques des roches du massif. Elle est menée dans le but de prévenir une rupture des flancs de
ce massif. Deuxméthodesontétéutilisées :
- La méthode analytique ou méthode d’équilibre limite qui consiste à déterminer les conditions limites de
la stabilité. Dans ce cas, le coefficient de sécurité est un rapport entre les forces ou moments résistants
et les forces ou les moments moteurs.
- La méthode numérique ou par analyse des contraintes et déformations. Pour celle-ci, le code éléments
finis sous le logiciel Plaxis a été utilisé.
L’analyse comparative de ces deux méthodes a montré que les valeurs du coefficient de sécurité calculé par
les méthodes numériques sont faibles par rapport à celles des méthodes analytiques avec un écart compris entre
15 et 18 %. De plus, les méthodes numériques, en plus de la détermination du coefficient de sécurité, donnent
des informations supplémentaires sur l’état de la stabilité du massif qui nous permettent de connaitre le
comportement global du talus à la rupture et nous permettent de bien le gérer en nous mettant loin des risques.
Mots clés : Stabilité des talus, stéréogramme, coefficient de sécurité, éléments finis, discontinuités

1 INTRODUCTION

La problématique de stabilité des talus trouve toute son importance avec l’approfondissement de la carrière ou
de la mine à ciel ouvert. Plusieurs études ont montré qu’au-delà de 100 mètres de profondeur, il y a lieu de
s’interroger sur la garantie sécuritaire offerte par une exploitation à ciel ouvert (Hoek, 2002).Bien évidemment,
les enjeux économiques liés à l’exploitation d’un gisement exigent d’éviter tout danger qui peut compromettre la
poursuite normale des travaux miniers. Parmi ces dangers, les glissements de terrains sont à notre sens ceux dont
il faut éviter à tout prix ou au mieux d’en limiter l’ampleur.
Cet article s’intéresse plus particulièrement à la stabilité d’un flanc d’un massif kimberlitique situé dans une
ville de la République démocratique du Congo exploité par une entreprise diamantifère appelée MIBA qui a
jadis présenté d’inquiétants signes d’instabilité. Il consistera à analyser la stabilité du massif avec les méthodes
analytiques et numériques.
Pour atteindre notre objectif, nous avons d’abord mené un travail minutieux de récolte d’informations tant
géologiques, mécaniques que miniers sur le terrain et au bureau afin de nous approcher le plus près possible des
conditions réelles des roches sur le terrain.

471
2 LOCALISATION DE LA ZONE D’ETUDE

Le gisement qui fait l’objet de cet article se trouve à Mbujimayi, chef-lieu de la province du Kasaï-Oriental
(RDC) dans la concession de la MIBA, communément appelé «Polygone minier» qui se trouve au Sud-est de la
ville de Mbujimayi.
Le gisement du Massif 1 se situe dans la zone comprise entre les X (7.500 – 8. 400) d’Ouest à l’Est et les Y
(2.800 – 4.000) du Sud au Nord en coordonnées locales. Son altitude varie de Z: 610 – 600 en surface à Z: 390
au fond et la longueur totale suivant le grand axe est de 750 m et le petit axe mesure environ 450 m. Signalons
par ailleurs que la ville de Mbujimayi se situe entre 6° et 7° de latitude sud et entre 23° et 24° de longitude Est.

Figure [Link] géographique de la zone d’étude

3 CONSIDERATION GEOLOGIQUE DE LA ZONE D’ETUDE

Le massif1 fut découvert en 1946 par puits manuels confirmés par des levés géophysiques. La prospection en
galerie s’est poursuivie jusqu’en 1950; sa structure a été déterminée au moyen de 107 sondages carottés dont le
plus profond a atteint 310 mètres, il est le premier massif à être découvert et sa mise en place a été faite par
coalescence de 3 périodes d’éruptions.; l’éruption du Massif 1 a formé son cratère entièrement dans les dolomies
(Figure2). Ces dolomies sont l’encaissant prédominant et compétent, bien que le sable brun-rouge et le grès fin
rouge constituent sporadiquement à l’Est l’[Link], le faciès des roches du Massif 1 sont: la kimberlite
massive, la xéno-kimberlite, les épiclastites, le calcaire et enfin le sable, gris et brèches regroupés en une même
unité lithologique.

472
Figure [Link] Flanc dolomitique du massif 1 de la Minière de Bakuanga

La division de géologie de la société Mibaa déduit, à partir des résultats des sondages, la modélisation de
l’encaissant dolomitique qualifiée de la plus probable [Link] gisements diamantifères exploités par cette
société sont de deux types : les gisements primaires ou kimberlitiques et les gisements secondaires ou
détritiques. Le massif 1 qui fait l’objet de cette étude est l’un desdouze gisements primaires du polygone minier
de la société.

4 MATERIELS ET METHODES

4.1. Matériel

Le levé structural a été réalisé sur les flancs Sud et Ouest du massif 1 à l’aide d’une boussole gé[Link]
total de 192 discontinuités a été levé en direction et pendage dont 128 au flanc Ouest et 64 au flanc Sud.
Ces discontinuités ont été représentées à l’aide d’une projection stéréographique utilisant le canevas de WULF.
Le logiciel DIPS 5.1 ([Link]) a servi à cette projection ainsi qu’à l’identification des modes de
rupture probables en associant les données structurales aux propriétés de résistance (angle de frottement et
cohésion) de la [Link] de traction et de compression simple au laboratoire ont permis de déterminer les
propriétés mécaniques et de résistance de la matrice rocheuse. Le logiciel ROC DATA ([Link])
nous a permis d’évaluer les propriétés géomécaniques du massif à partir de celles de la matrice rocheuse.
L’analyse de la stabilité a été réalisée par la méthode analytique d’équilibre limite et par la méthode d’éléments
finis implémentée sous le logiciel Plaxis 8.x ([Link]).

4.2. Analyse des discontinuités par diagramme stéréographique

Figure 3. Détermination du mode de rupture probable au Flanc Ouest du massif 1 de la MIBA.

Le traitement par analyse stéréographique conduit à une possibilité de rupture plane à l’Ouest du massif1
(Figure3), le pôle de la famille préférentielle dont la direction est 61° et le pendage 50°SW est dans la zone

473
d’instabilité. Par conséquent, il peut occasionner une rupture plane, cette famille sera alors utilisée comme
référence dans le calcul de la stabilité de talus.

5. ANALYSE DE LA STABILITE DU MASSIF

5.1. Géométrie du talus

La géométrie du modèle étudié est présentée sur la figure 4. La hauteur du talus H = 190m; la pente β = 75° et la
base du prisme d’éboulement B= 163 m. Le talus étudié est composé d’une couche de roche (Dolomie).

Figure 4 : Géométrie du modèle

5.2. Calcul du coefficient de sécurité par la méthode analytique

Les hypothèses de base retenues pour cette analyse sont :

1. La rupture se produit par glissement le long d’une surface de discontinuité généralisée mais plane
(hypothèse conservatrice car on a toujours des redents).
2. Le prisme glissant est monolithique c.à.d. qu’il n’y a pas de déformation ni de rupture au sein du
prisme (hypothèse optimiste car il y a toujours des ruptures internes).
3. Les forces agissant selon une résultante dont le vecteur est inconnu (mais pas la grandeur). On néglige
les moments.

On va faire une analyse bidimensionnelle (considérer une épaisseur unitaire) qui est justifié par le fait qu’on va
considérer le talus pour lequel la force géométrique restera constante suivant la direction. Dans ce cas la relation
utilisée est :

tg c
F  (1)
tg H sin 2  cot g  cot g 
2
Le calcul par la méthode analytique donne une valeur de 1,9. En variant certains paramètres
notamment la cohésion et l’angle de frottement interne on trouve les résultats qui sont repris dans le tableau 2.
On constate que le coefficient de sécurité est largement supérieur à 1,3 (valeur admissible pour une mine à ciel
ouvert) donc aucun risque de glissement plan n’est à craindre.

5.3. Calcul du coefficient de sécurité par la méthode numérique

L’utilisation des méthodes numériques permet de simuler le comportement du massif du fait de leur géométrie
complexe. Le maillage est utilisé pour définir la géométrie du domaine réel à étudier et permet de représenter un
milieu continu sous une forme spatiale par un nombre fini de subdivisions élémentaires.

5.3.1. Caractéristiques du matériau et conditions aux limites

474
Des conditions aux limites en déplacement sont utilisées sur les bords latéraux et sur la base. La partie
supérieure est laissée libre, car le modèle est sollicité par gravité.Les propriétés des couches de la roche sont
résumées dans le tableau ci-dessous:

Tableau [Link] propriétés caractéristiques de la roche

Paramètres Désignations Couche de roches

Etat de la roche sec
Poidsvolumique apparent  28 kN/ mଷ

Poidsvolumiquesaturé  sat 29 kN/ mଷ

Module de Young Em 8234100 kPa

Coefficient de Poisson  0,37
Cohésion C 1215kPa
Angle de frottement  37,02
Résistance à la compression  cm 3,763 MPa

Résistance à la traction  tm 0,246 MPa

Angle de dilatance  7,02

5.3.2. Génération du maillage

Le maillage est constitué d’élémentstriangulaires à 15 noeuds. Nous avons préféré un maillage moyen afin
d’optimiser le rapport temps/qualité des calculs. Ainsi après discrétisation, le nombre d’éléments est 71 et le
nombre de nœuds623.
5.3.3. Simulation numérique
Les simulations numériques sont faites au moyen du code de calcul éléments finis Plaxis version [Link] calcul
comporte deux phases, lapremière est en calcul plastique et la seconde prend en compte la réduction des
paramètres mécanique C et  de la roche pour le calcul du coefficient de sécurité. Les calculs s’effectueront en
déformation plane et le comportement du massif est décrit par le modèle de Mohr-Coulomb.
Le coefficient de sécurité calculé pour ce talus est ∑ ‫ =݂ݏ ܯ‬1,640. La stabilité du talus est assurée, car la valeur
du coefficient de sécurité est supérieure à 1,[Link] montre que le talus résistera aux efforts qui lui sont transmis
par son propre poids ainsi que les efforts de sollicitation qui proviennent du terrain environnant.
En plus du calcul du coefficient de sécurité, la méthode numérique fournit des informations sur l’état de
contraintes et déplacements dans le massif (Figures 5).

Figure 5. Déplacements verticaux dans le massif

La figure5 montre que les déplacements sont généralement faibles dans le massif sauf au niveau de
l’affleurement, où l’on peut enregistrer des mouvements de soulèvements plus importants en fonction de l’allure

475
des cercles de glissement (surface de glissement représenté par les lignes multicolores). Nous recommandons
une surveillance de ces zones pendant la durée de l’exploitation de la mine.
En variant les mêmes paramètres de cohésion et l’angle de frottement interne on trouve les valeurs
correspondant aux méthodes numériques. Cette situation se présente dans le tableau 2 de comparaison du
coefficient de sécurité.

5.4. COMPARAISON DE RESULTATS ET INTERPRETATION

Le tableau 2 présente les coefficients de sécurité calculés par les deux méthodes. Nous constatons que les
valeurs du coefficient de sécurité calculées par les méthodes numériques sont faibles par rapport à ceux des
méthodes analytiques avec un écart comprit entre 15 et 18 %.

Tableau 2. La comparaison du coefficient de sécurité du talus calculé en utilisant différentes méthodes

C  FS FS Écarts
(kPa) (degrés) méthodeanalytique méthodenumérique (%)

1069 37,02 1,829 1,496 18%

1482 37,02 2,29 1,910 17%
1660 37,02 2,491 2,052 18%
1215 43,44 2,15 1,827 15%

Les méthodes numériques, en plus de la détermination du coefficient de sécurité, nous donnentdes

informations supplémentaires sur l’état de la stabilité du massif, sur la distribution des contraintesdans le massif,
les déplacements et les déformations engendrées par poids propre du massif.
Les résultats de calcul de ces deux méthodes nous montrent que le massif Kimberlitique de la MIBA peut être
considérer globalement comme étant stable.

5 CONCLUSION

L’article a présenté une étude de la stabilité d’un talus minier en RDC. Pour ce faire des levés des discontinuités
et les essais de laboratoire ont été effectués en vue de déterminer les familles des fractures dominantes,
lesquelles sont susceptibles d’entrainer la rupture du massif. Les essais géomécaniques ont permis d’approcher
la loi du comportement de la matrice rocheuse et de déterminer les propriétés physiques et mécaniques de
roches. Les simulations du comportement des talus ont été faites avec le logiciel PLAXIS. Nous avons considéré
que le milieu est sec et que l’importance des pressions interstitielles générées par l’eau des nappes est
négligeable.
L’étude comparative entre les deux méthodes nous a montré que les coefficients de sécurité des méthodes
numériques sont faibles par rapport à ceux des méthodes analytiques avec un écart moyen de 17%.En outre les
résultats de ces deux méthodes nous ont montré que la stabilité du massif est assurée et l’angle de talus opté pour
le projet d’exploitation permet de faire face aux enjeux technico-économique du projet. Nous avons également
montré l’efficacité et les performances des méthodes numériques par rapport aux méthodes analytiques.

REFERENCES

Bieniaswski Z.T. (1974). The geomechanics classification in rock engineering applications. N.Y., Wiley.
Brawner C.O, Golder H.Q (1976). Stability in open pit Mining. Ed. Brawner C.O and Milligan V.
BrinkGereve R.B.J, Vermeer P.A (2003). Plaxis version 8, Tutorial manual. DELFT, University of Technology, Pays-Bas.
BrinkGereve R.B.J, Vermeer P.A (2003). Plaxis version 8, scientific manual. DELFT, University of Technology,Pays-Bas.
Duncan C.W, Christopher W. M (2004), Rock Slope Engineering. Spon Press, London and New York.
Hoek E. (2002). Practical Rock Engineering; Rocsience, Toronto, Canada.
Hoek E., Brown E.T (1980): Underground Excavations in Rock, London Institute Mining Metallurgy.
Hoek E., Brown E.T (1980). Empirical Strength criterion for rock masses. Geotechnical Engineering Div. ASCE 106 (GT9).

476
GEOTECHNIQUE ET ENVIRONNEMENT

ENVIRONMENTAL GEOTECHNICS

Faisabilité de la récupération d’énergie à partir du biogaz dans la décharge de

Sousse, Tunisie
Biogas energy recovery in the Sousse Municipal Solid Waste landfill, Tunisia
Moncef Zairi*1, Amal Kammoun1, Nihed Hcine1, Youssef Frikha1, Sana Abid1,
National School of Engineers of Sfax,BP: 1173, 3038 Sfax, Tunisia, [Link]@[Link]

RÉSUMÉ -Une enquête a été menée entre 2008 et 2014, afin de déterminer la composition des déchets solides
municipaux et son influence sur la production de biogaz dans la décharge de Sousse. Cette dernière se trouve à
proximité de la localité d’Oued Laya, 17 km à l'ouest de la ville de Sousse. Elle est en exploitation depuis juillet
2008 et reçoit environ 410 Mg/jour de DSM du gouvernorat de Sousse.
La fraction organique est le composé prédominant dans les DSM, suivie de papier, fines, plastique, cuir,
caoutchouc, métal, textile, verre et céramique. La composition des DSM est bien représentative par rapport à
celle des villes des pays en voie de développement.
Basé sur les quantités annuelles de DSM enfouis et en utilisant le modèle LandGEM, la production de gaz de
décharge maximale attendue se produira en 2020, un an après la fermeture de la décharge avec un taux de 8 x
106 m3/an. L'analyse de la conversion potentielle du biogaz en énergie électrique montre un maximum de
conversion autour de 42 GWh en 2020 et diminue pour atteindre 24 GWh après 20 ans de fonctionnement. Cette
estimation est basée sur une efficacité de captage du biogaz de 33 % et une efficacité énergétique de 33 %, ce qui
donne une faisabilité économique pour une installation de puissance 10 MW.

ABSTRACT– A survey was conducted between 2008 and 2014 in order to identify MSW composition and its
influence on biogas yield from the Sousse landfill. The landfill of Sousse is located in OuedLaya, 17 km west of
Sousse city. It is in operation since July 2008 and receives about 410 Mg/ day of MSW from Sousse
[Link] organic fraction isthe predominant compound in the MSW, followed by paper, fine elements,
plastic, leather, rubber, metal, textile, glass and ceramic. The recognised MSW composition is well
representative compared to that of cities in developing [Link] on the annual MSW landfilled quantities
and using the LandGEM model, the expected peak landfill gas production is estimated to occur in 2020, one year
after the landfill closure with a rate of 8 x106 m3/year. The analysis of the potential conversion of LFG to electric
energy shows a maximum LFG-to-electricity energy around 42GWhin 2020 and decreases to reach 24 GWh
after 20 years of operation based on an LFG collection efficiency of 33% and energy efficiency of 33% giving an
economic feasibility of a 10 MW power plant.

1INTRODUCTION

La mise en décharge est l'une des méthodes les plus économiques courantes d'élimination de déchets solides
municipaux (DSM) dans les pays en développement. Les principaux impacts environnementaux de ces
décharges, contenant des quantités élevées de matières organiques biodégradables, sont causés par lelixiviat et le
biogaz. La quantité et la qualité du lixiviat et du biogaz dépendent des caractéristiques des DSM, de la
conception et dufonctionnement de la décharge etdes conditions climatiques (El-Fadel et al ., 2002).
La composition des déchets urbains solides dépend de nombreux facteurs, tels que les conditions socio-
économiques, climatiques et géographiques (Sharholy et al. 2008). La connaissance de la composition de MSW
est cruciale pour la planification et la conception de l'équipement nécessaire, les systèmes de traitement et des
programmes et plans de gestion (Tchobanoglous et al. 1993, Guermoud et al.,2009).
Le biogaz est généré par la dégradation anaérobie des déchets urbains solides. Outre les principaux composants
de méthane (50 à 65 % du volume) et le dioxyde de carbone (volume de 30 – 40 %), le biogaz contient plus
d'une centaine de composés chimiques potentiellement toxiques (Chiriac , 2007).
Le méthane est une des plus importantes émissions de gaz à effet de serre avec un potentiel de réchauffement
planétaire de 25 fois celuidu dioxyde de carbone (Ehrig, 1983). En raison de la contribution à l'effet de serre, la
toxicité, le danger d'explosion et la gêne olfactive, la collecte et la combustion du biogaz pour produire de
l'énergie pourraient résoudre la pollution environnementale et contribuer à la limitation de la pénurie d’énergie
(IPCC, 2007).
En Tunisie, la génération des DSM est passée de 0,5 kg/habitant/jour en 1996 à 0,6 kg/habitant/jour en
2005(MEDD, 2005). Cette augmentation est due à la croissance urbaine rapide et démographique que connaît le
pays (Tizaoui, 2006). Parallèlement, un manque persiste dans lagestion et lavalorisation des déchets urbains
solides ainsi que dans la planification et l'exploitation de décharges en raison de l'information insuffisante sur la

479
caractérisation des déchets et des contraintes financiè[Link] décharge contrôlée de Sousse située dans la région
d’Oued Laya, à environ 17 km à l’ouest de la ville de Sousse est en exploitation depuis Juillet 2008 et reçoit
environ 410 t/j de déchets ménagers et assimilés en provenance du gouvernorat de Sousse.
Dans le présent travail, il s’agit de quantifier le biogaz issu de la décharge de Sousse en utilisant le modèle
LandGem et d’étudier la faisabilité économique de la valorisation du biogaz en électricité.

2DESCRIPTION DE LA DECHARGE

La décharge contrôlée de Sousse (fig. 1), exploitée depuis Juillet 2008, couvre une superficie de 47 ha, dont 20
ha sont aménagés. Autour du site, la principale occupation du sol est agricole et représentée par des oliviers
cultivés en sec. Les réseaux de distribution d’eau potable et les canalisations publiques des eaux usées sont à
environ 15km du site. Le réseau électrique à haute tension est distant de 5km tandis que l’électricité à moyenne
tension est à environ 3Km.
La décharge contrôlée de Sousse comporte deux grands casiers de superficie 6 ha chacun, séparés par une
digue centrale stable et étanche. Chaque casier est subdivisé en 3 alvéoles, aménagés et pourvus d’un système de
drainage des lixiviats et d’un système d’étanchéité assurant une protection du sol et de la nappe vis-à-vis des
infiltrations. L’étanchéité à la base des casiers est composée de trois couches : une couche d’argile compactée de
30 cm d’épaisseur, une couche de béton d’asphalte de 20 cm d’épaisseur et une couche drainante de gravier de
30 cm d’épaisseur, étalée sur le fond du casier et dont le rôle est d’assurer le drainage du lixiviat vers les
conduites d’évacuation. La couverture finale des déchets est composée d’une couche de matériaux semi-
perméables de 60 cm d’épaisseur, un géo-composite constitué par une membrane alvéolaire imperméable en
PEHD d’épaisseur de 0,8 mm assurant les fonctions d’âme drainante et d’imperméabilité, et d’un filtre géo-
synthétique non tissé assurant la protection physique de la membrane et une couche de terre végétale de 30 cm
d’épaisseur.

Figure1. Localisation et infrastructure de la décharge contrôlée de Sousse.

3. QUANTITE ET COMPOSITION DES DECHETS

La quantité de déchets totale enfouis dans la décharge de Sousse durant la période de Juillet 2008 à Décembre
2013 est de l’ordre de 822933 tonnes. La quantité des déchets enfouis dans le casier N°2, ayant un volume total
de 526141 m3, est de l’ordre de 586457 tonnes, la densité des déchets enfouis au niveau du casier N°2 est donc
égale à 1,119. Ce dernier a été fermé en Mai 2012.
A partir de Juin 2012, l’exploitation du casier N°1 a commencé et sa fermeture est prévue pour l’année 2019.
La quantité de déchets enfouis dans ce casier entre Juin 2012 et Décembre 2013 est de l’ordre de 236475 tonnes.
La composition des déchets ménagers varie d’une ville à une autre et d’une saison à une autre. Les déchets
ménagers et assimilés générés dans le gouvernorat de Sousse sont essentiellement composé de 67% de déchets
organiques, 11% de plastiques, 10% de papiers et carton, 3% de métaux, 3% de textiles, 4% de fines et 2% de
divers autres constituants (ANGeD, 2007).

480
La collecte du biogaz, à partir des casiers de la décharge contrôlée de Sousse, nécessite la réalisation de puits
forés, la mise en place de réseaux de collecte et des stations de pompage. Le biogaz collecté est ensuite
transférépour la combustion dans une torchère. Trente forages de captage du biogaz, répartis entre les casiers 1 et
2 sont réalisés à la tarière hélicoïdale de 600 mm de diamètre.
Après le captage du biogaz au niveau des puits, il est acheminé via le réseau de transport vers un sur-presseur
jusqu’à la torchère de type ouverte et de capacité 250 m3/h. Le système de torchage est relié à un système
d’information permettant de collecter et d’archiver l’ensemble des données d’exploitation.

4. SIMULATION DE LA PRODUCTION DU BIOGAZ

4.1 Mesure de la production de Biogaz dans la décharge

L’unité de torchage fonctionne depuis le 1er Février 2014 avec un régime de fonctionnement de 24h /24h. Le
volume de biogaz cumulé jusqu’à la fin de Mars 2014 est de l’ordre de 469982 Mm3. La moyenne du débit du
biogaz sur les deux mois Février et Mars est de l’ordre de 7956 Mm3/j.
Des analyses de la qualité de biogaz à l’entrée de la torchère et au niveau des 20 puits sont réalisées au moyen de
la baie d’analyse pour le premier cas et au moyen d’un appareil de mesure appelé Trigaz pour le deuxième cas.
Ces appareils permettent de mesurer les pourcentages de CH4, du CO2 et d’O2. Les mesures effectuées pour le
mois de Février 2014 sont présentés par la figure 2.

Figure 2 Pourcentages de biogaz au niveau de la torchère et des puits de dégazage en Février 2014.

4.2. Simulation de la production de Biogaz dans la décharge

De nombreux modèles existent pour évaluer les émissions de biogaz des décharges contrôlées. Nous nous
intéressons ici au modèle LandGem (LandfillGas Emission) de version 3.02 qui est très utilisé dans la littérature
et recommandé par l’Agence des Etats Unis pour la Protection de l’Environnement (USEPA). C’est un modèle
de 1er ordre qui permet d’estimer la production de biogaz. Ce modèle tient compte de l’historique de remplissage
du CET ou des apports moyens annuels et la durée d’exploitation du site. Il est construit sur une équation
cinétique du premier ordre. De plus amples détails sur la dérivation du modèle sont décrits par Paraskaki et
Lazaridis (2005).L’équation de base est donnée par

n 1  M i   ktij
 e
   kL (1)
Q 0  10 
CH4 i  1 j  0.1  
Avec:
- QCH4 = Production annuelle de méthane au cours de l'année de calcul (m3/an) ;
- i = Incrément de temps (1année) ;
- n = Age de la décharge (an);
- j = Incrément de temps (0,1 année) ;
- k = Taux de production de méthane (année-1) ;

481
- L0 = Capacité potentielle de production de méthane (m3/tonne) ;
- Mi = Masse de déchets acceptés au cours de l’ième année (tonne) ;
- tij = Age de la jème fraction de la masse des déchets Mi acceptée au cours de l’ième année.
La fraction de carbonre organique dégradable est (DOC) est introduite dans l’équation (2) pour donner L0
(Guermoud N et al., 2009, Zairi et al., 2014).

L0=MCF × DOC × DOCf × F × 16/12 (2)

Avec:
- MCF : Facteur de correction du méthane (= 1 pour décharges contrôlées et 0,4 à 0,8 pour décharges
sauvages) ;
- DOC : Carbone organique dégradable exprimé en kg/Tonne ;
- DOCf : Fraction assimilée au carbone organique dégradable (égale à 0,5 par défaut) ;
- F : Fraction de méthane dans le gaz d’enfouissement (égale à 0,5 par défaut) ;
- 16 /12 : facteur stœchiométrique (rapport entre la masse moléculaire de méthane et de carbone).
Le taux de décomposition k est déterminé en fonction de la précipitation moyenne annuelle P selon
l’équation (3) suivante :
k = 3,2 × 10-5 P + 0,01 (3)
Le torchage du biogaz de la décharge de Sousse a commencé depuis Février 2014. Ainsi, les mesures
disponibles correspondent aux deux premiers mois de dégazage. De ce fait, la valeur adoptée pour le calage du
modèle est la moyenne des mois de Février et Mars 2014, supposée constante tout au long de l’année 2014.
Lors de l’application du modèle dans la décharge contrôlée de Sousse, les paramètres d’entrée utilisés sont de
type inventaire pour une décharge contrôlée située dans une zone aride. Les valeurs utilisées sont les suivantes :
- Constante relative au taux de génération du méthane k=0,02 an-1 ;
- Potentiel de production du méthane L0= 100 m3/tonne;
- La concentration de composé organique non méthanique NMOC= 600 ppmv.
Pour le paramètre d’entrée ‘’le pourcentage de méthane par rapport au biogaz’’, la valeur retenue est de 55 %.
La projection de la production du biogaz dans la décharge contrôlée de Sousse est estimée par le modèle
LandGem (figure 3), avec les paramètres calés (k=0,037 an-1 et L0=80 m3/tonne) et la quantité de déchet déjà
enfouie jusqu’à 2013 ainsi que la quantité estimée pour la période 2014-2019.
Ainsi, l’apogée de production du biogaz correspondra à l’année 2020 avec un débit maximal de 8 Mm3/an. Le
flux du méthane maximum sera de l’ordre de 4,5 Mm3/an pour la même année. A partir de l’année 2020, le flux
du biogaz va diminuer jusqu’à son atténuation après une centaine d’années environ.

Biogaz total (Mm3/an) CH4 (Mm3/an) C02 (Mm3/year)

9,00

8,00

7,00

6,00
Biogaz (Mm3/an)

5,00

4,00

3,00

2,00

1,00

0,00
2008 2028 2048 2068 2088
Année

Figure 3. Projection de la production du biogaz dans la décharge contrôlée de Sousse.

482
5 VALORISATION DU BIOGAZ

5.1. Potentiel énergétique

Le pouvoir calorifique d’un combustible, exprimé enKWh/m3 pour le gaz, est la quantité de chaleur dégagée par
la combustion complète de ce dernier. Le biogaz est un combustible dont la valeur énergétique se mesure en PCI
(Pouvoir Calorifique Inférieur) qui représente l'énergie produite par la combustion complète d'1 m3 de gaz sans
tenir compte de la chaleur latente dans la vapeur d'eau.
Le PCI du biogaz est proportionnel à sa teneur en méthane. Pour le biogaz de la décharge contrôlée de
Sousse, composé de 55% de méthane, le PCI est égale à 9,42 × 0,55= 5,181 KWh/m3.
Ainsi le potentiel énergétique du biogaz de la décharge contrôlée de Sousse est présenté par la figure 4. En 2020,
le potentiel énergétique maximum est de l’ordre de 42 GWh. Cette valeur diminue au cours du temps et atteint
24 GWh, après une durée de 20 ans.

45
Potentiel énergétique (en GWh/an)

40
35
30
25
20
15
10
5
0
2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035
Temps

Figure 4. Potentiel énergétique du biogaz de la décharge contrôlée de Sousse.

5.2. Faisabilité économique

Dans cette étude, un moteur de puissance 10 MW, avec un rendement de 33%, est choisi pour produire de
l'électricité.
Les hypothèses suivantes sont prises en compte dans l’analyse économique de la conversion du biogaz en
électricité par un moteur à combustion interne:
- L'analyse économique est réalisée pour une période de 15 ans ;
- Les coûts annuels d'entretien et de maintenance sont estimés à 10% du coût du moteur à combustion
interne ;
- Le taux d'actualisation est de 10%.
La production d'énergies renouvelables permet de réduire les émissions de gaz à effet de serre. En effet, un
kilogramme de méthane (CH4) a un Potentiel de réchauffement global (PRG) 21 fois supérieur à un kilogramme
de dioxyde de carbone. La valorisation du biogaz des décharges peut réduire les émissions de gaz à effet de
serre. Dans ce cas d’étude, les réductions d'émissions totales réalisables à partir de la récupération de biogaz au
cours de la période de 15 ans, 2015 à 2030, sont de l’ordre de 7,73 105 tonnes d'équivalent CO2.
Sur la base des données et hypothèses précédentes, l’évaluation de la rentabilité des investissements est
présentée par le tableau 1 pour la période de 15 ans considérée. Le Cash Flow est la trésorerie générée chaque
anné[Link] taux d’actualisation permet d’orienter les choix d’investissements. Le Cash Flow actualisé est donné
par la formule suivante :

Avec:
- CFA : Cash Flow Actualisé;CFt: le cash-flow de la période t;k: le taux d’actualisation (10%) ;
La Valeur Actuelle Nette, calculée par la différence entre la somme des cash-flows prévisionnels actualisés
(tableau 1) et le coût de l’investissement initial est supérieure à zéro, indiquant que le projet est rentable pour une
période de 15 ans.

483
6. CONCLUSION

La dégradation des déchets ménagers et assimilés dans la décharge contrôlée de Sousse produit entre autres du
biogaz. Ce dernier est formé par un mélange de méthane, de gaz carbonique et d’éléments en trace. Le biogaz a
des effets néfastes sur l’environnement et sur la santé humaine.
L’objectif de ce travail consiste à simuler la quantité de biogaz potentiellement générées dans la décharge
contrôlée de Sousse et l’évolution de cette production au cours du temps en appliquant le modèle LandGem. Il
s’agit aussi, d’étudier la possibilité de la conversion électrique du contenu énergétique du biogaz.
L’apogée de la génération du biogaz, est estimée pour l’année 2020, soit une année après la fermeture de la
décharge, avec un débit maximal de 8 Mm3/an. Le flux du méthane maximum sera de l’ordre de 4,5 Mm3/an
pour la même année. A partir de l’année 2020, le flux du biogaz va diminuer jusqu’à son atténuation après une
centaine d’années environ.
Le potentiel énergétique du biogaz atteint une valeur maximale de 42 GWh en 2020. Cette valeur diminue au
cours du temps et atteint 24 GWh, après une durée de 20 ans.
L’analyse économique de la valorisation électrique du biogaz par un moteur de 10 MW de capacité et de 33% de
rendement, a montré la faisabilité de l’option de la production d’électricité.

Tableau1. Evaluation de la rentabilité économique (en dollars)

Année Revenu Charges d’exploitation Cash flow CFt (1+k)^(-t)
1 733306 49074 684231 622028
2 925755 53981 871773 720474
3 1149362 59379 1089982 818919
4 1406776 65317 1341458 916234
5 1704352 71849 1632503 1013656
6 2046375 79034 1967340 1110512
7 2168922 86937 2081984 1068387
8 2300133 95631 2204501 1028416
9 2438310 105194 2333115 989468
10 2583777 115714 2468063 951545
11 2740102 127285 2612816 915776
12 2903291 140014 2763277 880465
13 3078365 154015 2924349 847079
14 3263307 169417 3093890 814717
15 3458731 186359 3272371 783379

REFERENCES

ANGeD (2007),Agence Nationale de Gestion des Déchets, Dossier conception construction de la décharge de Sousse.
Chiriac R, Carre J, Perrodin Y, Fine L, Letoffe JM (2007) Characterisation of VOCs emitted by open cells receiving
municipal solid waste. Journal of Hazardous Materials 149:249–263
Ehrig HJ (1983) Quality and quantity of sanitary landfill leachate. Waste Management and Research 1:53–68
El-Fadel M, Bou-Zeid E, Chahine W, Alayli B (2002) Temporal variation on leachate quality room pre-sorted and baled
municipal solid waste with high organic and moisture content. Waste Management 22:269-282.
Guermoud N, Ouadjnia F, Abdelmalek F, Taleb F, Addou A (2009) Municipal solid waste in Mostaganem city (Western
Algeria). Waste Management 29:896–902
Intergovernmental Panel on Climate Change (IPCC) (2007) Fourth Assessment [Link] Change 2007: Synthesis
[Link] 2. Causes of Change: 35–41.
MEDD-Ministère de l’Environnement et de Développement Durable (2005) Etat de l’environnement. Rapport national
surl’état de l’environnement en Tunisie.
Paraskaki I, Lazaridis M (2005) Quantification of landfill emissions to air: a case study of the AnoLiosia landfill site in the
greater Athens area. Waste Management and Research 23:199–208.
Sharholy M, Ahmad K, Mahmood G, Trivedi RC (2008) Municipal solid waste management in Indian cities. Waste
Management 28:459–67
Tchobanoglous G, Theisen H, Vigil SA (1993) Integrated Solid Waste management. Engineering Principles and Management
Issues; Mc- GrawHill, New York,USA.
Tizaoui C, Bouselmi L, Mansouri L, Ghrabi A (2006) Landfill leachate treatment with ozone and ozone/hydrogen peroxide
systems. Journal of Hazardous Materials 140:316-324
Zairi M., A. Aydi., H. Ben Dhia (2014). Leachate generation and biogas energy recovery in Jebel Chakir municipal solid
waste landfill, Tunisia. J Mater Cycles Waste Manag (2014) 16:141–150.

Valorisation de sable de dune pour la conception de barrières ouvragées dans

les régions arides
Enhancement of dune sand for the design of engineered barriers in arid regions

M.K. Gueddouda.a, I Goual.a, [Link] a, S. TAIBIb , N. ABOU-BEKRc , M. LAMARA

a
Laboratoire de recherche de Génie Civil, Université Amar Télidji, BP.37G. Laghouat, Algérie;
b
Laboratoire Ondes et Milieux Complexes, Université du Havre, 53 rue deProny, 76600 Le Havre, France;
c
Laboratoire Eau et Ouvrages dans Leur Environnement,Université de Tlemcen, , Algérie

RESUME- Le présent travail concerne une étude de valorisation de matériaux locaux en vue d’une utilisation
comme barrière ouvragée dans les fonds et les flancs d’installations de stockage de déchets (ISD). Les
matériaux de base étant le sable de dune de Laghouat et la bentonite de Maghnia. Une étude de formulation
basée sur le critère de perméabilité a permis d’adopter un mélange optimal constitué de 85% de sable et de 15%
de [Link] caractérisation complète du mélange adopté, les paramètres physiques et hydriques qui en
découlent montrent qu’il s’agit d’un matériau qui répond aux exigences des dispositions réglementaires pour une
utilisation dans les ISD.
Mots clés : sable de dune, bentonite, sol compacté, perméabilité, barrière ouvragée

ABSTRACT- The present work concerns a valorisation study of local materials in order to use them as an
engineered barrier in the bottom and sides of hazardous wastes repository (ISD). The basic materials are dune
sand and bentonite from Laghouat and Maghnia regions respectively. A design study based on the criterion of
permeability enable us to choose an optimal mixture composed of 85% sand and 15% [Link] complete
characterization of the mixture adopted, on the basis of physical, hydraulic and mechanical parameters, shows
that the material meets the requirements of the regulations for use in the construction of ISD.
Key words: dune sand, bentonite, compacted soil, permeability, worked barrier.

1 NTRODUCTION

Le développement des activités humaines et industrielles concourt inéluctablement à l’augmentation et la

production des déchets, et engendrant des impacts néfastes sur l’environnement. Au contact avec l’eau, ces
déchets génèrent des substances chimiques, qui entraînés par les écoulements peuvent polluer les eaux
souterraines.
L’une des solutions envisagée, pour réduire le risque de pollution des nappes souterraines, est de prévoir des
barrières d’isolation composée de mélange du sol compacté pour l’imperméabilisation le fond de site de stockage
des déchets. L’expérience a montré que le fond et les flancs de ces sites doivent être dotés d’une conductivité
hydraulique inférieure à 10-9 m/s.
L’article présente une étude de valorisation de matériaux locaux à base de sable de dune et de bentonite de
carrière pour la conception de barrières ouvragées destinées à la réalisation des fonds des centres
d’enfouissement technique (CET). Les gisements de bentonite les plus importants se trouvent dans l’ouest
Algérien. On relève en particulier la carrière de Maghnia (Tlemcen)) dont les réserves sont estimées à onze
millions de tonnes (Bental, 2002). Le sable de dune est largement disponible dans les régions du sud de
l’Algérie. La démarche scientifique intègre la caractérisation physico-chimique et hydrique de ce matériau, ce
qui a abouti à proposer une formulation répondant aux critères réglementaires. La modélisation des points
expérimentaux de la conductivité hydraulique par plusieurs modèles a été indiquée.

2 MATERIAUX ET METHODES

2.1. Matériaux

Deux matériaux à caractéristiques radicalement différentes ont été utilisés. Le premier matériau est une bentonite
originaire d’une carrière de la région de Maghnia (ouest algérien). Elle est extraite du gisement de Hammam
Boughrara (Tlemcen). Elle est non traitée et finement broyée. Ses principaux constituants minéralogiques sont:

485
la silice (65,2 %) et l’alumine (17,25 %). Les résultats des analyses chimiques sont résumés dans le tableau 1.
Avec une limite de liquidité (wL) de 141 % et un indice de plasticité (Ip) de 93 %, la bentonite de Maghnia est
classée parmi les sols très argileux très plastiques (At) selon la classification LCPC/USCS. D’après la
classification de Skempton et al. (1953) basée sur l’activité (A), la bentonite de Maghnia présente un fort
pourcentage en Montmorillonite calcique (Ca 2 +). Les résultats des essais d’identification de la bentonite de
Maghnia sont résumés dans le tableau 2.
Le second matériau est un sable de dune de la région de Laghouat ; cette région est considérée comme semi-aride
à aride, située à 400 km au sud de la capitale Alger. Le sable de dune de la région de Laghouat est un sable
siliceux. Les résultats des analyses chimiques du sable de dune sont résumés dans le tableau 1. Les résultats des
analyses granulométriques résumés dans le tableau 2, ce sable se classe comme propre mal gradué (Sm (LCPC)
= SP (USCS)).

Tableau1. Composition chimique de la Bentonite de Maghnia et du sable de dune

Bentonite de Maghnia
SiO2 Al2O3 Na2O CaO K2O MgO Fe2O3 Autre%
% % % % % % %
65,2 17,3 3 5 1,7 3,1 2,1 2,6
Sable de dune
SiO2 SO3 Cl – CaCO3 Autre M.O
% % % % % %
96 1 0,4 2,5 0,1 0

Tableau 2. Propriétés physico-mécaniques du sable de dune et de la bentonite

%< % wL Ip Ss Dmax
80m (mm) Cu CZ
2 µm (%) (%) (m2/g)

B 60 85% 141 93 462 0,2 --- ---

2%
S --- --- ---- 1,4 0,5 1,7 1,1

Normes NFP 94-56/57 NFP 94-51 NFP 94-68 NFP 94-56

Des mélanges sable de dune-bentonite ont été formulés avec différents pourcentages et pour chaque formule,
des essais d’identification ont été élaborés avec les mesures préliminaires de la perméabilité saturée qui ont été
effectuées à l’œdomètre sur des échantillons compactés à l’OPN (P 94-090).
Les mélanges sable de dune- bentonite utilisés dans cette étude sont :3% B + 97% S, 5% B + 95% S, 10% B +
90% S, 12% B + 88% S, 15% B + 85% [Link] résultats des analyses granulométriques des différents mélanges
sont représentés graphiquement dans la figure [Link] remarque que les taux des particules inférieures à 2m et à
80 m sont proportionnels au pourcentage de bentonite additionnée.L’essai Proctor Normal est réalisé sur les
différents mélanges sable-bentonite. Les résultats sont présentés dans la figure [Link] résultats montrent une
diminution du poids volumique sec maximum des mélanges sable-bentonite avec l’augmentation du pourcentage
de bentonite ajouté. Contrairement, on remarque une augmentation de la teneur en eau optimale des mélanges
sable-bentonite avec l’augmentation du pourcentage de bentonite.

486
Teneur en eau optimale wopt %
100 20 20,0
w%opt
Poids cumulés (%) 18 d 19,5

d (kN/m3)
80
16 19,0

60 18,5
14
18,0
40
100 % S
100 % B
12
17,5
97 % S + 3 % B
95 % S + 5 % B 10 17,0
20 90 % S + 10 % B
88 % S + 12 % B
85 % S + 15 % B
8 16,5
0
100 10 1 0,1 0,01 1E-3 1E-4
6 16,0
0 5 10 15
Diamétre équivalent (mm) PourcentagedeBentonite%
Figure1. Courbes granulométriques des mélanges sable de Figure 2 .Effet du pourcentage de bentonite sur les
dune- bentonite paramètres Proctor Normal des mélanges

Les valeurs obtenues des limites de consistances représentent la moyenne de trois essais. Pour la limite de
plasticité des deux mélanges 97 % +3% B et 95% S +5% B, elle est mesurée mais deux essais ont échoué vu la
difficulté d’avoir des rouleaux de 3 mm de diamètre, c’est pour cela que leurs valeurs sont négligées. D’après les
résultats obtenus (figure 3a), on constate que, les limites de consistance des différents mélanges augmentent avec
l’augmentation du pourcentage de bentonite ajoutée. Pour l’indice de plasticité, et par définition, il représente la
différence entre la limite de liquidité et la limite de plasticité, nous avons calculé cet indice pour les différents
mélanges sable - bentonite. Les résultats sont représentés graphiquement sur la figure 3b.
50 20
Limite de liquidité, limite de plasticité %

Indice de plasticité Ip

wL
40 wP
15

30
10
20

5
10

0 0
0 5 10 15 20 0 5 10 15 20
Pourcentage de bentonite % Pourcentage de bentonite %

Figure 3 . 3a) Effet du pourcentage de bentonite sur wL etwp ;3b) Effet du pourcentage de bentonite sur IP

Le potentiel de gonflement est l’un des propriétés fondamentales pour la réalisation des barrières ouvragées.
Les propriétés de gonflement permettent le remplissage des fissures existantes ou qui se développeraient a
[Link] d’un sol est caractérisé par deux paramètres : le potentiel et la pression de gonflement.
Les essais de gonflement libre dans cette étude sont effectués dans un œdomètre classique, dont les dimensions
des échantillons sont respectivement :Hauteur : H = 20 [Link]ètre : D = 50 mm.L’essai est réalisé selon la
méthode du gonflement libre. En premier lieu, on procède à la préparation de l’échantillon (mélange sable de
dune - bentonite) suivi d’un compactage statique à des teneurs en eau et densités sèches correspondant à
l’Optimum Proctor Normal pour chaque mélange. L’échantillon est placé dans une cellule cylindrique entre deux
pierres poreuses, après qu’il est lissé et arasé au niveau supérieur et inférieur. Puis, suite à l’inondation, il est
autorisé de gonfler verticalement sous la pression du piston durant plusieurs jours, jusqu’à la stabilisation. Le
gonflement total (G%) est déterminer d’après la relation suivante :

H
G (%)   100 (1)
H
ΔH=Hf –H0 ; H0 : hauteur initiale ; Hf : hauteur finale après stabilisation.

487
L’évolution du gonflement libre des différents mélanges en fonction du logarithme du temps sera représentée
graphiquement sur la figure4.D’après les courbes, plus le pourcentage de bentonite ajoutée augmente, plus le
taux de gonflement augmente. 4. Pour la mesure de la pression de gonflement, on a utilisé la méthode du
volume constant. Pour la préparation des échantillons, les mêmes procédures effectuées pour mesurer le
gonflement libre sont considérées en gardant les mêmes caractéristiques. La pression de gonflement de la
bentonite est de l’ordre de 850 kPa. On note une faible pression de gonflement pour des pourcentages de
bentonite ajouté inférieur à 5%. La pression de gonflement dépasse 100kPa pour les pourcentages de bentonite
supérieurs à 10%.

12
Gonflement libre G%
3%
10
5%
10%
8 12%
15%
6

0
0,1 1 10 100 1000 10000
Temps (min)
Figure 4. Evolution du gonflement libre en fonction du temps des mélanges S/B

2.2. Méthodes

L’estimation indirecte de la conductivité hydraulique à partir de l’essai oedométrique sur des échantillons
remaniés basée sur des résultats graphiques (courbes de compressibilité (Cv) et de consolidation (mv)) peut
donner une estimation de la valeur de la conductivité hydraulique « ksat ». Les résultats de la conductivité
hydraulique des mélanges sable de dune- bentonite en fonction des contraintes verticales appliquées sont
présentés graphiquement dans la figure [Link] valeur de la perméabilité ksat en fonction des paramètres de
compressibilité et de consolidation est donnée parl’équation 2. La conductivité hydraulique diminue avec la
contrainte verticale appliquée; Pour des contraintes verticales appliquées supérieures à 200 kPa, la conductivité
hydraulique tend vers un [Link] peut avancer que le mélange adopté est 85% de sable de dune + 15% de
bentonite, c’est une première proposition, il reste de mesurer la conductivité hydraulique saturée par des
méthodes directes et de confirmer ce pourcentage de bentonite.

k  Cv  mv  w............................................................................( 2 )
0,1 0% B+100% S
3% B+97% S
0,01 5% B+95% S
1E-3 10% B+90% S
12% B+88% S
1E-4
15% B+85% S%
1E-5
log k (m/s)

1E-6
1E-7
1E-8 -9
1E-9 k =10 m/s
1E-10
1E-11
1E-12
1E-13
10 100 1000
log v(kPa)

Figure 5 : Conductivité hydraulique des mélanges S/Ben fonction de la contrainte verticale appliquée

488
3 RESULTATS

3.1. Mesure direct du coefficient de perméabilité saturé en régime permanent du mélange sable de dune-
bentonite (s85b15)
Pour mesurer le cœfficient de perméabilité saturée direct, nous avons utilisé un Perméamétre à paroi flexible.
Le dispositif expérimental est composé d’une cellule triaxiale de révolution de type Bishop- Wesley, pilotée à
l’aide de contrôleurs pression- volume (type GDS) et permettant de tester des échantillons de 35 ou 50 mm de
diamètre et de hauteur H/D variable, dans notre cas D= 35mm et H = 70 mm. En plus, elle permette d’appliquer
une contrainte de confinement isotrope (jusqu'à 1700 kPa). L’écoulement se fait verticalement (du bas en haut
dans notre cas). Le contrôle et /ou la mesure de pression et/ou du volume d’eau est assuré par trois contrôleurs.
Dans le cadre de la mesure du coefficient de perméabilité saturée (k) du mélange en régime permanent
(charge constante), on’ a étudié les effets des contraintes de confinement (3) sur le coefficient de perméabilité.
Pour déterminer la perméabilité en régime permanent, on applique une contrainte de confinement 3, puis on
applique des pressions à l’entrée ueet à la sortie de l’échantillon usde telle manière que ue soit supérieur à us, la
contrainte de confinement est supérieure aux pressions de l’eau appliqué[Link] détermine la valeur de k à partir
de loi de Darcy (Eq. 3):

v  k .i ( 3)

On constate à partir de cette figure6 que, l’effet de la contrainte de confinement sur la perméabilité est
remarquable pour les contraintes inférieures à 400 kPa, au-delà de cette contrainte, on constate que la
perméabilité varie peu, cela est dû à la forte consolidation de l’échantillon(figure 6).En représentant la variation
de l’indice des vides en fonction du logarithme du coefficient de perméabilité, (figure 7), on constate que l’indice
des vides décroît quand la contrainte moyenne effective augmente. Cette diminution est due à la fermeture de
l’espace poreux et entraine la diminution de la perméabilité. On constate également que cette diminution est
linéaire dans les deux plans.
1E-7
permèabilité k(m/s)

k =20 - 20 log ' + 3,4 log '

2 0,9
Indice des vides e

1E-8 e= -21,26 + 19,14 Log k

1E-9

1E-10 0,6

1E-11

1E-12 0,3
100 1000 10000 1E-8 1E-9 1E-10 1E-11
contrainte moyenne effective '3 (kPa)
permèabilité k(m/s)
Figure [Link] de perméabilité en fonction de(3’) Figure 7. Indice des vides en fonction de k

3.2. Modélisation du coefficient de perméabilité saturé

Il existe dans la littérature plusieurs modèles reliant la perméabilité des sols sableux aux caractéristiques
granulométrique (Hazen, 1911 ; Loudon, 1972 ; Kenny et al., 1984). Dans le cas des sols fins, peu de modèles
existent. La perméabilité de ce type de sol est fonction de plusieurs facteurs tels que la taille des pores, la
tortuosité, la nature du sol, le gradient hydraulique appliqué et la connectivité des pores. Ce dernier aspect est
étudié en examinant l’évolution de la perméabilité en fonction de l’indice des vides et donc en fonction de la
contrainte isotrope effective de confinement.
Tavenas et al., 1983 ont constaté une relation linéaire entre la perméabilité et l’indice des vides en testant des
échantillons d’argile provenant du Canada, des Etats Unis et de Suède. Cette relation est sous la forme :

e= e0 +Ck Log (k/k0) …………….………………………………………..(4)

489
k0 est la perméabilité à l’indice des vides initial e0, k est la perméabilité à l’indice des vides e et Ck l’indice de
variation dee la perméabilité. Il vaut à peu près la moitié de l’indice des vides initial du sol (Tavenas et al., 1983 ;
Leroueil et al., 1990). Modèle de Taylor (1948) :Ce Ce modèle s’écrit sous forme d’une équation cubique en e,
applicable à tous types de sol. Il est exprimé par l’expression suivante :

e3
k C .......... .......... .......... .......... .......... ........( 5)
1  e 
Avec : C : coefficient de Taylor qui représente la valeur de la perméabilité saturée correspondant à l’indice des
vides initial.

Nous avons superposé aux points expérimentaux (régime permanent) les modèles de Tavenas et de Taylor
sur la figure 8.. L’évolution de la perméabilité semble bien décrite par ces modèles. En revanche, cette évolution
semble être bien décrite par une loi de type logarithmique. L’ordre de grandeur des valeurs du coefficient de
perméabilité est conservé.

4CONCLUSIONS

Plusieurs conclusions peuvent être tirées de cet ce article.. Les essais préliminaires basés sur le critère de
perméabilité (essais œdométriques) ont fait apparaître que la formulation composée de 85 % de sable de dune et
15 % de bentonite satisfait ce critère pour la conception des barrières
barrières ouvragées des fonds des (ISD). Les
mesures du coefficient de perméabilité saturée sur le mélange S85B15 en régimee permanent ont confirmé le choix
de la formulation adoptée. Par ailleurs, le coefficient de perméabilité saturée du matériau compacté est très peu
sensible à la contrainte de confinement dans le domaine 0,4 -1,4 MPa. La superpositiondes points expérimentaux
aux modèles de Taylor et de Tavenas et al. al On constate que ces modèles décrivent vent bien l’évolution de la
perméabilité dans ce plan. Ces évolutions
évolution semblent être bien décrites par une loi de type logarithmique. L’ordre
de grandeur des valeurs du coefficient
coefficie de perméabilité est conservé.

Figure 8 : Evolution de la perméabilité en fonction de l’indice des vides

On en conclut que la formulation à base de sable de dune et de bentonite adoptée répond parfaitement aux
exigences de la réglementation en vigueur et constitue un produit économique à base de matériaux locaux
disponibles en abondance pour la conception
onception des
de barrières des fonds des CET.

490
REFERENCES

Bental., Projet de développement des gisements d’argile bentonitique du district de Hammam Boughrara, Maghnia,
(Tlemcen), Algérie, 2002.
Chapuis R.P., Sand - Bentonite liners: predicting permeability from laboratory tests, Canadian Geotechnical Journal,
27,1990, p. 47-57.
Kenney T.C., Veen W.A., Sallow M.A., Sungaila M.A., “Hydraulic conductivity of compacted bentonite-sand mixtures”,
Canadian Geotechnical Journal, vol. 29, 1992, p. 364-374.
Parker R.J., Bateman S., Williams D., Design and management of landfills , Geotechnical management of waste and
contamination, Fell, Phillips & Gerard (eds), Balkema, Rotterdam, 1993, p. 209-252.
Skempton.A.W. The collaidal activity of clays .Proceedings of the Third International Conférence on Soil Mechanics and
Foundations [Link].1, 1953. pp 57-61.
Tavenas, F., Jean, P., Leblond, P., Leroueil, S. The permeability of natural soft clays. Part II : permeability characteristics.
Canadian Geotechnical Journal, vol. 20, n°4, 1983. p. 645-660.
Terzaghi, K. Theoretical Soil Mechanics. New York, Wiley, (1943). 510p.
Van Impe, W.F., Bouazza, A., Barrières d’étanchéité dans les sites de décharges : développement récents ,Revue Française
de Géotechnique, 74 : 1996, p. 3-18.

Geogrid reinforced structures for important infrastructure projects

Remblais renforcés en géogrilles pour la construction des projets d’infrastructures

O. Naciri1 and L. Vollmert1

1
BBG Bauberatung Geokunststoffe GmbH & Co. KG, Espelkamp-Fiestel, Germany

ABSTRACT -Efficient traffic routes form the basis for a brisk economic trade and a close economic and social
relationship of the different countries. Thus, the massive development of the national and international traffic
routes is correspondingly important. Especially in densely populated areas the routes have to be designed and
constructed in such a way that emissions remain compatible for the direct surroundings. In this case the modern
construction management requires solutions which are well-engineered and which can be carried out
economically. The visible parts of the constructions should correspondingly suit to the landscape in an ideal way.
Thus, geosynthetics are used for the static design of high dams and noise protection barriers, in case of a
foundation on extremely weak subsoil and for ground water protection measures. Large-scale projects actually
realized as well as constructions monitored since years stand representatively for the possibilities and the
importance which geosynthetics provide for the use in important infrastructural projects.
The first part of this paper gives information about the current guidelines for the design of geosynthetics
reinforced structuresin Europe. Part two of the paper provides interesting case study of infrastructure projects
which have been realized with geogrid reinforced retaining structures.

RÉSUMÉ-L’efficacité d’un réseau routier est la base d’échanges économiques rapides et de relations étroites en
matière économique et sociale entre les différents pays. Ainsi, le développement massif des infrastructures
routières nationales et internationales est par conséquent important. En particulier dans les zones fortement
peuplées, les routes doivent être conçues et construites de sorte que les nuisances restent acceptables par le
milieu environnant. Dans ce cas, la programmation de construction moderne exige des solutions qui soient bien
conçues et réalisables de façon économique. Les parties visibles d’ouvrages devraient, également, s’adapter au
paysage environnant de manière la plus naturelle possible. Pour ce faire, les géosynthétiques entrent dans la
conception statique de grands ouvrages de murs de soutènements et des ouvrages anti-bruit, dans la construction
de remblais sur des sols d’extrêmement faible portance et dans les ouvrages de protection des eaux souterraines.
De grands projets aujourd’hui achevés ainsi que des ouvrages sous monitoring depuis plusieurs années sont
représentatifs de l’étendue des possibilités qu’offrent les géosynthétiques dans les grands projets d'infrastructure.
Dans la première partie de cette communication les recommandations actuelles en Europe pour la conception des
remblais renforcés par les géosynthétiques vont être présentées. La deuxième partie montrera une étude de cas
intéressante des murs de soutènements réalisés dans un projet d’infrastructure en remblais renforcés par les
géogrilles.

1 INTRODUCTION

The reconstruction of existing and the design of new roads have to accept the geographic situation and have to
take political and social decisions into consideration. So, not only bridges and tunnels, but also dams and
cuttings are required in areas where the subsoil is not well suited and noise barriers have to be built using local
soils.
Geosynthetics as modern construction material are relatively new in terms of understanding and are still not
part of the standard education. Problems of understanding synthetics are often linked to the fact that synthetics
behave different compared to well-known materials as concrete and are not ideal elastic as e. g. steel. On the
other hand, synthetics and wood, one of the eldest construction materials ever used by civil engineers, are both
polymers and comparable in many aspects. Additionally, synthetics are already used in many applications where
concrete is not suitable and has to be protected against chemicals e. g. in pipelines.
It will be shown that geosynthetics have already become an important construction material in infrastructure
applications and allow for modern and economic constructions, saving costs by lean structures, combining local
soils, concrete and steel.
The significant reduction of the carbon footprint in many cases seems to be a future topic, but it has started
right now. Several studies have currently been worked out, comparing classical solutions and structures using

493
geosynthetics, as e. g. by Corney et al. (2009). Not only the economic effect becomes clear in this study by
requiring less energy, but also the reduction of CO2 during the whole process.
Egloffstein (2009) shows an example for an executed steep slope with an inclination of 60° for a road in a
hilly region of Germany, using geosynthetics (Figure 1). This construction had been planned with conventional
cantilever walls, but had to be redesigned using a fully greened facing due to political reasons. Nevertheless, the
budget for the construction has been reduced by factor 1.6 (Wessling & Vollmert). The carbon footprint - not
being a topic in 2004 - has been calculated for the well documented wall respectively slope by Egloffstein (2009)
as given in Figure 2. The result 6:1 stands for its own.

Figure [Link] slope in Idstein, under construction 2001 (left) and in service 2004 (right).

Cantilever Wall Geogrid-reinforced steep slope

CO2: 542 t CO2: 101 t

Production Secugrid
Production Steel-Elements
Production Contrete B25
Installation
Production Steel ST500/550
Concrete Beam
Transport Concrete / Steel
Brickwork
Installation
Transport
Soil and Transport
Soil and Installation
Transport Soil
Transport Soil

Figure [Link] footprint comparison, example Idstein; cantilever wall (left) and executed geogrid reinforced steep slope (right).

2 EUROPEAN REGULATIONS –CURRENT STATUS AND LINK TO GEOSYNTHETICS

Eurocodes for civil engineering structures, using the limit state approach with the method of partial factors, are
introduced by European countries.
Eurocode 7 on Geotechnical Design is now actively implemented throughout Europe. Each country is free to
supplement the general rules of Eurocode 7 (EC7) applications by national standards, to define methods of
calculations in his country, but these national standards should respect the principle of Eurocode 7 and have to be
reduced to additional regulations, not being in any conflict with EC7.
The design of reinforced soil is not specifically covered in Eurocode 7 and the values of the partial factors
given in EN 1997-1 have not been calibrated for reinforced fill structures. Some countries that have introduced
Eurocode 7 have recently published additional national annexes to provide design requirements for the design of
reinforced steep slopes and walls. This is the case in Germany with the annex of the National Standard DIN 1054
and the “Recommendations for Design and Analysis of Structures using Geosynthetic Reinforcements -
EBGEO”, with the NF P94-270 National Standard in France, and with the British Standard BS 8006 in the UK.

494
2.1 German Design Standard EBGEO (2010)

For Germany, the current status for the normative range of regulations is given in Figure 3.
The current German National Standard DIN 1054 will be used as supplementary rule, but being reduced to
fragments. The three parts of the rules will be combined in a normative handbook with blended text for practice
aspects. The Standard DIN 1054:2010 refers to recommendations published by the German Geotechnical Society
(DGGT).
The latest recommendation EBGEO dealing with design with geosynthetics is directly linked to DIN
1054:2010 and therefore also according to EC7; special hints are given how to use EBGEO in the EC7 concept.
This recommendation is available as English translation (EBGEO, 2011). Substantial design instructions which
meet the requirements of numerous practical applications are offered. These reflect the state-of-the-art
considering the proven scientific findings.
Eurocode 1
Actions on Structures

EC 5 EC 6 EC 7-1: EC 8 EC 9
DIN EN 19971-1
....
Geotechnical Design
-General Rules

National Annex:

DIN EN 1997-1/NA

DIN 1054:2010
Supplementary EAU
Rules to
DIN EN 1997-1 EAB

DIN 4084:2009-1
Soil – Calculation of EA-Pfähle
Embankment Failure […]
EBGEO
DIN XXXX
Merkblätter
DIN YYYY

Figure [Link] range designing with geosynthetics (according to Kempfert, 2011).

2.2. French Design Standards NF P94270 (2009) and XP G380-64 (2010)

In France, the standard for the design of reinforced soil structures is divided into two standards depending on the
inclination of the front facing of the structure (Figure 4), according to the European standard EN 14475 (standard
execution of reinforced soils structures). In July 2009, AFNOR published the standard NF P94-270 which
constitutes the national implementation of Eurocode 7 for geotechnical design of reinforced retaining walls and
soil nailing (with an angle of the front face of ≥ 76°) and in 2010 the standard XP G380-64 for steep slopes and
inclined walls (with an angle of the front face of < 76°). The latter concerns the design of reinforcements with
only geosynthetics.
Both standards are completely consistent with the principles of limit states with partial factors as defined in
the NF EN 1990 Standard.
b = 90° 90° > b ≥ 76° 76°>b ≥ 45° b < 45°

NF P94-270 XP G38-064
(Geotechnical design - Retaining (Use of geosynthetics in reinforced
structures - Reinforced and structures - Inclined walls and
soil nailing structures tan 1≤ 1/4) steep slopes- Justification of dimensioning
and design elements tan 1>1/4)
Figure [Link] for reinforced structures according to the inclination of the facing.

495
2.3 UK Design Standard BS8006 (2010)

In Great Britain, the comparable recommendation is called BS 8006 “Code of practice for
strengthened/reinforced soils and other fills” published by the British Standards Institution (BSI). BS8006 was
initially published in 1995 with an amendment issued in 1999. The Standard uses a limit state format and
detailed reinforced soil design methods for retaining walls, steep slopes and weak foundations.
So, actually three finalized guidelines dealing with the use of geosynthetics are available. Recently, the
European Technical Committee TC250/SC7 for EC 7 has founded an evolution group EG5 for reinforced soil to
compare and to assess the recent European practice for the design of reinforced soil structures, with the objective
to try to determine, if the design for reinforced soil structures could be brought in accordance with EC7.

3. CASE STUDY FOR INFRASTRUCTURE PROJECT: MEDITERRANEAN COAST ROAD WITH

GEOGRID REINFORCED RETAINING STRUCTURES

3.1. Description of the project

Infrastructure improvement and the promotion of regional development are important issues throughout the
world.
In Morocco this was the case with the traditional “Rocade Méditerranéenne" road. The Rocade is situated in
the northern part of Morocco. It is a traditional coast road which runs along the Mediterranean from the port city
of Tangier in northern Morocco to Saidia in the east of Morocco on the border with Algeria. The aim of this
project was the construction of a Mediterranean Highway in mountainous areas.
This important coast road winds for 510 km along the Mediterranean to the eastern border of the country.
Since the complete stretch was in a very poor overall condition, the Ministry of Transport decided to undertake
extensive repairs which included widening and partial re-routing of the road. The principal objectives were to
optimize the alignment and improve the quality of the pavement in order to shorten journey times along the
Mediterranean coast and to facilitate the development of the region. To increase safety and raise average speed
on the road, extensive slope-stabilization measures were called for.
The project is divided into several sections. This case study describes the first phase of the construction, i. e.
the 46 km connection between Tetouan to Oued Lau.
Northern Morocco is mountainous and rugged. Slope failures, undercutting by the sea and landslips are
characteristics of the area, and these contribute to making road construction all the more difficult. The re-
alignment was planned in an extremely efficient manner. While the predominantly rugged landscape captivates
tourists with its fascinating views, it represents a major challenge for road designers and builders. Complicated,
expensive structures were rejected in favor of economical, safe alternatives with geogrid-reinforced soil
embankments. These structures permitted optimum road alignment and resulted in savings in construction time.
The design engineers selected a WRAP solution using gabions. NAUE Secugrid® PET geogrids were used in
combination with local soil in some sections of the project, yielding a long-term, safe and economical
reinforcement solution. Geogrids Secugrid® with tensile strengths between 40 and 200 kN/m were used. The
installed geogrid is a strip geogrid for soil reinforcement; it is made from pre-stressed, monolithic flat bars and
possesses a high connection strength. A total of over 300,000 m² of geogrids were installed in the section
described.

Figure [Link] on one constructed structure.

496
The use of locally available soils in the structures made for a solution which was both economical and
ecologically sound. Today, the completed route is a time-saving, high-quality coast road that allows more
intensive access by tourists to these very scenic areas.
The construction of the geogrid reinforced structures reduces travel times on the “La Rocade Mediterranean”
in northern Morocco. The local people appreciate the ease of travel on the new road, which has seen an increase
in traffic. The Rocade Mediterranean road is now wider and safer.
3.2 Geometry

The existing terrain at the rough coast line was planned to be in cut. However, to ensure a balance between cut
and fill, several embankments up to 30 m high having an inclination of 65° were also necessary. A general
overview of the geometry can be found in the following Figure 6.

Figure [Link] Cross sections.

3.3. Design method

After consulting with the contractor it was decided to conduct the design according to EC 7 plus the German
recommendation for the design and construction of geosynthetic reinforced structures (EBGEO 2010) and the
German national annex DIN 1054 as well as the French standards for the design of steep slopes and walls with
geosynthetics. The region is an earthquake area. Thus, appropriate seismic acceleration coefficients had to be
taken into consideration in the design. Here, BBG Bauberatung Geokunststoffe (Geosynthetics consulting
specialist based in Germany) was charged with the design of a total of 11 sections of geogrid-reinforced slopes
for the project Rocade Mediterranean, with an overall length of approximately 1.4 km. The final plans called for
road embankments with slope angles of around 65° and heights of up to 30 m.
3.4. Gabion facing system

Gabions are manufactured from steel wires woven to form baskets. The baskets are filled with stones which can
as well be taken from quarries close by. This gabion facing is a perfect combination of a natural and man-made
element.

Figure [Link] facing system of the construction.

497
Usually the gabions are placed at the front / facing on top or rather between the geogrid reinforcement. Due
to the boundary conditions and sufficient stability of the facing system the intersection with the geogrid layer
was not required. In this case the geogrids were installed horizontally up to the back of each gabion and then
wrapped around over a soil layer at a spacing of 40 cm. A tie back length of 1.50 m was installed in each case.
Within the wrap-around a Secutex® filtration and separation geotextile was installed to prevent any loss of fill
material due to leakage water.
The slope facing utilized a wrap-around system (WRAP) and gabions. Crucial in the choice of this type of
system was the fact that it allowed the retaining structures to be traversed by culverts, which were required to
allow passage of surface runoff from the mountains uphill of the structures. The collector system on the
upstream face, with numerous pipes penetrating the structure, and the cascades on the downstream face
necessitated an extremely robust, erosion-resistant outer facing system combined with a stilling pool at the foot
of the steep slopes.
In order to maintain the undisturbed major water flow from the mountains during the infrequent but severe
rainstorms, large hydraulic pipe penetrations were incorporated in the middle of the structures.
3.5. Special case (pipeline penetration)

As the new road embankment was constructed on the existing natural slope, there was a problem with regard to
surface water. This surface water increases significantly with heavy rainfalls. To reduce the risk of hydraulic
failure in the embankment and to reduce the risk of heavy erosion, BBG carried out recommendations
concerning a water collection system between the natural slope and the embankment. It was planned to collect
the water in front of the embankment and to drain it off slowly through the embankment. Therefore large
concrete culverts were installed. These penetrate across the embankments (Figure 8).
In the case of heavy rainfall the amount of water draining through the embankment would naturally increase
the facing system requirements to be very robust to mitigate against the suction effect of the resulting cascade at
the end of the culvert.
Further measures for the facing as well as for the foundation needed to be planned. Finally, a double walled
system consisting of two gabion baskets in front of the facing as well as a gabion layer on the bottom of the
slope was installed to reduce the risk of hydraulic failure.

Figure [Link] culvert.

4. CONCLUSION

Extending and reconstructing the infrastructure within the European countries as well as the international routes,
structures for noise barriers and embankment foundations are required, fulfilling the economic requirements.
Geosynthetics, namely geogrids, allow to use local soils and can be combined with steel elements as well as
concrete and wooden elements. The major project Mediterranean coast road demonstrates once again how the
use of geosynthetics enables the construction of sophisticated structures even under difficult local conditions.

498
REFERENCES

AFNOR (2009) Calcul géotechnique - Ouvrage de soutènement - Remblais renforcés et massifs en sol cloué, norme NF P 94
270,
Juillet 2009, 186 pages.
AFNOR (2010) Murs inclinés et talus raidis en sols renforcés par nappes géosynthétiques – Justification du dimensionnement
et éléments de [Link] G 38-064, Août2010, 69 pages.
British Standards Institution, (2010) Code of practice for strengthened/reinforced soils and other [Link] 8006-1, 2010.
London, 260 pages
Deutsche Norm (DIN) (2008) Eurocode 7 Geotechnical design – Part 1: General rules; German version EN 1997-1:2004, 175
pages.
Deutsche Norm (DIN)(2007) Eurocode 7 Geotechnical design – Part 2: Ground investigation and testing. EN 1997-2:2007
Deutsche Norm (DIN) (2010) Ground - Verification of the safety of earthworks and foundations. 1054-2010-01, 123 pages.
Egloffstein, T. (2009) Bauverfahren mit mineralischen Baustoffen und Bindemitteln im ökologischen Vergleich mit dem
Einsatz von Geokunststoffen.6. Geokunststoff-Kolloquium, NAUE GmbH & Co. KG, Bad Wildungen, January 2009.
German Society for Geotechnics (DGGT) (2011) Recommendations for Design and Analysis of Structures using
Geosynthetic Reinforcements - EBGEO, 2011. Ernst & Sohn, Berlin, 313 pages
Kempfert, H.-G. (2011) Das Normenhandbuch. RuhrGeo Tag: Eurocode 7, Einführung in Deutschland, TU Dortmund, Heft
30, 2011.
Naciri, O., Psiorz, C., Salek, M., & Bretin, P. (2014) Mediterranean coast road with geogrid-reinforced retaining
structures.10th International Conference on Geosynthetics (ICG), Berlin.
Wessling, A. & Vollmert, L. (2004) Geogitter bewehrte Steilböschung spart Kosten. Straßen & Tiefbau, Heft 11/2004.

Case history report: Geotextile encased columns (GEC’s) - Highway A2 in

Poland
Retour d´expérience: Colonnes encastrées géotextiles (GECs) - Autoroute A2 en Pologne

Laurent Sakou1, Johann Bruhier2, ARAB Rabah3, Peter Assinder4, Janusz Sobolewski5
1
Huesker Synthetic GmbH, Germany; Email: sakou@[Link]
2
Huesker SAS, France; Email: [Link]@[Link]
3
Huesker SAS, France; Email: [Link]@[Link]
4
Huesker Synthetic GmbH, Germany; Email: assinder@[Link]
5
Huesker Synthetic GmbH, Germany; Email: sobolewski@[Link]

ABSTRACT -Since 1995, the Geotextile-Encased Columns (GECs) founding system has been used and
continually developed to found earth structures on low-bearing soils. The first professional application of GEC’s
in Poland occurred in 2010-2011 during the construction of the Highway A2 (between km 60+225 and km
60+450). In this section the A2 crosses an area of very low-bearing organic soils with the thickness up to 28m. In
addition to the difficulties expected when introducing a system into a new market, this project included boundary
conditions for production which redefined the limits of the GEC system’s use. Illustrative of this is a column
length of around 30m, which had never been achieved before. This meant that the way the system functions,
theoretically described in the conceptual design, had to be verified through extensive measurements. A detailed
design and comprehensive monitoring of all construction stages ensured all requirements defined by Client
concerning the allowable post-construction settlements (≤ 10cm) and the allowable long term settlement
difference (≤ 15mm/10m) during 30 years under operation, could be achieved.
The following paper outlines the experience gained through the realisation of a demanding road project on
organic soils with extremely low shear strength.

Keywords: Geotextile-Encased Columns, embankment on soft soil, monitoring, innovation, economic

foundation system

1. INRODUCTION

Since 1995, the Geotextile-Encased Column (GEC) founding system has been used and continually developed to
found earth structures on low-bearing soils. The system’s first professional use in Poland was for a motorway
project in 2010. In addition to the difficulties expected when introducing a system into a new market, this project
included boundary conditions for installation which redefined the limits of the GEC system’s use. Illustrative of
this is a column length of around 30m, which had never before been achieved. This meant that the way the
system functions, theoretically described in the conceptual design, had to be verified through extensive
measurements. The following article outlines the experience gained through the realisation of a demanding road
project on organic soils of extremely low shear strength.

[Link] OVERVIEW

The new four-lane Motorway A2 between Frankfurt/Oder and Poznan crosses several regions with low-bearing
organic soils. Due to the difficult geological boundary conditions, various ground improvement or founding
methods were examined for the 4m-high road embankment on peat area 15.2, between km60+220 and
km60+450: Vibro stone columns; Stabilisation columns; driven precast concrete piles; vertical drainage with
preload, and GEC. Finally a foundation on geotextile-encased columns was chosen as the most technically and
economically advantageous solution.
Key points of GEC’s foundation:
 number of columns: 3400
 diameter of columns: 780mm and 800mm
 soil exchange ratio = 15%
 axial spacing = 1.97m
 casing: Ringtrac® 100/300 (circular-woven PET material)
 filling: closely-graded sand with silt content <5% by weight
 horizontal reinforcement: Stabilenka® 800/100 (PET material)

501
 maximum column length = 30m
 embedment in natural foundation: min. depth = 0.20m
 installation of columns: displacement method

2.1. Bearing system, design model and predicted values

GEC’S system has been successfully used since 1995 in: Germany, The Netherlands, Sweden, Italy, Russia and
Brazil, (Brokemper et al. 2006, Alexiew et. al. 2010). In the case of the described section of the A2 in Poland
some new developments were needed due to extremely long columns, installed in soft soil deposits down to 28m
under the level of a working platform, see Figure 3.1. Some properties of soft soil strata and sand used in the
design for the embankment and columns are shown in the Table 3.1. The design of columns was based on the
rules given in EBGEO 2010, (DGGT 2010). The stability and settlement analyses of columns were performed
using the self-written program SAND38BETA. This program enables analysis of the stability of columns in the
ultimate limit state and settlements in the serviceability limit state according to EC7- EN 1997-1.
For the A2 the following out-put data was predicted using the software:
 Required design value of long term tensile strength of the encasement: Fr,d = 122 kN/m
 settlement of columns heads after 210 days of preloading: 2.33 m
The stability of the embankment during preloading and in the post-construction stage was analyzed using GGU-
Stability, see Figure 3.2. After translation from the 3D real model into a 2D analytical model, Bishop’s method
was used for the estimation of the value of utilization factor (Ed/Rd). Directly, after placement of the second stage
of embankment (see Figure 4.1) the calculated capacity utilization factor was exactly equal to 1.00. In the post-
construction time due to consolidation and self-improvement of soft soil the estimated value of utilization factor
was reduced to 0.76, see Figure 3.2. Based on the results of stability analyses Ringtrac® 100/300 (PET
encasement) and Stabilenka® 800/100 (PET fabric for basal reinforcement) were chosen.

Figure 2.1: Geological longitudinal profileFigure 2.2: Properties of soft soils, sand used for
columns and embankment

Figure 2.3: Stability analysis, post-construction stage, Bishop’s method

502
3. CONSTRUCTION WORKS

The preparation of the working platform, installation of columns and basal reinforcement & installation of
monitoring devices was undertaken from: April 2010 to October 2010, see Figure. 4.1.

Figure 3.1: Designed and realized schedule of construction works

The working platform with the thickness of 1.0m was placed on a non-woven and was only able to support light
equipment. The heavy equipment used for the installation of columns (Liebherr LRB 255 and Liebherr LRB 155,
weight ca. 1,100 kN) moved on a wooden platforms (6,0m x 6,0m), which were always replaced and positioned
over already intalled columns, which meant that new columns were installed ahead the working front. In this
way the vertical stresses equal to 30,5 kN/m² could be safetly distributed over the newly installed columns. Two
heavy machines were used for pilling works, see Figure [Link] the installation of columns a displacement
method was used, which involves insertion of a closed bottom steel pipe, vibrated down into firm stratum
through soft soil layers. The closed flaps form a cone at the bottom during driving of the pipes. After the
placement of the geotextile encasement into the pipe, sand will be filled up to the top of the pipe and the
geotexile loosened. The flaps will be opened by the initial withdrawal of the steel casing. The sand in column
will be continuosly compacted whilst withdrawal under vibration and after removal of the steel pipe a column
with a given diameter and length is created.
To install the longest GEC, the process had to be adapted as follows:
 displacement pipe with base flaps exchanged for open pipe with lost base plate
 superimposed hydraulic vibrator replaced by ring vibrator as shown in Figure 4.3
 use of water/bentonite slurry to reduce skin friction in pipe during withdrawal process

Figure 3.2: Liebherr rams during installation of GEC´s Figure 3.3: Moveable ring vibrator used for the longest pipes

503
After installation of the columns a sand leveling layer with a thickness of 30cm was placed and basal
reinforcement (Stabilenka® 800/100) installed. In this period many monitoring devices were assembled or
attached to geotextiles. After the installation of all monitoring devices the first base measurements readings
were taken. The main earth works (embankment with preload) were started on 29.10.2010. The embankment was
placed in two stages: the 1st stage: sand layer with thickness 3m , the 2nd stage: sand layer with thickness 2.4m,
up to the level +76.0m ASL. A 35 day break for the partial dissipation of excess pore water pressure between
both steps was needed. The difference between the planned level of preload of 77.0m ASL and the executed on
76.0m ASL was as a result of construction settlements during the earth works, see Figure 4.1. This settlement
was neglected in the theoretical analysis performed in the design (assumption on the safe side).

4. MONITORING
The project was accompanied by a comprehensive geotechnical measuring program (see Figure 5.1) to verify the
assumptions made in static calculations and the achieved results of the final design. The predicted development
of the consolidation process, settlements and tensile forces in the column encasement were of primary
importance. Finally, the results of monitoring were decisive for the time point of removal of the preload. The
measurements are scheduled to continue for five years after the start of highway operation.
The monitoring system consists of:
 29 bench marks with telescopic connected pipes,
 6 hydrostatic lines for settlement measurements (3 in the longitudinal direction and 3 in the crosswise
direction),
 4 sensors for measurements of vertical stresses on column heads and between columns,
 4 columns with installed ring sensors for measurements of column perimeter,
 10 vertical inclinometer pipesinstalled at the embankment toe for the observation of the spreading behavior
of the embankment base and deeper subsoil.

Figure 4.1: Plan of measuring configuration

4.1. Settlement measurements

Line-shaped settlement measurements were conducted at a total of six gauging sections using the Lhotzky
Hydrostatic Settlement Lines, see Figure 5.1. This innovative method is based on the intermittent measurements
of a hydrostatic pressure difference between the given point and the reference level. The locations of points in
the flexible tube are known, due to the use of step pumping system. In this way the longitudinal position of fluid
head will be known and the settlement profile can be determined. The total length of the six installed flexible
tubes (lines) was 882 m. The number of observed points was equal to 3,528 (it means, that the points were
located in the linear spacing of ca. 0.25 m).
Additionally, the settlement process was controlled by geodetic measurements on 29 bench marks located on the
top of a leveling layer placed above the columns heads.
4.2. Horizontal deformation

Ten Vertical inclinometers (V01 to V10) were used to measure horizontal deformationof the embankment at the
northern and southern slope toes (see Figure 5.1). Up to 30m long, theinclinometers were installed in the natural
subsoil between the GEC. Readings weretaken together with those of the hydrostatic settlement lines.

504
4.3. Column circumference and distribution of stress

In two measuring fields as shown in Figure 5.1, vertical stress (DD1 – DD6) and the circumference of the GEC
at seven levels (GEC-D1, GEC-D2) were measured continuously. Consulting engineers Lhotzky + Partner
developed a system to measure column circumference with the aid of a draw-wire displacement sensor, which is
coupled with a hydrostatic measuring tube to determine the elevation and inclination of the measuring planes.
The six earth pressure cells were installed in pairs, one on the head of a column and one in the surrounding area
to measure the differential load transfer.

5. MONITORING RESULTS

5.1. Settlement measurement results

The results of 7 measurements sets performed in MQ 4 (longitudinal profile along the edge of embankment) are
presented in Figure 6.1. The horizontal marked line presents the start configuration of monitoring dated October
29 in 2010. The additional seven measurements sets present the deformed base of embankment along the
installed hydrostatic lines.
In the building permit application, the maximum amounts of primary settlement forecastwere 1.28m for the
first fill stage (4m-high embankment) and 2.36m forthe second fill stage (7m-high embankment including
preload).Beneath the load of the first fill stage, a maximum of 0.62m was measured. The maximal value of
settlement measured on June 28 in 2011(by thetime the preload was removed) was equal to 1.05m. The predicted
design value of settlement was clearly higher (s = 2.36m). It can be explained by neglected improvement of soft
soil in the analytical model. Furthermore the monitoring did not consider settlements caused by the self weight
of the working platform and the leveling layer and settlements which occurred under the moving weight of the
column installation equipment. This large difference between predicted and observed settlements shows a large
reserve in the use of the analytical model. It means that further improvements in assumptions can be undertaken.

5.2. Horizontal deformation results

In the building permit application, a value of 0.40m was defined as a threshold for horizontal deformation,
dependent on the expected extension behaviour of the horizontalgeotextile above the column heads. 0.55m was
defined as the limit value. Figure 6.2 shows the progress of the greatest horizontal deformation of the in-situsoft
soil measured at the slope toe. According to this data, the maximum value, 0.28min the stratum boundary area
between the peat and the gyttja, is only 70% of thethreshold value.

Figure 5.1: Settlement under southern embankment shoulder Figure 5.2: Maximum horizontal deformation of slope
(Line measurement MQ4) toe perpendicular to road axis longestpipes
5.3. Circumferential tensile forces in geotextile casing and distribution of vertical stress

505
The diameter of the geotextile encasement was equal to the inner diameter of steel pipes used in the
displacement method for the column installation. This means, that at the moment the pipe is lifted, the tensile
force in the casing should be theoretically equal to zero (no tension in geotextile casing). After pipe withdrawal,
the tensile force in casing depends on the changes in the perimeter of columns, which can be caused by
horizontal stresses in soft soil and later by vertical load on column heads and soft soil in between. As shown in
Figure 6.3, immediately after the pipe is lifted, the diameter of the geotextile casing was larger than the diameter
®
of Ringtrac at three levels. This indicates that the diameter of the observed GEC-D1 generally widened. Only at
the level -18m below the column head, the circumference of the column was lower than the circumference of
Ringtrac®. This could be caused by a stiffer interlayer located at this depth, which mobilized higher radial earth
pressure after the pipe withdrawal. The maximum tensile force observed in the device ME-2, located at the depth
of 8.0m (near the bottom of the peat layer). Directly after the pipe was lifted, the elongation of the geotextile
casing was equal to 1.4%(diameter of casing = 0.791m). At the final stage of preloading, the elongation was
3.97%(diameter of casing = 0.811m). Based on the isochrones of Ringtrac®, (made from polyester), a tensile
force equal to 75 kN/m was estimated, taking into account the real loading period of approximately 8 months.
The predicted in service tensile force for the final construction stage was calculated at 122 kN/m. It can be said,
that the design overestimated the tensile force by around 33%.
As expected, the load transfer from embankment weight and traffic (or preload) was strongly affected by the
very high difference in the stiffness between columns and surrounding soft soils. At the final stage of preloading,
the vertical stress on the column head was 320 kN/m². The predicted design value for this stage was estimated at
794 kN/m². Correspondingly the observed vertical stresses on top of the soft soil were observed in the range 100-
120 kN/m² (see Figure 6.4).The large difference between measured and predicted valuescould be due to the
growing stiffness of soft soil layers during consolidation of the soft soil. This selfimprovement of soft soil layers
during the preloading was not taken into account inthe design. The oedometric modulus of the soft soil was
assumed to be constant duringthe entire construction period.
In the final stage, a small growth of stresses on the column heads and a slow reduction of stresses acting on the
soft soil were noticed.

Figure 5.3: Changes of column diameter during construction Figure 5.4:Development of the column diameter during
time the earth works, column GEC-D1

Three interlinked measurements, namely: settlements, tensile forces in geotextile and vertical stresses on column
heads were clearly lower than predicted in the design. This large difference highlights a large reserve in the
analytical model used. It means that further improvements in assumptions of the model are possible.

506
6. CONCLUSIONS

One section of the new road embankment constructed in 2010-2011 for the A2 Highway in Poland is located in
an area with very deep organic subsoil, which reaches the depth of 28m. After consideration of many design
options the GEC system was used due to its ability for a controlled load transfer into deep subsoil. As a decisive
criterion for the proof of serviceability, the client specified the maximal allowable post-construction settlement
difference between two points spaced 10 m apart over a period of 30 years ≤ 15 mm. The total post-construction
settlement should be ≤ 10 cm over the period of 30 years. Based on a detailed design and an extensive
monitoring it was possible to give the proof, that the designed and executed GEC-system for this embankment
fulfills these requirements.
In conclusion, the founding of the embankment on GEC’s under extremely difficultground conditions can be
described as a complete success, in terms of both executionand function.

REFERENCES

Raithel. M. 1999. Zum Trag- und Verformungsverhalten von geokunststoffummantelten Sandsäulen, Schriftenreihe
Geotechnik. Heft 6. Universität Kassel
Brokemper. D. et al. 2006. Design and construction of geotextile encased columns supporting geogrid reinforced landscape
embankments. Proceedings of the 8th International Conference on Geosynthetics, Yokohama. pp. 889-892
Raithel. M. 2008. Calculation techniques and dimensioning of encased columns. Proceedings of the 4th Asian Regional
Conference on Geosynthetics, Shanghai. pp. 718-723
Alexiew. D. et al. 2010. Foundation of a coal coke stockyard on soft soil with geotextile encased columns and horizontal
reinforcement. Proceedings of the 9 th International Conference on Geosynthetics, Brazil, pp. 1095-1099
DGGT. 2010. Empfehlungen für den Entwurf und die Berechnung von Erdkörpern mit Bewehrungen aus Geokunststoffen-
EBGEO. Ernst & Sohn, Berlin
Lhotzky, K.; Friedl, G.; Lhotzky, S. (2008) Hydrostatische Offshore – Setzungsmessungenbeim Bau des JadeWeserPorts,
Interdisziplinäre Messaufgabenim Bauwesen, Weimar 2008

Expérience rwandaise dans les ouvrages de soutènement en sol renforcé

Rwandese experience in retaining structures with reinforced soil

RESUMÉ - Les géosynthétiques (étymologiquement, matériaux synthétiques utilisé en contact avec les sols)
sont des produits apparus dans les années 50. Ils sont utilisés de nos jours dans tous les domaines du Génie Civil.
Rares sont les chantiers/projets où leurs propriétés de séparation, de filtration, de drainage, de confinement, de
contrôle érosif ou de renforcement ne sont pas mises à profit pour régler les problèmes de venue déau, de
contamination des sols et nappes phréatiques, de colmatage ou de renforcement. Malgréces applications
nombreuses, courantes et prouvées sur les plans techniques, économiques et environnementales, les
géosynthétiques restent des matériaux peu utilisés par les maîtres d´œuvre, les entrepreneurs, les bureaux
d´études, les instituts et écoles díngénieries en Afrique et plus particulièrement en Afrique Subsaharienne. Des
tentatives déxplication peuvent être liées au mot “synthétique”qui porte généralement auprès du public une
connotation négative qui a été donnée au mot “plastique”, et aussi la non-implémentation jusqu’ici de lápport
des géosynthétiques dans les programmes de formation des ingénieurs en Afrique en général. Les
géosynthétiques étant la plupart du temps incorporés à un sol pour former une espèce de « géostructure»
nécessitent une connaissance appropriée de leurs modes dàctions, de leurs dimensionnements (ULS, SLS) qui
sont intimement liés non seulement aux propriétés géologiques et géotechniques des sols en place et des
matériaux de remblayage utilisés, mais aussi aux propriétés du géosynthétique choisi.
Le présent article présente les différentes fonctions que peuvent assurés les géosynthétiques dans le Génie Civil.
La fonction de renforcement dans les ouvrages de soutènement est illustrée. Un cas pratique dóuvrage de
soutènement en sol renforcé au Rwanda dans le cadre de la réhabilitation de láutoroute joignant Ruhengeri à
Gisenyi est illustré. Le challenge et les objectifs, la conception et le dimensionnement, les dispositions
constructives prises pour les ouvrages de soutènement en sol renforcédans le projet sont présentés.
Mots clés: Mur de soutènement, sol renforcé, renforcement, dimensionnement

ABSTRACT - Geosynthetics (etymologically, synthetic materials used in contact with soils) are products which
started to emerge in the [Link] they are used in many areas of civil engineering and it is rare to see
sites/projects where their separation, filtration, drainage, containment, erosion control or reinforcement
properties are not used to overcome various problems including water flow, contamination of soil and
groundwater, clogging and reinforcement. Regardless of the current and proven geosynthetic applications in
technical, economic and environmental plans, geosynthetics remain little used by main contractors, sub-
contractors, consultants, engineering institutes and schools in Africa and particularly in Sub-Saharan Africa.
Attempts at an explanation to why their usage is not so prevalent in Africa may be linked to the word “synthetic”
which usually carries to the public a negative connotation that has been given to the word “plastic”, and also the
non-implementation to date of the geosyntheticscontributionin the engineering education programs in Africa in
general. Geosynthetics being usually incorporated into the soil to form a kind of “geostructure” require
appropriate knowledge of their modes of actions, their designs analysis (ULS, SLS), which are closely related
not only to the geological and geotechnical properties of the soils in place and the backfill materials used, but
also to the properties of the selected geosynthetic.
This article introduces the different functions of geosynthetics in civil engineering. The function of
reinforcement in reinforced soil retaining structures is illustrated.A practical case of a reinforced soilretaining
structure in Rwanda as part of the rehabilitation of the highway joining Ruhengeri to Gisenyi is illustrated. The
challenges, objectives, conception and design methodologies, together with the construction details for the
reinforced soil retaining structure for the project is presented.

Keywords: retaining wall, reinforced soil, reinforcement, design

509
1. INRODUCTION: FONCTION DES GEOSYNTHETIQUES

Les géosynthétiques comprennent une variété de matériaux en polymères synthétiques (Polyester PET,
Polyvinylalcohol PVA, Polyamide PA, Polyéthylène PE, Polypropylène PP) spécialement manufacturés pour
être utilisés dans les différents domaines du génie civil (terrassements et fondations, génie environnementale,
génie hydraulique, etc…). Ces géosynthétiques sont appelés, en fonction de leurs caractéristiques mécaniques,
leurs structures et/ou de leurs modes d’assemblage, à remplir diverses fonctions telles que séparer, filtrer et/ou
drainer, confiner, protéger, renforcer. Dans la plupart du temps, ces fonctions peuvent être combinées en
fonction du type et des exigences de l’ouvrage. Dans le cadre des ouvrages de soutènement en sol renforcé par
les géosynthétiques, la fonction primaire requise à assurer par les géosynthétiques est la fonction de
renforcement. Les autres pouvant être classées comme secondaires.

[Link] de renforcement dans les ouvrages de soutènement

Contrairement aux sols, les géosynthétiques/géogrilles disposent d’une résistance en traction élevée. Leurs
propriétés de contact et de frottement avec les sols (interaction sol/géosynthétique) permettent à ces matériaux
d’assurer à la fois une bonne transmission et une répartition optimale des contraintes/charges avec le milieu
environnant. L’insertion des nappes géosynthétiques dans un remblai permet de renforcer globalement l´ouvrage
(mur de soutènement, culée de ponts, viaduc, etc.). Les géosynthétiques de renforcement permettent dès lors la
construction des remblais à grande hauteur avec des parements presque verticaux. La figure 1 pressente une
coupe transversale typique d´un ouvrage de soutènement en sol renforcé par des géogrilles.

Figure 1. Coupe transversale d’un ouvrage de soutènement en sol renforcé par des géogrilles

2. REMBLAI RENFORCÉ SUR LA ROUTE RUHENGERI-GISENRYI, RWANDA

2.1 Situation générale / Challenge

La route Ruhengeri à Gisenyi dans le nord-ouest du Rwanda est la principale voie de transit des marchandises
vers la République démocratique du Congo (RDC). Cette route traverse une région montagneuse. Il est important
de mentionner que la route originale a été construite il y´a environ 30-40 ans. L´augmentation du volume du
trafic des poids lourds et la détérioration générale de l´état de la route existante ont entraîné des ruptures et des
déplacements en bordure de chaussée sur une section de la voie principale. Par ailleurs, certaines parties de cette
section de la chaussée présentent une largeur de chaussée insuffisante et un manque de structure de sécurité
appropriée.

2.2 Objectifs et Solution retenue

Le financement a été obtenu de la commission européenne avec comme principal objectif l´amélioration de la
qualité de la voie principale tout en respectant sa stabilité et sa sécuritéà long terme. La réhabilitation de cette

510
section implique entre autre l'élargissement de la chaussée existante d'environ 1 m, la création supplémentaire
dúne « épaule » dénviron 2 m et la construction dúne barrière de sécurité appropriée. Ces objectifs ont
conduit à la comparaison de différentes solutions au bout desquelles la solution de remblai de sol renforcé par
des géogrilles avec un parement raidi végétalisable a été retenue.

Cette solution de remblai de sol renforcé par des géogrilles a été retenue pour les raisons suivantes:

 La possibilité de réemploi des matériaux du site,

 la souplesse et la rapidité de la mise en œuvre, la main d´œuvre locale
 l’adaptation du système avec les phases de construction,
 le respect du temps de construction,
 le faible coût de cette solution par rapport aux solutions traditionnelles,
 le respect des mitoyens et de l’environnement.
En bref, cette étude comparative a montré que la solution de remblai de sol renforcé par des géogrilles était plus
appropriée, plus économique, plus écologique et durable dans un contexte de développement durable. Les
travaux de réhabilitation de la voie Ruhengeri-Gisenyi ont débuté en juillet 2007 et ont été achevé en mars 2010.
La section incluant le remblai de sol renforcé a été construite en 2009.

3. CONCEPTION ET DIMENSIONEMENT

La conception et le dimensionnement ont été réalisés conformément aux normes et guides européens. Les calculs
géotechniques ont été effectués conformément à la norme allemande DIN 1054(Vérification de la sécurité des
travaux de terrassement et des fondations), le guide allemand EBGEO (Recommendations for Design and
Analysis of Earth Structures using Geosynthetic Reinforcements), et l´Eurocode EC7 à travers les facteurs de
sécurité partiels. L’analyse de l´ouvrage consiste à vérifier d’une part sa stabilité interne, externe et mixte (États
limites ultimes ELU), et d’autre part les déformations / distorsions, déplacements et tassements du remblai
renforcé en service normal (États limites de services ELS). Les calculs ont été réalisés avec le logiciel de
conception GGU Stability(Slope stability analysis and analysis of soil nailing and reinforced earth walls to DIN
4084 and EC 7).

3.1 Eléments généraux de conception

Les données du projet fixées contractuellement par le maître d’œuvre recouvrent le site d'implantation et la
catégorie de l’ouvrage (ouvrage permanent). Les autres données nécessaires à la justification de l'ouvrage de
soutènement étaient les suivantes:
 Données géométriques:
Parement incliné de 20 degrés par rapport à la verticale
Hauteur maximal du remblai de sol renforcé H= 8 m.
 Données géotechniques:
Le sol de fondation et le matériau du remblai présentent les valeurs caractéristiques (angle de frottement,k,
cohésion c,k, poids volumique ɣ,k) suivantes :

 Surcharge et/ou charges concentrées:

Une charge uniformément répartie de 20 kN/m² a été retenue pour tenir compte de la circulation routière.
Une charge piétonne uniformément répartie de 5 kN/m² a été retenue sur les accotements.
 Actions sismiques:
Aucune action sismique n’a été considérée dans le dimensionnement.

a. Justification, Choix des géogrilles et Résultats

511
La justification de la stabilité interne, externe et mixte du remblai renforcé par géogrilles consiste à vérifier le
choix des géogrilles en type, nombre, longueur et espacement vertical à l'intérieur du remblai, de façon à assurer
l’équilibre pour toutes les lignes de ruptures potentielles. Cette justification consiste à vérifier que, pour chaque
combinaison d'actions, les résistances de calcul des éléments constitutifs et celles mobilisées par les phénomènes
mis en jeu (interaction sol/géogrille, cisaillement du sol) génèrent un degré dútilisation inférieur à 1,0. Le degré
dútilisation µ (Équation 1) est défini tel que ce degré dútilisation µ multiplié par la somme des valeurs de
calculs de résistances de dimensionnement (Rd) corresponde exactement à la somme des valeurs de calculs
dáctions de dimensionnement (Fd).

Équation 1 (Ziegler, 2005)

Le choix de la géogrille s’effectue en fonction de la résistance en traction de dimensionnement à long terme du

géogrille, du coefficient d’interaction, du polymère constitutif qui doit être compatible en termes de durabilité
avec le matériau de remblai, et le cas échéant, avec les matériaux constitutifs du parement. Les différents
coefficients partiels de réduction utilisés dans ce projet sont présentés dans le Tableau 1.

Tableau 1. Facteurs partiels de réduction appliqués sur le géogrille utilisé dans le projet

A1 A2 A3 A4 A5 M
RB,k0
grain Neutr
grain mix grain gros Acide Alcalin
1 2 5 10 60 120 fin e CC CC
[kN/m] D90< D90< 63 (2<pH< (9<pH<1 CC 3
an ans ans ans ans ans D90< 2 (4<pH 1 2
32mm mm 4) 0)
mm <9)

Fortrac
1,3 1,4 1,4 1,4 1,5
55 9 1 3 5 0
1,52 1,09 1,11 1,14 1,0 1,15 1,03 1,10 1,0 1,4 1,3 1,2
55 T

A1: Facteur de réduction lié à la déformation par fluage en fonction de la durée de charge
A2: Facteur de réduction lié aux dommages causés lors de l'installation, le transport et le compactage
A3: Facteur de réduction lié aux traitements (coutures, assemblage, jointures)
A4: Facteur de réduction lié aux impacts environnementaux (résistance aux intempéries, produits chimiques,
micro- organismes, animaux)
A5: Facteur de réduction lié à l'impact de l'action dynamique
M: Coefficient partiel de sécurité pour la résistance structurelle des éléments de renforcement flexibles
conformément à la norme DIN 1054
RB,k0: Valeur caractéristique de la résistance en traction "à court terme" du géogrille conformément à la norme
EN ISO 10319
Les facteurs de réductions sont définis par des essais normalisés et validés par des avis techniques. La résistance
en traction de dimensionnement à long terme RB,d(EBGEO 2010) est définie l'équation suivante:

On obtient donc pour la géogrille ayant une résistance en traction à court terme de 55 kN/m une résistance en
traction de dimensionnement à long terme de 22 kN/m.

La Figure 2 présente la surface de rupture la plus défavorable selon la méthode de Bishop, le degré d´utilisation
maximale obtenu sur plus de 3000 cercles de rupture potentielle examinés est de 0,96 (< 1,0). La Figure 3
présente la surface de rupture la plus défavorable selon la méthode des blocs glissants, le degré d´utilisation
maximale obtenu sur plus de 2000 lignes de rupture potentielle examinées est de 0,95(< 1,0).

512
Figure 2. Examen des cercles de rupture potentielle Figure 3. Examen des cercles de rupture potentielle
selon la méthode de Bishop (GGU-Stability) selon la méthode de Bishop (GGU- Stability)

4. PROCÉDÉ DE CONSTRUCTION

Il est rappelé que la Norme européenne NF EN 14475 (Exécution des travaux géotechniques spéciaux -
Remblais renforcés-) établit de manière générale les principes généraux pour la réalisation de ce type de travaux.
Le procédé de construction est un processus répétitif (Figure 4).

La première phase de construction comprend l'excavation de la pente existante jusquáu niveau requis (Figure 5
& 6). Un géotextile non-tissé a été installé le long de la coupe pour assurer la séparation et la filtration (Figure
7). Le support du remblai a été soigneusement compacté afin dóbtenir une plate-forme uniforme. Lúniformité
et lángle désiré du parement sont réalisés à láide dún coffrage provisoire fixé et maintenu en place par des
tiges métalliques (Figure 8). Un géotextile anti-érosif est placé en face avant du géogrille pour maintenir la terre
végétale (Figure 8). Le géogrilles avec la longueur requise dáprès les calculs de stabilité est mis en œuvre avec
une longueur suffisante en avant du coffrage provisoire pour assurer un retour de nappe/ancrage dáu moins 1,50
m (Figure 9). La couche de matériau de remblai est ensuite apportée jusqu´à la hauteur spécifiée et compactée à
une densité Proctor dáu moins 98% de lÓPN.
La couche de remblai étant complète, le coffrage provisoire est enlevé et replacé au dessus pour la couche
suivante. Le même procédé est répété jusqu'à ce que la hauteur maximale de la structure soit atteinte.

Figure [Link] du site et début des travaux de terrassement

513
Figure 6. Excavation de la pente existante

Figure 7. Installation du Géotextile de séparation

Figure 8. Installation d´un coffrage provisoire, du géogrille et du géotextile anti-Erosion

514
Figure 9.(a) Procédé général de construction d’un ouvrage de soutènement et (b) Ancrage du géogrille par retournement en
sol renforcé par géogrilles

Figure 10. Vue de l´ouvrage avant le démarrage des travaux Figure [Link] globale de l´ouvrage en cours de construction

Figure 12. Vue de l´ouvrage achevé

5. CONCLUSION

Léxpérience Rwandaise avec le remblai renforcé par nappes de géogrilles dans le cadre de la réhabilitation de la
voie principale Ruhengeri-Gisenyi a montré que cette solution, lorsquélle est retenue par les maîtres d´œuvres
présente des avantages très significatifs tant sur le plan économique, écologique et quénvironnemental avec une
parfaite intégration paysagère. Les solutions traditionnelles sont souvent coûteuses et peu adaptées dans des
contextes géotechniques et géologiques difficiles. Les recherches continues des solutions techniques en
accordance avec les principes du développement durable étant de plus en plus exhortées par les états rendent ces
types de structures particulièrement intéressants. Face à l'urgence de la crise écologique et sociale qui se
manifeste désormais de manière mondialisée, le développement durable est une réponse de tous les acteurs
(états, marché, société civile) pour reconsidérer la croissance économique dans chaque pays afin de prendre en
compte les aspects environnementaux et sociaux du développement. Le choix dúne telle solution au détriment
des ouvrages traditionnelles (par exemple murs poids en béton armé, murs poids en maçonnerie) présente

515
d´énorme avantages au regard du développement durable. Grâce à la réduction importante du volume et du poids
des matériaux transportés, ces types de structures permettent en général de réduire les émissions de gaz à effet de
serre de manière significative. Ils permettent le réemploi desmatériauxdu siteouà proximité du site tout en
réduisant le temps de construction et les coûts de chantier.

BIBLIOGRAPHIE
DIN 1054 (2005) Baugrund; Sicherheitsnachweise im Erd- und Grundbau
DIN EN 1997-1 (2004) Eurocode 7: Geotechnical design - Part 1: General rules
EBGEO (2010) Recommendations for Design and Analysis of Earth Structures using Geosynthetic Reinforcements, German
Geotechnical Society (DGGT), Ernst & Sohn, Essen-Berlin.
NF EN 14475 (2007) Exécution des travaux géotechniques spéciaux - Remblais renforcés. Indice de classement: P 94-326
NF EN ISO 10319 (2008) Géosynthétiques - Essai de traction en bande large
M Geok E 2005 (2005) Merkblatt über die Anwendung von Geokunststoffen im Erdbau des Straßenbaus, Regulations for the
application of geosynthetics in earthworks of road building, FGSV - Köln, 2005
Ziegler M. (2005) Geotechnische Nachweise nach DIN 1054: Einführung mit Beispielen, Ernst & Sohn

Traitement de talus marneux par un geosynthétique alvéolaire : cas de l’auto

route est ouest en Algérie.
Marl slope reinforcement by alveolar geosynthetic studied:case of east west Algeria highway.
Said Tabti1, Farid Cherifi 1

1Afitex Algerie, Algerie.

RÉSUMÉ - Le projet d’Autoroute Est-Ouest Algériens’étend sur un linéaire total de 1216 km entre la frontière
marocaine et la frontière [Link] comporte plusieurs contraintes géotechniques dont la stabilité des
talus.
Cette région se caractérise par son terrain accidenté qui influe directement sur le réseau routier. Plusieurs
glissements recoupent le tracé des routes existantes. Cesglissements ont été confortés provisoirement (solution
basée sur l’expérience, pas d’étude de dimensionnement) et certains d’entre eux se manifestent encore
régulièrement.
Pour stabiliser les talus marneux de l’autoroute Est Ouest Algérien, une solution mixte géocomposite pour le
drainage et géosynthétique alvéolaire pour la stabilité des talus a été proposée. Cette solution a été approuvée par
le maitre de l’ouvrage et le bureau d’études.
Mots-clés : géocomposite, alveoter, anti érosion, drainage, glissement, stabilité de terres.

ABSTRAST-TheEast-West highwayprojectin Algeriacoversatotal lengthof1216km betweenthe Moroccan

borderandthe Tunisian border. This project involves several geotechnical constraints. Especially slope stability
of marl and clays material located in eastern part of Algeria.
This region is characterized by its rugged terrain that directly affects the road [Link] slope slides have
been pointed on existing road. Reinforcement of these slope slides are based on experiment without taking into
account any circulation or design and some slope slides still occurs regularly.
In order to stabilize skin fill slope of Algerian East-West highway, a mixed solution composed of geocomposite
for drainage and alveolar geosynthetic for slope stabilization (skin slide)was proposed.
Keywords: geocomposite,alveoter, anti erosion, drainage, sliding,slope stability.

1. INTRODUCTION

L’itinéraire imposé de la section autoroutière situé dans la partie de l’Est Algérien emprunte des dépressions
topographiques importantes et une géomorphologie complexe due à la conjugaison de dislocations liées aux
mouvements tectoniques et à la nature argileuse ou marneuse des terrains en place.
Ces terrains sont difficiles à drainer et les processus d’érosion par ruissellement rendent les pentes très escarpées.
Il existe de nombreux exemples de pentes trop abruptes dans un état de stabilité limite qui lâche (glissement)
suite à une élévation des niveaux de nappes phréatiques au dessus de la moyenne. L’eau réduit la résistance au
cisaillement du matériau et entraine des glissements des terrains.
Les solutions présentées dans cet article visent à protéger les talus de l’érosion superficielle (glissement de peau)
(Watn A., Oiseth E. 2000).
Les talus marneux étant supposés auto-stable à court et long terme sur les grands glissements, la figure 1 illustre
un talus marneux à traiter par solutions géosynthétiques.

517
Figure [Link] marneux à traiter.

2. DESCRIPTION DE LA SOLUTION DE BASE

Le talus est protégé de l’érosion et des variations hydriques par une géomembrane étanche. Le placage d’une
couche de terre végétale ou de concassé sur cette géomembrane risque de la détériorer (poinçonnement) et sera
instable (glissement à l’interface). La mise en place d’un géotextile renforcé ou non suivant la pente du talus et la
longueur du rampant permet de répondre à ces deux contraintes. L’ensemble du dispositif est ancré en tête dans
une tranchée d’ancrage.
Les avantages de la solution de base
 Protection complète du talus.
Les inconvénients de la solution de base
 Solution assez lourde et complexe.
 Enracinement très superficiel de la végétation.
 Risque si venue d’eau importante sous la géomembrane.
 Difficulté d’entretien

[Link] DE LA SOLUTION VARIANTE PROPOSEE.

La solution géosynthétique (Tableau 1) proposée en remplacement de la solution de base a été retenue et

appliquée sur tout l’itinéraire de l’auto-route au niveau des talus marneux, qui sont caractérisés par leurs pentes
et hauteur des talus suivant trois cas . Ledispositif est illustré dans la figure 2.

Tableau 1. Dispositif géosynthétique de drainage et retenue de terres sur talus

Pour assurer la fonction

Pente du talus Hauteur de talus Pour assurer la
fonction de drainage de retenue de terre

Geocomposite de
≤ 2H /1V ≤5m Geosynthétique
drainage Alvéolaire
Geocomposite de
≤ 2H/1V 5 m< H <10 m Geosynthétique
drainage Alvéolaire
Geocomposite de
2H/1V ≤ p ≤ 3H/2V ≤ 10 m Geosynthétique
drainage Alvéolaire

518
Structure de drainage et retenue de terre sur talus

Structure du DRTT sur talus marneux

TERRE VEGETALE

GEOSYNTHTETIQUE DE RETENUE DE TERRE SUR

TALUS

GEOCOMPOSITE DE
DRAINAGE

Figure 2. Dispositif de retenue de terres sur talus (DRTT).

3.1 Le géocomposite de drainage

Le drainage des eaux internes est assuré traditionnellement par des matériaux granulaires de bonne qualité La
préservation des ressources naturelles étant un objectif du Grenelle de l’environnement, on trouve des
alternatives à l’utilisation du tout granulaire grâce à l’emploi de géocomposites drainants. Le produit
géocomposite de drainageproposé en variante à une solution traditionnelle, présente la caractéristiqued’associer
des géotextiles drainants avec des géotextiles filtres et des mini-tubes perforés. Il est utilisé dans le drainage sous
remblai en talus ou sous chaussée, que ce soit pour évacuer les eaux de la structure ou des eaux extérieures à
l’ouvrage.
Très ancien sur le marché, ce produit a suivi de très nombreuses évolutions (nombre de drains, propriétés des
filtres…) preuve de son adaptation mais également d’une certaine complexité.
L’instruction du dossier a montré que la complexité du produit nécessite une étude hydraulique spécifique pour
son dimensionnement et un Contrôle rigoureux lors de la mise en œuvre sur chantier. Moyennant ces
précautions, le produit peut prétendre à remplir la fonction de drainage attendue.
Le talus marneux est stable sans apport de poids supplémentaire mais les infiltrations et les arrivées d’eaux
peuvent être sources d’instabilités et de ruptures.
Le matériau d’apport est capable de remplir la fonction mécanique mais pas la fonction hydraulique.

[Link]ésentation du produit et du procédé

Le géocompositemis en œuvre qui assure les fonctions de filtration et de drainage norme NF EN
13252),appartient àune gamme de produits manufacturés, souples et équipés de mini-drains régulièrement
perforés. Les différents composants des géocomposites de la gamme (nappe drainante, nappes filtrantes, mini-
drains) sont en polypropylène (PP).
L’ensemble des éléments (nappe drainante, nappes filtrantes et mini-drains) est assemblé industriellement par le
procédé d’aiguilletage.
La nappe composite (nappe drainante et nappes filtrantes) est appelée partie non tissée dans la suite du texte.

519
Figure 3 .Structure du géocomposite de drainage

3.1.2 Dimensionnement
Pour ce projet une note de calcul a été établie afin de déterminer le flux drainé par le géocomposite situé sous
un geosynthétique alvéolaire et une épaisseur de terre végétale, notamment pour définir sa capacité à drainer des
eaux pluviales.
Le dimensionnement est réalisé à l’aide d’un logiciel (lymphia) permettant de modéliser le comportement d’un
géocomposite posé sur le talus marneux et recouvert du complexe géosynthétique alvéolaire plus terre végétale.
Il est important de noter que le calcul est mené pour le plus grand talus (rampant le plus long), soit 16,20 m à
2.5H/1V , ces calculs sont fait conformément aux procédures d’utilisation des geosynthétiques sur talus (Myhre
J. 1985).
Le logiciel utilisé permet de déterminer les capacités drainantes en fonctions de différents paramètres.
Par sécurité, le géocomposite est considéré en régime d’écoulement libre non saturé, c’est-à-dire sans pression au
droit des mini drains.
L’introduction de ces paramètres dans le logiciel permet de déterminer le flux drainé, exprimé aussi en débit de
pied par mètre linéaire, permettant de dimensionner le drain routier collecteur.
Les hypothèses et résultats sont les suivantes :
 Le flux maximal drainé par le géocomposite est de 1,36.10-5 m²/s.
 Ce flux correspond à une hauteur d'eau de 1 180 mm/jour.
 La pluviométrie centennale (100 ans) fournie est de 300 mm/h.
 La transmissivité équivalente (m2/s) 9,64.10-3
 Le coefficient de ruissellement et d'évapotranspiration de la couche de terre végétale est supérieur
à 85 %.
 Il y aura donc moins de 15% de la pluviométrie à drainer, soit de l'ordre de 1 000 mm/jour.
 La pression maximale entre mini-drains de 11 mm de hauteur d'eau, celle-ci est négligeable vis-à-
vis du poids de remblai, elle n'a donc aucune influence sur la stabilité de ce remblai au contact du
géosynthétique.

3.1.3 Avantages du géocomposite de drainage

Les avantages du géocomposite dans le cadre de l’utilisation sur talus marneux sont :
 la présence des mini-drains incompressibles qui permettent une évacuation rapide et
monodirectionnelle de l'eau vers les tranchées collectrices.
 l'aiguilletage dense des filtres géotextiles à la nappe drainante qui permet d'éviter tousdéplacements
filtre/nappe drainante.
 la souplesse du géocomposite lui permet d'épouser les irrégularités du fond de forme.
Ces deux dernières caractéristiques optimisent la fonction filtre en limitant les vides au
contact du filtre, donc l'apparition de sol en suspension.

520
En
n effet, les sols en suspension présentent la particularité d'avoir des particules fines indépendantes les unes des
autres à leur arrivée au contact du géotextile, ce qui rend la fonction filtre moins performante, donc le risque de
colmatage plus important(Watn
Watn A. 2005).
Pour cette raison, il est important que le filtre soit le plus possible en contact avec les sols et avec la nappe
drainante.

Figure [Link]éma
.Schéma du géocomposite sur talusFigure
talus 5. Comparatif avec ou sans géocomposite

Figure6. Mise en œuvre du géocomposite de drainage

3.2 Le géosynthetique alvéolaire

Le géosynthétique alvéolaire mis en œuvre sur le géocomposite de drainage est destiné à prévenir du ravinement
et des glissements de peaux sur les talus naturels.
Il se présente sous forme d’une grille tridimensionnelle en polyester (figure 7) constituée d’alvéoles hexagonales
de 40 cm x 45 cm de large pour une hauteur de 15 cm.
Le géosynthétique alvéolaire est simplement posé sur un sol préalablement préparé dont on a arasé les
irrégularités, afin que chaque alvéole soit en contact continu avec le sol support,Le géosynthétique est retenu en
tête de talus par une tranchée d’ancrage. Cette fixation est complétée en pleine pente par des fers à béton dont la
densité et la dispositionn dépend de la pente et de la nature du terrain.
Les liaisons entre lés se font par agrafage,soit par liens synthétiques mais en tout état de cause, ces fixations
n’ont qu’un rôle temporaire,dans l’attente de la mise en œuvre du remblai de terre.

521
Figure 7. Structure du géosynthétiques alvéolaire.

3.3 Les géosynthétiques de drainage et alvéolaire

La figure 8 présente une illustration sur les deux produits mis en œuvre sur talus marneux.

Figure8. Les deux géosynthétiques mis en œuvre.

4. RESULTATS DE MISE EN ŒUVRE DE LA DOUBLE SOLUTION GEOSYNTHETIQUE

Les figures 9 et 10 illustrent, après une période de deux mois le talus traité par la double solution
geosynthétique.

Figure 9,10 les talus marneux après traitement par les deux géosynthétiques.

522
5. CONCLUSION

Le contexte géologique et topographique de la section, particulièrement difficile et très mal connu à l’origine du
contrat d’études et de travaux, pour les deux fonctions (drainage et anti érosion), de l’application de la stabilité
des talus, a nécessité une reprise complète de la
Conception initiale ; la réactivité de la maitrise d’ouvrage a été sans faille pour valider cette conception.
La double solution proposée pour assurer les fonctions drainage par le géocomposite et la retenue de terre
par le géosynthétique alvéolaire sur talus marneux a permis la stabilité des terres sur talus sur des tronçons du
projet auto route Est Ouest en Algérie.
Le géosynthétique alvéolaire mis en œuvre sur le géocomposite de drainage est destiné à prévenir du ravinement
et des glissements de peaux sur les talus naturels.
Le recul de plus de quatre ans sur cette solution géocomposite a permis à la maitrise d’œuvre de la préconiser
pour la suite du chantier.

REFERENCES BIBLIOGRAPHIQUES

Myhre J. (1985). Strengthening of a road on peat using Tensar geogrid and Terram Geotextile. SINTEF report STF61
F85011.
Watn A., Oiseth E. (2000). Geosynthetic reinforcement in roads – test road at Hitra, Sør-Trøndelag,Norway. SINTEF report
STF22 F00604.
Watn A. (2005). Geosynthetic reinforcement for pavement systems – European [Link] Bearing Capacity of
Roads and Airfields (2005).

Formulation of a new road material based on treated dredged marinesediments

Formulation d'un nouveau matériau routier à base de sédiments de dragage marins.
Maherzi Walid1,BenAbdelghani Farouk*2
1
Département de Génie civil et Environnemental, École des mines de Douai, Douai, France.
2
Département d'Hydraulique et Environnement, ISTEUB, Cherguia II, Tunisie.
*
Corresponding author

ABSRACT-In France,dredging operations generate, each year, more than 50 million m3of marine sediments.
These raw sediments can't be used in road construction without a specific treatment process. In order to improve
sediments compacity and their mechanical performance, addition of a granular material is [Link] use
of a dredged sand, to improve the granular mixture containing sediments, allows a better management of the two
types of dredge materials (sand and sediment). In this study, a new road material containing dredged marine
sediments and dredged sand is formulated and treated by adding various binders. Mechanical performance
investigation of different mixtures by measuring Proctor-IPI values and simple compressive strengths is realized.

RÉSUMÉ-En France, les opérations de dragage génèrent chaque année plus que 50 millions de m3 de sédiments
[Link] sédiments ne peuvent pas être réutilisés à leur état brut en techniques routières sans un traitement
spé[Link] d'améliorer leur compacité et leurs performances mécaniques, l'addition d'un granulaire est
recommandée. L'utilisation du sable de dragage pour améliorer la granulométrie des mélanges contenant les
sédiments permet une meilleure gestion des deux matériaux, le sable et sédiments. Dans cette étude, un nouveau
matériau routier a été formulé à base des sédiments et du sable de dragages et en ajoutant des liants hydrauliques.
Les performances mécaniques des différents mélanges ont été étudiées en mesurant les paramètres Proctor et les
résistances à la compression simple.

1 INTRODUCTION
Nowadays, more than 90% of the economic exchanges are performed by sea [Link] order to improve
harbor infrastructures, dredging and clearing out operations are organized and generate a large quantity of
sediments, around 600 Mm3/year (Boutin, 1999). These important volumes of sediments must be managed
according to the new environmental [Link] that civil engineering material resourcesbecome rarely, one of
the innovative solutions for dredged sediments management is their valorization as a reused material in road
[Link] sediments are characterized by their high initial water content and high organic matter
content, which causes difficulties for their reuse at their raw state in road construction.

Recent studies were realised on treated sediments with hydraulic binders (1 to lime 4%) to evaluate the
modification of their behavior according to the hydraulic binder(Hake, 2003; Tran, 2009; Wang, 2011).Chauvel
etal.(1985) show that a clay soil treatment will only modify the Proctor optimum parameters (water content;
density) from (27%;1.62 t/m3) to (31%; 1.47 t/m3) and will increase the bearing pressure of the ground from 10%
to 54%.The influence of lime on dredged sediments (sampled from Le Havre port, in France)was studied by
Colin (2003). Obtained results show that a treatment with only 4 % of lime increases the optimal water content
by 1% (from 38 to 39%).The density, whereas, decreases from 1.21 to 1.14 t/m3. The bearing pressure increases
from 8% to 10%. By using a mixed treatment composedof 2% of lime and 6 % of cement (CEMI), the author
notes that the water content reaches a value of 37%, the dry density reaches a value of 1.19 t/m3 and the bearing
pressure is 12%.The author concludes that treatment with only lime does not improve the bearing pressure of
sediments. This is due to the important organic matter content of the studied sediments which consumes a part of
the pesticides and could resist tohydration reaction (Rekik, 2007).In this order,a mixed material containing
marine sediments and dredged sand treated with two types of road hydraulic binders with different mineralogical
compositions is formulated. Mechanical performance of different formulations is investigated.

2. MATERIALS AND METHODS

2.1 Sampled sediment
The studied sediment is sampled from the Brest port located at the West of France. Physical and geotechnical
characteristics of the sample are listed in Table [Link] these results and according to the GTR French guide
(1992), the studied sediment is considered as a fine material and classified in F11subclass with high organic
content. With such GTR classification, this sediment can't be used in road construction and hydraulic binders
must be used to improve its mechanical performance.

525
Table [Link] and geotechnical characteristics of the sampled sediment
Sediment parameter Value (%)
Organic matter (O.M) 5.4
Argilosity fraction (0/2 µm) 19.6
Siltyfraction (2/63µm) 31.4
Sandy fraction (63µm/2mm) 49.0
Fine fraction (0/80µm) 77.0
Absolute density (ρsmg/m3) 2.5
Liquid limit (LL) 57.0
Plasticity index (IP) 20.0
Methylen Blue Value VBs (g/100 g dry) 1.4
GTR classification F11

2.2 The granulometric correctors

In this experimental work two types of granulometric correctors were used to study their influence onsediments
compactness and treatment efficiency. The first one is constituted by adredged sand (class 0/2 mm) (SD) and the
second is apitsand (SC)composed of 98% of particles having a diameter higher than 63µm.

2.3 Hydraulic binders

In this study two types of road hydraulic binders were used, Ligex FPl1 (composed of 66 % C3S) andRolacPi
LH (62 % C3S). Obtained treated sediments were prepared according to the following steps (see Figure 1): the
sediments are humidified 24 hours before pesticides addition; than quicklime addition; after a period of1 h,
which is the needed time for lime action (NF EN 14227-11), the hydraulic binder is incorporated; and finally the
unit is mixed during 3 minutes before compaction test.

Figure [Link] of treated mixtures.

Mechanical performance of treated samples isinvestigated through simple compression and diametrical
compression [Link] compressive strength test was carried out on 2 test tubes (5×10 cm) and the diametrical
compression tests were carried out on 1test-tube (5×5).After specimen preparation,the test-tubes are preserved at
a constantwater content in hermetic membranes. The cure period is 7 and 28 days.

3. RESULTS AND DISCUSSION

3.1 Proctor test results
Obtained results forthe optimum Proctor Normal test and IPI the treated formulations with lime and hydraulic
binders are summarized in Table [Link] mixtures treated with HB2, dry densities vary between 1.52 and 1.70t/m3,

526
for water contents between 14.4 and 20.5%.Corresponding IPI values are between 21.0 and 29.1%.It can be seen
from these results that IPI values are improved by various treated mixtures.
For mixtures treated with the LIGEX, dry densities are between 1.49 and 1.66t/m3, for water contents ranging
between 17.1 and 23.1%. Corresponding IPI values are between 19.8 and 26.8%.In comparison with obtained
results for mixtures with 100% of raw sediment and 70% sediment+ 30%SD, a treatment with lime and with
road hydraulic binders could improve the IPI [Link] results show that all studied formulations respond
to the criterion of a minimal bearing pressure (for A1 Class, IPImin=20%) for a use in subgrade layer (GTS.
2000).
Table [Link]-IPI results forthe mixtures treated with cement.

Formulation WOPN (%) ρd OPN (t/m3) IPI (%)

Untreated sediments(G001) 21.5 1.51 10.5
G001+SD30 15.0 1.73 17.5
G001+SC30 16.0 1.76 19.0
G001+Lig6 23.0 1.49 19.8
G001+Lig15 23.1 1.51 22.1
G001+SC+Lig6 17.1 1.65 19.8
G001+SD+Lig6 18.3 1.64 23.3
G001+SC+Lig15 17.7 1.65 26.8
G001+SD+Lig15 18.0 1.66 26.8
G001+Rol6 20.5 1.52 24.5
G001+Rol15 18.7 1.54 25.6
G001+SC+Rol6 15.7 1.66 24.5
G001+SD+Rol6 16.1 1.64 21.0
G001+SC+Rol15 14.4 1.70 26.1
G001+SD+Rol15 16.5 1.66 29.1

[Link] performance results

Results of the simple compressive strength (RC) (or the unconfined compressive strength) are an average of
three [Link] standard deviation of measurements varies between 0.01 and 0.09 [Link] 2 and 3 show
the evolution of the simple compressive strength of the treated mixtures at a constant water content. These
figures show that the UCS increases between 7 and 28 days of cure for all formulations. Values obtained with
the LHR are higher than those with the rough sediment treated with CEMII (NF EN 197-1, 2012) for the mixture
treated with 6% of ligex. A value of 1 MPa for the UCSneeded for the circulation of machines (GTS, 2000) was
obtained for all formulations treated with 15 % of Ligex after 28 days of cure. Also it was the case for the
formulations containing a granular corrector treated with 6% of Ligex. Mechanical performance results are better
when using Rolac hydraulic binder. In fact, and for both times cure, Rc values are higher for Rolac binder.

Figure 2. Results of simple compressive strength for mixtures treated with lime and Ligex at 7 and 28 days.

527
Figure 3. Results of simple compressive strength for mixtures treated with lime and Rolac at 7 and 28 days.

4. CONCLUSION

The main objective of this study is to formulate a new road material based on dredged marine sediment, sand and
hydraulic [Link] results highlight the use of treated mixtures containing sand and dredged marine
sediments for their reuse in road construction. From this study, we can conclude that addition of sand todredged
sediments can increase the compactnessof mixtures. This can have a significant influencein the exchange process
between treated materials and their environment. Also, mixtures containing dredged sediments, sand and
hydraulic binders with high lime contentcan increase the mechanical performances,particularly the compressive
strengths.

Finally, using differentgranular corrector is interesting in order to evaluate their influence on the compactness of
obtained mixtures.

REFERENCES

Boutin R. (1999)Improvement knowledge about the behavior of sea rejections by dredging of vase. Thesis,INSA Lyon,
France, 527p.
Colin D. (2003) Valorization of fine dredged sediments in road construction. Thesis, Université de Caen, France.
Chauvel A., Nobrega M., Tessier D. (1985)The effect of compaction and addition of lime on [Link] Micromorphology,
1, 10-18.
Hake D. (2003)Valorization of fine dredged sediments in road [Link], Université de Caen, France, 180 p.
LCPC-SETRA (1992)Construction of the embankments and the subgrades - technical Guide, 204 p.
LCPC-SETRA (2000) Soil stabilization to limes and/or hydraulic [Link] to the realization of the embankments
etdes subgrades - technical Guide, 240 p.
Rekik B. (2007) Geotechnical properties of dredged sediments treated with cement. Thesis, Université de Caen, France.
Sannier L., Levacher D., Jourdan M. (2008)Economic approach and validation of treatment methods of contaminated marine
sediments treated with hydraulic [Link] Paralia, 2, s-2.1-s2.15.
Tran N.T. (2009) Valorization of marine and fluvial sediments in road [Link], ENSM de Douai, France, 159p.
Wang D. (2011)Solidification and valorization of Dunkerke sediments in road [Link], école Nationale
Supérieure de Mines de Douai, France, 156 p.

Chemical and physical characterization ofGabes marine sediments

Caractérisation physique et chimique des sédiments marins du port de Gabes

Amel MISSAOUI*1,2, Imen SAID1, Essaieb HAMDI1, Zoubeir LAFHAJ2

1
Université de Tunis El Manar – Ecole Nationale d’Ingénieurs de Tunis. LR14ES03-Ingénierie Géotechnique.
BP 37 Le Belvédère, 1002 Tunis. Tunisie
2
Ecole Centrale de Lille, Villeneuve d’Ascq, France.
* Corresponding Author

ABSTRACT -Sediments are dredged periodically (every 10years) from the eight Tunisian harbors in order to
ease ships navigation. According to the Tunisian office of merchant navy and ports,about 3.6 million cubic
meters (Mm3) of marine sediments are dredged each two years. Gabes harbor is one of the biggest commercial
harbors in Tunisia. It is located in the south east and the center of gulf of Tunisia. About 2.2 Mm3 of sediments
were dredged from Gabes harbor in 2009 before discharging in the sea. The pollution of the latter is mainly due
to the important industrial activities around the harbor’s area. Therefore, a treatment process, to clean up present
pollutants, is necessary.
This articleaims to evaluate the main characteristics of Gabes harbor sediments and to study the feasibility of
their most suitable and cost-efficient pollution treatment. In this study, physical and chemical characterization
tests were carried out on seven samples, extracted from different locations in the harbor, in order to evaluate the
quality of Gabes marine sediments.

RESUME: The second part of the article includes an overview on several possible existing treatments’ technics.
The last part includes a proposition of a potential treatment of these polluted sediments.
Key words: Marine sediments, pollution, characterization, treatment, heavy metals.
ABSTRACT - Des sédiments marins sont dragués régulièrement (toutes les 10 années) des huit ports tunisiens
afin de faciliter la navigation des navires. Selon l’office de la marine marchande et des ports Tunisiens, environ
3,6 millions de mètres cubes (Mm3) de sédiments marins sont dragués, des ports Tunisiens, chaque deux ans. Le
port de Gabes présente un des plus grands ports commerciaux en Tunisie. Il est situé dans le sud-est et le centre
du golfe de la Tunisie. Environ 2,2 Mm3 des sédiments ont été dragués de ce port en 2009. Les sédiments
dragués présentent une pollution qui est principalement due aux activités industrielles importantes autour de la
zone du port. Par conséquent, un procédé de traitement, pour nettoyer les polluants présents, est nécessaire.
Cet article présente une étude de caractérisation des sédiments portuaires de Gabes et d’introduire la faisabilité
du procédé de traitement, le plus approprié et rentable. Dans cette étude, des essais de caractérisation physiques
et chimiques ont été effectuées sur des échantillons des sédiments marins de Gabes.
Une deuxième partie de cet article présente un aperçu sur quelques techniques de traitement des sédiments
existantes. La dernière partie comprend une proposition d'un procédé de traitement potentiel d’être appliqué sur
les sédiments de Gabes.

1. INTRODUCTION

Marine sediments are deposited insoluble materials, organic matter in various stages of decomposition and
remains of marine organisms. Marine sediments vary widely in composition and characteristics in accordance
with water depth, distance from land, sediments’ source, physical, chemical, and biological characteristics of
their environment (Romero et al, 2008).
Sediments problematic is presented on its accumulation on the bottom of channels and harbors, resulting
dysfunctioning of marine commercial activity. Consequently, the dredging of sediments from harbors becomes a
regularly and compulsory action.
Due to sediments’ grain size and composition, pollutants enter and trap in sediments’ matrix in different ways.
Pollution can be released in waterways several years ago, and then lodged in sediments in different

529
concentrations. These latters vary in relation with the depth of the sediment layer (Junakova & Balintova, 2012).
Other pollutants may be entered in sediments daily from permanent sources then circulate directly from
industrial discharges (Hloušková et al., 2014), municipal waste discharges especially in very high urbanized sites
(Acquavita et al., 2014) and/or intensified agriculture areas (Zhang & Shan, 2008). Pollutants could be also
transported through the atmosphere (air, water …) landing in waterways far from their original sources (Nouira
et al, 2013)
According to United States Environmental Protection Agency, five major types of pollutants are found in
sediments (i.e. Nutrients, Bulk Organics, Persistent Organics, Polycyclic Aromatic Hydrocarbons (PAHs) and
Heavy Metals). These major pollutants have shown in different studies (Nakata et al., 2014) to be toxic to marine
ecosystems and harmful for benthic species. Occasionally, they can be toxic to human beings (Antizar-Ladislao,
2008). Due to the negative influence of these compounds on the environment as well as on humans, a quick
monitoring is compulsory.
Monitoring of these polluted sediments was mainly presented by a pretreatment followed by a treatment process.
Due to its high costs (Lafhaj & Husson, 2011), this solution is not economically profitable, even if it clearly
reduces the toxicity and the pollution of sediments. This fact underlines the necessity to find economic and
efficient treatments or to find ecological reuse paths for treated sediments to make their global monitoring costs
competitive when compared to sea disposal and landfilling alternatives.
This paper presents and discusses some of the results of the geotechnical and environmental characterization of
sediments dredged from Gabes harbor. These latters present a critical ecologic issue for marine environment that
could be due to industrial wastes discharged, by neighboring factories, in the marine water surrounding the
harbor. Physical and chemical characteristics were determined and discussed in order to propose finally a
suitable treatment process.

2. OVERVIEW ON TREATMENT PROCESSES

According to the type of contamination, management options for sediments’ cleaning were determined. Three
principal categories, according to the way of pollutants’ treating, can be defined for existing treatment processes.
The first category is the cleaning by contaminants’ removing, the second is treatment by destruction and the last
family is the immobilization [Link] are treatment processes that combine two categories such as
physico-chemical treatments that include destruction and immobilization. Commonly, destruction methods are
used for organic contaminants that are sensitive to high temperatures while removing and immobilizing methods
are commonly used for heavy metals (inorganic contaminants). Considering the contaminants’ concentrations
removed (via emissions of effluents) or destroyed in the feedstock, the effectiveness of the treatment process is
determined. For immobilization processes, where the contaminants are trapped in sediments’ matrix, the
effectiveness is measured as the difference of contaminants ’ concentrations in leachates, extracted from
sediments, before and after the application of the remediation process (Tech & Averett, 1994). The effectiveness
could be also related to the physical characteristics of the sediments, especially its gain size distribution. Table 1
presents examples of studied treatments, a general description of the used processes, pollutants concerned in the
studies, the effectiveness of the processes, and the sediments’ origins.

530
Table [Link] of some existing treatment processes
Treatment Sediments
General description Treated pollutant Effectiveness Reference
Process origins
Two-step: Marine
At 900°C - Combustion
Thermal Inert (pyrolysis) or Sediments (Rausa et
Organics (a complete destruction
treatment oxidative (combustion) (Italian al., 1999)
of all organics)
atmosphere Harbor)
Polychlorinated
dibenzo-p-dioxins
Charcoal (As both a Greater than 96% and River/ Canal
Thermal (PCDDs) and (Zhao et
thermal source and 90% for PCDD and Sediments
(Heat) treatment polychlorinated al., 2012)
pollutants adsorbent) PCDFs respectfully (Japan)
dibenzofurans
(PCDFs)
Above 80% for both 3-
Polycyclic aromatic
Sediment Slurry rings PAHs (fluorene River (Chiavola
Biological hydrocarbons (PAHs)
Sequencing Batch and anthracene) and 4- sediments et al.,
treatment (fluorene, anthracene,
Reactor (SS-SBR) ring PAHs (pyrene and (Italy) 2010)
pyrene and crysene)
crysene)
(Clément,
Canal Vaille,
Physico-
Good results for PHAs ; sediments Moretto,
chemical Novosol Process* PAHs and Metals
Mercury ; Copper wasquehal- Vernus, &
treatment
Lille Abdelghaf
our, 2010)
Stabilization (In-
Situ) Stabilization using
High phosphate Dewatering
Cu, Zn and Pb River
Temperature = by injecting lime and Heavy metals (Cu, Zn, (Yu et al.,
Reduced by 65%, sediments
Aerobic flocculants. Pb) 2009)
82.2% and 90% (China)
Composting HTAC (T° to 72°C for
(HTAC) (Ex- 20d)
situ)
Different enhancing
electrolytes (solutions of
High extraction yields Marine (Rozas &
Electrokinetic different nature (distilled Heavy metals (Cu, Cd,
for the six heavy metals sediments Castellote,
treatment water, citric acid, acetic Cr,Zn, Ni and Pb)
in one individual test (spain) 2012)
acid, humic acid and
EDTA))
Marine
Remediation using Organic & inorganic 65% reduction of TPH (Masciand
Phytoremediatio sediments
plants (shrub and grass contaminants Decrease in Zn, Cu and aro et al.,
n (port of
species) Cr concentration. 2014)
Livorno, Italy)
* Process proposed by a Belgian chemical company named Solvay S.A. It provides an efficient answer for the
treatment of adverse minerals residues, polluted with organic and inorganic compounds, such as sediments, fly
ash, automobile shredder residues, and contaminated soils, industrial sludge. The two main steps in Novosol
process are: Phosphating and Calcination (Lafhaj et al., 2007).

3. SAMPLING SITE DESCRIPTION

Sediments studied were collected from the harbor of Gabes, which is located in the middle of the Gulf of Gabes,
in the South East of Tunisian coast as shown in figure 1. This harbor is situated next to the most important
chemical group in Tunisia. In 2002, it was confirmed that about 4 to 4.5 Mm3/ year of industrial wastes, mainly
phosphogypsum, were discharged in sea. Thanks to navy currentology, these wastes full of contaminantsmay be
transported into the harbor and stuck in its basin and/or between its channels (CJB & EAM, 2007). In Addition,
Gabes harbor was constructed in 1974 for commercial and mainly industrial purpose. Its activity is characterized
by an industrial vocation that essentially assures the transit of chemicals on behalf of neighboring plants located
in the industrial zone of Gabes. This commercial traffic consists primarily of sulfur and ammonia import and
phosphoric acid and phosphate fertilizer for export.

Seven sediments samples were collected from this harbor in two different periods (figure 1). Samples SBA and
SC2 were collected during November 2013. Samples SG5, SG6, SG7, SG9 and SG10 were collected during
August, 20014. The collected sediments were stored in hermetic barrels and transported to the laboratory for
further characterization. The sediment were sampled by divers to a depth of 50 cm, using a plastic buckets.

531
Figure [Link] of Gabes harbor and sediments' samples

4. METHODS

4.1. Physical characterization

In this paper, water content, grain size distribution, bulk density, and organic matter content were reported for
Gabes’ sediments. Each sample was tested at least in triplicates and the average values were resulted. Water
content was determined by oven drying at 50°C for 72 hours (until having a constant mass) according to the
European standard XP CEN ISO/TS 17892–1. The Grain size distribution (0 to 2mm) wasanalyzed using
Mastersizer MS 2000 Particle Size Analyzer (Malvern Instruments Ltd., UK). Organic matter content was
determined by Loss On Ignition (LOI) method, i.e., the sample was combusted at 550°C according to French
standard NF EN 12 879.

4.2. Chemical characterization

For the determination of heavy metals concentrations, sediments’ samples were digested with aqua regia by the
use of hot plate heating (NF EN 13346- Method B). The obtained solutions were analyzed by inductively
coupled plasma mass spectrometry (ICP-MS) for Cadmium and Chromium and by inductively coupled plasma-
atomic emission spectrophotometry (ICP-AES) for Arsenic, Copper, Lead, Zinc, Nickel and Phosphorus.
Tunisian authorities use the Netherlands standards for sediments’ environmental quality assessment. Netherlands
is the unique country using only one set of action levels, which is regulated by the ZBT- Zoute Bagger Toets
(translation: Saline dredged material test).Threshold values are treated as strict limit values without exceptions.
The ZBT is introduced for assessing whether dredged material can be dispersed in the marine environment or
not. (DGE, 2007; Röper & Netzband, 2011). In this study, heavy metals in Gabes sediments are assessed
according to the ZBT values that are reported in Table 2.

532
Table [Link] threshold values(DGE, 2007)
Metal Unit ZBT value

Arsenic mg/kg 29

Cadmium mg/kg 4

Chromium mg/kg 120

Copper mg/kg 60

Mercury mg/kg 1.2

Lead mg/kg 110

Nickel mg/kg 45

Zinc mg/kg 365

5. RESULTS

This part is dedicated for the presentation, interpretation, and analysis of physical and chemical results for seven
sediments’ samples collected from Gabes harbor and denoted as SBA, SC2, SG5, SG6, SG7, SG9, and SG10.
Generally, these sediments present a visual aspect characterized by very dark grayish color and bad smell. The
shell debris content varies from one sample to another.

5.1. Physical characterization

Table3 presents physical characterization results of Gabes’ marine sediments’ samples. They were characterized
by relatively high water content that ranges from 87.2% to 533.2%. This high moisture content is related to the
handling and sampling of sediments. The difference between values could be related to the handmade sampling
(using the diver) that could not ensure the same seawater quantities in barrels. According to the grain size
distribution results (table 3), Gabes sediments present a fine particles fraction of ~ 40%. They are considered as
soils mainly composed of sand and silt. Bulk density of Gabes marine sediments varies from 1.12 g/cm3 to 1.74
g/cm3. Soils with high in organic matter contents may have a bulk density well below 1 g/cm³ and this fact is
confirmed by organic matter content results. For example, sample SG6 presents a low bulk density (1.12 g/cm3)
and relatively high organic content (17.5%).

Table [Link] results of Gabes sediments

Sample
Test Unit
SBA SC2 SG5 SG6 SG7 SG9 SG10

Water content % 87.2 180.0 99.4 533.2 222.2 219.2 108.9

Fine fraction (≤ 63µm) % 39.3 25.0 31.5 27.1 42.8 51.6 38.1

Silt (2µm - 63µm) % 35.6 23.3 28.8 25.2 40.1 46.7 34.8

Clay (≤ 2µm) % 3.7 1.6 2.6 1.9 2.7 4.9 3.4

Bulk density g/cm3 1.66 1.43 1.60 1.12 1.43 1.28 1.74

Organic content % 6.1 17.1 6.2 17.5 18.4 17.1 5.2

533
5.2. Chemical characterization

Concentrations of heavy metals in Gabes harbor sediments are given in Table 4. The maximum concentrations
values of every metal are noted with a star on the top. It comes from Table 4 that the maximum values of
concentrations of every metal are distributed between four samples SBA, SG6, SG7 and SG9, which are located
in the left side of the harbor. Marine currentology may be the factor of this distribution. Table 4 also presents a
comparison of Gabes sediments’ heavy metals concentrations with Netherlands ZBT standard. The ratio of
heavy metals to the ZBT value of each metal is introduced and put between brackets in Table 4. Concentrations
that exceed threshold values are presented in colored cells. The concentrations of Arsenic (As) exceed the ZBT
threshold value for SG6 and SG7. Both latter samples are from the same side of the harbor (the left breakwater).
These high concentrations of Arsenic could be related to the important handling of sulfur in Gabes harbor, which
is one of the chemical elements that comes in conjunction with Arsenic. The Arsenopyrite, FeAsS, is the
dominant arsenic in most As-bearing natural occurrences and is therefore the mineral most commonly implicated
at surface sites (Smedley, 2002). The Copper (Cu) concentrations exceed the threshold in SBA, SC2, SG6 and
SG7. Sediments SBA present more than twice of the ZBT value. Copper concentration could be originated from
phosphate fertilizer production (ATSDR, 2004). The Cadmium (Cd) concentrations are greatly higher than ZBT
value. For SG9, the Cd concentration value exceeds 52 times the threshold value. The origin of these high
concentrations could be related to the trade exchange of natural phosphate. The Zinc (Zn) concentrations exceed
the ZBT value by a factor of 5 in the 7 studied samples. The (Zn) concentration in SG9 exceeds 3 times the ZBT
value. The high concentration is probably due to the fact that Zinc is naturally occurring in phosphogypsum
(Azabou et al., 2007). For all samples, metals’ concentrations exceed of threshold values. This makes these
sediments considered as very polluted soils and rich with toxic compounds. This contamination seems to be
principally related to the industrial wastes that are discharged directly to the sea. These wastes are mainly
phosphogypsum, which is a waste by-product of the phosphate fertilizer industry. The high contamination level
of Gabes sediments make their dumping in the sea strictly prohibited. For all these reasons, a treatment of Gabes
’ marine sediments after their dredging is necessary.

Table 4. Concentration of heavy metals in Gabes’ marine sediments and ratio to ZBT threshold (*: Maximum concentration).
Metal Unit SBA SC2 SG5 SG6 SG7 SG9 SG10 ZBT
*
Arsenic (As) mg/kg 17.8 (0.6) 25.6 (0.9) 10.2 (0.4) 36.2 (1.3) 38.6 (1.3) 26.3 (0.9) 8.1 (0.3) 29
*
Copper (Cu) mg/kg 166.0 (2.7) 63.1 (1.1) 19.6 (0.3) 77.3 (1.3) 79.5 (1.3) 56.4 (0.9) 12.8 (0.2) 60
*
Nickel (Ni) mg/kg 22.5 (0.5) 27.2 (0.6) 11.0 (0.2) 44.6 (1.0) 55.9 (1.3) 46.2 (1.0) 8.8 (0.2) 45
*
Lead (Pb) mg/kg 159.0 (1.5) 32.7 (0.3) 9.49 (0.1) 39.5 (0.4) 28.2 (0.3) 18.9 (0.2) 7.6 (0.1) 110
*
Zinc (Zn) mg/kg 490 (1.3) 580 (1.6) 225 (0.6) 721 (2.0) 1020 (2.8) 1240 (3.4) 192 (0.5) 365
*
Mercury (Hg) mg/kg 0.25 (0.2) 0.29 (0.2) 0.16 (0.1) 0.51 (0.4) 0.22 (0.2) 0.17 (0.1) 0.15 (0.1) 1.2
*
Cadmium (Cd) mg/kg 50.7 (12.7) 81.3 (20.3) 37.5 (9.4) 139.0 (34.8) 183.0 (45.8) 208.0 (52.0) 35.5 (8.8) 4
*
Chromium (Cr) mg/kg 118.0 (1.0) 178.0 (1.5) 57.9 (0.5) 234.0 (2.0) 221.0 (1.8) 176.0 (1.5) 38.2 (0.3) 120

6. CONCLUSION

This study has investigated some physical characteristics and concentrations of heavy metals in seven different
located samples of sediments from Gabes harbor. Results showed that these sediments present a medium fine
fraction, silty sands particles and important organic matter content (5.23 - 18.40 %). Results of chemical
characterization, revealed very high concentration of heavy metals in Gabes’ sediments, especially for Cadmium,
Zinc, Copper and Chromium. Compared to Netherlands threshold (ZBT), concentration of all heavy metals,
except Mercury, exceeds in different samples the regulatory limit. These results make the dispersion, of dredged
marine Gabes’ sediments, forbidden. Consequently, a treatment of these dredged materials is required.
According to the high concentration of heavy metals in sediments, a process from the categories of removal or
immobilization is necessary, such as chemical or electrokinetic treatments. A study on the organic pollutants in
Gabes’ sediments is in progress in order to decide the entire treatment process.

534
REFERENCES

Acquavita, A., Falomo, J., Predonzani, S., Tamberlich, F., Bettoso, N., & Mattassi, G. (2014). The PAH level, distribution
and composition in surface sediments from a Mediterranean Lagoon: The Marano and Grado Lagoon (Northern
Adriatic Sea, Italy). Marine Pollution Bulletin, 1–8. doi:10.1016/[Link].2014.01.041
Antizar-Ladislao, B. (2008). Environmental levels, toxicity and human exposure to tributyltin (TBT)-contaminated marine
environment. a review. b_antizar@[Link]. Environment International, 34(2), 292–308.
doi:10.1016/[Link].2007.09.005
ATSDR Agency for Toxic Substances and Disease Registry. (2004), TOXICOLOGICAL PROFILE FOR COPPER, p. 314.
Azabou, S., Mechichi, T., & Sayadi, S. (2007). Zinc precipitation by heavy-metal tolerant sulfate-reducing bacteria enriched
on phosphogypsum as a sulfate source. Minerals Engineering, 20(2), 173–178. doi:10.1016/[Link].2006.08.008
Chiavola, A., Baciocchi, R., & Gavasci, R. (2010). Biological treatment of PAH-contaminated sediments in a Sequencing
Batch Reactor. Journal of Hazardous Materials, 184(1-3), 97–104. doi:10.1016/[Link].2010.08.010
CJB Environnement INC, Environmental assessment and management EAM. (2007), Etude d’impact sur l’environnement du
dragage du complexe portuaire La Goulette-Radès, Office de la Marine Marchande et des ports, p. 109.
Clément, B., Vaille, G., Moretto, R., Vernus, E., & Abdelghafour, M. (2010). Effects of a physico-chemical treatment of a
dredged sediment on its ecotoxicity after discharge in laboratory gravel pit microcosms. Journal of Hazardous
Materials, 175(1-3), 205–15. doi:10.1016/[Link].2009.09.150
DGE - Dutch-Gerlan Exchange. (2007) Status of ecotoxicological assessment of sediment and dredged material in Germany
and the Netherlands with a short description of the situation in Belgium, France and grat Britain.p 104
Hloušková, V., Lanková, D., Kalachová, K., Hrádková, P., Poustka, J., Hajšlová, J., & Pulkrabová, J. (2014). Brominated
flame retardants and perfluoroalkyl substances in sediments from the Czech aquatic ecosystem. The Science of the
Total Environment, 470-471(May 2013), 407–16. doi:10.1016/[Link].2013.09.074
Junakova, N., & Balintova, M. (2012). Assessment of Nutrient Concentration in Reservoir Bottom Sediments. Procedia
Engineering, 42(August), 165–170. doi:10.1016/[Link].2012.07.407
Lafhaj Z., and Husson B. 2011, Foreword of the European. Journal of Environmental and Civil Engineering, Vol. 15, No. 2,
p. 302.4
Lafhaj Z., Adeline, S., Laurent, C. S., Yoann, C., Truc, H. T., Benoît, L. G., & Federico, A. (2007). The use of the Novosol
process for the treatment of polluted marine sediment. Journal of Hazardous Materials, 148(3), 606–12.
doi:10.1016/[Link].2007.03.029
Masciandaro, G., Di Biase, A., Macci, C., Peruzzi, E., Iannelli, R., & Doni, S. (2014). Phytoremediation of dredged marine
sediment: Monitoring of chemical and biochemical processes contributing to sediment reclamation. Journal of
Environmental Management, 134, 166–74. doi:10.1016/[Link].2013.12.028
Nakata, H., Uehara, K., Goto, Y., Fukumura, M., Shimasaki, H., Takikawa, K., & Miyawaki, T. (2014). Polycyclic aromatic
hydrocarbons in oysters and sediments from the Yatsushiro Sea, Japan: comparison of potential risks among PAHs,
dioxins and dioxin-like compounds in benthic organisms. Ecotoxicology and Environmental Safety, 99, 61–8.
doi:10.1016/[Link].2013.10.005
Nouira, T., Risso, C., Chouba, L., Budzinski, H., & Boussetta, H. (2013). Polychlorinated biphenyls (PCBs) and
Polybrominated Diphenyl Ethers (PBDEs) in surface sediments from Monastir Bay (Tunisia, Central Mediterranean):
occurrence, distribution and seasonal variations. Chemosphere, 93(3), 487–93.
doi:10.1016/[Link].2013.06.017
NF EN 12879, (2000) Characterization of sludges - Determination of the loss on ignition of dry mass.
Rausa, R., Mascolo, G., & Bassetti, A. (1999). Thermal treatment of sediments as a function of temperature and reacting
atmosphere. Journal of Analytical and Applied Pyrolysis, 52, 115–135.
Romero M., A. Andre´s, Alonso R., Viguri J., Ma. Rinco´n J. (2008) Sintering behaviour of ceramic bodies from
contaminated marine sediments, Ceramics International 34, ,1917–1924.
Röper H., Netzband A. (2011) Assessment Criteria for Dredged Material with special focus on the North Sea Region. p. 36
Rozas, F., & Castellote, M. (2012). Electrokinetic remediation of dredged sediments polluted with heavy metals with
different enhancing electrolytes. Electrochimica Acta, 86, 102–109. doi:10.1016/[Link].2012.03.068
Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and distribution of arsenic in natural waters.
[Link]. 17, 517 – 568.
Tech, T., & Averett, D. E. (1994). Options for Treatment and Disposal of Contaminated Sediments from New York / New
Jersey Harbor (p. 156).
XP CEN ISO/TS 17892-1 (2005). Geotechnical investigation and testing - Laboratory testing of soil - Part 1: determination
of water content -Reconnaissance et essais géotechniques.
Yu, G., Lei, H., Bai, T., Li, Z., Yu, Q., & Song, X. (2009). In-situ stabilisation followed by ex-situ composting for treatment
and disposal of heavy metals polluted sediments. Journal of Environmental Sciences, 21(7), 877–883.
doi:10.1016/S1001-0742(08)62357-8
Zhang, H., & Shan, B. (2008). Historical records of heavy metal accumulation in sediments and the relationship with
agricultural intensification in the Yangtze-Huaihe region, China. The Science of the Total Environment, 399(1-3), 113–
20. doi:10.1016/[Link].2008.03.036
Zhao, L., Hou, H., Iwasaki, K., Terada, A., & Hosomi, M. (2012). Utilization of recycled charcoal as a thermal source and
adsorbent for the treatment of PCDD/Fs contaminated sediment. Journal of Hazardous Materials, 225-226, 182–9.
doi:10.1016/[Link].2012.05.018

MSE wall foundation composed of granular soil mattresses reinforced with

geogrids: the Herb Grey Parkway (Windsor, Canada) case study

Fondation d’un mur en terre armée composée par un matelas granulaire renforcé par des
géogrilles : le cas d’étude du Herb Gray Parkway (Windsor, Canada)

P. Rimoldi1 and A. Grimod*2

1
Officine Maccaferri S.p.A., Zola Predosa (Bo), Italy
2
Maccaferri France S.A., Valence (26), France
*Auteur correspondant

ABSTRACT– Several researches have shown that geosynthetics can be used in the construction of load transfer
platforms. The purpose of these elements is to provide reinforcement, separation, drainage or filtration. This
paper presents an innovative approach to design load transfer platforms on weak soils for situations where a
uniformly distributed surcharge is applied on a rectangular area. The design method adopted was developed
based on the theory of static load distribution in unreinforced and reinforced cohesive soil, while the multi-layer
geogrid design was based on a specific load distribution model which allowed the working strain level and
tensile forces in reinforcement to be calculated and compared with set design values. The Herb Grey Parkway
case study will be presented. Different geogrid reinforced granular mattresses (RGM) were designed and
installed in order to transfer and reduce the applied load associated with the mechanical stabilized earth wall
(MSE wall) onto soft silty clay/clayey silt foundation soils with low undrained shear strength.

RÉSUMÉ –Plusieurs études ont démontrées que les géosynthétiques peuvent être utilisés pour la construction de
plateformes pour le transfert des charges. La fonction de ces éléments polymériques est de pouvoir renforcer,
séparer, drainer ou filtrer. Cet article présente une méthodologie novatrice pour le dimensionnement des
plateformes dans le transfert des charges basées sur des terrains meubles, et dans les conditions de charges
distribuées appliquées sur une surface rectangulaire. La méthode de calcul adoptée a été développée sur la
théorie de la répartition de la charge statique dans un sol cohérent non renforcé et renforcé. Les géogrilles ont été
ensuite dimensionnés sur la base d’un model spécifique de répartition des charges, qui permet le calcul de la
déformation et de la résistance à la traction de la géogrille afin de pouvoir la comparée avec les valeurs défini
dans le projet. L’Etude de Cas du Herb Grey Parkway sera présenté. Dans ce projet plusieurs matelas granulaires
renforcés (RGM) ont été dimensionnés et réalisés comme fondations de mur en terre armée basés sur des sols
limoneux/argileux ayant une cohésion non drainée et donc une faible résistance au cisaillement.

1. INTRODUCTION

The construction of a large transportation project at the Herb Grey Parkway (Ontario, Canada) resulted in the
installation of numerous mechanically stabilized earth (MSE) walls on soft glacial till soils. The installation of
MSE walls on a glacial deposit of silty clay/clayey silt soils required the design and construction of gravely sand
mattresses, reinforced with multiple layers of polyester geogrids. The purpose of the reinforced granular mattress
(RGM) was to transfer and reduce the applied loads associated with the MSE wall onto soft silty clay/ clayey silt
foundation soils with low undrained shear [Link] soft soils were also subjected to heavy loads associated
with pieces of large construction equipment. For this specific project, an innovative methodology has been
usedto design granular foundations reinforced with wovenand extrudedbiaxial geogrids according to the high
loads applied on the structures: 110 t of crawler [Link] used as reinforcement has been shown to
fall within two main mechanisms: tensioned membrane and lateral base course restraint. Reinforcement design
involves the assessment of a load and its distribution to the subgrade. While the tensioned membrane mechanism
can be produced either by geotextiles and geogrids, base course lateral restraint is mainly produced by geogrids
through an interlocking mechanism. In parking decks, container storage areas, construction sites and other
situations where a load transfer platform is required, traffic is randomly distributed; therefore, the number of
passages of a wheel on any given point is very limited. Moreover, in these areas vehicle speeds are very low;
hence, in these conditions even the wheel load can be treated as a static load on a rectangular area, which greatly
simplifies both theory and design [Link] are two typical soil conditions that can produce inadequate

537
bearing capacity for a parking deck or a soil platform: soft silty/clayey soils and loose sand. The present paper
will concentrate only on soil decks/ platforms on soft clay subgrade, characterized by the following geotechnical
parameters in undrained conditions: φ = φu = 0 and c = cu.

2. SOFT CLAY SUBGRADE

When a vertical load is applied to a load transfer platform involving a gravel aggregate – clay subgrade system,
horizontal and vertical stresses are generated. The horizontal stresses in the aggregate result in outward shear
stresses on the surface of the subgrade. These shear stresses may reduce the bearing capacity of the clay to as
little as one half the value for purely vertical loading. If a reinforcing geosynthetic is present, these shear stresses
can possibly be carried by the reinforcement (depending on interface friction), allowing the full bearing capacity
of the clay to be mobilized (Rimold et al., 2013).

There are three different reinforcing mechanisms which a geosynthetic is commonly believed to provide: the first
mechanism is that of membrane support or membrane reinforcement (Giroud and Noiray, 1981); the second
reinforcing mechanism is shear reinforcement. Shear stresses exist at the base of a gravel layer under loading;
and, the third type of reinforcing mechanism is lateral confinement, which is related to the shear reinforcing
mechanism.

In 1980 Barenberg (Barenberg, 1980) showed that when a load is applied to a soil-geotextile-aggregate
system, the normal stress transmitted to the subgrade may exceed the allowable stress. As the rut deepens, the
geotextile deforms and takes on a permanent tension, and a portion of the load is then carried by the geotextile.
Rutting continues to deepen with an increase in the geotextile tension until the normal stress transmitted to the
subgrade is equal to the permissible subgrade stress. At this point, the system becomes stable and no further
rutting occurs. Barenberg (1980) recommended that a permissible stress on the subgrade, σ, be adjusted to
compensate for the effect of the geotextile modulus on the failure criterion:

σ = A · π · cu (1)
Where: A is a coefficient related to the confining effects of the geotextile on the soil (A equals 1 when no
geotextile is used and varies up to 2 when geotextile is used) and Cu is the undrained cohesion of the soil.

Several researches (Giroud and Noiray, 1981; Barenberg, 1980; Steward et al., 1977) defined that the
increase in the bearing capacity factor for a gravel deck with reinforcement is directly related to the interaction
of shear stress and normal stress existing at the surface of the subgrade. If shear stress is absent, then the full
value of the bearing capacity (the maximum normal stress) can be realized. If shear stress is then added to the
loading, a reduction in bearing capacity occurs. With the maximum shear stress of the undrained shear strength
present, the bearing capacity factor reduces to approximately one half of the value for purely vertical loading.
Thus, if reinforcement is introduced, these shear stresses can be carried by the geosynthetic, which is put into
tension, and the full bearing capacity of the clay can be mobilized.

Summarizing, we can assume that the bearing capacity of an unreinforced platform on a soft clay subgrade is:

qu = π cu (2)

While we can assume that the bearing capacity of a geosynthetic reinforced platform on a soft clay subgrade
is:

qu = 2 π cu (3)

3. STATIC METHOD FOR RECTANGULAR LOADED AREA

The static method is based on the distribution of pressures generated at the surface by a rectangular loaded area
through a geogrid reinforced layer. The design method ensures that the pressure at the top of subgrade is less

538
than the allowable bearing pressure of the subgrade soil divided by a proper Factor of [Link] technique
assumes that the vertical pressures are distributed through the platform soil layer according to Boussinesq theory
for rectangular loaded area (Facciorussoet al., 2011). As shown in figure 1, if a vertical pressure q is applied,
evenly distributed on a rectangular area, to a homogeneous elastic half space, since the overburden stress γh is
constant, the application of the pressure q generates equilibrium conditions in every point of the medium. Hence
the vertical stress is independent from the medium characteristics. The Boussinesq theory provides the stress
components in every point along the vertical line passing by one corner of the rectangle.

Figure 1. Scheme for Boussinesq Equation (left), and scheme for calculating the stress along the vertical passing by the point
M (right)

It is possible to calculate the stress along a vertical line from a point inside the rectangular loaded area by
dividing the whole area into 4 rectangles (figure1-right). By applying the principle of superposition, the total
stress at a point will be the sum of the stresses generated by the rectangles 1, 2, 3, 4.

Figure 2 shows the scheme of the load distribution for the calculation of the bearing capacity.

Figure [Link] for bearing capacity calculation

The required foundation thickness (h) can be plotted versus the undrained cohesion (Cu) of the subgrade for both
reinforced and unreinforced soil (figure3). If a geogrid is introduced in the system, the thicknesses of the
foundation decrease dramatically.

Figure [Link] plot of the required platform thickness h vs the undrained cohesion cu of the subgrade considering an
reinforced (red line) and unreinforced (blue line) foundation

539
3. DESIGN OF THE GEOGRID REINFORCEMENT

The design methods for paved and unpaved roads on soft soil usually assume that the road base is stabilized
with just one geogrid layer, but the actual required geogrids needs to be furtherly designed.
Geogrids provide the following stabilizing mechanisms:
 base course lateral restrain mechanism for horizontal stresses generated by the soil self weight;
 base course lateral restrain mechanism for horizontal stresses generated by wheels loading;
 tensioned membrane mechanism at the base or subbase – subgrade interface.
Each of these three mechanisms produce tensile forces in the geogrid layers.
The dynamic / cyclic effect of wheel loading is always considered in the presently available methods for
paved or unpaved road design (AASHTO method, Giroud – Han method, Leng - Gabr method, etc.): therefore
the thickness of base and / or subbase afforded by such methods is already appropriate for providing the
structural capacity of the road to resist the design number of wheel passages for the whole design life of the road.
When the applied load is static (like in case of an RGM below a retaining wall) or quasi-static (like in case of
a crawler crane), then the required RGM thickness can be calculated with the Static method, above illustrated, to
prevent bearing [Link] the RGM thickness, by considering separately the effect of the static loads (applied
static or quasi-static loads, soil self weight and tensioned membrane mechanism) and the instant effect of wheel
load, it is then possible to calculate the distribution of the horizontal tensile forces in the whole road structure
and the overall tensile forces generated in each layer of geogrids, and then to select the appropriate geogrid for
each layer based on a limit state criterion.
When we are dealing with paved or unpaved roads, the limit state criterion cannot be the failure (which is
anyway prevented by the calculation of the thickness according to AASHTO method, Giroud – Han method,
Leng - Gabr method, Static method, etc.), but rather the operating condition criterion, that is the deformations
shall be [Link] achieve this goal, it is necessary to limit the geogrid strain. Theory and practical experience
suggests that geogrid strain should be limited to a maximum of 5 %. Depending upon the function of the
platform, lower strain values can be selected. For important or critical structures, the geogrid strain should be
limited to 2 %. For less critical structures, or when the design conditions afford slightly larger deformations,
geogrid strain values between 3 – 5 % can be [Link], a practical way to apply the mentioned limit strain
criterion is to limit the short term tensile strength of geogrids, as measured in a wide width tensile test according
to EN ISO 10319 standard.
When the platform is subject to long term loads, the geogrid strengths for the various strains can be obtained
from isochronous curves from creep testing.
The tensile forces produced on the geogrid layers by the three active mechanisms must now be defined. A
model consisting of four soil layers and any number of reinforcement layers has been developed for the geogrid
design (figure 4):asphalt course (AC);base course (BC);subbase course (SB); and subgrade (SG).
The model assumes that the load is applied as a uniform vertical pressure σv0 = p on a rectangular area with
half length L and half width B; this load spreads in the 3 layers of the platform structure (AC, BC and SB)
according to their load spreading angles α1, α2, α3.

Figure [Link] scheme of the 4-layers model and of the first layer of geogrid

540
The model affords the calculation of the tensile force in each geogrid, produced by the following mechanism:

 Force due to the horizontal soil trust:

(4)

Where: Tzi is the tensile force generated in the i-geogrid in the base course;Z is the reference depth; 1 is
the unit weight of the asphalt (1); 2is the unit weight of base (2); and, K is the coefficient of thrust.

 Force due to the horizontal stresses generated by uniform loads:

(5)

Where: Tpi is the tensile force generated in the i-geogrid in the base course by the load; and is the
horizontal stress
 Forces due to membrane mechanism at the interface with the subgrade: the first geogrid layer (at the
interface with the sub-grade) is the most deformable one due to the compaction of the first layer of soil at
the top of the geogrid (figure 4). The upper layers of geogrids are less subjected to vertical displacements.

The tension in the reinforcement is determined from the following equation (Giroud et al. 1990):

Tmj = WTCj. Ω.rf

(6)
where: Ω is the dimensionless factor function of reinforcement strain; WTCis the uniform vertical load,
which is a function of the load cone volume below the wheel, fill density, wheel load, subgrade reaction
R, and area of the reinforcement layer.

4. CASE STUDY: THE HERB GREY PARKWAY (WINDSOR, CANADA)

4.1 Overview of the project

The Herb Grey Parkway project started in 2012 with the purpose to build 11km of new highway connecting
southwestern Ontario (Canada) with the USA though a new bridge. The owner of the project, Windsor Essex
Mobility Group, assigned the construction of the overall project to Parkway Infrastructure Contractors that was
able to finish the project at the end of 2014. The total value of the project was estimated in 1.4 billion
CAN$.Two concrete facing MSE walls were built to support precast concrete girders, which were installed to
support the 6 tunnels for a total length of approx. 220 m over a portion of a 4 lane highway (Fig. 5). The pressure
from the 7.8 m high MSE wall to the foundation was estimated by designers at 160 kPa.
According to the geotechnical analysis, the in-situ silty/clay soil (figure5) presented the following:
 CBR = from 1 to 2.5%
 Undrained shear strength = 42 kPa

541
Figure [Link]-situ soil and cross section for RGM

4.2 Design of the reinforced granular mattress foundation for MSE wall

Using the methods described above, the first step in the design of the RGM below the retaining wall (figure 5)
was to back-calculate a Factor of Safety against bearing failure as the thickness of the RGM had been
predetermined. The Factor of Safety was back-calculated to be [Link] a thickness of 1.5 m, the geogrid was
designed for 3 layers set within a mattress of compacted, well graded, crushed gravelly sand (Ɣ = 21 kN/m3; φ =
35º). The layers of geogrid were positioned as follows: 0.5 m below top of RGM; 0.75 m below top of RGM
and 1.25 m below top of RGM. The positioning of the geogrid was dictated by the contractor’s construction
operations. A maximum strain of 2 % was set as the design strain in order to limit deformations under the MSE
walls. A load spread angle of 45º was used for the load distribution through the granular mattress.
The tension in the base layer of geogrid was calculated to be 8.53 kN/m at 1 % strain. The tension in the mid
layer was calculated to be 5.07 kN/m at 1 % strain. The upper layer of geogrid had a tensile load of 16.37 kN/m
at 1 % strain.
Woven polyester PVC coated geogrids were selected for use. Thanks to the polyester core, the geogrid can reach
high tensile strength, while the PVC coating gives the protection increasing the lifespan of the reinforcement.
The geogrid was installed with the machine direction oriented perpendicular to the centre line of the roadway, as
shown in figure 6.

Figure [Link] foundation with multiple layers of woven geogrid

542
4.3 Load transfer platform for cranes

Where roadway subgrade soils had an undrained shear strength of less than 75 kPa (CBR = 2.5 %), geosynthetic
reinforcement was designed and installed to support construction traffic loading and permanent service traffic.
For construction traffic, a single layer of extruded biaxial geogrid was used, placed at the subgrade/subbase
interface.
Sections of the roadway would also experience loading from crawler cranes. These crawler cranes operated in
pairs to erect precast girders used for bridges and tunnel sections. The weight of the largest crane (110 tonnes)
plus the maximum lift load resulted in an applied pressure transmitted to through the crawler track of 140 kPa.
This load was applied over an area of 0.91 m x 5.44 m. The edge to edge separation distance between the tracks
was 3.6 m. As the movement of the crane is very slow, it is possible to treat this load as a static load over a
rectangular area.
The crane load was first used to calculate the thickness of granular soil required to prevent bearing failure using
a Factor of Safety of 1.5. Granular thicknesses were calculated for the range of CBR values from 1 % to 2 % at
0.5 % increments. A maximum thickness of 0.85 m was calculated for CBR = 1 % (Fig. 7). For CBR=2 %, the
thickness of granular material required was 0.6 m.
Once these values were calculated, the number of layers of geogrid required and the associated strengths were
calculated. Unlike MSE wall applications, the track load is not confined in a direction along an axis, like a strip
footing. As extruded polypropylene biaxial geogrids were used for construction traffic purposes, it was decided
that the additional layers of geogrid required for crane loads would also be extruded polypropylene. A check
performed for CBR = 2.5 % and the proposed construction granular thickness of 425 mm with a single layer of
geogrid found that the construction traffic reinforcement was suitable for the additional use as a crane load
transfer platform.

Figure [Link] arrangement for CBR=1%

For CBR = 1 - 1.5 %, four layers of bi-axial extruded geogrid were used (figure 8). The strains were limited to a
maximum of 3 %, with a maximum single layer strength requirement of 11.5 kN/m. The vertical spacing on the
geogrids was approximately 200 mm.

For CBR = 2 %, two layers of bi-axial extruded geogrid were required. The geogrid layers were positioned 425
mm apart. The maximum tensile strength at the subgrade/subbase interface was 13.5 kN/m at 3 % strain.

Figures 8 shows the RGM for crawler cranes.

Figure [Link] crawler crane used to install girders and the extruded biaxial geogrids for the RGM

543
5. CONCLUSION

A design method has been presented for geogrid reinforcement of load transfer platforms on weak soils for
situations where a uniformly distributed surcharge on a rectangular area is to be transferred to a weak subgrade.
This design method was developed based on the theory of static load distribution in unreinforced and
reinforced cohesive soil, while the multi layer geogrid design was based on a specific load distribution model
which allowed the working strain level and tensile forces in reinforcement to be calculated and compared with
set design values.
Two examples were presented where the design method utilized different types of loads and different types
of geogrid.

REFERENCES

Barenberg, E.J. (1980). Design procedures for soil-fabric-aggregate systems with Mirafi 500X [Link] Engineering
[Link] of Civil Engineering, University of Illinois.
Bender, D.A. and Barenberg, E.J. (1978).Design and behaviour of Soil-Fabric-Aggregate [Link] Research
Record 671, [Link] Research Council, Washington DC, USA.
Facciorusso, J, Madiai, C., Vannucchi, G. (2011). Pressioni di contatto e diffusione delle tensioni in un semispazio
[Link] di Geotecnica. Università di Firenze.
Giroud, J.P., Bonaparte, R., Beech, J.F., and Gross, B.A. (1990). Design of soil layer-geosynthetic systems overlying
[Link] and Geomembrane, Elsevier, 9(1).
Giroud, J.P. and Noiray, L. (1981).Design of geotextile reinforced unpaved [Link] of Geotechnical Engineering,
ASCE, 107, 1233-1254.
Hudec, P. P. (1998). Geology and Geotechnical Properties of Glacial Soils in [Link] Paper (42), 225-236.
Rimoldi, P. and Simons, M.J. (2013).Geosynthetic Reinforced Granular Soil Mattresses used as Foundation Support for
Mechanically Stabilised Earth [Link] ID 623. Montreal Sept. 29th-Oct. 3rd, 2013..
Rodin, S. (1965). Ability of a clay fill to support construction [Link] of Terramechanics, 2, 51-68.
Steward, J., Williamson, R., and Mohney, J. (1977).Guidelines for the use of fabrics in construction and maintenance of low-
volume [Link] Department of Agriculture, Forest Service. Portland, Oregon. Also published as Report [Link]-TS-
78-205 (1978).

Influence of Earth Pressure Cells Installation Methods on Measurement

Accuracy
Influence de la mis en place des cellules de pression sur la précision des mesures
ALI ASGHAR MIRGHASEMI and DANIAL SAFAIFARD
School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran.

ABSTRACT -Differences in stiffness between zones in Core-type rock-fill dams causes arching-actions which
occasionally result in hydraulic fracturing. In order to determine stress redistributions occurring within the dam,
earth pressure cells are installed inside the dams. However the accuracy of measured pressure is often inadequate
due to the various influencing factors such as large differences between the stiffness of pressure cells and the
surrounding media, mechanical interaction between the earth pressure cell and the surrounding fill material, and
embedding methods of the earth pressure cells. The main reason of this problem is due to Earth pressure cells
installation procedure. Therefore different installation methods are suggested to increase accuracy of measured
pressure. In this paper three installation methods are evaluated. For determination of earth pressure cells
accuracy three methods have been used. In one method, at the end of construction, the vertical and horizontal
stresses at the center of clay core can be considered as principal stresses therefore Using Mohr circles, stresses at
45 degree planes have been calculated, and the results compared with the stresses measured in those planes. In
another method with the known stresses measured by means of horizontal, vertical and 45 degree upstream cells,
stresses at 45 degree downstream plane have been calculated, and the result compared with the stresses measured
in 45 degree downstream cells. In the last method calculated vertical pressure is compared with measured
vertical pressure obtain from earth pressure cells .In this paper earth pressure cells installation methods are
evaluated and it is understood however third installation method increases the accuracy of measured pressure but
still it has some defects.

RÉSUMÉ–En raison de la différence des rigidités au sein des barrages du type à enrochements avec noyau la
fracturation hydraulique se produit. En vue de reproduire les redistributions de contraintes des cellules de mesure
de pressions sont installées au sein du barrage. Cependant, plusieurs facteurs influencent la précision des
mesures de pression, notamment l’interaction aussi bien entre les matériaux et les cellules de pression. Pour cette
raison, le présent article suggère une investigation sur les méthodes de mise en place des cellules de pression en
vue d’augmenter la précision des mesures. En se basant sur une répartition des contraintes basée sur l’usage du
cercle de Mohr une prédiction est faite pour les contraintes s’exerçant sur différents plans aussi bien dans les
zones amont et aval du barrage. Le degré de précision des mesures obtenues à l’aide des cellules de mesures est
comparé avec les prédictions.

1. INTRODUCTION

One of the most common causes of central core-type rock-fill dam failure is hydraulic fracturing due to arch-
actions. Arch-actions are caused by differences in stiffness between zones.
The assessment of the variation of total earth pressure is one of the most important parameters in the
evaluation of the safety of geotechnical structure. In order to determine stress redistribution occurring within the
dam, earth pressure cells are installed inside the dams. This instrument has a less accuracy in comparison with
other instruments since the reliability of earth pressure measurements is significantly reduced by different factors
such as: placement effects, size of cells, temperature, calibration of cells, thickness of cells, and etc.
Dunnicliff (1993) discusses the different parameters influence the earth pressure accuracy and gives
suggestion to minimize those effects. Dibiagio (2006) believe that cells installations conditions effects the
accuracy of cell measurement. Dibiago (2006) studied the effect of difference factors on instrument readings
were placed in embankment and finally suggested an installation method to minimize the effect of these
parameters.
Mirghasemi (2006) described the instrumentation scheme of Karkheh embankment dam and discussed the
uncertainties found in measurement of earth pressure cells in the clay core. Mirghasemi (2006) understood that
no consistent data was achieved from earth pressure cells. Matsuu (2008) introduced a new installation
procedure to reduce the installation effects on earth pressure cell measurements.
In a research project the data measured from earth pressure cells installed in three different Iranian
embankment dams were used to evaluate the effect of earth pressure cells installation methods on the internal
consistency of the pressure cells.

545
2. OVERVIEW OF THE SELECTED EMBANKMENT DAMS

Three embankment dams namely Cheraghveys , Gotvand and Karkheh have been chosen for this research. In the
following sections brief description of these dams and their earth pressure cells installation methods are
presented.

2.1 Karkheh dam

Karkheh dam is a central core zoned embankment dam with an embankment volume of 32 million cubic meters.
The height of this dam is 127 m and its length is 3030 m. A total of 102 clusters, each of five earth pressure cells
have been installed to determine the total stresses in embankment at various locations.
The installation method includes excavation of trenches as a conventional installation method of earth
pressure cells. The trench is excavated to make holes as installation pockets. The minimum length and width of
pockets should be six times and three times of cell's diameter, respectively (Figure 3). In order to avoid any
damage, over 5mm grain soils must be withdrawn from bottom of the pocket by hand. Then earth pressure cells
are installed into pockets. Excess loads by heavy machines and large stones are likely to cause damage on the
installed earth pressure cells. Hence, they are embedded in small grain size soils which are compacted using
light-weighted machines in order to solve this issue. Therefore the fill stiffness between embedded and
surrounding is different.

2.2 Gotvand dam

Gotvand dam is central core zoned embankment dam on the Karun River in Khuzestan province, Iran. The
height of this dam is 180 m and its crest length is 760 m. the number of earth pressure cells that have been
installed in this dam is 409.
In order to prevent possible rotation of the pressure cells, new approaches are adopted to install earth pressure
cells in Gotvand dam. A specific type of blades with complementary tools is used to groove the regions
compressed by heavy compactors. The height and width of those blades are 30 cm. Then cells are embedded into
the trench. (Figure 1)

Figure 2embedding cell into the trench

2.3. Cheraghveys dam

Cheraghveys is a central core-type rock-fill dam with height of 64.5m. The crest length of this dam is 275 m and
the reservoir volume is 85,000,000 m3. An instrumentation system including 43 earth pressure cells is installed in
three different dam cross sections.
Earth pressure installation method is based on new installation procedure that introduced by Matsuu (2008).
In which installation method improved by not using light-weight compactors on embedded sections and not
using small grain size soil in the embedded section. The specifications of new method are shown in Figure 2.

546
Figure 3The new installation method (after Matsuu , 2008)

3. EVALUATION METHODS OF EARTH PRESSURE CELLS

In this paper three different methods are used for evaluation of measured data. In some sections, earth pressure
cells are installed in quintuplet pressure cells cluster as shown in Figure 3. The directions of cells as shown as in
Figure 3 are vertical parallel to dam axis, 45 degree upstream, horizontal, 45 degree downstream and vertical
normal to dam axis, respectively from right to left.
The pressure cells in one cluster can be considered as stresses at different directions for a single point in the
embankment because they are close enough to each other.

Figure 4Quintuplet pressure cells cluster

3.1. Method 1

Based on Mohr circle the summation of vertical stress and horizontal stress is equal to the summation of 45
degree upstream and 45 degree downstream. In this method with the known stresses measured by means of
horizontal, vertical and 45 degree downstream cells, the pressure acting on the 45 degree upstream cell can be
calculated. Then calculated pressure is compared with the measured stress of cell.

3.2. Method 2

At the end of construction and before impounding, there is no horizontal force acting on the embankment from
reservoir water pressure. Also if the dam cross section is almost symmetrical, at the center of the clay core the
maximum principal stress is near vertical. Therefore at the center of the clay core the vertical and horizontal
stresses can be assumed as principal stresses.

547
Using Mohr circle the stress oriented 45 degree to upstream is calculated and compared with the
corresponding stress measured by pressure
ressure cell. In this method the calculated pressure is equal to average of
vertical and horizontal stresses.

3.3. Method 3

In the lower portion of an embankment dam, the stress field during construction is essentially 22-dimensional if
the valley is wide and not predominately V-shaped.
V shaped. In this case, it is possible to calculate from the density and
thickness of the fill what the measured stresses should be. This procedure is commonly used to check the
performance of pressure cell installed in high embankments
embankme (Dibiagio, 2006).
Therefore in this method vertical stress is calculated from density and thickness of the fill (up to 20 meters of
fill over the instrument) and compared with vertical stress measured by means of horizontal earth pressure cell.

4. EVALUATION
VALUATION OF EARTH PRESSURE CELLS

4.1. Evaluation of Karkheh earth pressure cells

Total earth pressure cells installed in Karkheh dam are evaluated and the overall result is presented in Table 1.
As shown, 89% of cells have differences greater than the 10% between measurement and calculated stresses
using method 1. This number for method 2 and 3 are 74% and 97%, respectively.
The main reason of such great error in Karkheh dam is local arching phenomenon. It is happened due to the
differences in stiffness between the ordinary and embedded sections. And because of local arching phenomenon,
earth pressure cells measured less pressure than actual one.

Table 1 Overall evaluation result of Karkheh dam

Method 1(%) Method 2(%) Method 3(%)
Cells have error more than 10 percent 89.01 74.02 97
Cells have error more than 20 percent 63.74 45.17 92
Cells have error more than 40 percent 28.57 25.81 72
Cells have error more than 50 percent 23.08 19.36 43
Cells have error more than 80 percent 9.89 3.23 1
Cells have error more than 100 percent 8.79 3.23 0

4.2. Evaluation of Gotvand earth pressure cells

Stresses at plane 45 degree to upstream were calculated and compared with corresponding measured stress in the
same direction. Figure 4 show the stresses measured by means of 45 degree upstream earth pressure cells and its
corresponding calculated stresses obtain from method 1and 2.

Figure 5 Comparing measurement stresses with calculated stresses using method 1 and 2 in section (1
(1-3)

548
In Table 2 total earth pressure cells installed in Gotvand dam are evaluated. As shown 88/9% of cells have
the differences more than the 10% between measurement and calculated stresses using method 1. This number
for method 2 is 72.2%. And 72.5% of cells have differences greater than 10% between measured and calculated
vertical stress.
In installation procedure used in Gotvand dam, however it reduced the errors associated with rotation of
Earth Pressure Cells, but it would add another source of errors correspond to the imposed stresses. And it is due
to process of placing cells into the grooves.

Table 2overall
2 evaluation result ofGotvand dam
Method 1(%) Method 2(%) Method 3(%)
Cells have error more than 10 percent 88.9 72.2 72.5
Cells have error more than 20 percent 69.8 61.1 55.1
Cells have error more than 30 percent 58.7 22.2 33.3
Cells have error more than 40 percent 46 5.6 15.9
Cells have error more than 50 percent 30.2 5.6 4.3
Cells have error more than 80 percent 17.5 0 1.4
Cells have error more than 100 percent 17.5 0 1.4

4.3. Evaluation of Cheraghveys earth pressure cells

At Cherghveys dam most of earth pressure cells are single pressure cell. And the other pressure cells are installed
in triplet pressure cells cluster. But in triplet pressure cells cluster, the measured stresses are not in one plane
therefore method 1 and method 2 are not applicable.
Method 3 was the only method which could be used for all functioning earth pressure cells at Cheraghveys
dam. In Table 3 the overallall evaluation result is presented. As shown more than 70% of the cells have difference
greater than 10% between calculated and measured vertical stresses. Also it can be seen that no cell has error
more than 50%.

Table 5evaluation result of Cheraghveys dam

Method 3(%)
Cells have error more than 10 percent 70
Cells have error more than 20 percent 32.5
Cells have error more than 30 percent 12.5
Cells have error more than 40 percent 5
Cells have error more than 50 percent 0
Cells have error more than 80 percent 0
Cells have error more than 100 percent 0

4.4 Comparison of result in selected dams

The Comparison of Gotvand dam with Karkheh dam using results obtain from method 1 are shown in Figure 5.
It shows that installation method used in Gotvand dam not only improved the results but also reduced the
accuracy.

Figure 6comparison
comparison of Gotvand dam with Karkheh dam using results obtain from method 1

549
Figure 6 presents the influence of installation method employed for three dams on accuracy of the results
using evaluation method 3. It is obvious that results in Cheraghveys dam are more accurate than two other dams.

Figure 7comparison
comparison of three dams using results obtain from method 3

5. CONCLUSION

In this paper the effect of earth pressure cells installation method on measurement of total stresses are evaluated.
It is understood that installation method used in Gotvand dam comparing with the procedure of installation on
Karkheh dam, makes no difference nce to obtain more consistent stress data. This installation method could be
improved by new instruction for placing cells into the grooves in order to decrease imposed stresses. In
Cheraghveys results are better than other dams but they are not enough reliable.
rel
Also it is obvious that earth pressure cells installation methods have great influence on accuracy of
measurements. Therefore special care must be given to details of installation methods.

REFERENCES

DiBiagio, E. (2006), Discussion of Karkheh Dam

D Instrumentation System-Some
Some Expriences, Geotechnical News, Vol.24
No.1,March 2006, pp 35.
Dunnicliff J., Green, Gordon E. (1993), Geotechnical Instrumentation for Monitoring Field Performances, Wiley & Sons
Matsuu , M., Takahashi A. , Uchita Y. , Ohta H. (2008) “Improvement Of Earth Pressure Measuring Method In Rock Rock-Fill
Dams” international commission on large dams 76th Annual meeting, Sofoa ,Bulgaria , June.
Mirghasemi , A.A (2006) , "Karkheh Dam Instrumentation System-Some
System Some Experiences" Geotechnical N News , Vol. 24, No. 1,
32-36, March.
Mirghasemi , A. A., Shoghli, B.(2009) , "Evaluation of Earth Pressure Measurements in Three Iranian Embankment Dams "
Proc. of International Symposium on Ground Improvement Technologies and Case Histories (ISGI09), Dec. 9-12 ,
Singapore.

550
SOLS LATÉRITIQUES

LATERITIC SOILS

The Effect of Pretreatment on Index and Compaction properties of a Lateritic

Gravel

L'effet du prétraitement sur l'index et compactage propriétés d'un gravier latéritique

S.I.K Ampadu*(1), P. Ackah (2), F. Owusu-Nimo (3), F. J. F. Ayeh (4)
(1)
College of Engineering, KNUST, Kumasi Ghana
*Corresponding Author

ABSTRACT - The index properties of laterites and lateritic soils can be very sensitive to the type of
pretreatment given to the sample prior to testing. Pretreatment is defined as the method of pre-test drying and the
extent of manipulation during testing for the Atterberg limits. By their mode of formation, laterites and lateritic
soils are rich in iron and aluminium oxides. These oxides provide characteristic bonding between the soil
particles and pretreatment of these laterites and lateritic soils can lead to breakage of and/or irreversible changes
in these bonds. However, laterites and lateritic soils formed under different climatic conditions exhibit different
degrees of sensitivity to pretreatment. The Atterberg limits are useful indicators of the engineering behavior of
soils and in many cases, construction material acceptance criteria are based on them. Therefore, for a given soil,
obtaining consistent values of the Atterberg limits from laboratory testing is important in making reliable
prediction of their engineering behaviour. This study is part of a larger study that seeks to determine the
optimum testing conditions under which consistent results can be obtained from laboratory testing of laterites
and lateritic soils. Bulk samples of a lateritic soil were prepared by oven drying, air-drying and also at natural
moisture content. For each method of sample pre-drying, the Atterberg limits were determined after mixing the
soil sample for durations ranging from 5minutes to 90minutes. The paper discusses how the degrees of soil
manipulation prior to testing influences the Atterberg limits for the different methods of pre-drying.
Keywords: Lateritic soil, Atterberg limits, grading, compaction, pretreatment, oven drying, air drying, natural
moisture content, manipulation.

RESUMÉ - Les propriétés d'index des latérites et les sols latéritiques peuvent être très sensibles au type de pré-
traitement donné à l'échantillon avant le test. Le prétraitement est défini comme étant le procédé de séchage pré-
test et la mesure de la manipulation au cours de l'essai pour les limites d'Atterberg. Par leur mode de formation,
les latérites et les sols latéritiques sont riches en fer et en oxydes d'aluminium. Ces oxydes fournissent liaison
caractéristique entre les particules de sol et prétraitement de ces latérites et les sols latéritiques peuvent conduire
à la rupture des changements et / ou irréversibles dans ces obligations. Cependant, les latérites et les sols
latéritiques formés dans différentes conditions climatiques présentent différents degrés de sensibilité à un
prétraitement. Les limites d'Atterberg sont des indicateurs utiles du comportement de l'ingénierie des sols et dans
de nombreux cas, matériaux de construction critères d'acceptation sont basés sur eux. Par conséquent, pour un
sol donné, obtenir des valeurs cohérentes des limites d'Atterberg de tests de laboratoire est important pour faire
une prévision fiable de leur comportement mécanique. Cette étude fait partie d'une étude plus vaste qui vise à
déterminer les conditions de test optimales dans lesquelles des résultats cohérents peuvent être obtenus à partir
de tests de laboratoire des latérites et les sols latéritiques. Des échantillons en vrac d'un sol latéritique ont été
préparés par séchage, séchage à l'air du four et aussi à la teneur en humidité naturelle. Pour chaque méthode de
pré-séchage échantillon, les limites d'Atterberg ont été déterminées après mélange l'échantillon de sol pour des
durées allant de 5 minutes à 90 minutes. Le document explique comment les degrés de manipulation du sol avant
l'essai influe sur les limites d'Atterberg pour les différentes méthodes de pré-séchage.

1 INTRODUCTION

The index properties of a soil consist of its particle size distribution and for the fines component the consistency
limits made up of the liquid limit, the plastic limit and the shrinkage limits. These properties are measured using
disturbed samples and the test procedures completely re-mould the samples. Consequently the index properties
are independent of the condition of the soil in-situ and are not influenced by the microstructure. Yet due to their
mode of formation, for sedimentary temperate zone soils, the index properties through their use in soil
classification, not only provide an excellent guide of the properties of disturbed soils used in earthworks but they
also are a good indication of the in-situ mechanical properties such as compressibility, permeability,
compactibility, shrink-swell and shear strength.

553
For tropical residual soils, however, the formation processes, dominated by chemical weathering of the parent rock
result in a material deficient in silica, but rich in the sesquioxides of iron and aluminium and containing
substantial amounts of clay minerals. This material, known as laterites and lateritic soils has a microstructure made
up of aggregations of these sesquioxides and clay minerals “cemented” to one another as “peds’ and also to
the individual grains. Consequently, the results obtained from following conventional procedures for index
properties tests on laterites and lateritic soils depend on the extent to which the pre-test sample preparation and
the testing procedures themselves have affected the structure of the clay minerals and also have been able to de-
structure the soil. Various studies into sample preparation procedures of laterites and lateritic soils (Gidigasu
1970; Gidigasu et al., 1973; Northmore et al., 1992; Basma et al., 1994; Ababew 2005; Sunil and Krishnappa,
2012) have concluded that sample preparation techniques can influence the index and some engineering
properties of laterites to various extents. Most of the pre-test sample preparation procedures involve drying.
However, in laterites and lateritic soils, drying can lead to irreversible changes in soil properties that are reflected
in the values of the index properties. Despite these difficulties, it has been suggested (Northmore et al., 1992)
that once consistent index property values have been obtained, conventional approaches to using the index
property values to predict engineering behavior may also be applicable to residual soils. Compaction is a very
common soil improvement technique used in earthworks and in highway pavement construction. In order to
prepare the specifications for quality control during earthworks, compaction parameters are determined in the
laboratory. Again, sample preparation procedures for compaction test in the laboratory are known to affect the
maximum dry density and the corresponding optimum water content values. However depending on the climatic
condition, different lateritic soils have different sensitivity to drying and to manipulation (Lyon 1970; Ababew
2005). There have even been attempts in the past (Gidigasu 1970; Blight 1997) to develop procedures to
determine which soils that are likely to be sensitive to pretreatment.
This paper investigates how the Atterberg limit tests, the grading properties and compaction characteristics of
a local lateritic soil from the Kumasi area are affected by the method of pre-test drying and the degree of de-
structuring during testing. Samples of a local lateritic soil were prepared by air drying, oven drying and at
natural moisture content and subjected to the Atterberg limit tests using different mixing times ranging from
5minutes to 90 minutes prior to the testing. The grading was also performed on the different drying methods.
The proctor compaction test was also performed on fresh and on re-used samples that have been subjected to the
different pre-test drying methods. The results are presented and discussed.

2 METHODOLOGY

Sampling and Sample Storage

The lateritic soil for the study is designated Y-soil and was obtained from a gravel borrow pit located outside the
Kumasi metropolis. The material was being borrowed for various construction works in the Kumasi metropolis.
Part of the gravel material stockpiled at the borrow pit was loaded into a truck using a wheel loader and delivered
to the KNUST materials bins, where it was covered with black polythene bag to preserve the in-situ soil
moisture content.

Sample Pretreatment

The pre-test drying methods made up of oven drying (OD), air-drying (AD) and testing at natural moisture
content (NMC) without any pre drying were investigated. For OD, the material was drawn from the bin, spread
out on a tray and then placed in an oven maintained at a temperature of 105oC for 24 hours. The sample was
then allowed to cool and used for the test. The AD sample was prepared by drawing samples from the bin and
spreading it out in trays and leaving it open to the air in the laboratory for 24 hours. The room temperature
was in the range of 28oC to 32oC. For the NMC method, samples were drawn straight from the bin without any
pre-drying and used for testing.

2.3 Test Procedure for Atterberg Limits

The test procedure used for the Atterberg limits sought to investigate, in addition to the effect of pre-test drying,
the effect of the extent of manipulation of the soil-water mix before testing on the test results. For the Atterberg
limits, after the pre-test drying, the samples were sieved and the portion passing 425m was used for the
plasticity parameters. For these tests, portions of the sieved soil sample for each pre-test drying type was taken
and then mixed with distilled water and left overnight in an air-tight container for moisture content equilibration.
The mixed sample was then placed on a flat glass plate and mixed with water into a smooth consistent paste
using palette knives for 5 minutes. The paste was then used for the liquid limit test in the Casagrande Apparatus

554
in accordance with BS 1377: Part 2, 1990 clause 4.5. The water content for each test point was determined using
oven drying method. The procedure was then repeated except that the duration of the mixing was increased to 10,
15, 30, 45, 60 and 90 minutes.
The plastic limit test was done following the conventional thread rolling procedure given in BS 1377: Part 2,
1990 clause 5.3 using part of the sample which had been manipulated for the liquid limit test. The sample was
hand rolled on a glass plate until it started to crumble on drying at a diameter of 3mm. The same mixing duration
used for the liquid limit test was used. The linear shrinkage test was conducted in accordance with BS 1377 Part
2 1990 Test 6.5 as also using the same mixing times.

2.4 Compaction

Bulk samples were drawn from each dried soil storage sack and divided into two portions. The Standard Proctor
compaction specification was used for all tests and this involved filling a 101mm diameter mould with soil in
three equal layers, applying a compactive effort consisting of 25 blows to each layer using a 2.5 kg rammer with
a free fall of 300mm. On the first bulk sample referred to as “re-used sample” the same sample was used for
each compaction trial which consisted of increases in water content of about 3%. For the “fresh sample”, for
each compaction point, fresh sample was drawn from the sack and pre-calculated water was added and
compacted and discarded.

2.5 Test procedure for Grading

The grading was carried out for each of the pre-dried sample following the description in the test procedure in BS
1377-2 1990 Clause 9, using the wet sieving method and sodium hexametaphosphate as the dispersing agent. The
same procedure was followed for the sample after each compaction trial.

3 DISCUSSION OF RESULTS

3.1 Characteristics of Soil Investigated

The Kumasi region is in the forest zone of Ghana and has a mean annual rainfall of the order of 1400mm with a
double maxima in June/July and in September. The terrain is hilly and undulating. The site had an elevation of
the order of 280m above mean seal level. A summary of the index properties of the Y-soil used for the
investigation as measured on the NMC, AD and OD samples are shown in Table 1. The material is described as
a sandy gravel with intermediate plasticity. For purposes of this investigation, the stockpiled material is
classified as a material in its natural moisture content. However, in reality the material exposed to the weather
can be considered to have undergone some drying under its field conditions.

Table 1. Summary of Results of Investigation Properties

Pre-test Drying Atterberg Limits (%) Grading (%)
Method LL(2) LL(1) PL LS(3) PI(1) PI(2) Gravel Sand Silt Clay
NMC 40 41 21 9 20 19 68.7 10.2 7.6 13.5
AD 38 40 22 9 18 16 69.1 12.7 8.8 9.4
OD 35 39 22 8 17 13 66.2 14.4 13.7 5.7
(1)
Determined by Cone Penetrometer after 10minutes manipulation
(2)
Determined by Casagrande Method after 10 minutes manipulation
(3)
Linear Shrinkage after 10 minutes of manipulation

3.2 The Effects of Pretreatment on Atterberg Limits

The liquid limit defines the water content beyond which the soil passes from plastic to the liquid state. The liquid
limit values for all pre-test drying methods and for mixing times ranging between 5 and 90 minutes are plotted in
Figure 1 and values at conventional 10 minutes of manipulation are tabulated in Table 1. The Casagrande
method of determining the liquid limit is the definitive method specified in ASTM, but BS1377-1990 rather
prefers the cone penetrometer method. The liquid limit values determined using the cone penetrometer method
after 10 minutes of manipulation are also tabulated in Table 1 and superimposed on Figure 1 for comparison
purposes. The results suggest that the cone penetrometer gives slightly higher values and the difference increases
from 3% in NMC to 11% in OD samples. Similar results were reported on a Brazilian laterite containing only
kaolinite clay mineral (Queiroz de Carvalho, 1986) where the cone penetrometer method gave values about 10%
higher than that obtained by the Casagrande method.
The plastic limit is the water content beyond which the soil moves from a plastic to a semi-solid state. The
difficulty of obtaining consistent results using the conventional thread rolling method has prompted attempts in

555
the past to define it in terms of a cone penetration (Harison, 1998). Shrinkage limit is defined as the water content
below which there is no further volumetric reduction in the sample. In this investigation, the linear shrinkage test
which measures the percentage reduction in length of a soil dried from the liquid limit is used to represent the
shrinkage limit. The Plasticity Index, which is the difference between the liquid limit and the plastic limit is a
useful index for describing the range of water contents over which the soil remains plastic. The effects of pre-
treatment on these index parameters are shown in Figure 2 which plots the liquid limit, the plastic limit and the
linear shrinkage against the duration of manipulation for each of the three methods of pre-test drying.

3.2.1 The Effect of Pre-test Drying

The results in this investigation show a reduction in the liquid limit when the samples are pre-dried with OD
samples giving the largest reduction of 13% while AD gave only 5% reduction from the NMC value. The results
suggest a small reduction in the plastic limit though within the variability of test results this will be considered
insignificant. The implication for these effects is that NMC samples, gave the highest PI values while for normal
levels of manipulation, OD samples gave the lowest values. The literature (Sunil and Krishnappa, 2012) suggests
that pre-test drying normally resulted in a reduction in the liquid and plastic limit results though the magnitude of
reduction depended on the type of soil. The sensitivity of a soil to pre-test drying depends on the clay mineral
present and the state of hydration. Soil containing allophane and hydrated halloysite are known to show the most
sensitivity to pre-drying (Morin 1982; Newill 1961) while those containing kaolinite showed the least sensitivity.
This is because pre-test drying can lead to removal of water of hydration from these clay minerals causing
irreversible changes to the structure of the clay minerals which can results in stronger bond between the particles
of the sesquioxides making it resistant to penetration by water.
This limited study has confirmed the recommendation in both ASTM and BS 1377-1990 that pre-test drying
of any form should not be used for lateritic soils for Atterberg limit determination. However, given the need to
ensure that sample preparation procedures model as much as possible, the field condition for which the test
targets, (De Graft Johnson and Bhatia, 1969) it seems that for earthworks, the processes of winning and
stockpiling material leads to some amount of air-drying. Therefore it appears air-drying will be a more suitable
pretreatment procedure that better reflects the field condition.

Figure 1. Variation of Atterberg Limits with Pre-treatment Figure 2. Variation of Plastic Limit with Pre-treatment

3.2.2 Effects of Manipulation

The underlying assumption in the liquid and the plastic limit tests is that the tests are done on completely re-
moulded soils. The extent to which the structure of a lateritic soil is broken down during testing depends on the
amount of manipulation that the sample is subjected to during re-moulding. In this investigation the amounts of
manipulation is measured by the duration of mixing using palette knives on a glass plate. The results show that
the liquid limit increases with increasing manipulation for all pre-test drying methods. In fact there is a 10%
increase in liquid limit as manipulation time increases from 5 minutes to 90 minutes for NMC samples. For
plastic limit, whereas increasing manipulation leads to increasing plastic limit values for NMC and AD samples,
for OD increasing manipulation leads to a reduction in plastic limit.
The results obtained in this investigation show insignificant effect of manipulation on the plasticity index
values of NMC samples. The AD samples registered an initial reduction in PI and then became stable. For OD
sample however the Plasticity Index continued to increase even after 90 minutes of manipulation. The results
suggest that a stable value of Plasticity Index for NMC and AD is obtained at a manipulation time of about 45

556
minutes. However the increase of less than 5% in liquid limit (from 39.6 to 42 for NMC and from 34.1 to 36.3
for OD) resulting from the 45 minutes manipulation is considered insignificant. For plastic limit on the other
hand, the increase of 11% (from 20.5 to 23.3) and 17% (20.5 to 24.5) for AD and OD sample may be significant.
Increasing manipulation during testing breaks down the aggregations and the sesquioxide coatings producing
finer particles. This increases the specific surface area and thereby increases the surface charges and hence
increasing the amount of water absorbed, which leads to a rise in the liquid limit values.
The results in this investigation suggest that for AD and NMC the aggregation of clay size particles break up
sufficiently after 45 minutes of manipulation. However, whereas the resulting increase in liquid limit may be
considered insignificant, the effect may be significant on the plasticity index values.

3.2.3 Effect of Pretreatment on Compaction

For compaction, the effect of the level of manipulation during compaction was investigated using “fresh
samples” and “re-used samples” for each compaction trial. The results are summarized in Table 2 and plotted in
Fig. 4 and Fig. 5.

Table 2. Summary of Compaction Characteristics

Method Fresh Samples Reused Samples
MDD (Mg/m3) OMC (%) MDD (Mg/m3) OMC (%)
NMC 1.95 14.05 1.98 13.99
AD 1.98 13.96 1.99 12.75
OD 2.06 12.49 2.12 11.90

The results show that for both fresh and re-used samples, OD samples give the highest MDD and the
corresponding lowest OMC while NMC gives the lowest MDD and highest OMC with AD samples lying
somewhere between the two. The difference between OD and NMC samples is largest at water contents around
the optimum. The results also show that using fresh samples each time leads to about 2% reduction in MDD and
the optimum moisture contents are slightly higher. Conversely, reusing the samples for each compaction point
give slightly higher dry densities with corresponding lower OMC compared with that of fresh samples. This may
be attributed to the progressive breakdown of the concretionary coarse fractions under the impact of the rammer.
Though the magnitude of the difference between the maximum dry densities for the AD fresh and reused
samples is small, the results obtained were consistent with findings so far which show that oven drying gives the
highest MDD and lowest OMC while compaction at natural moisture content gives the lowest (Gidigasu 1970;
Ababaw 2005)

Figure 3. Effect of Pre-test drying on compaction characteristics of Figure 4. Effect of pre-test drying on compaction characteristics
fresh sample of re-used samples

3.2.4 Effect of Pretreatment on Grading

Figure 5 shows the particle size distribution curves for the “as-received” samples alongside the curves after
compaction. The fines content is defined as particles smaller than 75µm. The results show that for the “as-
received” samples, the difference among the grading curves for particles larger than about 75m for OD, AD
and NMC sample is very small to insignificant. However, for particles smaller than 75m, drying leads to a
reduction in the percentage passing so that for the clay sized fraction i.e. particles smaller than 2µm, whereas OD
samples had only 5.7% clay sized fraction NMC sample had 13.5%. Overall the results show that the effect of

557
oven drying is to increase the clay content by about 7%. This is due to aggregations of soil particles due to
drying as evidenced from the results of scanning electron micrographs from Sunil and Krishnappa, 2012.

Figure 5 Effect of pretreatment on grading

A comparison among the “after compaction” curves for OD, NMC and AD curves shows that the effect of
drying method on the grading is small for particles larger than about 75m. For particles smaller than 75m,
however, OD samples give the highest percentage passing of 22% while NMC and AD samples give about 15%
passing. When these are compared with the “as received” sample sieved through 4.25mm the results show that
the fines content increases from about 34.1% for un-compacted “as-received” samples to an average of about
41% for compacted samples. This shows that compaction leads to particle breakdown and increases the fines
content by about 7% (Ampadu and Arthur, 2006). This is explained by the brittleness of the concretionary
coarse particle created by oven drying (Gidigasu, 1970).

4 CONCLUSION

On the basis of this study for lateritic sandy gravel with intermediate plasticity obtained from the Kumasi
district, it can be concluded that even though pre-treatment affects the index properties, the effect is small to
insignificant. Specifically, the
1. Pre-test drying reduces the Atterberg limits
a. Oven-drying reduces the liquid limit by 13% below the value for natural water content.
b. The reduction in the plastic limit and in the linear shrinkage values is considered insignificant.
c. For lower levels of manipulation, oven drying give the lowest values of plasticity index
2. Increasing degree of manipulation increases the liquid limit for all pre-test drying methods. The effect on
the plastic limit and on the linear shrinkage limit is small and not consistent.
3. For air-dried samples and samples at the natural moisture content, a minimum manipulation time of about
45 minutes is required to sufficiently break down the aggregations to give consistent Plasticity Index
values.
4. Pre-test drying leads to small increases in Maximum Dry Density and a corresponding reduction in OMC
for both re-used and fresh samples.
a. OD samples give the highest MDD and a corresponding lowest OMC while NMC gives the lowest MDD
and highest OMC.
b. Reused sample for each compaction point give slightly higher dry densities with corresponding lower
OMC compared with that of fresh samples.
5. Pre-test drying has insignificant effect on the coarse fraction i.e. fraction larger than about 75m.
However, pre-test drying leads to a reduction in the fines content so that oven drying reduces the clay
fraction by about 7%.

ACKNOWLEDGEMENT

558
The authors are grateful to Mr. Gilbert Fiadjoe and Solomon Gidigasu of the Soil Mechanics laboratory of
KNUST and to the GeoGroup for their support.

REFERENCES

Abebaw Z. (2005) Basic Engineering Properties of Lateritic Soils Found in Nejo – Mendi Road Construction Area, Welega,
MSc thesis, Addis Ababa University, Ethiopia.
Ackah P., Ampadu S., Owusu-Nimo F., Ayeh F. (2014) The effect of pretreatment on compaction characteristics of a lateritic
gravel, XVII Congresso Braseileiro de Mecânica dos Solos e Engenharia Geotécnica., Goiania, Brazil.
Ampadu S.I.K., Arthur D.T. (2006) The Dynamic Cone Penetrometer in Compaction Verification on a Model Road
Pavement, Geotechnical Testing Journal, Volume 29, No. 1, GTJ 12306, pp 70-79, January 2006.
Basma A.A., Al Homoud A.S., Al-Tabari E.Y. (1994) Effects of methods of drying on the engineering behavior of clays. J
Appl Clay Sci 9 (3):151–164.
Blight G.E. (1997) Mechanics of Residual soils. A.A Balkema, the Netherlands.
British Standards Institution (1990) Method of test for soils for civil engineering purposes, British Standards Institution.
De Graft Johnson J.W.S., Bhatia H.S. (1969) Engineering properties of lateritic soils. Gen. Rept. Spec. Sess. Lateritic Soils.
Proc. Int. Conf. Soil Mech. Found. Eng., 7th Mexico, 2: 13-43.
Gidigasu M.D. (1970) The effect of pretreatment on the compaction characteristics of laterite soils. Build. Road Res. Inst.,
Kumasi, Ghana, Proj. Rep., SM. 6: 12pp.
Gidigasu M.D., Bani S.K. (1973) Effect of preheating on compaction characteristics of laterite: Geotechnical properties of
troublesome laterite materials. Proc. Intern. Conf. Soil Mech. Found. Eng. 8th, Moscow, Vol. 2, pp: 89-96.
Harison J.A. (1998) Using the cone penetrometer for the determination of the plastic limit of soils, Geotechnique, 38, 3, 433-
438.
Lyon Associates Inc. (1971) Laterite and lateritic soils and other problem soils of African, Kumasi Ghana.
Queiroz de Carvalho J.B. (1986) The applicability of the cone penetrometer to determine the liquid limit of lateritic soils,
Technical note, Geotechnique.
Morin W.J. (1982) Characteristics of tropical red residual soils, Proceedings of the ASCE Geotechnical Engineering
Division, Specialty Conference on Engineering and Construction in Tropical Soils, Honolulu, Hawaii 172-198.
Newill D. (1961) A laboratory investigation of two red clays from Kenya, Geotechnique 11, 4, 302-318.
Northmore K. J., Entwisle D.C., Hobbs P.R.N., Culshaw M.G., Jones L.D. (1992) Engineering geology of tropical red clay
Soils: geotechnical characterization: index properties and testing procedures, British Geological Survey Technical Report
WN/93/12
Sunil B.M., Krishnapa H. (2012) Effect of drying on the index properties of lateritic soils. Geotech Geol Eng, 30: 869-879.

Basic engineering properties of laterites in Muglad basin, Western Kordofan,

Sudan
Propriétés géotechniques de base des latérites dans le bassin de Muglad, dans l'ouest du
Kordofan, au Soudan
1 2 3
Tarig Elkhawad , Hussein Elarabi , Salah Bashier
1
[Link]. Building and Road Research Institute, University of Khartoum, Khartoum,
[Link]@[Link]
2
Associate Professor & Head Department of Geotechnical Engineering, Building and Road Research Institute,
University of Khartoum, Khartoum, Sudan. husseinelarabi@[Link]
3
Associate Professor& Dean of Faculty of Science, University of Khartoum, Khartoum,
[Link]@[Link]

ABSTRACT - This paper reports and evaluates the geotechnical properties and corrosion of laterites in study
area in western Kordofan State, as part of Muglad basin, Sudan. Field investigation include geophysical
investigation using electrical resistivity method, test pits, borehole, in situ testing Standard Penetration Test
(SPT), Dynamic probe light Test (DPL) in different locations and depths. Laboratory tests include sieve analysis,
consistency test, compaction test, California Bearing Ratio (CBR) and chemical analysis. The corrosion in soil
increases as soil resistivity decreases and the subsoil Corrosivity in the study area increases in South West
direction with depth so it is recommended that the pipelines in the area should be protected by proper protective
system to avoid any corrosive interaction between steel surface and soil environment. Laterites in study area
have been classified as Clayey Sand rich in iron oxides and contain appreciable amount of silica. According to
laboratory tests laterites are proven to be good as construction materials.

RÉSUMÉ – Ce papier rapporte et évalue les propriétés géotechniques ainsi que la corrosion des latérites dans la
zone d'étude de l'État du Kordofan, dans le bassin de Muglad au Soudan. Les investigations sur terrain
comprennent des essais géophysiques utilisant la méthode de résistivité électrique, des puits d'essai, des forages,
des essais in situ (SPT) et des essais dynamiques (DPL) dans différents endroits et à des profondeurs différentes.
Les essais de laboratoire comprennent l'analyse granulométrique, les essais de consistance, les essais de
compactage et les analyses chimiques. La corrosion dans le sol augmente quand la résistivité du sol diminue et
la corrosivité du sous-sol dans la zone d'étude augmente dans la direction du Sud-Ouest avec la profondeur de
sorte qu'il est recommandé que les pipelines de la région soient protégés par un système approprié pour éviter
toute interaction corrosive entre la surface en acier et le sol. Les Latérites dans la zone d'étude ont été classés
comme des sables argileux riches en oxyde de fer et contenant une quantité appréciable de silice. Selon les essais
de laboratoire, il a été prouvé que les latérites sont de très bons matériaux de construction.

1 INTRODUCTION

This Study investigated the engineering properties and soil corrosion along the pipelines to evaluate subsurface
soil condition and degree of corrosion in the study area which lies in the north-eastern part of the Muglad Rift
Basin, which is the largest and most important oil-producing basin of the Sudan. The Central and Southern
Sudan Interior Rift Basins evolved during Late Jurassic to Early Tertiary time in response to rifting processes.
Similar processes were active in other parts of Central Africa as well as along the eastern and western
continental margins of Africa, and caused the formation of major rift basins (Sayed, 2003).
The geologic information from the previous data and recent work show that rocks units include the
metamorphic and intrusive igneous rocks (Precambrian-Cambrian),uncomformably overlain by the Mesozoic
Nubian sandstone Group (Mesozoic-Early Tertiary) partly intruded by Tertiary volcanic (Mid Tertiary), covered
by laterite (Mid-Tertiary - Oligocene), covered by Umm Ruwaba formation (Pliocene to Pleistocene) and super
facial deposits (Pleistocene to Recent).
Laterites as highly weathered and altered residual soils formed by the in-situ weathering and decomposition
of rocks under tropical condition (Blight, 1997). Laterites are formed from the leaching of parent sedimentary
rocks, metamorphic rocks and igneous rocks, and mineralized proto-ores, which leaves the more insoluble ions,
predominantly iron and aluminum, the mineralogical and chemical compositions of laterites are dependent on
their parent rocks (Tardy, 1997).

561
2 METHODOLOGY OF INVESTIGATION

2.1 Field work

In this study the field work include geophysical investigation using electrical resistivity method, test pits and
borehole drilling at different locations to different depths and performing in situ testing (Standard Penetration
Test (SPT), Dynamic probe light Test (DPL). The location of different tests shown in Figure (1).
Eight vertical electrical sounding (VES) were applied in study area and resistivity measurements were
conducted by using SAS300c resistivity meter. Four electrodes were inserted into the soil utilizing the Wenner
Arrangement. Resistivity have been used to identify corrosion rate depends on current flow between a metal and
the adjacent medium. (Table 2) ;( Baeckmann and Schwenk, 1975).

Figure 1 The location of boreholes, test pits, VES and DPl.

2.2 Laboratory work

Laboratory tests including sieve analysis, consistency test, compaction test, California Bearing Ratio (CBR) and
chemical analysis. Procedures were in general conformance with those recommended in British Standard
(B.S.1377:1990), (ASTM) and (AASHTO). The soils were classified according to the unified soil classification
system (USCS). The results of testing are shown in Table 5.

3 RESULTS

Resistivity method is very useful for the early detection of incipient corrosion. Eight vertical electrical
sounding (VES) have been applied in study area, the final result of apparent resistivity values at different
depths shown in Table 1. The corrosion potential on metals in soil increases as soil resistivity decreases.

562
Table 1 Electrical resistivity Results at each VES locations

Depth(m) Resistivity (Ω.m)

1 VES#1 VES#2 VES#3 VES#4 VES#5 VES#6 VES#7 VES#
8
2 742 3159 2418 5219 799 1363 14 252
3 1031 1678 1262 1331 142 424 59 127
4 1068 785 641 644 50 184 48 114
5 924 426 469 397 48 100 43 99
6 644 312 424 323 51 62 38 82
7 367 319 451 261 47 47 39 76
8 237 313 470 241 39 47 45 74
9 186 294 412 226 47 49 48 73
10 195 309 363 226 53 50 53 74
11 206 297 314 217 51 47 55 78

Table 2 Resistivity values and degree of Corrosivity after (Beckmann and Schwenk, 1975)

Soil Resistivity (ohm-m) Degree of Corrosivity/aggressively

Up to 10 Very Strongly Aggressive (VSA)
10 – 60 Moderately Aggressive (MA)
60 – 180 Slightly Aggressive (SA)
180 – above Practically Non-Aggressive (PNA)

The corrosion in soil increases as soil resistivity decreases and the subsoil Corrosivity around the study area
increases south western ward with depth
Standard Penetration Test (SPT), Dynamic probe light Test (DPL) have been used to evaluate soil
density/hardness. The number of blows required to give a tube penetration of 300mm will be taken as SPT-N
value of the soil tested at a specified borehole depth with interval 1.5m and classified according to N-values
(Table 3).

Table 3 Properties of cohesive/no-cohesive soil after (ASCE, 1996)

cohesive soil Non-cohesive soil

Soil consistency N values Soil consistency N values
Very soft 0-2 Very loose 0-4
Soft 3-4 Loose 5-10
Medium 5-8 Medium dense 11-30
Stiff 9-15 Dense 30-50
Very stiff 16-32 Very dense >51
Hard >32

3.1 Correlations between SPT and DPL tests:

Penetration tests have been used to evaluate soil density and consistency. The aim of the dynamic probing tests
is to determine the resistance of soil and/or soft rock in-situ to the dynamic penetration of a cone. Standard
Penetration Test, commonly known as the (SPT) is another in situ dynamic testing method. This test is one of
the most popular and widespread methods evaluate soil density and consistency of in situ soil investigation all
over the world. The dynamic probing test (DPL) calculate from the field and the Standard Penetration values
Test (SPT) they are estimated from the correlations between them (DIN4094) shown in Table 4.

563
Table 4 SPT blows corresponding to DPL penetrations in study area

DPL ID Depth DPL (mm/blow) SPT(N)

1.5 16.7 10 – 30
1
3 7 30 – 50
1.5 11.1 30 – 50
2
3 2.4 >50
1.5 8.3 30 – 50
3 3 6.7 30 – 50
4.5 7.1 30 – 50
1.5 6.3 30 – 50
4 3 5.9 30 – 50
4.5 7.1 30 – 50
1.5 6.3 30 – 50
5 3 5.9 30 – 50
4.5 7.1 30 – 50
1.5 10 30 – 50
6 3 2.4 >50
4.5 6.3 30 – 50
1.5 14.3 10 – 30
7 3 5 >50
4.5 6.3 30 – 50
1.5 7.1 30 – 50
8 3 4.3 >50
4.5 5 >50
1.5 3.2 >50
9 3 8.3 30 – 50
4.5 4.3 >50
1.5 2.7 >50
10 3 8.3 30 – 50
4.5 4.3 >50
1.5 2.4 >50
11 3 8.3 30 – 50
4.5 4.3 >50
1.5 1.7 >50
12 3 8.3 30 – 50
4.5 4.3 >50

The data of DPL deduced from study area and the values of SPT which were calculate , the number of blows
corresponding to DPL penetration (Table4) indicated the lateritic soil density vary from dense to very dense and
hard layers.

3.2 Engineering properties of laterites

Six test pits were excavated within the study area at different locations (Figure 1) and samples from different
depths were taken for the laboratory testing Sieves Analysis, Atterberge limits, Compaction test and California
bearing ratio test.

564
Table 5 Common properties of laterites in study area
Test pit PI
Depth(m) M.D.D (gm/cm3) O.M.C (%) CBR @ 2.5 %
No
1.0 9 2.10 9.70 9
T.P#1
2.0 11 2.12 9.94 5
1.0 11 2.05 10 7
T.P#2 Non -
2.0 2.14 7.5 14
plastic
T.P#3 1.0 23 2.13 9.55 70
T.P#4 1.0 6 2.11 9.5 7
T.P#5 1.0 15 2.18 7.44 52
T.P#6 3.0 9 2.03 10.5 39

3.3 chemical Analysis

Atomic absorption spectroscopy (AAS) is a spectral analytical procedure for the quantitative determination of
chemical elements employing the absorption of optical radiation (light) by free atoms in the gaseous state. In
analytical chemistry the technique is used for determining the concentration of elements and oxides in sample,
which were taken from slope and Wadies along Pipe line path and analyzed, after that measured major oxides
for five samples wish were taken from shallow depth along the pipe line. The results shown in (Table 6)
indicated low amount of sesquioxide due to washing process during rainy seasons.

Table 6 chemical analysis results

Test pits Depth
SiO2 Fe2O3 AL2O3 K 2O Na2O MgO CaO
No (m)
1 92.3 5.41 1.72 0.87 Nil Nil Nil
2 89.6 8.60 1.00 0.37 Nil Nil Nil
3 89 9.87 1.32 0.48 Nil Nil Nil
4 87.5 9.42 3.56 0.72 Nil 1 Nil
5 94.3 5.71 1.00 0.10 Nil 1.22 Nil

4 DISCUSSION

Laterites are light brown to reddish brown in colour. It is classified as Clayey SAND according to USCS
(Elarabi and Elkhawad, 2014). The corrosion in soil increases as soil resistivity decreases and the subsoil
Corrosivity around the study area increases in South West direction with depth. Chemical analysis indicates
that laterites are rich in quartz; contain appreciable amounts of iron and aluminium oxides, poor in potassium
oxide and very poor in sodium, magnesium and calcium oxides. Laterites in study area is good as construction
materials, and can be used for embankment purposes; on the other hand can improve the low CBR values by
blending with gravels.

5 CONCLUSION

laterites are highly weathered and altered soils formed by the in-situ weathering and decomposition of
rocks (igneous, metamorphic and sedimentary) derived from high terrene area and bolder of Muglad basin under
tropical condition, their ware some variation in fine materials, sand, and sesquioxide contain according to
topography and drainage system.
Laterites in main drainage system and slope containing high amount of sand and low amount of sesquioxide
(light colour) due to washing process. According to (SPT) and (DPL) the hardness of lateritic soil increase with
depth. Top soils (1-2 meter) are loose and with depth soil become dense to very dense. The resistivity method
very useful for the early detection of incipient the corrosion which will assist in the prevention of the associated
consequences of pipelines failure. It is recommended the pipelines in the area of study should be protected by
proper protective system to avoid any corrosive interaction between steel surface and soil environment. The
Corrosivity in the study area increases South- East direction with depth. Also

565
Laterites are valuable road pavement materials, can be used in the tropics countries as sub-base and base
material and also for gravel roads.

REFERENCES

AASHTO. (2004) Standard specifications for Transportation materials and Methods of Sampling and Testing.
ASCE, (1996) American Society of Civil Engineers. Available at: [Link]/lib/copyright/[Link].
ASTM, (2004). Special procedures for testing soil and rock for civil engineering purpose.
Baeckman W.V. and Schwenk W. (1975) Handbook of Cathodic Protection: The theory and Practice of Electrochemical
Corrosion Protection Techniques. Portcullis press Survey. 396.
British Standard, (1990) Method of Test for Soils for Civil Engineering Purposes, BS 1377.
DIN4093, German Code, (2006).
Elarabi H., Elkhawad. T (2014) Evaluation of Subsoil Corrosivity Condition around Baracaia Area Using the Electrical
Resistivity Method a Case Study from the Muglad Basin, South western Sudan. Journal of earth science and
engineering. David publisher. 4 (2014) 326-330.
Sayed A. (2003) Sedimentology and Reservoir Geology of the Middle-Upper Cretaceous Strata in Unity and Heglig Fields
in SE Muglad Rift Basin, Sudan Ph.D. thesis, Germany, Freiberg: university of Freiberg.
Tardy Y. (1997) Petrology of Laterites and Tropical Soils, Elsevier Publishing, Amsterdam. 978-9054- 106784, 419.

A laboratory investigation of the effect of used oil contamination on the direct

shear behavior of a lateritic soil
Une enquête de laboratoire de l' effet de la contamination de l'huile usée sur les propriétés de
cisaillement directs d'un sol latéritique
F.J.F. Ayeh*1, S.I.K. Ampadu1, F. Owusu-Nimo1, F.K. Boadu2 and A. Lawer1
1
Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
2
Duke University, Durham NC, USA
*
Corresponding Author

ABSTRACT - This study is part of a larger study to understand the effect and extent of used motor oil
contamination on soil and water and its impact on human health in an auto mechanic village in Ghana. This
paper focuses only on the effect of used motor oil contamination on the shear strength of a lateritic soil obtained
from the auto mechanic village as part of the development of a methodology for assessing the extent of
contamination. Soil samples were obtained from the site but away from the main influence area of the
contamination. The samples were characterized and the Standard Proctor compaction characteristics were
determined. Using the standard proctor compaction, the samples were prepared for direct shear test. The
samples were mixed with water corresponding to the optimum moisture content and then contaminated with
various percentages of used motor oil ranging from 0% to 16% dry weight of soil. The mixture was thoroughly
mixed and compacted using three different compactive efforts. Samples were then cut out and the direct shear
test was performed using normal loads of 50kPa, 100kPa, 150kPa and 200kPa. The results indicate that low and
medium density oil-contaminated samples have small to negligible cohesion suggesting that the material behaves
as cohesionless soil. At higher densities, however oil-contaminated samples show a c- behavior with different
values of cohesion at low (0-2%), medium (4-10%) and high (>10%) oil contamination levels. The study also
showed that increasing oil contents increases the tendency of the soil-oil mixture to dilate with the stress-strain
curves showing an increasingly less defined peak.

RÉSUMÉ - Cette étude fait partie d'une vaste étude pour comprendre l'effet et l'étendue de la contamination de
l'huile de moteur utilisée sur le sol et de l'eau et son impact sur la santé humaine dans un village de mécanique
automobile au Ghana. Cette étude se concentre uniquement sur l'effet de la contamination de l'huile moteur
utilisée sur la résistance au cisaillement d'un sol latéritique obtenu à partir du village de mécanique automobile.
Des échantillons de sol ont été obtenus à partir du site, mais loin de la principale zone d'influence de
contamination. Les échantillons ont été caractérisés et les caractéristiques de compactage de l’essai Proctor
normal ont été déterminées. En utilisant l’essai Proctor normal, les échantillons ont été préparés pour l'essai de
cisaillement direct. Les échantillons ont été mélangés avec de l'eau correspondant à la teneur en eau optimale et
ensuite contaminé avec divers pourcentages d’huile de moteur usée allant de 0 % à 16 % en poids sec de sol.
L’ensemble a été vigoureusement mélangé et compacté en utilisant trois différents efforts de compactage. Les
échantillons ont ensuite été découpés et l’essai de cisaillement direct a été réalisé en utilisant des charges
normales de 50 kPa, 100 kPa, 150 kPa et 200 kPa. Les résultats indiquent que pour les faibles et moyennes
densités l'effet de la teneur en huile sur la cohésion est négligeable ce qui suggère que le matériau se comporte
comme pulvérulent. La résistance est définie uniquement par l'angle de frottement interne. À des densités plus
élevées, la cohésion est faible de contamination d'huile bas (0-2 %), mais augmente pour des teneurs plus élevées
de pétrole (entre 4 et 10 %). L'augmentation de la teneur en huile augmente la tendance du mélange huile - sol
pour dilater les courbes contrainte-déformation montrant un pic de plus en plus moins défini.

1 INTRODUCTION

Soil may become contaminated with oil as a result of accidental oil spills that occurs during the oil drilling
processes or during transportation of the oil by sea and on land and also from oil storage tanks. In some cases, it
occurs due to deliberate human activity such as the destruction of oil wells in Kuwait during the Gulf War (Al-
Awadhi et al. 1992) and also from terrorist activities. Apart from these, another major and widespread source of
oil-contaminated soils is the human activity of changing of engine oil and other vehicle lubricants that occur
during repair and maintenance of vehicles.

567
Used motor oil, however, introduces another dimension to the problem of contaminated soils. Apart from the
hydrocarbon content, used motor oil also contains heavy metals such as magnesium, copper, zinc, lead and
chromium. In most developing countries due to non-existent regulation or their lack of enforcement, the used
motor oil is indiscriminately disposed of in nearby drains and on the bare ground with the result that these
compounds ultimately find their way to contaminate soil and water resources. Apart from the obvious effect on
ecological life in the environment, such contamination also poses serious health risks to inhabitants living and
working nearby. Where auto-mechanic villages exist that congregate large numbers of such auto mechanics over
a long period of time, the contamination may be severe. In order to develop a methodology for assessing the
extent and implications of such contamination and to design remediation measures for the contaminated soils, it
is important to understand the effect of such used motor oil contamination on various properties of soil. Previous
investigations (Ervin and Das 1992, Alsanad et al 1995, Shin and Das 2001, Khamechchiyan et al 2007) into the
effect of oil-contamination have suggested that oil contamination leads to a reduction in shear strength of the
contaminated soil but the extent of reduction depends on the chemical composition of the contaminant and on the
type of soil.
Suame Magazine is an auto-mechanics village in the city of Kumasi, Ghana, where the practice of
indiscriminate disposal of use motor oil has been going on since the 1960’s. This study sought to investigate the
effect of such used motor oil contamination on the shear strength of the soil from the auto mechanic village. Soil
samples from an uncontaminated area in the auto mechanics village was collected and artificially contaminated
with different percentages of used motor oil ranging from 0 to 16% in the laboratory. The contaminated soil was
then subjected to the direct shear test under normal stresses of 50kPa, 100kPa, 150 and 200kPa. Three different
levels of compaction were used. The paper discusses the results of the test.

2 TEST METHODS

2.1 Sampling and sample preparation

A trial pit was manually excavated at an uncontaminated section of the auto-mechanic village to a depth of 3m.
Disturbed samples were taken in 25kg sacks. The samples were then transported to the laboratory, air dried for 4
days bagged in 25kg sacks and then stored for further use. The sample, Suame Magazine Uncontaminated soil
sample obtained at a depth of 200cm below ground level is designated SMU200.

2.2 Index Properties and Direct Shear Tests

Part of the sample was taken and then characterized using procedures in accordance with BS1377, Part 2,1990.
For the characterization, the uncontaminated soil was subjected to wet sieving where a specified amount of soil
was weighed and sieved through the 2mm sieve. A portion (50g) of the soil passing was taken for the
hydrometer test. The portion retained on the 2mm sieve was soaked for 24 hours and then washed. In the
hydrometer test, sodium hexametaphosphate was used as the deflocculating agent. The grading of the soil is
represented by the grading curve in Figure 1. The liquid limit was determined using the cone penetration method.
The compaction characteristics were also determined.
Used motor oil was obtained from an auto-mechanic shop where vehicles were brought for servicing. The
air-dried soil samples were then artificially contaminated with used motor oil in the amount of 0%, 1%, 2%, 4%,
6%, 8%, 10%, 12% and 16% by dry weight of the soil samples and each was thoroughly mixed in a pan. The
mixture was then bagged in plastic bags and left for 24 hours to allow for equilibration. After the equilibration
period, for each percentage of oil-contaminated soil, water equivalent to the Proctor optimum moisture content
was added to the specimen and compacted using the Standard Proctor compaction specification to obtain
samples at 100% MDD. Additional sets of samples were prepared using 15 blows and 10 blows per layer
corresponding to targeted relative compaction levels of 80% and 90% of the respective MDDs. Then 10cm x
10cm samples were cut and assembled in the direct shear box and tested in accordance with BS1377 Part 7:1990
at a rate of 0.9mm/min.
The testing program adopted for the direct shear test is shown in Table 1. Three sets of nine (9) samples each
were prepared with different compactive efforts and at different oil contents. Each specimen was divided into
four (4) portions where they were subjected to normal stresses of 50kPa, 100kPa, 150kPa, and 200kPa. In total
108 specimens were tested. In order to determine the residue content in the used motor oil, samples of oven-
dried soil were contaminated with various percentages of used oil, dried in the oven for 24 hours and weighed.

568
Table 2 Testing programme for direct shear.

Test Series Number of Blows Oil Content (%) Normal Stress (kPa)
SM-80-X 10 0, 1, 2, 4, 6, 8, 10, 12, 16 50, 100, 150, 200
SM-90-X 15 0, 1, 2, 4, 6, 8, 10, 12, 16 50, 100, 150, 200
SM-100-X 25 0, 1, 2, 4, 6, 8, 10, 12, 16 50, 100, 150, 200

3 DISCUSSION OF RESULTS

3.1 Properties of Tested Soil

The grading curve of the test sample is presented in Figure 1. The soil was made up of 25.1% gravel, 26.9%
sand, 33.6% silt and 14.4% clay. The test sample had a Liquid Limit of 49.6 and a Plasticity Index of 20.6.
Figure 2 presents the compaction curve of the uncontaminated soil together with the starting states of the
contaminated soil specimen for the direct shear test. Each state represents the dry density and the corresponding
optimum water content for each of the oil contents studied. The water content values have been corrected for the
oil residue. These initial water content dry density values are shown in Tables 2 to Table 4.

100
SMU200
Liquid Limit = 49.6%
80
Plasticity Index = 20.6%
Percentage passing (%)

0
1E-3 0.01 0.1 1 10
Particle size (mm)
Fine Medium Coarse Fine Medium Coarse Fine
Clay Silt Sand Gravel

Figure 1. Grading curve of uncontaminated soil.

1.8
LL = 49.6%, PI = 20.6%
SM-100-X OMC = 24.6%,
3
MDD = 1.548Mg/m

1.7 w = w'(1+mn)-n(1-m)
where:
w - actual water content
SM-90-X w' - apparent water content
n - oil content
1.6
Dry density (Mg/m )
3

1.5
SM-80-X

1.4

1.3

1.2
0 10 20 30 40
Moisture content (%)
Figure 2. Compaction characteristics of uncontaminated soil showing initial sample conditions for direct shear

3.2 Direct shear test results

Table 2, Table 3 and Table 4 show a summary of the sample initial condition and failure strength parameters for
test series SM-100-X, SM-90-X and SM-80-X respectively. The initial sample condition made up of the water
content and the resulting dry density is the average for the four test specimens in the particular batch. The

569
average dry density for each test series is also indicated in the Table. The results show that the 100%MDD had
an average dry density of 1.645Mg/m3, while the 90%MDD and 89%MDD had average dry densities of 1.510
Mg/m3 and 1.331 Mg/m3 respectively. The individual test dry density values are all within 5% of these average
test series values indicating that the dry density variation within a test series is relatively small. The water
content however varied from 19.66% for 100%MDD at 2% oil content to as low as 6.23% for 80%MDD at 16%
oil content.

Table 3 Summary of Strength Parameters of high density samples (SM-100-X)

Test Average Initial Condition Peak Strength
Designation Water content (%) Dry density (Mg/m3) Cohesion (kPa) (deg)
SM-100-0 19.06 1.615 22.5 35.2
SM-100-1 19.66 1.588 20.0 33.4
SM-100-2 18.04 1.606 19.8 36.1
SM-100-4 16.11 1.667 38.2 35.3
SM-100-6 14.62 1.630 32.8 35.8
SM-100-8 13.20 1.670 35.8 32.0
SM-100-10 11.13 1.685 31.5 30.8
SM-100-12 9.71 1.665 2.0 36.7
SM-100-16 7.53 1.683 3.8 40.0
Average Dry Density 1.645

Table 4 Summary of test results for medium density samples (SM-90-X)

Test Designation Average Initial Condition Peak Strength
Water content (%) Dry density (Mg/m3) Cohesion (kPa) (deg)
SM-90-0 17.5 1.471 4.3 39.4
SM-90-2 16.32 1.474 3.6 39.5
SM-90-4 13.93 1.529 -0.2 39.0
SM-90-6 13.38 1.482 1.2 39.2
SM-90-8 12.23 1.478 2.7 36.2
SM-90-10 9.48 1.540 -1.8 36.5
SM-90-12 6.93 1.569 9.0 34.7
SM-90-16 6.34 1.533 1.6 36.7
Average Dry Density 1.510
Table 4. Summary of Strength Parameters for low density samples (SM-80-X).

In this study, failure was defined as the maximum shear stress measured. In the case where a defined peak
was not observed, the shear stress corresponding to 15mm horizontal displacement was used to define failure.
The typical shear stress-horizontal displacement and corresponding vertical deformation for samples tested at
100% relative compaction and for those tested at 90%MDD compaction are shown from Figure 3 to Figure 8.
A comparison between the control sample shown in Figure 3 and samples with small and medium oil
contents shown in Figure 4 and Figure 5 respectively, illustrates the effect of oil content. The figures show that
for a given level of compaction, there is no significant difference in the failure shear stress for soils with and
without oil content. However, the vertical displacement during shearing shows that there is increasing tendency
for dilation as the oil content increases. The stress-strain curves show a clearly defined peak for lower levels of
oil content but as oil content increases the curves do not show a clear peak.

200 200
SM-100-0-50 SM-100-2-50
180 SM-100-0-100 180 SM-100-2-100
SM-100-0-150 SM-100-2-150
SM-100-0-200 SM-100-2-200
160 160

140 140
Shear stress (kPa)

Shear stress (kPa)

120 120
-1.5 -1.5
100 100

80 80
-1.0 -1.0
Vertical deformation (mm)

60 60
Vertical deformation (mm)

40 -0.5 40
-0.5
20 20

0 0.0 0 0.0
0 5 10 15 20 0 5 10 15 20

Horizontal displacement (mm)

0.5 Horizontal displacement (mm) 0.5

1.0 1.0
Figure 3. Stress-displacement curves for SM-100-0 samples. Figure 4. Stress-displacement curves for SM-100-2 samples.

570
200 200
SM-100-8-50 SM-90-8-50
SM-100-8-100 180 SM-90-8-100
180 SM-90-8-150
SM-100-8-150
SM-90-8-200
SM-100-8-200 160
160
140
140

Shear stress (kPa)

120
120
-1.5 100
100
80
-1.0
80
-1.0 60
60

Vertical deformation (mm)

40
-0.5
40

Vertical deformation (mm)

-0.5 20

20 0 0.0
0 5 10 15 20
0 0.0
0 5 10 15 20
Horizontal displacement (mm) 0.5

0.5
1.0

1.0
1.5

Figure 5. Stress-displacement curves for SM-100-8 samples Figure 6. Stress-displacement curves for SM-90-8 samples.

200 200
SM-90-0-50 SM-90-2-50
SM-90-0-100 180 SM-90-2-100
180
SM-90-0-150 SM-90-2-150
SM-90-0-200 SM-90-2-200
160 160

140 140

Shear stress (kPa)

120 120
-1.5 -1.5
100 100

80 80
-1.0 -1.0
60

Vertical deformation (mm)

60
Vertical deformation (mm)

40
40 -0.5 -0.5

20 20

0 0.0
0 0.0
0 5 10 15 20
0 5 10 15 20

Horizontal displacement (mm) 0.5 0.5

1.0
1.0

Figure 7. Stress-displacement curves for SM-90-0 samples Figure 8. Stress-displacement curves for SM-90-2 samples

The variations of the strength as defined by the peak cohesion intercept and the peak angle of internal friction
with oil content are shown in Figure 9 and Figure 10. Figure 9 shows that for the high density samples (SM-100-
X, i.e., 100% MDD), the cohesion initially reduces slightly (from 22kPa to 19kPa) as the oil content increases
from 0 to 2%, then increases sharply to 38kPa at 4% and remains almost constant till 10% oil content beyond
which the cohesion drops sharply to about 2kPa. This may correspond to different mechanisms depending on the
degree of saturation. Airey and Prathapa, (2013) observed that asphaltic concrete was very sensitive to grading
and bitumen content when they compared the behavior of dense asphaltic concrete (5% bitumen and well-graded
aggregates) and stone mastic asphalt (gap-graded aggregates and 6% bitumen) in drained triaxial compression.
For the samples prepared at medium and low densities (SM-90-X and SM-80-X, i.e., 90% and 80% MDD), the
cohesion values were very low to negligible and sometimes even negative. This suggests that the contaminated
soil at medium to low densities behave like a cohesionless material. Figure 10 shows the relationship between
the oil content and the angle of internal friction. The figure shows that for low densities (80%MDD) to medium
densities, the angle of internal was fairly constant oscillating between the narrow values of 35o to 40o. For high-
density samples (100% MDD) the angle of internal friction was about 35o from 0%-8% oil content, drops
between 8% and 10% to about 30o and then rises sharply to about 40o at 16% oil content. This may correspond to
three different zones of saturation which may present different shear mechanisms.

571
Figure 9. Variation of cohesion intercept with Oil content. Figure 10. Variation of angle of internal friction with Oil content

4 CONCLUSION

Direct shear box tests were conducted on artificially waste oil contaminated soil prepared at different dry
densities and sheared at normal stresses of 50kPa, 100kPa, 150kPa and 200kPa.
1. The results suggest that for a given density the oil content has relatively small effect on the failure shear
stress.
2. For dry densities of 1.331Mg/m3 and 1.510 Mg/m3 the cohesion intercept is very low and negligible,
suggesting that the material behaves as cohesionless with strength defined only by the angle of internal
friction of 37o and 38o respectively.
3. For higher density samples of 1.645Mg/m3, however, the contaminated soil has a c- behavior which is
different at very low oil contents of up to about 2% from the behavior at higher oil content of 4 to 10%.
4. However, as the oil content increases, there is increasing tendency towards dilation and the stress-strain
curves show increasingly less defined peak.

ACKNOWLEDGEMENT
The authors gratefully acknowledge the GeoGroup of the Kwame Nkrumah University of Science and
Technology - Kumasi for the support during the period of this research.

REFERENCES
Al-Awadhi N.M., Abdal M.S., Briskey E.J., Kenneth W. (1992) Assessment of technologies for the remediation of oil-
contaminated soil resulting from exploded oil wells and burning oil fires in Kuwait. Proc. 85th [Link] Meeting and
Exhibition, Air and Waste Mhmt. Asso,. Kansas City, Mo.
Al-Sanad H.A., Walid K.E., Nabil F.S. (1995) Geotechnical Properties of Oil Contaminated Kuwaiti Sand, Journal of
Geotechnical Engineering, May 1995. 121:407-412
Airey D and Prathapa R, (2013), Triaxial Testing of Asphalt, Proceedings of 18th International Conference on Soil
Mechanics and Geotechnical Engineering edited by Pierre Delage et al. Paris 2013, Vol 1. pp. 301-304
BS1377, Part2: 1990 Method of test for soils for civil engineering purposes. Part 2. Classification tests.
BS1377, Part 7:1990 British Standard Methods of test for soils for civil engineering purposes. Part 7. Shear strength tests
(total stress).
Shin E. C., Das B. M., (2001) Bearing capacity of unsaturated oil-contaminated sand. International Journal of Offshore and
Polar Engineering, Vol. 11, No. 3
Das B. M., Sobhan K. (2014) Principles of Geotechnical engineering. 8th edition SI
Ervin E., Das B. M., (1992) Mechanical behavior of an oil-contaminated sand. Usmen, Acar (Eds), Environ. Geotechnology
Proc., Mediterranean Conference. Balkema Publishers, Rotterdam, The Netherlands.
Gillott J. E. (1987), Clay in Engineering Geology. Developments in Geotechnical Engineering 41. Elsevier.
Khamechchiyan et al (2007) Effect of crude oil contamination on geotechnical properties of clayey and sandy soils
Engineering Geology 89 (2007) 220-229
Meegado N. J., Ratnaweera P., (994) Compressibility of contaminated fine-grained soil. Geotechnical Testing Journal 17.

Quasi-static numerical modeling of an ore carrier hold

Modélisation numérique en quasi-statique d’une cale de minéralier

S. Daoud*1, S. Ennour2 , I. Said3 and M. Bouassida4
1
Université de Tunis El Manar – Ecole Nationale d’Ingénieurs de Tunis. LR14ES03-Ingénierie Géotechnique.
BP 37 Le Belvédère, 1002 Tunis / Mecater Ingénierie, Hammamet, Tunisie
2
Mecater Ingénierie, Hammamet, Tunisie
3
Université de Tunis El Manar – Ecole Nationale d’Ingénieurs de Tunis. LR14ES03-Ingénierie Géotechnique.
BP 37 Le Belvédère, 1002 Tunis, Tunisie
4
Université de Tunis El Manar – Ecole Nationale d’Ingénieurs de Tunis. LR14ES03-Ingénierie Géotechnique.
BP 37 Le Belvédère, 1002 Tunis, Tunisie
*
Corresponding Author

ABSTRACT - The problems associated with ore carriers shifting are well known. In fact, there have been
increases in the rate of bulk carrier casualties and crew fatalities since 2005. This issue brings up to wonder
about the causes of these incidents and how to avoid them? This study presents a brief review about the possible
origins of cargoes instability. Afterwards, it analyzes the ore carrier hold stability and examines the stress
distribution in this latter by means of a quasi-static numerical model. Finally, an evaluation of the shear ratio
variation in terms of the hold inclination is established in order to prevent sliding failure of a bulk cargo. This
work is a preliminary study of a project which aims to develop a new in-situ test procedure for evaluating the
shear strength of loaded ore in view of its liquefaction prevention. According to this research, in a 15° inclined
hold, the maximum shear ratio is less than 0.2 in all areas of the ore pile, except under the residual slope and at
the surface. Therefore, these parts are assumed to be the most vulnerable areas to the hold inclination under static
loading.

RÉSUMÉ – Les problèmes associés au naufrage des minéraliers ne sont pas nouveaux. En fait, il y a eu une
augmentation du taux des accidents de vraquiers et des décès de l'équipage depuis 2005. Cette question nous
amène à s'interroger sur les causes de ces incidents et comment les éviter. Cette étude présente un bref aperçu sur
les origines possibles d’instabilité des cargaisons. Ensuite, on analyse la stabilité de la cale du minéralier et
examine la distribution des contraintes dans ce dernier au moyen d'un modèle numérique quasi-statique. Enfin,
une évaluation de la variation du taux de cisaillement en termes de l'inclinaison de la cale est menée afin de
prévenir la rupture des talus de la cargaison. Ce travail est une étude préliminaire qui s’inscrit dans le cadre d'un
projet qui vise à développer une nouvelle procédure d’essai in situ pour évaluer la résistance au cisaillement du
minerai chargé en vue de la prévention de sa liquéfaction. Selon cette recherche, dans une cale inclinée à 15 °, le
ratio de cisaillement maximal est inférieur à 0,2 dans la totalité du tas de minerai, sauf au niveau des talus
résiduels et la surface sommitale. Par conséquent, ces parties sont supposées être les zones les plus vulnérables à
l'inclinaison de la cale sous chargement statique.

1 INTRODUCTION

Although there have been reductions over time in bulk carrier casualties and crew fatalities since the early 1980s,
with an interim peak during the early and mid-1990s, there have been increases since 2005. In fact, three bulk
carriers foundered in 2010 resulting in the loss of 45 lives while carrying Nickel ore from Indonesia (Roberts et
al., 2013).
But, this issue has recently returned to sharp focus with the sinking of the bulk carrier, Vinalines Queen, on
25 December 2011, whilst on a voyage from Indonesia to China also carrying nickel ore with the loss of all but
one of the ship’s 23 crew.
According to Mecater (2011), causes could be attributed to two main phenomena. First, a cargo shift which
results from an irreversible deformation of the ore pile surface, under the rolling effect. Indeed, for ores with
high humidity, when the incline exceeds a certain limit (>15°), the fast shear induced on the surface of the ore
exceeds the ore undrained cohesion and causes rupture of the ore structure. If the deformations of the surface are
important, the center of gravity of the cargo moves and causes a severe list. Second, ore liquefaction produced
under the effect of cyclical shear due to the ore carrier agitation in the form of engine vibrations, ship’s motions
and swells impact. This leads to a reduction of the inter-granular spaces. If compaction is such that there is more

573
water inside the cargo than there are spaces between the particles, the water pressure inside the cargo can rise
and press the particles apart. This has the effect of reducing the friction between particles, and thus the shear
strength of the cargo. This may produce a transition of all or part of the cargo from a solid state to a viscous
fluid.
These risks often create dangerous angles of list and in some instances the loss of stability can be such that
the ore carrier and the lives of those onboard are lost. It is therefore imperative that seafarers are aware of the
types and conditions of cargo that may undergo one of these hazards. Unfortunately, despite the provisions of the
international convention for the Safety of Life at Sea (SOLAS, 1974) and the International Maritime Solid Bulk
Cargoes Code (IMSBC Code, 2013), cargo unsuitable for transportation still continue to be loaded and carried.
The Japanese authorities have been concerned about the propensity of nickel ore cargoes to shift, and a Study
on Prevention of Sliding Failure of Nickel Ore in Bulk was elaborated by Susumu et al., (2000). The study
details a new procedure for evaluating the shear strength of nickel ore and thus the suitability for carriage of the
cargo. The test procedure utilizes a ‘cone penetrometer’ to measure the shear strength of a graded sample of
nickel ore suitably compacted in a standard container.
The incidents cited above were mostly linked to bulk carriers that foundered when carrying nickel ore.
Actually, according to Roberts et al (2013), cargoes of nickel ore are prone to liquefaction under certain
atmospheric conditions, which can then lead to cargo shifts, loss of stability and capsize. According to the
IMSBC code, nickel ores are classified as liquefiable (group A).
In the first part of this research, a quasi-static numerical model is established to analyze the hold stability and
the stress variations. Afterwards, an examination of shear ratio and of the stress path at a point of the hold were
conducted in order to enable an estimation of the maximum shear ratio and analyze its sensitivity according to
the rolling movements that the ore carrier can undergo.

2 QUASI-STATIC NUMERICAL MODELING OF AN ORE CARRIER HOLD

In order to study the stress distribution in the hold, a quasi-static numerical modeling was used to reproduce what
may happen to an ore carrier during its journey.
Preliminary consolidated undrained test with measure of pore water pressure (CU+U) were elaborated on
saturated nickel ore samples. The response to the applied consolidation stresses of 65kPa, 130kPa and 260kPa,
showed that the nickel ore sample is a slightly delating material with an apparent cohesion C’ of 5kPa and an
internal friction angle φ’ of 33°. The results of the geotechnical characterization of this sample indicated it is a
highly plastic silt and looks like a fine lateritic soil, with the presence of many small friable aggregates that can
be broken by hand.
Since the rolling of the ore carrier appears to be the most dangerous motion, we limited this study to analyzing
its effect on the ore carrier. In quasi-static, this movement has been reduced to a progressive inclination of the
hold relatively to the vertical axis passing through the center of gravity of the hold. Thus, we compared the ore
stress states in a horizontal case and a 15° inclined case.
The numerical modeling of the hold is conducted by PLAXIS software based on the Finite Elements Method
and established using the geometry of the central section of the hold. The model used is two-dimensional and
supposes that the deformation perpendicular to the model plan is zero (plane strain).
The mesh used is fine with 15-nodes elements and the material model is Mohr-Coulomb.

2.1 Stability analysis of the hold

2.1.1 Horizontal hold

Under drained conditions, the application of the Phi/C reduction procedure in the PLAXIS code, allowed us to
determine the potential sliding surfaces.
Figure 1 shows that the mass of potential instability is concentrated under the residual slopes with circular iso-
values curves. It extends from the slope foot up to 80 cm upstream of the slope crest.

574
Figure 1. Areas of large deformations in case of a horizontal hold loaded with laterites (C’=5kPa, φ’=33°); Fs=1.97.

Under undrained conditions, calculations based on the method of vertical slices at limit equilibrium
(calculation code TALREN) show that the undrained cohesion required to ensure the stability of the residual
slope with a factor of safety near 1 is equal to 6 kPa as illustrated in Figure 2. This situation corresponds to a
very fast loading of a very wet ore pile (close to saturation).

Figure 2. Areas of large deformations in case of a horizontal hold under undrained conditions (C=6kPa, φ=5°).

2.1.2 15° inclined hold

Figure 3 indicates that the sliding surfaces are concentrated under the left slope and the obtained safety factor is
near 1.2 when the hold is 15° inclined. Thus, it is deduced that under drained conditions, the 35° residual slope
with over 3 m height can withstand a quasi-static 15° inclination.

Figure 3. Areas of large deformations in case of a 15° inclined hold loaded with laterites (C=5kPa, φ=33°); Fs=1.19.

Considering undrained conditions and corresponding to a fast inclination of the hold, by means of calculations
code TALREN, Figure 4 demonstrates that the required cohesion to ensure the stability with a safety factor of 1,
is close to 12 kPa in a case of 15° inclined hold.

575
Figure 4. Areas of large deformations in the case of a 15° inclined hold under undrained conditions (C=12 kPa, φ=5°).

Thus, it is deduced that for ores with high humidity, a rapid inclination of the hold at 15° can cause a
remarkable deformation of the pile; this deformation includes both the residual slopes and the summit platform.
In fact, depending on the saturation state of the loaded ore, there are two possible deformation scenarios:
 Partially saturated ore:
When the loaded cargo is partially saturated, the rolling of the ore carrier can induce the deformation of the
pile geometry by flattening its slopes. Considering the symmetry of this motion, the hypothetical final ore pile
geometry may be as illustrated in Figure 5.
It is noted that the inclination of the ore carrier leads to formation of a disturbed soil area without total loss of
its cohesion.

Figure 5: Hypothetical final ore pile geometry after the ship rolling in the case of a partially saturated ore

 Fully saturated ore :

However, the deformation becomes more notable when the loaded ore is fully saturated. The disturbed soil
formed in this case loses its cohesion and may get liquefied.
The ore pile geometry deformation induces the migration of gravity center and as a result dangerous list
angles which can lead to the capsizing of the ore carrier.
The hypothetical final ore pile geometry is illustrated in figure 6.

Figure 6: Hypothetical final ore pile geometry after the ship rolling in the case of a fully saturated ore

576
3 STUDY OF THE SHEAR RATIO

Since the shear ratio is the most important parameter to characterize cyclic excitation, this part of the study is
dedicated to analyze its variation in terms of the hold inclination.

3.1. Shear ratio calculation

When comparing the case of an ore carrier hold exposed to swell motions and a specimen under cyclic triaxial
test, we need to mention that:
In the case of a cyclic triaxial test, the variation of the deviatoric stress q is compared to a reference value
fixed at a null deviator, which corresponds to an isotropic consolidation of the specimen under a stress equal to
σc. Thus the cyclic shear ratio R is calculated using the following formula (1):

(1)

For the hold case, the consolidation is not isotropic in particular under the residual slopes, thus the variation of
the deviator stress Δq is compared to a reference value which corresponds to deviator q0. The deviator stress qi
for an inclination i of the hold starting from the horizontal is Δq = qi - q0. On the other hand, as the main stresses
are not equal (σ2 ≠ σ3), the consolidation stress σc becomes equal to the mean stress p . The shear ratio is
determined as follows (2):

(2)

3.2. Analysis of shear ratio variation in terms of hold inclination

Based on the calculation method defined above, the shear ratio R was calculated in all nodes (Plaxis Calculation)
then shear ratio iso-values of the horizontal hold was elaborated. These values provide the distribution of the
initial shear ratio at the end of the loading.
It is noticed according to Figure 7 that the initial ratio (horizontal hold) is close to 0.5 in terms of absolute
values in the whole ore mass except near the surface of the residual slopes where it is above 0.7. The high shear
ratio under the slope normally increases with an increasing angle and height of the residual slope which is of 35°
to a height of 3 m.

Figure 7. Iso-values of initial shear ratio for a horizontal hold

The same calculations of shear ratios were conducted for a 15° inclined hold relative to the horizontal case
(Figure8). For the calculation of these shear ratio variations, we use the formula (2) mentioned above. It is
assumed that:
 Consolidation is under main stress p0 (horizontal hold)
 The shear induced by a defined inclination (i) corresponds to the difference between the calculated shear of
a horizontal hold and the calculated shear for an inclined hold at the angle (i).

577
Figure 8. Iso-values of shear ratio variation for a 15° inclined hold

These results show that it is possible to distinguish 3 parts in the mass of ore loaded into the hold:
 The central and lower parts, which represents nearly 80% of the ore mass, where the average variation of
the ratio is limited to 0.1 (in absolute value). The maximum point value calculated in these areas is 0.2.
 The summit surface of the ore pile, where the average ratio variation is limited to 0.1 (in absolute value)
with a maximum value that may exceed 0.3 to less than 50 cm above the free surface.
 The residual slopes, where the average ratio variation can reach 0.7 for a 15° inclination.

The high value of the ratio variation is due to the steep residual slopes. Thereby, the residual slopes are the most
vulnerable areas to the hold inclination whether under static or dynamic loading (cyclic).

4 CONCLUSION

This study allows concluding the following statements:

 When analyzing the stability of the ore carrier hold using the quasi static numerical modeling, it is deduced
that under drained conditions, the 35° residual slope with over 3 m height can withstand a quasi-static 15°
inclination. However, under undrained conditions, this inclination can cause a notable deformation of the
ore pile.
 The examination of the mean and deviatoric stress iso-values indicated that a 15 ° inclination of the hold
causes an increase in both stresses under the residual down dip slope and a reduction in the upstream dip.
 The analysis of shear ratio variation in terms of hold inclination showed that the residual slopes are the
most vulnerable areas to the hold inclination under static loading. Thus, it is believed that in case of high
waves, the slopes will deform under the effect of static rupture (undrained shear) and hence the slope will
be reduced.
 Assuming an inclination of 15 °, subject to confirmation by the results of the second ore carrier
instrumentation, the maximum shear ratio is less than 0.2 in all areas of the hold, except under the residual
slope and at the surface.
 The stress path crossed by a point of the hold carrier is in accordance with a stress path resulting from a
cyclic triaxial test.

ACKNOWLEDGEMENT

This study was carried out under the cooperation of MECATER Engineering and Laboratory of Research in
geotechnical engineering. I gratefully acknowledge the support and generosity of my supervisors and I would
like to express sincere thanks for people involved in this research.

REFERENCES

Roberts, E. Stephen [Link], [Link], 2013, Casualties and loss of life in bulk carriers from1980 to 2010, Marine
Policy 42, 223–235.
MECATER, 2011, Rapport de synthèse Rhéolat 1.
SOLAS, international convention of safety of life at sea, Adoption: 1 November 1974; Entry into force: 25 May 1980.
Intercargo, Lloyd’s Register and the UK P&I Club, IMSBC code, 2013.
Susumu Ota, Tamaki Ura, Masaki Murayama, Masato Tanaka, 2000, Study on Prevention of Sliding Failure of Nickel Ore in
Bulk, New Procedure for Evaluating Shear Strength, the Journal of the Society of Naval Architects of Japan, vol 187,
301-307.

Correlations between CBR index, modified proctor’s characteristics, and

identification parameters of lateritic soils in Cameroon

Corrélations entre l’indice portant CBR, les caractéristiques à l’optimum de l’essai Proctor
Modifié, et les paramètres d’identification des sols latéritiques du Cameroun

E. M. KANA*1, M. MAMBA2, T.R.F. FANGLEU3

1 National Civil Engineering Laboratory, Yaoundé, Cameroon

2 National Advanced School Polytechnic, Yaoundé, Cameroon
3 Consultant Office SADEG, Yaoundé, Cameroon
* Corresponding Author

RÉSUMÉ - La caractérisation physico-mécanique des sols support d’ouvrages de génie civil, ou la sélection de
matériaux destinés à leur construction, exige généralement de très longs délais d’attentes qui sont incompatibles
avec les stratégies de productivité de plus en plus exigeantes des entreprises en général, et de celles de Bâtiments
et Travaux publics en particulier.
Pour la construction routière par exemple, la détermination de l’indice portant du « Californian Bearing Ratio »
(CBR) à j jours d’imbibition (Norme NF P94-078), avec j généralement supérieur ou égal à quatre, à l’optimum
de la compacité donnée du matériau.
Au Cameroun, 1500 résultats d’essais de laboratoire obtenus à partir d’échantillons de latérites prélevés sur
toute l’étendue du pays ont été étudiés. Afin de réduire ces délais et par conséquent optimiser le coût des projets
concernés, il est proposé dans le cadre de ce travail, deux modèles de prédiction : l’un pour estimer les valeurs
du CBR à 4 jours d’imbibition et l’autre pour donner valeurs de densité sèche (γdopm) et de teneur en eau (Wopm)
à l’optimum Proctor modifié (Norme NF P94-093), lorsqu’on connait les paramètres d’identification des sols
latéritiques.
Les corrélations obtenus et testés sur une trentaine d’échantillons, donnent une prédiction avec une erreur
inférieure ou égale à 6 %, entre les valeurs estimées et celles obtenus après essais de laboratoire.
Mots clés : Indice CBR, Proctor, corrélation, sols latéritiques, limites d’Atterberg, granulométrie

ABSTRACT- For earthworks related to constructions, laboratory tests carried out for physical and mechanical
characterization of soils take a lot of time. Due to the continuous search for an increase of productivity and
competitiveness for enterprises in general, and particularly for those in the sector of buildings and public works,
the long duration of those necessaries tests, is generally considered as negative on the planning program of the
projects related to.
For instance, for the roads works, one of the main characteristics used is the Californian Bearing Ratio (CBR). In
conformity with the French norm (NF P94-078), it takes almost four or more days, to have one value of CBR at
the Optimum water content. To reduce the time taken for geotechnical investigations, this paper is focused on
findings related to analysis on 1500 laboratory tests results on different lateritic soils samples which were picked
up all over the country of Cameroon since 1954. It tries from some identifications’ tests results, to propose a
prediction, firstly of the CBR index after 4 days of humidification and secondly, of the dried density (γdopm) and
the moisture content (Wopm) at the level of the Modified Optimum Proctor. The correlations functions’ obtained,
verified on almost 30 samples, gave predictions for Cameroonian lateritic soils, with a relative error situated
around 6%.
Keywords: CBR, Proctor, correlation, lateritic soils, Atterberg Limits, sieves analysis

1 INTRODUCTION

En construction routière, l’un des paramètres les plus utilisés est l’indice CBR (Californian Bearing Ratio), qui
est un indicateur de résistance du sol aux sollicitations multiformes dont il est l’objet. En effet, une valeur élevée
de cet indicateur signifie que le sol utilisé a de bonnes caractéristiques qui confèrent un gage de longévité à
l’infrastructure routière concernée. Par ailleurs, pour les barrages en terre ou des remblais et plateformes routiers,
les sols mis en place doivent être compactés jusqu’à atteindre l’optimum de leur densité. Les paramètres les plus
usités en Afrique subsaharienne pour mesurer le niveau de compacité des sols sont : les caractéristiques à

579
l’optimum Proctor modifiées (γdopm,Wopm ). Une valeur élevée de γdopm étant un gage de faible déformabilité et
de résistance mécanique élevée.
La prédiction des propriétés d’un sol à partir de certains paramètres d’identification ou d’état n’est pas
quelque chose de nouveau. En parcourant la littérature on observe que plusieurs corrélations empiriques ont été
développées entre eux et : l’angle de frottement (Mamba et al, 2013), la résistance au cisaillement (Goktepe et al,
2008), l’indice de gonflement et de compressibilité d’une part, les caractéristiques granulométriques et les
paramètres d’état des sols d’autre part.
Pour les caractéristiques d’un sol à l’optimum Proctor, les auteurs ont proposé des corrélations (Alvi et
al2009 ; Sulewska, 2009 ; Ugbe, 2011 ; Gupta, 2008 ; Goma et al 2007 ; et Ohandja 2011). S’agissant des
indices CBR, plusieurs types de corrélation sont proposés dans la littérature (Saklecha et al, 2011 ; NCHRP,
2001 ; Satynarayana et al, 2006 ; Gregory et al, 2007; Vinol et al, 2008 ; Breytenbach, 2009 ; Patel et al, 2010 ;
Yildirim et al, 2011 ; Ohandja, 2011 ; Datta et al, 2011 ; Saklecha et al, 2011 ; Ramasubbaro et al, 2013.
La caractéristique commune à toutes ces corrélations simples ou multiples résulte du fait qu’elles sont
données sous la forme d’une régression linéaire. Ces modèles ont l’avantage d’être simples et n’utilisent que
quelques caractéristiques granulométriques des sols. De plus, ils supposent que l’effet de chaque variable
explicative sur la variable expliquée (CBR4j, γdopm et Wopm ) est linéaire.
Le reproche que l’on peut entre autre leur faire, est leur non prise en compte de l’effet de non linéarité entre
certains paramètres tels que le pourcentage des particules fines et les limites d’Atterberg.

2 MODELISATION

Dans le modèle proposé, il est pris en compte l’ensemble des variables utilisées pour la classification des sols
(paramètres granulométriques et de plasticité) et les variables d’état (ou paramètres d’assemblages) utilisées
pour décrire la façon dont les échantillons de sol sont mis en place (pour la prédiction des indices CBR4j).
La granulométrie d’un sol (Norme P94-056) dans notre modèle sera décrite par : le diamètre maximum des
grains (Dmax), le pourcentage des graves (G), le pourcentage de sable (S), le pourcentage de limon (L), le
pourcentage des fines (P80µ = % passants au tamis de 80µm), le pourcentage d’argile (Arg) et le coefficient
d’uniformité C  d 60 (où dn désigne le diamètre correspondant à n% des passants). Les paramètres utilisés pour
u
d10
la description de la plasticité des sols (NF P94-051) sont l’indice de plasticité IP et la limite de liquidité WL. Les
paramètres d’assemblage ou d’état retenus, qui caractérisent la compacité du sol à l’Optimum Proctor Modifié
sont : la densité sèche optimale (γdopm) et la teneur en eau optimale (Wopm).
Nous cherchons à prédire l’indice CBR à 4 jours d’imbibition et les caractéristiques de compacité d’un sol à
l’optimum Proctor Modifié à l’aide d’un modèle non linéaire et multi variables donnés par les relations (1),
(2) et (3):

CBR4 j  F  Dmax , G , S , Cu , P80 µ , L, A, I P ,WL ,  dopm ,Wopm  (1)

CBR4 j  Dmax
n1
 G n2  S n3  P80n4µ  Ln5  An6  I Pn7 WLn8   dopm
n9
Wopm
n10
 CUn11

Où nj(j-1 à11) est réel.

 dOpm  F  Dmax , G, S , Cu , P80 µ , L, A, I P ,WL ,  dopm ,Wopm 

(2)
 dOpm  Dmax
l1
 G l2  S l3  P80l4 µ  Ll5  Al6  I Pl7  WLl8 CUl9

Où lj(j-1à9) est réel.

WdOpm  F  Dmax , G , S , Cu , P80 µ , L, A, I P , WL ,  dopm ,Wopm  (3)
WdOpm  Dmax
m1
 G m2  S m3  P80m4µ  Lm5  Am6  I Pm7  WLm8 CUm9

Où mj(j-1à9 est réel.

2.1 Les données et outils

Les données utilisées pour ce travail proviennent du LABOGENIE-CAMEROUN, et du Laboratoire

Géotechnique et Matériaux de l’École Nationale Supérieure Polytechnique de Yaoundé. Ces données ont été
obtenues à partir tests réalisés sur des échantillons de sol prélevés sur l’ensemble du territoire du Cameroun. Les
résultats de chaque échantillon de sol ont dû être ré exploité afin que leur indentification puisse se faire par la
donnée des valeurs des douze variables suivantes :

580
CBR 4 j , Dmax , G , S , Cu , P80 µ , L, A, I P ,WL ,  dopm ,Wopm . (4)
Au terme de ce travail, il a été obtenu un échantillon de 1500 individus identifiés par la donnée de ces
douze variables. C’est sur cet échantillon que nous avons réalisés notre étude statistique en utilisant comme outil
le logiciel SAS 9.2(2009).
2.2 Les critères de sélection des variables explicatives et corrélations
Nous avons utilisé comme critère pour la sélection des corrélations les valeurs des paramètres suivants : le
coefficient de corrélation multiple R² et le coefficient de Mallow C(p).
Pour la construction des matrices des corrélations nous avons utilisés le coefficient de corrélation au sens de
Pearson rXY (entre les variables X et Y) et la probabilité associée, lorsqu’on admet que les deux variables(X et
Pearson

Y) sont indépendantes ( rXY  0 ). Cette matrice de corrélation permet de cerner les interactions entre les
Pearson

différentes variables du modèle. Les variables X et Y désignent les paramètres utilisés pour l’identification d’un
sol (4). Certaines variables explicatives de la corrélation on dû être éliminées, du fait que leurs apports étaient
non significatifs.

3 PRÉSENTATION DES RÉSULTATS

L’analyse des données, nous a permis de ressortir dans le Tableau 1 : les valeurs maximales, moyennes,
minimales, les écarts types ainsi que les unités des différentes variables utilisées pour l’élaboration des différents
modèles. L’histogramme des Figures 1, 2 et 3 illustrent la distribution des variables expliquées que sont: l’indice
CBR4j, la densité sèche γdopm et la teneur en eau Wopm de l’Optimum Proctor Modifié (OPM) de l’échantillon
étudié.
La matrice de corrélation obtenue est présentée dans le Tableau 2.
Tableau 1 : Valeurs moyennes, minimales, maximales, écart type et unités des variables utilisées

Variables Effectifs Minimum Maximum Moyenne Écart type

(mm) 1500 5,00 20,00 19,72 1,353
(%) 1500 5,20 59,30 21,89 6,645
(%) 1500 17,70 87,60 51,42 10,279
(%) 1500 8,23 80,73 50,08 11,954
(%) 1500 14,31 91,70 41,21 12,868
(%) 1500 0,52 13,54 6,62 2,087
(%) 1500 19,09 9399,00 1127,00 1215,000
S(%) 1500 8,77 72,88 29,23 10,589
CBR4j(%) 1500 8,00 150,00 56,86 22,315
(kN/m3) 1500 16,4 23,6 20,18 1,25
(%) 1500 4,40 29,00 13,48 3,018
L(%) 1500 0,42 14,70 5,50 1,824

Figure 1 : Histogramme des valeurs de l’indice CBR4j de Figure 2 : Histogramme des valeurs de la densité seche à
l’optimum Proctor modifié de l’echantillon étudié
l’echantillon étudié

581
POURCENTAGE

Figure 3 : Histogramme des valeurs de la teneur en eau optimale à l’optimum proctor modifié de l’echantillon étudié

Tableau 2: Matrice de corrélation ( rXY )

Pearson

1ère ligne : coefficient au sens de Pearson, 2ème ligne : probabilité associée lorsque les deux variables sont supposées
indépendantes
S CBR L
1 0.025 0.04732 0.109 -0.12777 -0.13127 0.00391 -0.17606 0.04155 0.01160 0.04946 -0.16525
0.00 53 0.1166 81 <0.0001 0.0001 0.8968 <0.0001 0.1683 0.7005 0.1010 <0.0001
0.000 0.000
1 3
1 0.81078 0.278 0.15901 0.16611 0.03115 -0.24129 - -0.10743 0.14162 0.04758
0.00 <0.0001 62 0.0001 0.0001 <.0001 0.23610 0.0004 0.0001 0.1146
<0.00 0.3017 0.0001
01
1 0.237 0.15238 0.16541 0.04281 -0.24695 - -0.22864 0.19814 0.10145
0.00 50 0.0001 0.0001 0.1557 <.0001 0.24382 <0.0001 0.0001 0.0007
<.000 0.0001
1
1 -0.29977 -0.14886 - -0.81454 0.18956 0.37714 - -0.54145
0.00 <.0001 <.0001 0.03332 <.0001 0.0277 0.0001 0.17004 <0.0001
0.2694 0.0001
1 0.93004 - 0.29943 0.26560 -0.05908 0.17952 0.69171
0.00 <.0001 0.10413 <0.0001 0.0001 0.0500 0.0001 0.0001
0.0005
1 - 0.54841 - -0.04425 0.15590 0.59783
0.00 0.08139 <.0001 0.13318 0.1423 <.0001 0.0001
0.0069 <.0001
1 0.25216 - -0.14698 0.03284 0.08320
0.00 0.0032 0.06702 0.0001 0.2763 0.0001
0.0262
S 1 - -0.12521 0.02642 0.37862
0.00 0.15306 0.0001 0.3811 <0.0001
0.0001
CBR 1 0.33650 - -0.33044
4j 0.00 <.0001 0.12521 <0.0001
0.0008
1 - -0.41540
0.00 0.68405 <0.0001
<0.0001
1 0.27594
0.00 <0.0001
L 1.

La relation (5) établie et indiquée ci-après, donne l’indice CBR à 4 jours d’imbibition, lorsqu’on connait
les variables Dmax , G, S , P80 µ , L, A, I P ,WL ,  dopm , Wopm , avec un coefficient de corrélation multiple de R=99.17%.

0,36923 0,31503 0,01789 0,09676 0,04768 0,33196

CBR4 j  Dmax G Cu S A IP WL0,03486Wopm  opm
0,45179 2,1942
(5)
Avec R²=0.9917 ; C(P)=9 ; et RMSE= 0.3610

582
Figure 4. Courbe de tendance de l’indice CBR 4j à 4jours d’inbibition

La relation (6) établie, donne la densité sèche optimale γdopm lorsqu’on connait les
variables Dmax , G , S , P80 µ , L, A, I P , WL , avec un coefficient de corrélation multiple de R=99.17%.

 dopm  Dmax
0,13629 0,10104 0,00186 0,0638 0,01425 0,0500 0,00496
G Cu S A L IP WL0,06699 (6)
Avec R²=0.9917 ; C(P)=8 ; RMSE= 0.5997

La relation (7) donne Wopm lorsqu’on connaît les variables Dmax , G , S , P80 µ , L, A, I P , WL , avec un coefficient
de corrélation multiple de R=99.37%.
0,48006 0,06917 0,01338 0,01431 0,03294 0,07816 0,00597
Wopm  Dmax G Cu S A L IP WL0,30853 (7)
Avec R²= 0.9937 ; C(p) = 8 ; RMSE = 0.2226
Densité mesurée

Densité estimée

Figure 5 : Courbe de tendance de la densité optimale à l’optimum proctor modifié

583
19
17

Wopm mesurée
15
13
11 Formule Cameroun
9
Linéaire (Formule
7 Cameroun)
5
5 10 15 20
Wopm estimée

Figure 6 : Courbe de tendance de la teneur en eau optimale à l’optimum proctor modifié

4 VALIDATION DES CORRELATIONS

Pour tester la validité des modèles proposés, nous avons comparé les résultats théoriques à ceux des essais de
laboratoire pour une trentaine d’échantillons de sol latéritiques provenant des localités de : Dschang, Lom
Pangar, Kumba, Yaoundé, Bakebe et Fontem, dont les roches mères sont très distinctes tant sur le plan
pédologique et géomorphologique. Les résultats obtenus illustrés par les Figures 7 à 8, montrent que les modelés
de prédiction proposés sont très intéressants. En effet, on constate que pour ces cas, les corrélations proposées
donnent les valeurs de l’indice CBR4j et γdopm avec une précision respective de 6 % et 1,7%. Les valeurs de
Wopm ont été obtenues quant à elle, avec une précision de 4,70%.

Figure 7: Variation de l’indice CBR à 4 jours d’imbibition mesuré et estimé

Figure 8: Variation de la densité sèche à l’optimum Proctor modifié mesurée et estimée

584
5 CONCLUSION
En définitive, les corrélations proposées peuvent être testées, en phase d’études conceptuelles, pour de la
sélection des sites d’emprunts de sols latéritiques destinés à la construction routière ou de barrages en terre. Ces
caractéristiques devant toutefois être validés par des essais de laboratoire en phase d’exécution. Les travaux pour
affiner ces corrélations sont en cours et concernerons à terme, un échantillon de matériaux latéritiques d’autres
régions au climat tropical d’Afrique et d’autres continents. Il y a également lieu de chercher la dépendance du
CBR dans les paramètres mécaniques comme, la compressibilité, la résistance au cisaillement avec une
méthodologie appropriée.

REFERENCES BIBLIOGRAPHIQUES

ASHOK Kumar Gupta, (2008) Relationship between the mean particle size factor, optimum moisture content and
permeability of sandy soil. 12th Conference of International Association for Computer Methods and Advances in
Geomechanics. Goa, India.
BOHI B. (2008) Caractérisation des sols latéritiques utilisées en construction routière : le cas de la région de l’Agnéby
(Côte d’Ivoire), Thèse de doctorat, École Nationale des Ponts et chaussées, Paris, France, p 143
DATTA T. and CHOTTOPADHYAY B.C. (2011) Correlation between CBR and Index Property of soils. Proceeding of
Indian Geotechnical Conference. Kochi, 131-133.
FANGLEU T. F. (2013) Validation des corrélations multiples entre les paramètres d’identification des sols et ses
caractéristiques CBR et à l’Optimum Proctor. Mémoire de fin d’étude, Ecole Nationale Supérieure Polytechnique de
Yaoundé, 2013.
Massamba NDIAYE (2013) Contribution à l’étude des sols latéritiques de l’Afrique et du Brésil, Thèse de Doctorat,
Université de Paris-Est et Cheik Antar DIOP, Dakar
MAGDI M. E. Zumrawi, (2012) Prediction of CBR value from index properties of cohesive soils, Annual conference of
Postgraduate studies and scientific Research. Hall, Khartoum, 111-117.
METCALF John B and ROMANOSCHI Stefan A, Predictions of maximum dry density and optimum moisture content from
simple material properties.
NCHRP, (2011) Guide for mechanistic empirical design of new and rehabilitated pavement structures. Final document.
Appendix CC-1: Correlation of CBR values with soil index properties. p204
NUGROHO S. A. and HENDRI A. (2012) Correlation between index property and California bearing ratio test of
Pekanbaru soils with and without soaked. Canadian journal on Environmental Construction and Civil Engineering. Vol.
3, N°1, 7-17.
OHANDJA N. (2011) caractérisation géotechnique et essais de litho stabilisation des graveleux latéritiques du département
du Dja et Lobo (Région du Sud) pour leurs utilisation en construction routière. Mémoire de Master II, de la faculté des
sciences de l’Université de Yaoundé I.
RASHMI Patel S. and DESSAI M. D: (2010), CBR predicted by index properties for alluvial soils of South Gujarat. Indian
Geotechnical conference 2010, Bombay, 79-83.
SAKLECHA P.P, KATPATAL Y.B., RATHORE S.S. and AGARAWAL D.K.: (2011), Spatial correlation between
properties of subgrade soil for foundation characterization. International Journal of Computer Application. Vol. 36,
N°11, 21-25.
F.C. UGBE: (2011) Estimating Compaction characteristic from fines in A-2 type lateritic soils. Research Journal of
environmental and earth Sciences 3 (4), 433-437.
YOUSSEF G. and G. ABDERAHMEN: (2007), Correlations between relative density and compaction test parameters.
Twelfth, International Colloquium on Structural and Geotechnical Engineering, Cairo-Egypt. p 8.

Interaction sols latéritiques-rejets des concentrateurs: quel(s) impact(s)?

Interaction lateritic soil–concentrators’discharges: what impact(s)?
MUKOKO KALENDA G.1, 2 *; THIMUS J-F.1; NGOIE NSENGA D.2
1
Université Catholique de Louvain (UCL), Pôle Génie-Civil et Environnemental (GCE)/ Louvain-la-
Neuve/Belgique
2
Université de Lubumbashi (UNILU), Département des Mines, Service de Géotechnique/ Lubumbashi/ RD
Congo
* [Link]@[Link],

RESUME - Dans la plupart des cas de ruptures des digues, le claquage hydraulique est la cause principale
évoquée. Plusieurs auteurs ont analysé les propriétés des sols sous sollicitations mécanique et hydrique.
Toutefois, ces analyses ne tiennent pas compte de l’influence chimique du liquide percolant la matrice de sol
chimisorbé. L’article analyse l’impact des effluents liquides émanant des concentrateurs sur le comportement des
sols latéritiques compactés dans les digues à rejets miniers. Il met l’accent sur les propriétés hydromécaniques
des latérites compactées et percolées par des effluents liquides de nature chimique différente. L'eau
déminéralisée a été prise comme prototype. Les tests ont inclus les mesures de la consistance, de la
compressibilité, et de la conductivité hydraulique quand les éprouvettes ont été saturées avec les différents
liquides. Les résultats montrent que les effluents liquides acide et basique affectent la consistance et les
paramètres de compressibilité. La perméabilité et le module de déformation sont sensiblement modifiés. L'effet
des succion et densité initiales avant l'imbibition a été aussi examiné, ce qui révèle que les latérites sont
susceptibles à l’effondrement lors du passage de l’état partiellement saturé à l’état saturé. Cependant, des
changements structurels causés par la saturation et la densité initiales des éprouvettes semblent être plus
importants pour expliquer le comportement volumétrique que l'interaction chimique.

ABSTRACT – In most of the cases of breaks of dikes, hydraulic fracturing is the evoked main cause. Several
authors analyzed the properties of soils under mechanic and hydric solicitation. However, these analyses do not
take into account the chemical influence of the liquid percolating the matrix of soil. The article analyzes the
impact of the liquid effluents emanating from concentrators on the behavior of compacted lateritic soil in
restraint dikes of mining discharge. It emphasizes the hydromechanical properties of the compacted laterites and
percolated by liquid effluents with different chemical nature. Demineralized water was taken as prototype. The
tests included the measurement of compressibility, shear strength and saturated hydraulic conductivity when the
specimens were imbibed with the different fluids. The results show that the acid and basic liquid effluents affect
the consistency and the parameters of compressibility. The permeability and the module of deformation are
appreciably modified. The effect of the initial suctions and the density before the imbibition was also
investigated, what reveals that laterites are susceptible to collapse (show potential collapse) during passage of the
partially saturated state to the saturated state, (on wetting path). However, structural changes caused by the initial
saturation and density of specimens seem to be more important for explaining the volumetric and deviator
behavior than chemical interaction.

1 INTRODUCTION

En dépit de retombées économiques dues à l’exploitation minière au Katanga méridional en RDC, l’industrie
minière génère de quantités importantes de rejets qui sont potentiellement polluants [01]. Ce sont notamment les
résidus de procédés de traitement minéralurgique du minerai qui sont les plus problématiques. En général, ces
résidus sont acheminés, sous forme de pulpe, vers des aires de stockage aménagées en surface, idéalement dans
les vallées où ils sont endigués. Dans ces situations, des digues de retenue des rejets miniers, en sols latéritiques,
sont mises en place pour assurer la gestion et la restauration des sites et préserver l’environnement.
Les effets chimiques de la percolation des effluents liquides sur quelques propriétés hydromécaniques de
latérite compactée avec et sans succion initiale sont examinés dans cet article. La conductivité hydraulique
saturée a été mesurée en utilisant de l'eau déminéralisée, et de l'eau polluée acide ou basique. Le comportement
volumétrique a été examiné en analysant la compressibilité et le gonflement œdométriques.

587
2 MATERIELS ET MATERIAUX UTILISES DANS CETTE ETUDE

2.1 Matériels et méthodes utilisés

Pour la caractérisation physique, nous avons effectué une analyse granulométrique par voie sèche et par voie
humide selon la norme CME 01.01. La teneur en eau a été déterminée par la méthode d’étuvage selon les normes
NF P94-050 et NBN 589-203 respectivement pour les latérités de Kakanda, de Kipushi et les latérites des
carrières. La détermination de la masse volumique absolue s’est faite au pycnomètre selon la procédure: N03-09.
Les limites d’Atterberg ont été obtenues selon la norme NF P94-051 et la limite de retrait suivant la procédure
Peltier 108. Le poids volumique maximal sec ainsi que la teneur en eau optimum ont été obtenus par un essai
Proctor Normal selon la norme NF P 94-093. Les matériels utilisés sont des appareils classiques de laboratoire
qui ne nécessitent aucune description particulière.
2.2 Matériaux utilisés dans cette étude: latérites et effluents liquides
Le sol étudié est la latérite utilisée dans la construction des remblais routiers et des digues de retenue des
rejets miniers par les entreprises minières et de génie civil du Katanga en R.D. Congo. La densité moyenne de
particules solides est de Gs = 2,77. Les échantillons ont été prélevés à Kakanda sur la grande digue de Kakanda,
à Kipushi sur la digue de Kipushi et dans des carrières où les matériaux de remblais routiers sont extraits.
Les effluents liquides utilisés proviennent des bassins de retention des digues de Kakanda et de Kipushi, ils
émanent des concentrateurs de Kipushi et de Kakanda. Ils ont un pH faiblement basique se situant entre 7,9 et
8,7 pour Kakanda et faiblement acide pour Kipushi avec des valeurs comprises entre 5,9 et 6,5.
La composition chimique des effluents liquides utilisés dans cette étude est présentée sur l’histogramme de la
figure 1. Le dosage des éléments chimiques (chimie totale) a été fait par fusion au métaborate et tétraborate de
Lithium suivie d’une mise en solution et une analyse des éléments par spéctrométrie d’emission atomique ICP
AES (Inductively Coupled Plasma Absoption Emission Spectral), modèle ICAP 2500.
La figure 1 montre la concentration en éléments polluants, ces éléments qui polluent l’eau des bassins des
digues ne sont tout de même pas en quantité faible au regard de la législation en vigueur en RDC (annexe XI du
code minier de la RDC) [03]. Ces ions contenus dans les effluents liquides ont une inﬂuence sur les propriétés de
la fraction argileuse de la latérite.

Figure 1. Composition chimique des effluents liquides des bassins

Une classification usuelle des polluants est présentée par Delage P. & Schrefler B. (2005), [02], elle
comprend: les polluants chimiques solubles; les métaux lourds; et les composés hydrocarbonés dérivés du
pétrole brut.
La même source indique trois catégories de polluants à savoir: métalliques (métaux lourds), minéraux (Ca,
Mg, Na, K, Fe, Si, Sr, Ti, Al, N, Cl, F, P, S et cyanures) et organiques (comprenant entre autres les
hydrocarbures, les alcools, les phénols, les esters, les sulfures, les pesticides…).
Les minéraux cristallins ont été identifiés par diffraction de rayons X (DRX). L’appareillage utilisé
comprend le diffractomètre D8 Advance de Brüker (cuka, I = 0,15418 nm, 40 kV, 30 mA et 2 thêta = 1/2
min), muni d’une anticathode en cuivre et une baie de mesure (Philips PW 1710) pilotée via un ordinateur
par le logiciel PC-APD.

588
Les minéraux argileux ont été identifiés sur la fraction < 2 microns. L’argile a été respectivement saturée par
les ions K et Mg. L’argile traitée au potassium a été soumise au chauffage à 105, 300 et 550°C, celle traitée au
magnésium a suivi le traitement à la solvatation à l’éthylène glycol (Mg eg). Les spectres ont été obtenus sur les
échantillons orientés, leurs interprétations montrent que les minéraux argileux sont principalement de la kaolinite
et de l’illite présents dans tous nos échantillons. La Capacité d’Echange Cationique (CEC) mesurée par la
méthode à l’acétate d’ammonium à pH 7 confirme la présence des minéraux argileux non gonflants et donc
moins actifs.
 Composition minéralogique des échantillons des latérites
En se basant sur les formules chimiques des minéraux secondaires néoformés lors de processus de
latérisation, les résultats des analyses chimiques élémentaires (chimie totale, DCB) et minéralogiques (DRX,
MEB), nous avons pu déterminer par calcul les quantités approximatives des minéraux majeurs contenus
dans nos échantillons. L’approche consiste à considérer le bilan de masse entre les compositions chimiques
mesurées exprimées en oxydes d’éléments chimiques contenus dans le sol et les compositions chimiques
classiques des minéraux monophasiques, en principe, de même nature que ceux identifiés par la DRX dans
les échantillons analysés [04]. Le Tableau 1 présente la composition minéralogique (en % massique).

Tableau 1. Composition minéralogique des échantillons de latérite et oxydes contenus dans le sol
Minéraux Kakanda Kipushi Oxydes Kakanda Kipushi
Quartz [%] 32,90 45,70 SiO2 [%] 56,27 66,34
Kaolinite et Halloysite [%] 34,87 19,77 TiO2 [%] 1,10 1,06
Illite [%] 11,64 16,24 Al2O3 [%] 16,31 15,62
Muscovite ou feldspath 2,23 4,93 Fe2O3 [%] 15,29 8,78
Hématite [%] - 8,78 MgO [%] 0,38 0,53
Gœthite [%] 17,02 - MnO [%] 0,02 0,01
Corindon [%] - 3,02 CaO [%] 0,08 0,06
Anatase [%] 1,10 1,06 K2O [%] 1,15 1,82
Calcite [%] 0,18 0,14 Na2O [%] 0,29 0,59
Pyrolusite [%] 0,02 0,01 PF* [%] 8,83 5,01
Total 99,96 99,65 Total [%] 99,72 99,82
PF*: Perte au Feu
3. IMPACTS DES INFLUENTS LIQUIDES SUR LES PARAMETRES DES LATERITES

3.1 Influence sur les limites de consistance et indices d’état

Les résultats des limites de consistance déterminées en utilisant les différents liquides (l’eau déminéralisée,
l’effluent acide du bassin de kipushi et l’effluent basique du bassin de kakanda) montrent que la plasticité du sol
est affectée par la pollution. Nous schématisons sur la figure 2, l’évolution de la consistance des sols latéritiques
en fonction du liquide de percolation.

Figure 2. Evolution de la consistance des latérites en fonction de l’effluent liquide

La littérature renseigne que la limite de liquidité, pour des argiles très actives telles que la montmorillonite et
la bentonite, est réduite pour des liquides de nature chimique différente que l’eau [05]. On peut de façon

589
simpliste expliquer ce changement du comportement d’argile en raison de l’interaction chimique avec le liquide
de percolation par adsorption d’ions du liquide par les minéraux d’argile et des changements conséquents de la
distance entre des plaquettes parallèles d’argiles en raison des forces répulsives ou attractives [05].
Les limites de consistance (wL, wP, ws) et les indices d'état (lP, lc,) qui en découlent traduisent de façon
quantitative des états physiques assez imprécis. Cependant, leur intérêt majeur est d'être reproductibles puisque
mesurés à partir d'essais normalisés. Les valeurs de ces limites et indices sont directement fonction de tous les
paramètres physico-chimiques (nature minéralogique, géométrie, hydratation, ions et électrolyte), aussi peut-on
considérer qu'ils constituent des paramètres de synthèse. Leur principal inconvénient découle d'ailleurs de cette
multidépendance qui amène les plages de valeurs à se recouper largement quand on compare plusieurs argiles
[06]
Les résultats des limites et indices de consistance de latérites montrent bien l’impact de la pollution sur la
consistance de latérite. D’une manière générale, nous avons observé une diminution de la limite de liquidité en
milieu acide et une légère augmentation en milieu basique. L’indice de plasticité croit sous pollution acide,
tandis qu’il diminue avec l’effluent liquide basique. Les valeurs de bleu de méthylène (VB) obtenues vont dans
le même sens que la limite de liquidité (wL): une légère augmentation dans le cas de l’effluent basique du bassin
de Kakanda où la VB passe de 4,03 ± 0,01 g/100 g pour le sol sain à 4,21 ± 0,03 g/100 g pour le sol pollué soit
une augmentation de 4,5%; et une légère diminution d’environ 3,9% dans le cas de l’effluent acide du bassin de
Kipushi (soit 2,98 ± 0,04 g/100 g contre 3,1 ± 0,03 g/100 g pour le sol sain). Ceci confirme les observations
faites sur les limites d’Atterberg, et montre que l’argile est consommée en milieu acide.
En effet, lorsque la concentration en ion augmente, les argiles assurent la fonction de tampon et les
protons peuvent séparer les cations etc… des argiles suivant la réaction:

(1)
où M+ représente un cation.
Lorsqu’on rajoute plus de protons (pH < 4,2), ce sont les complexes hydroxo d’aluminium qui jouent le rôle
de tampon ou alors des hydroxydes de fer pour un pH voisin de 2 [07].
En résumé, la présence de cations et leur échangeabilité modifie le développement de la double couche,
modifie la texture de l'argile, conditionne l'équilibre «eau-particules» et par conséquent, le comportement
mécanique du matériau.

3.2 Influence sur le pH et la CEC

La détermination du pH de sol est importante, comme l’acidité excessive ou l’alcalinité peuvent être
nuisibles. Les résultats expérimentaux indiquent que les valeurs de pH des échantillons de sol latéritique non
contaminés varient de 4,76 à 5,15 respectivement pour les échantillons de Kipushi et de Kakanda. Sunil
B.M., Shrihari S., Sitaram N., (2009) [08] trouvent pour le sol latéritique de Karnataka en Inde, des valeurs
de pH comprises entre 4,31 à 5,28; une latérite est donc un sol acide.
Quant à la CEC, les sols mouillés aux effluents liquides des bassins et séchés à l’air libre ont été emiettés et
écrêtés au tamis de 2 mm. Les sols ainsi pollués ont été soumis aux analyses chimiques. Le tableau 2 présente les
résultats de la CEC. De ces résultats, on peut voir que la CEC a augmenté d’une valeur initiale de 8,75 méq/100g
à 9,23 méq/100g, pour le sol pollué par l’effluent basique tandis qu’elle reste presque inchangée dans le sol
pollué par l’effluent acide. (6 à 5,98 méq/100g).
L’échange d’ions est le plus important de tous les processus se produisant dans le sol. La capacité du sol pour
adsorber et échanger des ions varie grandement avec le type et la quantité d’argile. Les phyllosilicates de type
kaolinite présentent sur les bordures des cristaux, là où les liaisons sont rompues, des sites pouvant
éventuellement interagir avec les cations présents en solution. Il s’agit de groupements hydroxylés de type
silanol (Si-OH) et aluminol (Al-OH). Quand l’effluent liquide basique percole le sol, les cations monovalents et
divalents comme les ions Ca2+, Mg2+, Na+ sont adsorbés sur la surface. N’importe quel processus qui favorise la
présence de hauts niveaux de bases échangeables réduira l’acidité et augmentera l’alcalinité [09] [10].
Tableau 2. Cations échangeables et Capacité d’Echange Cationique (CEC)
Echantillon Kakanda non pollué Kipushi non pollué Kakanda pollué Kipushi pollué
Ca [mg/kg] 243,95 51,55 256,76 69,39
K [mg/kg] 54,25 56,48 77,21 56,29
Mg [mg/kg] 89,05 35,44 95,63 54,63
Na [mg/kg] 11,84 6,34 12,47 8,32
Ca [ECEC] 2,14 2,01 2,21 2
CEC [méq/100g] 8,75 6 9,23 5,98
Ca/Mg 2,74 1,46 2,68 1,45
Nbre d’essais 3 3 5 5

590
La dissociation de l’hydrogène présent dans les groupes SiOH aux bords des couches tétraédriques et
AlOH aux bords des couches octaédriques conduit à l’acquisition de charges négatives supplémentaires selon
ces réactions:
et (2)
Les réactions chimiques entre ions de la solution et groupements fonctionnels ionisables de la surface
dépendent du degré d’ionisation et, de ce fait, du pH de la solution. La dissociation augmente avec le pH
croissant et aboutit à une augmentation de la capacité d’échange de cations [10].
3.3 Influence sur la conductivité hydraulique saturée de latérite compactée
Les mesures ont été effectuées à l’aide d’un perméamètre à circuit ouvert et à charge hydraulique
variable. Les éprouvettes ont étés compactées à 100, 95, 90 et 85 % de l’OPN, et saturées en eau distillée ou
polluée. Les résultats de ces tests qui sont résumés sur la figure 3 montrent la variation de la conductivité
hydraulique saturée du sol en fonction de la compacité pour les différents liquides saturant (à gauche),
ainsi que celle de la porosité des éprouvettes après percolation (à droite).

Figure 3. Variation de des sols pollué et non pollué et de la porosité après essais en fonction de l’état initial
La conductivité hydraulique saturée du sol latéritique percolé par l’effluent liquide acide augmente quand
on la compare à sa valeur de référence (essai avec l’eau déminéralisée).
La conductivité hydraulique saturée de sol latéritique compacté à des densités différentes et percolé par
l’eau déminéralisée croît dans la gamme de 50 à 150 % lorsque le liquide percolant est remplacé par
l’effluent liquide acide; la porosité des éprouvettes augmente après percolation par l’effluent liquide acide.
La conductivité hydraulique saturée et la porosité de sol sont probablement altérées en raison des réactions
chimiques de l’effluent liquide acide avec le sol.

3.5 Influence sur la compressibilité à l’œdomètre de latérite compactée

Dans le but d’analyser le comportement du sol en présence d’un milieu favorable à l’échange ionique, les
éprouvettes ont été confectionnées avec les mêmes paramètres initiaux et saturées à l’eau polluée issue des
bassins de rétentions ou à l’eau distillée. L’effet des succions et densités initiales avant l’imbibition a été aussi
examiné en préparant certaines éprouvettes plus ou moins sèches avec une teneur en eau très faible et en les
soumettant à des incréments de contrainte verticale à sec jusqu’au palier de 1 MPa; la saturation est faite à la fin
de ce palier. Le palier suivant est appliqué 72 heures après remplissage de la cellule œdométrique en liquide. Les
résultats des essais œdométriques exécutés sont présentés sur les figures 4 et 5.

 Effet de l’effluent liquide acide ou basique des bassins de rétention sur la compressibilité de latérite.
Dans chaque cas, on remarque que l’effluent acide entraine un gonflement initial plus faible et un tassement
final plus important pour les échantillons pollués (figure 4 a, b, c et d). L’allure des courbes est également
légèrement différente, ce qui indique des paramètres de compressibilité différents. Sous l’effluent basique, les
courbes montrent d’abord une phase de gonflement due à la saturation du sol sous faible charge appliquée au
premier palier; l’échantillon saturé en eau déminéralisée gonfle plus que son équivalent saturé en eau polluée et
se tasse plus que ce dernier (figure 4-e-f). Les deux courbes de gonflement présentées à la figure 4-g montrent
que le gonflement est atténué en milieu pollué basique, ces résultats concordent avec ceux de la figure 4-e-f.

591
Figure 4. Inﬂuence de l’effluent liquide acide (a, b, c, d) ou basique (e, f, g) sur les courbes œdométriques.

 Effet des effluents liquides acide ou basique sur l’effondrement des sols latéritiques compactés
Quelques résultats d’une vingtaine d’essais d’effondrement réalisés à l’œdomètre sont présentés sur les
graphiques de la figure 5. L’effondrement observé sous la contrainte de 1MPa sous l’eau déminéralisée est
inférieur à celui observé sous l’effluent liquide acide pour les mêmes états initiaux des éprouvettes et les mêmes
états de chargement (figure 6-a-b-c-d). Ce comportement pourrait être dû à une forte et rapide destruction des
liants (carbonates, hydroxydes de fer et d’aluminium) par l’acidité du fluide saturant comparativement à
l’hydratation par l’eau distillée. En effet, la kaolinite qui est l’une des argiles qui assurent dans les sols la
cohésion entre les constituants de tailles plus importantes grâce au colmatage établi par les aggrégats au niveau
de la jonction entre les grains, n’est pas soluble en un seul acide, mais adsorbe en surface des ions aluminium
issus de l’environnement acide. Il en résulte des réactions très complexes induisant la déstabilisation du
colmatage ce qui conduit à l’effondrement supplémentaire de la structure. Par contre en milieu basique,
l’effondrement est inférieur à celui enregistré sous l’eau déminéralisée (figure 6-e-f); ceci pourrait s’expliquer
par l’apport de ciment (chaux éteinte).

592
Figure 5. Courbes d’effondrement en milieu acide en fonction du degré de saturation initiale

Le tableau 3 résume les caractéristiques mécaniques de compressibilité en milieu acide et le tableau 4

présente les valeurs chiffrées des tassements et paramètres mécaniques enregistrés sous l’effluent liquide
basique. Ces résultats révèlent l’impact de la pollution du sol latéritique par les effluents émanant des
concentrateurs sur la déformabilité et la résistance de ce sol.
Tableau 3. Caractéristiques mécaniques de compressibilité sous l’effluent liquide acide.
Eœd [MPa] av [MPa-1] C [-] A [-]
Moyenne de 10 essais déminéralisés 18,2 0,06 25,1 224,5
Moyenne de 10 essais pollutes 16,2 0,06 22,4 255

Tableau 4. Tassements et modules pour les sols sain et contaminé par l’effluent liquide basique
. In situ Proctor
Tassement Pollué Déminéralisé Pollué Déminéralisé
Rapport des tassements : 3,8 % 5,9 % 1,3 % 2%
pollué/déminéralisé 0,64 0,65
C [-] 60 33 169 111
Eoed [MPa] 8,44 4,65 23,71 15,56
E [MPa] 5,69 3,13 16 10,5

4. CONCLUSION

L’objectif de cette étude consistait, à l’analyse des paramètres hydromécaniques des sols latéritiques
compactés sous influence de la pollution par les effluents liquides qui émanent des concentrateurs.
L’augmentation de la concentration en ions Ca2+ et Al3+ due à l’effluent liquide acide diminue la limite de
liquidité. Ceci pourrait être dû à la diminution des forces d’attraction entraînant une plus grande mobilité des
particules. Par contre wL subit un léger accroissement à cause de l’augmentation de la concentration en ions Na+
en présence de l’effluent liquide basique, ce qui tend à diminuer les forces de répulsion entraînant une tendance à
la floculation et conjointement un accroissement de wL.
Le pH du sol latéritique acide est modifié en fonction de la nature chimique de l’effluent liquide. L’ensemble
des résultats des analyses sous pollution conduit aux mêmes constats: d’une part, une diminution des paramètres
mécaniques de compressibilité (C, A, Eœd) et du pH et une augmentation de la conductivité hydraulique saturée
et de l’indice de plasticité lorsque les spécimens ont été saturés à l’effluent liquide acide et d’autre part, un
comportement tout à fait inverse en milieu basique. Le gonflement est, lui aussi, fortement atténué par l’effluent
liquide basique.

593
L’effondrement observé sous l’eau déminéralisée est inférieur à celui observé sous l’effluent liquide acide.
Cela suggérerait une forte et rapide destruction des liants (carbonates, hydroxydes de fer et d’aluminium) par
l’acidité du fluide comparativement à l’hydratation par l’eau distillée. Il faut noter que la chute des
caractéristiques de résistance conduisant à l’effondrement est d’autant plus grande que la teneur en eau initiale
de mise en œuvre est faible. Par contre en milieu basique, l’effondrement est inférieur à celui enregistré sous
l’eau déminéralisée; ceci pourrait s’expliquer par l’apport de cations monovalents et divalents de la solution
environnante pour compenser les charges négatives supplémentaires dues à la dissolution des groupements
fonctionnels aluminol et silanol aux abords des feuillets de la kaolinite, suite à l’augmentation du pH, ce qui
freine tant soit peu la déformabilité.

REFERENCES BIBLIOGRAPHIQUES

[01] Kaniki T. A. (2008), - Caractérisation environnementale des rejets minero-métallurgiques du copperbelt

congolais. Thèse de Doctorat en Sciences de l’Ingénieur. Université de Liège, Belgique.
[02] Bouazza A., Delage P., Wojnarowicz M., (2005), - Nature et reconnaissance de la pollution des sols. Dans
Géomécanique environnementale sols pollués et déchets. Chap. 1, p 17- 45. Sous la direction de Delage P., Schrefler
B., Ed. Hermès-Science Publication, Lavoisier, Paris.
[03] Cabinet du Président de la RDC, (2003) - Décret n°038/2003 du 26 mars 2003 portant règlement minier. Journal officiel
de la RDC. Numéro Spécial, 44ème Année.
[04] Pialy P., (2009), - Étude de quelques matériaux argileux du site de Lembo (Cameroun): minéralogie, comportement au
frittage et analyse des propriétés d’élasticité. Thèse, Université de Limoges
[05] Cardoso R., Barbedo D., Ribeiro D., (2012), - Hydro-Mechanical Behavior of Compacted Kaolin Considering
Percolating Fluids with Different Chemical Nature. Dans Unsaturated soil: Reseach and application, volume 2
Springer, ISBN 978-3-642-31342-4 e-ISBN 978-3-642-31343-1, DOI 10.1007/978-3-642-31343-1, p439-446.
[06] Mukoko G. (Août 2014) – Comportement des sols latéritiques compactés dans les remblais et digues de retenue des
rejets miniers du Katanga (R.D.C),Thèse de doctorat, Presse universitaire de Louvain, Dépôt légal D/2014/9964/30,
ISBN 978-287558-314-7, Diffusion universitaire Ciaco, [Link], Belgique.
[07] Bliefert C., Perraud R., (2009), - Chimie de l’environnement. Air, eau, sol, déchets, 2è édition De Boeck Université.
Bruxelles. ISBN 978-2-8041-5945-0. 478p.
[08] Sunil B.M., Shrihari S., Sitaram N., (2009), - Shear strength characteristics and chemical characteristics of leachate-
contaminated lateritic soil. Engineering Geology 106 : 20–25
[09] Giroud J-P, Bottero A, (1972), - Influence des propriétés physico-chimiques des argiles monominérales sur leur
comportement mécanique. [Link] [Link] Ch.-62-nov.-déc.-Réf.1204.
[10] Tertre E., (2005), - Adsorption de Cs+, Ni2+ et des lanthanides sur une kaolinite et une smectite jusqu’à 150°C: Etude
experimentale et modélisation. Thèse de doctorat, Université de Toulouse III.

Prediction of Compaction Density of Lateritic Soil Base Pavement Layers from

Dynamic Cone Penetration Values in Uganda
Prédiction de la densité de compactage des couches d’assise de la chaussée à partir des
résultats d’essais de pénétration dynamique au cône
S. Jjuuko*1, D. Kalumba1, H.K. Mutaasa2 and J. Lukanda2
1
University of Cape Town, Cape Town, South Africa
2
Ndejje University, Kampala, Uganda
*
smljju001@[Link]

ABSTRACT – The study focused on obtaining a correlation between sand replacement method density values
and dynamic cone penetration (DCP) rate values (mm/blow) for base pavement layers of lateritic soils in
Uganda. A maximum depth of 500 mm of gravel of lateritic soil on existing roads, in Rubaga division in
Kampala (Uganda), was penetrated using the DCP to obtain the rate of penetration (mm/blow). They were also
investigated for their index properties and field density values using sieve analysis, Atterberg limits and sand
cone replacement methods. The gravel was generally classified as reddish-brown clayey sand with an AASHTO
class of A-2-6. The correlation obtained between dry density values from the sand-cone replacement method and
DCP rates showed an inverse relationship with R2 coefficient of 0.5999 indicating a moderate correlation
between the two variables. The correlation was validated with data from other lateritic base layers of roads in
Wakiso district, Uganda.

RÉSUMÉ – Cette étude a porté sur l'obtention d'une corrélation entre les valeurs de la densité de sol déterminées
avec le densitomètre à membrane et celles du taux d’enfoncement dans le sol du pénétromètre dynamique à cône
(PDC) (mm/coup). Une profondeur maximale de 500 mm de sols graveleux sur les routes existantes, dans la
division de Rubaga du district de Kampala en Ouganda (route de Mutesa 1, de Sentema et des environs de
l’hôpital Mengo), a été pénétrée à l'aide du PDC pour obtenir le taux d’enfoncement (mm/coup). Ils ont été
également étudiés pour leurs indices et valeurs de la densité de sol sur le terrain à travers les essais d'analyse
granulométrique, de limites d'Atterberg et de densitomètre à membrane. Le sol graveleux était généralement
classé comme sable argileux brun-rougeâtre avec une classe de l'AASHTO de A-2-6. La corrélation obtenue par
une simple analyse de régression linéaire, entre les valeurs de la densité sèche du densitomètre à membrane et
celles du taux d’enfoncement du PDC a montré une relation inverse avec un coefficient, R2, de 0,5999 indiquant
une forte corrélation entre les deux variables. La corrélation était validée avec les données de la route de
Matugga dans le district de Wakiso. La corrélation peut être utilisée pour prédire des valeurs de la densité sèche
à partir du taux d’enfoncement du PDC avec une précision de plus de 90%, supérieure à 50%, donc valide et
fiable.

1 INTRODUCTION

All civil engineering construction works including construction of dams, buildings, highways and other
structures are founded on or in soil. As such the stability and hence performance of such construction works are
dependent on the suitability of the foundation and quality of soils (Rahaman, 2010). However, uncertainties are
abundant in geotechnical engineering and possible sources include inherent variability, measurement errors and
modeling uncertainties. In recent years, the probabilistic framework has been adopted to rigorously quantify such
uncertainties. The standard way of achieving so is to transform the information of the field or laboratory test
data, denoted by “test indices” later, into the probability density functions (PDF) of the quantities of interest.
Quite frequent, transformation equations between soil parameters are very useful in converting test index
information into field density information through parametric correlation (Kulhawy and Mayne, 1990). Though
very much needed in the determination of parameters for pavement design and monitoring in Uganda, their
application is still very limited.
Pavement structure design in Uganda is based on three factors; loading (projected traffic), paving material
properties (strength, aging, environmental effects, etc.), and subgrade support. But many uncertainties exist in
pavement design; even when a road is opened to traffic, the engineer cannot verify the accuracy of the traffic
projection until the project has been through its design life. During the design stage the engineer selects a
subgrade support value based on a few samples taken from the project site, some engineering assumptions and

595
evaluation of indices. The engineer controls paving material properties through quality assurance/ control
(QA/QC) programs during construction. Most countries use density of the in- place subgrade and unbound base
for construction quality control, (Kulhawy and Mayne, 1995).
In Uganda, the dry density of the compacted soil or pavement material, calculated from field density, is a
common measure of the amount of the compaction achieved during construction. There are several methods for
the determination of field density of soils such as core cutter method, sand replacement method, rubber balloon
method and heavy oil method. Present practice mainly involves the use of the sand-cone method and a nuclear
gauge. Though the nuclear gauge is quick and very convenient to obtain the in-situ soil density and water
content, it is very expensive and radioactive. It also requires a special operator with a registered operating
license. The sand replacement method (SRM) is time consuming and demand significant effort, (Peck et al.,
1974). Therefore a study into a correlation between the values obtained from these methods and other frequently
utilized non-destructive insitu methods used in the determination of the strength of base layers is necessary.
In-situ penetration tests like DCP, SPT and CPT have been widely used in geotechnical and foundation
engineering for site investigation in support of analysis and design. The DCP is the most widely used, non-
destructive insitu, test in determining the strength of compacted base pavement layers in Uganda. In this study,
the correlation between sand replacement method density values and dynamic cone penetration (DCP) rate
values (mm/blow) in lateritic base pavement layers in Uganda was obtained. Laterite soils, otherwise called red
tropical soils, are extensively used in the tropics, specifically in Uganda, for road construction (Arulanandan and
Turnbridge, 1969), since they are readily available and relatively cheap where haulage distances are short.

2 MATERIALS AND METHODS

2.1 Sampling and sample preparation

Lateritic soil samples were collected randomly from different sections of existing roads around Ndejje
University, Kampala campus, located in Rubaga division Kampala district, Uganda. It involved digging shallow
trial pits and carefully taking disturbed bulk samples from depths between 0.1 and 0.3 m by means of pick-axe
and shovel. Depths between 0.1 and 0.3 m were chosen to limit the sampling within the base layer of each road.
The samples were then carefully put in black and airtight polythene bags to limit moisture loss and labelled
according to sampling location. 5 – 10 kgs of sample material were obtained at each location, deemed enough to
carry out all the required tests. Table 1 shows details in terms of sampling locations and depth, field soil
description and moisture content.
The bulk samples were brought to the laboratory and air-dried for two to three (2-3) days depending on the
initial moisture content. They were then riffled to obtain representative samples of the subsequent laboratory
tests. This initial preparation was done following the procedures described in BS 1377: part 1:1990.

Table 5. Sampling location and depth and description of sampled material

Moisture
Sampling depth
Sampling location Sample description content
(m)
(%)
Sentema Rd, CH 0+400 CL
0.1 – 0.30 Reddish brown clayey sandy gravel 6.4
32.525258E, 0.332647N
Sentema Rd, CH 0+800 LHS
0.1 – 0.30 Reddish brown very dense sandy clayey gravel 8.6
32.524082E, 0.333384N
Sentema Rd, CH 1+200 RHS
0.1 – 0.30 Reddish brown light clayey gravel 6.6
32.5021727E, 0.334408N
Mengo Rd, CH 0+050 CL
0.1 – 0.30 Reddish brown very dense sandy gravel 8.6
32.558293E, 0.311203N
Mengo Rd, CH 0+250 LHS
0.1 – 0.30 Reddish brown very dense sandy gravel 8.6
32.557669E, 0.311709N
Mutesa 1 Rd, CH 0+300 CL
0.1 – 0.30 Reddish brown clayey sandy gravel 6.3
32.558585E, 0.323248N
Mutesa 1 Rd, CH 0+800 LHS Light reddish brown, very dense sandy clayey
0.1 – 0.30 8.7
32.555872E, 0.326440N gravel.
Mutesa 1 Rd, CH 1+300 RHS
0.1 – 0.30 Reddish brown clayey sandy gravel 6.4
32.554825E, 0.316205N

596
2.2 Laboratory tests

2.2.1 Moisture Content test

The objective of the moisture content test was to determine the amount of water in the soil expressed as a
percentage of the dry mass of the soil sample based on BS 1377: Part 2:1990. Containers were cleaned and
weighed (M1), crumbled sample placed in the containers and weighed again (M3), and the weights recorded.
They were then placed in the oven for 24 hours at temperature of 1100C. After 24 hours, the containers with the
dried soil were removed from the oven, cooled to room temperature and weighed (M2).

2.2.2 Atterberg Limits test

These were done to determine the plasticity characteristics of the soil samples. The parameters that were
determined under the test included liquid limit, plastic limit and plasticity index. The test sample was obtained
by sieving the air- dried specimen through 0.425mm sieve. The sieved material was prepared and tested for
liquid limit in accordance with BS 1377: part 2:1990 clause 4.3 using the cone penetrometer method. On the
same sample, plastic limit according to BS 1377: part 2:1990 clause 5.3 was determined and finally plasticity
index (PI) was calculated as PI (%) = LL- PL.

2.2.3 Particle size analysis

The sieve analysis test was carried out to determine the particle size distribution and gradation properties of the
obtained soil samples. Wet sieving of the samples was done following procedures based on BS 1377:part2:1990
clauses 7.3 and 7.4.5.

2.3 Field Tests

2.3.1 Determination of in- situ Density Test

The sand replacement method was carried out according to BS 1377: part 9:1990 (Test 15). The surface of the
testing point was cleaned and leveled. A metal tray with a central hole of 100 mm was placed on the prepared
surface. A 150 mm deep hole was excavated, as material was collected in an airtight polythene bag. Weights of
excavated material were recorded and their moisture contents determined. A known weight of sand was poured
into the calibrating cylinder up to the top and placed centrally over the excavated hole. The shutter was opened
till the sand filled the excavated hole and the cone completely. The shutter was closed and the weight of the
remaining sand in the cylinder measured. The volume, bulk and dry densities of excavated soil were determined.

2.3.2 Dynamic Cone Penetrometer(DCP) test

The DCP test was carried out in accordance with BS 1377: part 4:1990. The steel rule attached to the guide foot
was placed through the slot in the hand guard. The foot was then placed on the surface to be tested and cone tip
passed though the guide hole. The entire apparatus was held by the handle perpendicular to the surface as the
technician observed the readings on the rule at the top of the hand guard and recorded as zero reading. The drop
weight of 8 kg was then raised to its maximum height and released through a falling height of 575 mm. Extreme
care was taken to gain maximum height for each drop but not to strike the handle base as this would cause the
instrument to withdraw and results affected. The number of blows and depth reading were recorded on the DCP
test forms. For this study, DCP testing reading required were up to a depth of 150mm corresponding to the depth
at which the sand replacement test was conducted. The field data was condensed in terms of penetration versus
corresponding cumulative number of blows, plotted vertically along the y- axis and horizontally along the x- axis
respectively. The slope values were then computed as the change in penetration versus the change in the number
of blows observed over the range – expressed as mm/blow.

3 RESULTS AND DISCUSSION

3.1 Index tests

The percentages of the lateritic soil samples passing BS Sieves 2.36 mm, 425 µm and 75 µm are shown in Table
2. The percentages passing through No. 200 (75 µm) BS sieve ranged between 0.3% and 38.1%. Samples from,
Mutesa 1 road CH 0+800 LHS and CH 1+300 RHS whose percentage finer than 75µm was greater than 35%
indicated fine grained soils while the rest were coarse grained soils. The soil samples were suitable for subgrade
construction since their percentages by weight finer than No. 200 BS sieve was less than 35%, according to the
Uganda Ministry of Works, Housing and Communication (MWH&C) general specifications for national roads of
2004. Soil samples from Sentema road CH 0+400 CL and CH 1+200 LHS belonged to A-7-6 class while the
other samples were classified as A-2-6 according to AASHTO method of classification.

597
Table 6. Index Test Results
Grading (% passing BS
Atterberg Limits
sieve) Grading AASHTO
Location
LL PL PI Modulus Classification
2.36mm 425µm 75µm
(%) (%) (%)
CH 0+400 CL
27.7 13.3 0.6 43.7 37.3 16.4 2.58 A-7-6
32.525748E, 0.332647N
Sentema

Road

CH 0+800 RHS
32.7 18.3 0.3 35.5 18.1 17.4 2.49 A-2-6
32.524082E, 0.333384N
CH 1+200 LHS
51.0 33.1 22.1 40.9 20.2 20.7 1.94 A-7-6
32.521727E, 0.334408N
CH 0+400 Cl
77.1 55.8 13.3 31.0 18.8 12.2 1.54 A-2-6
Mengo
Road

32.558293E, 0.311203N
CH 0+250 LHS
63.2 40.7 12.2 28.8 17.6 11.2 1.84 A-2-6
32.557669E, 0.311709N
CH 0+300 Cl
52.1 42.5 33.9 33.7 19.6 14.1 1.72 A-2-6
32.558585E, 0.323248N
Mutesa 1

Road

CH 0+800 LHS
36.2 22.5 35.9 35.9 21.4 14.5 2.36 A-2-6
32.555872E, 0.326440N
CH 1+300 RHS
53.5 36.6 38.1 38.1 22.4 15.7 1.84 A-2-6
32.554825E, 0.316205N

The liquid limits of the finer particles of soil samples ranged between 28.8% and 43.7% and plastic limits
between 17.6% and 37.3% while plasticity indices were between 11.2% and 20.7% as shown in Table 2. Soils
with liquid limit less than 30% are considered to be of low plasticity, those with liquid limit between 30% and
50% exhibit medium plasticity and those with liquid limit greater than 50% exhibit high plasticity. Other than
the sample obtained at Mengo road CH 0+250 LHS which exhibited low plasticity, the rest of the samples
exhibited medium plasticity primarily indicating that they were fair to good subgrade materials according
AASHTO classification.

3.2 Dry Density and DCP Test

Sand replacement and DCP tests were carried out at thirty six (36) different chainages along Sentema, Mengo
and Mutesa 1 roads (15 points, 6 points and 15 points for Sentema, Mengo and Mutesa 1 roads respectively).
The field dry densities as obtained from the sand replacement test ranged between 1701 kg/m3 and 2333 kg/m3
representing a range of typical dry densities usually obtained for lateritic soils (Ministry of Works, Housing and
Communications Specifications (Roads and Bridges), 2004) while the DCP values ranged between 0.45
mm/blow and 3.44 mm/blow as depicted in Figure 1. These values did not vary greatly on a given road tested.

3.3 Correlation between Dry Density and DCP Values

To obtain a good correlation, the dry density values of the lateritic soil base layers were divided with the
corresponding DCP values to get the ratio DDDCPRATE. The DCP rate values were then plotted against the
corresponding DDDCPRATE values on a natural logarithmic scale as shown by the blue line (Correlation) in Figure
1 with R2 of 0.5999. Hence the correlation between dry density and DCP rate was written as;

æ Dry density (kg/m 3 ) ö

DCP Rate (mm/blow)=-0.949ln ç ÷ + 8.7417
è DCP Rate (mm/blow) ø
(1)
Dry density and DCP penetration values from two different other roads were used to validate the data and plotted
as Validation data 1 and Validation data 2 in Figure 1.

598
Figure 6. Plot of Dry density vs DCP rate

3.4 Validation of the developed model

The model that was developed for the lateritic soils was validated using data from two other independent roads
(Matugga – Wakiso road and Lugoba road) with lateritic soil base layers in order to test its reliability and
accuracy. The obtained (estimated from equation 1) dry densities were compared with the actual dry density
values obtained by sand replacement method from these roads. The error was calculated as a magnitude of the
difference between the actual and estimated values as shown in Table 3 and 4. From statistical analysis carried
out on the errors, it was determined that the model can be used to predict values of dry density with an average
accuracy of 51.6%. This value is greater than 50%, therefore the model is valid and reliable.

Table 7. Matugga – Wakiso road validation

Actual Dry Density Actual DCP rate Estimated Dry Density (x-µ)2 Dry Density
Error Dry Density (x-µ) Dry Density
(kg/m3) (mm/blow) (kg/m3)
2087 2.00 2434 347 -160.6 25792.36
2078 2.20 2168 90 -417.6 174389.76
2043 2.50 1796 247 -260.6 67912.36
2138 1.50 3091 953 445.4 198381.16
2115 1.73 2798 683 175.4 30765.16
2060 2.09 2313 253 -254.6 64821.16
2012 3.04 1236 776 268.4 72038.56
2040 2.67 1604 436 -71.6 5126.56
2053 1.87 2610 557 49.4 2440.36
2050 1.87 2610 560 52.4 2745.76
2125 1.67 2877 752 244.4 59731.36
2114 1.73 2798 684 176.4 31116.96
2135 1.58 2993 858 350.4 122780.16
2055 2.47 1832 223 -284.6 80997.16
2061 2.76 1508 553 45.4 2061.16
2100 1.97 2474 374 -133.6 17848.96
2078 2.20 2168 90 -417.6 174389.76
2075 2.21 2155 80 -427.6 182841.76
2150 1.49 3103 953 445.4 198381.16
2115 1.73 2798 683 175.4 30765.16
µ = 507.6 0 sum = 1,545,326.80

Standard deviation = = = 285

Coefficient of Variance =

599
Table 8. Lugoba road validation
Actual Dry Density Actual DCP rate Estimated Dry Density (x-µ)2 Dry Density
Error Dry Density (x-µ) Dry Density
(kg/m3) (mm/blow) (kg/m3)
2080 2.10 2300 220 -610.85 373137.72
2090 1.70 2838 748 -82.85 6864.12
2155 1.30 3308 1153 322.15 103780.62
2078 2.40 1916 162 -668.85 447360.32
2160 1.10 3455 1295 464.15 215435.22
2060 2.90 1367 693 -137.85 19002.62
2060 2.50 1796 264 -566.85 321318.92
2135 1.40 3206 1071 240.15 57672.02
2053 3.20 1100 953 122.15 14920.62
2160 1.10 3455 1295 464.15 215435.22
2160 1.10 3455 1295 464.15 215435.22
2155 1.30 3308 1153 322.15 103780.62
2060 2.90 1367 693 -137.85 19002.62
2045 3.50 877 1168 337.15 113670.12
2061 2.70 1571 490 -340.85 116178.72
2120 1.50 3091 971 140.15 19642.02
2045 3.50 877 1168 337.15 113670.12
2100 1.90 2569 469 -361.85 130935.42
2110 2.00 2434 324 -506.85 256896.92
2160 0.60 3192 1032 201.15 40461.32
µ = 830.85 0 sum = 2,904,600.55

Standard deviation = = = 391

Coefficient of Variance =

3.5 Practical application

The developed model was validated and found to have an average accuracy of 51.6% just above 50%. Therefore,
it is advisable that the model be used only on rural low volume roads with small budgets in African countries. It
should be used with experienced engineers and professionals in lateritic soils.
It is also advisable that the model be used with DCP penetration rates in the range of 2.0 to 2.7 mm/blow.

4 CONCLUSION

The study was conducted to investigate the level of compaction (densities) of lateritic soil base pavement layers
using a correlation approach of sand replacement method (dry density) and penetration rate (mm/blow) from
DCP test. An inverse relationship existed between the dry density and DCP rate values (mm/blow). The
statistical parameters indicated that the model developed for correlating dry density values with DCP rate fitted
the data well with a coefficient of correlation of R2 of 0.5999 indicative of a moderate correlation between the
two variables. From statistical analysis carried out on the errors, it was determined that the model can be used to
predict values of dry density with an average accuracy of over 51.6%. This value is greater than 50%, therefore
the model is valid and reliable. With the additional economic benefits involved, this model is recommended for
experienced engineers.

REFERENCES

ASTM Standard D4318-05 (2005) Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils.
ASTM International, West Conshohocken, PA, USA.
Basiles C. (1991) A comparison of the Casagrande and fall cone penetrometer methods for liquid limit determination in marls
from Crete, Greece. School of Geology (Laboratory of Engineering Geology- Hydrogeology), Aristotle University,
Elsevier Science Publishers B.V., Amsterdam.
British Standard Institution (1990) BS 1377, Methods of test for soils for civil engineering purposes. General requirements
and sample preparation. British Standard Institution, London, UK.
British Standard Institution (1999) BS 5930, Code of practice for site investigations. British Standard Institution, London,
UK.
Kulhawy, F.H., Mayne, P.W. (1990) Manual on estimating soil properties for foundation design (No. EPRI-EL-6800).
Electric Power Research Inst., Palo Alto, CA (USA); Cornell Univ., Ithaca, NY (USA). Geotechnical Engineering Group.

600
Ministry of Works, Housing and Communications. (2004) General Specifications For Road and Bridge Works. Series 3000;
Earthworks and Pavement Layers of Gravel or Crushed Stone.
Peck, R.B., Hanson, E.W., Thornburn, H.T. (1974) Foundation Engineering. 2nd. Edition. New York: John Willy & Sons, 7-
8.
Rahman Q.B.A. (2010) Correlation between California Bearing Ratio (CBR) results and physical properties of soils.
Universiti Teknologi Malaysia.

601
602
ARGILES GONFLANTES

EXPANSIVE CLAYS

Duplicate testing conducted on the input parameters for the estimation of

potential expansiveness of clay.
Essais dupliqués sur les paramètres d’entrée pour l’estimation du potentiel de gonflement des
argiles.
Wessel BADENHORST1, Elizabeth THERON2, Philip STOTT3
1,2,3
Central University of Technology, Bloemfontein, South Africa

ABSTRACT – This paper presents a study conducted on three active clay soils analysed at seven commercial
laboratories in South Africa. Commercial test results are often used as input parameters for prediction models
used to estimate potential heave expected from active clays, especially when designing foundations for light
structures. This paper briefly looks at the typical results obtained from such laboratories and comments on the
correlation achieved.

RÉSUMÉ – Cet article présente une étude menée sur trois argiles actives testées dans sept laboratoires privés en
Afrique du Sud. Les résultats de ces essais sont d’habitude utilisés comme paramètres d’entrée de modèles de
prédiction en vue d’estimer le potentiel de soulèvement d’argiles actives, en particulier lors du dimensionnement
des fondations de structures légères. Le présent article évalue les résultats typiques de ces essais de laboratoire et
discute les corrélations proposées.

1 INTRODUCTION

The American Society of Civil Engineers estimates that of all homes in the United States, a quarter has had some
damage caused by expansive soils. The financial loss to property exceeds that of earthquakes, floods, hurricanes
and tornadoes combined.
In Africa there exists a widespread problem of providing economic housing for lower income communities.
In South Africa the government is attempting to provide small subsidised houses for the very poor. Most of
South Africa has semi-arid and sub-humid conditions (Weinert, 1980) which lead to generally shallow residual
soils subject to seasonal saturation and aridity. Such conditions are known for giving expansive foundation
problems.
Advanced mathematical solutions are available for heave prediction, but unfortunately they rely on the input
of the soil water characteristic curve (swcc) for which the typical cost is US $10 000 (Fredlund 2009) - more
than the cost of a low cost house. Besides the cost, the time needed to develop the swcc takes in excess of 6
weeks.
In practice almost all low-cost housing relies on empirical methods – estimation of heave potential based on
quick, simple and economical tests. The most popular method in South Africa (Van der Merwe 1964), uses
Atterberg limits and particle size distribution These tests are also known as “Foundation Indicators”. Predictions
of heave are only as reliable as the input parameters. This paper focuses on the reliability of these input
parameters in the form of test results received from seven premier commercial laboratories.

2 Particle size analysis of soil samples

Particle size analysis of soil samples is a combination of grading analysis using sieves with different size
apertures and hydrometer analysis which uses settlement and Stoke’s law to determine the finer fractions of a
soil sample.

2.1 Grading Analysis

Figure 1 shows results from a sample that was divided and sent to seven leading commercial laboratories. The
results show an acceptable variance. Results are closely grouped from 13.2mm to 0.425mm. Sample preparation
might have played a role in divergence of results from 0.425mm to 0.075mm. The authors have noted that
details of the preparation of fines samples are critical and have a profound influence on the end result obtained.

605
Figure 1. Grading analysis of Brandwag sample.

Six of the laboratories used in this study are accredited with SANAS and conform to the international ISO:17025
standards, suggesting that the quality is well managed and ensured.
For each of the three samples analysed, the grading analyses were reasonably consistent.

2.2 Hydrometer Analysis

Hydrometer analysis is used to determine the various fractions of fines in a sample. The method is based on
Stoke’s law and is flawed due to several of its assumption being dubious. Clay particles may be flakey and may
have very large specific surface areas (Whitlow, 2001). Hydrometer accuracy is doubtful for four reasons
(Savage 2007),: Stoke’s law assumes all particles are spherical, de-flocculation may not be complete at the time
of testing, clay particles are partially carried down by larger particles and a relative density of 2.65 is assumed
for all particles, which may not be true. More on this in section 2.3.
Figures 2a and 2b indicate the variance obtained from different commercial laboratories. Note that the values
vary between 63% and 18% on the Steelpoort sample (Figure 2b), a difference of 45%.

Figure 2a. Hydrometer analysis of Botshabelo Sample.

606
Figure 2b. Hydrometer analysis of Steelpoort Sample.

2.3 Savage’s method to determine the 0.002mm fraction

Savage suggested using Skempton’s activity formula to relate activity to the ratio (R) of the liquid limit (LL) to
the plastic limit (PL) of a soil sample. Savage’s analysis is as follows:

Activity = PI / P0.002 (1)

Where P0.002 refers to the percentage of material smaller than 0.002mm

Savage found an exponential relationship between the ratio LL / PL (R) from a table of the activity values for
Montmorillonite , Illite and Kaolinite published by Cornell University in 1951.

Activity = 0.16R2.13 (2)

Using formulas (1) and (2) the clay content can be established empirically as:

P0.002 = PI / Activity = PI / 0.16R2.13 = 6.25 PI.R-2.13 (3)

The P0.002 value obtained is based on the PI that was tested, typically at P0.425, and does not represent the
whole sample. The equation is adjusted to reflect the entire sample (PIGross):

PI0.425 x P0.425 = PIGross (4)

P0.002 = 6.25 PIGross.R-2.13 (5)

Table 1 draws a comparison between the hydrometer values for clay fraction and those derived from Savage’s
formula (5).

Table 1. Hydrometer P0.002 compared to Savage P0.002.

Sample LAB1 LAB2 LAB3 LAB4 LAB5 LAB6 LAB7
Steelpoort
Hydrometer 17 56 24 26
Savage 68 44 11 49
Brandwag
Hydrometer 49 58 33 44 56 50 47
Savage 51 27 18 44 40 43 36
Botchabelo
Hydrometer 35 56 29 43 44 40 35
Savage 31 26 22 30 31 37 34

In Table 1 it can be seen that some of the results compare favourably, while those in italics are unacceptable.
Savage’s method does not consistently predict lower or higher values and the values do not correlate well. Figure
5 shows a graphical representation of the Steelpoort sample comparison.

607
Figure 3. Steelpoort sample: Hydrometer values compared to Savage’s values.

2.4 Atterberg limits

Atterberg limits include the Liquid Limit (LL), Plastic Limit (PL), Plasticity Index (PI = LL-PL), the Shrinkage
Limit (SL) and Linear Shrinkage (LS).
Commercial laboratories in South Africa make use of the Casagrande –cup apparatus and typically perform
the LL test according to TMH1:1986 Method A2, which uses the same apparatus as the British Standard BS
1377-2. The authors have found that sample preparation has a significant bearing on the results, which may be
responsible for the variance between the different commercial laboratories. Figure 4 illustrates typical results
obtained for one of the three different active clay samples.

Figure 4. Liquid limit, plastic limit, plasticity index and linear shrinkage of Brandwag sample.

Based on figure 4, it is clear that the LS values are relatively consistent, with most laboratories within two per-
cent of the average. The LL, PL and PI values show a large variance, although there is a grouping of LAB2,
LAB3, LAB5 and LAB6 which seem to compare well.
Casagrande’s plasticity chart, as used by the Unified Soils Classification System can be derived from the
relationship between the LL and PI of a soil sample.
Figure 5 shows the relationship between the liquid limit and plasticity index for the Steelpoort sample.

608
Figure 5. Casagrande’s plasticity chart, Steelpoort sample.

3 Van der Merwe’s empirical method for the estimation of potential heave

Van der Merwe’s method relies on the PI of the gross sample and the P0.002 fines fraction, which according to the
British Classification is deemed to be the clay fraction of a soil sample. For Van der Merwe’s method to give
valuable output, valuable input is required. This paper does not focus on the prediction model itself, but rather
the input parameters.
For better understanding of the method, a simplified example of the process follows:
PIGross is plotted on the y-axis and P0.002 on the x-axis as the “classification of heave potential” curve, (after
Van der Merwe 1964). Figure 6 shows an example of such a plot with values obtained from various laboratories
using Hydrometer analysis to obtain the P0.002 fraction.
Table 2 and Figure 6 compare the resultant heave potential to that obtained using Savage’s formula to
determine the P0.002 fraction.
Skempton suggested using the relationship between the P0.002 fraction and the Plasticity Index to give an
indication of the heave potential of soils (Skempton 1953). He suggested using slopes of less than 0.75 to refer to
inactive clays, slopes of more than 1.4 would suggest active clays and everything in between would be referred
to as normal clays. In figure 6 those designations have been adjusted to reflect slopes of 2.0, 1.0, 0.7, 0.6 and 0.5,
with anything less than 0.5 considered inactive.

Table 2. Heave Potential using Hydrometer Analysis compared to using Savage’s Formula
Method LAB1 LAB2 LAB3 LAB4 LAB5 LAB6 LAB7
Hydrometer Medium Medium Medium High Low Very High Low
Savage Medium High Medium Medium Medium Very High Low

609
Figure 6. The classification of heave potential after Van der Merwe, with modified Skempton’s Activity lines: Hydrometer
values and Savage’s values for the P0.002 fraction.

4 CONCLUSION

Seven leading commercial laboratories were tasked with performing “Foundation Indicators”, which refer to the
Atterberg Limits and Particle Size Analysis. Although the physical sieving provided comparable results down to
the 0.425mm sieve, anything finer proved troublesome. The authors concluded that the problem probably lies
with the preparation of the samples, as some details of preparation were found to have a major impact on testing
done “in-house”.
The Atterberg Limits were performed on fractions passing the 0.425mm sieve, and preparation might have
played a role there also. A grouping of laboratories obtained results that compared well, while two laboratories
found substantially different results.
The finer fractions, those passing the 0.075mm sieve, proved problematic as not all of the laboratories use the
same method, although theoretically they should yield similar results. The results varied substantially and the
range between the highest and lowest P0.002 is alarming.
It can be concluded that using “foundation indicators” alone as the basis for empirical heave prediction
methods is a very risky approach and that other approaches need to be identified.

REFERENCES

Fredlund D.G., (2009) Paradigm Shifts to Facilitate the Practice of Unsaturated Soil Mechanics. 8th C.W. Lovel
Lecture. Perdue Geotechnical Engineering. 26 October 2009.
Nelson J.D., Miller D.J., (1992) Expansive Soils – Problems and Practice in Foundation and Pavement
Engineering. John Wiley and Sons, New York, Chicester, Brisbane, Toronto, Singapore.
Skempton, A.W., (1953) The colloidal “Activity” of clays. Proceedings of the 3rd International Conference of
Soil Mechanics and Foundation Engineering.
Van der Merwe, D.H., (1964) The prediction of heave from the plasticity index and the percentage clay fraction
of soils. The Civil Engineer in South Africa. June 1964, pp 103-107.
Weinert, H.H., (1980) The natural road construction materials of southern Africa. Pretoria: Academica.
Whitlow, R., (2001) Basic Soil Mechanics. Fourth Edition, first edition published in 1983. Prentice Hall,
Pearson Education.

Effect of cyclic loading on the behavior of swelling clays

Effet d’un chargement cyclique sur le comportement des argiles gonflantes
Amal Medjnoun *1 and Ramdane Bahar 1,2
1
Laboratoire Géomatériaux Environnement et Aménagement, Université Mouloud Mammeri, Tizi-Ouzou,
Algérie
2
Université des Sciences et de la Technologie Houari Boumediene, Faculté du Génie Civil, LEEGO, Bab-Ezzouar, Alger,
Algérie
* medjnounamel@[Link]

ABSTRACT – This study examines the effect of mechanical and seasonal cyclic loading on the behavior of
swelling clays. The used soils are highly plastic with low water content and high densities. The swelling
parameters recorded have been a significant decrease in the swelling potential during the loading-unloading
cycles. It is noted that the effect of the initial state in swelling clays disappears from a certain number of cycles.
In a case of high or very high swelling potential, the stabilization requires several cycles to be achieved. The
study of cyclic seasonal loading consists on humidification infiltration and wetting by capillarity to represent the
penetration of surface water and upwelling respectively. The seasonal cycle loadings are repeated several times
in a climatic chamber. Observations showed that the samples studied reacted to each water variation. For the
saturation infiltration, the shrinkage is expressed differently in the first four cycles, the aggregates gradually
crumbling. The shrinkage is observed in the horizontal and vertical direction. In the saturation by capillarity, a
burst of dry upper part occurred after the saturation of the lower part of the sample. Large particles also are
crumbling during wetting-drying cycles. The upwelling by capillarity causes the transport of water in dissolved
salts on the surface. The SEM observations show that the texture of studied clays has evolved considerably after
a cycle of shrinking and swelling. The effect of cycles of shrinking and swelling can cause problems in a large-
scale example of the ground movement observed in the sites of Médea.

RÉSUMÉ - Cette étude examine l'effet cyclique des sollicitations mécaniques et saisonnières sur le
comportement des argiles gonflantes. Les argiles utilisées sont très plastiques avec faibles teneurs en eau et avec
des densités élevées. Les paramètres de gonflement enregistrés montrent une décroissance du potentiel de
gonflement au cours des cycles mécaniques. Il est à noter que l'effet de l'état initial dans les argiles gonflantes
disparaît à partir d'un certain nombre de cycles. Dans un cas de potentiel de gonflement élevé ou très élevé, la
stabilisation nécessite plusieurs cycles pour être atteinte. L'étude des sollicitations saisonnières cycliques
comporte la saturation par infiltration et par capillarité, cela a pour objectif de représenter la pénétration et les
remontées des eaux. Les sollicitations cycliques saisonnières sont répétées plusieurs fois dans une chambre
climatique. Les observations ont montré la réaction des échantillons étudiés à chaque variation hydrique. Dans la
saturation par infiltration, le retrait est exprimé différemment dans les quatre premiers cycles, les agrégats
s’effritent d’avantage, le retrait est observé dans le sens horizontal et vertical. Dans la saturation par capillarité,
un éclatement de la partie supérieure sèche de l’échantillon s’est produit après la saturation de sa partie
inférieure. La remontée par capillarité provoque le transport des sels dissous dans l’eau à la surface des
échantillons. Les observations au MEB montrent que la texture des argiles étudiées a considérablement évolué
après un cycle de retrait-gonflement. Les cycles de retrait-gonflement causent des problèmes de grande échelle,
exemple du mouvement de terrains observé dans les sites de Médéa.

1 INTRODUCTION

The soil underwent a set of changes, over time, caused by cyclic loading in the form of loading-unloading and
seasonal variations. The expansive soils located on the surface of natural ground have the ability to change
volume because of the water variation. Several authors have studied the effect of loading and unloading cycles
of swelling clays (Alshihabi, 2002; Cuisinier, 2002; Delage, 1993; Yahia-Aissa, 1999), they found that the
unloading slope, at the saturated state, is highly related to soil type and it seems to increase with the proportion
of swelling clay in the soil. Other authors have researched the effect of wetting-drying cycles on swelling of
remolded and undisturbed clay samples (Chen & Ma, 1987); Subba Rao & Satyadas, 1987; Al-Homoud et al.,
1995; Basma et al., 1996; Tripathy et al., 2002; Day, 1994, 1995; Geremew et al., 2009). Their results showed
the disappearance of the inherited characteristics of the deposition conditions and the stabilization of
deformations from the third cycl of swelling-shrinking. These solicitations change, each time, the initial state of

611
the soil, by giving it a new density and a new saturation percent. The state characteristics, such as, a density,
moisture content and consolidation pressure change under the effect of various solicitations.
The paper examines the effect of a mechanical and seasonal cyclic loading on the behavior of swelling clays.
The cyclic seasonal loading consists on saturation by infiltration and wetting by capillarity to represent the
penetration of surface water and upwelling respectively. The tests are performed in climatic chamber.

2 EFFECT OF MECHNICAL CYCLIC SOLICITATIONS

2.1 Test procedure for loading-unloading cycles

The principle is to perform tests free swelling on samples of intact clay withdrawn from the Medea region.
The soil in place is located on the surface of natural lands and it is not submitted to an initial external stress,
hence the choice of the free swelling method of the French standard, AFNOR (1995).
The equipment used is a slaved device type oedometer Geocomp (Figure. 1.a). The scheme has boosted to
record several readings of the order of 1/100 second for the time variable and 1/10000 millimeter for
displacement. A program in the form of a table destined for the user is to fill it with the test data and loading-
unloading program with all the details (Figure1.a). The number of cycles is introduced into the program allowing
the direct passing in subsequent cycles. At the end of the test, the swelling results are given in table and
graphical form (Figure 1 c).

Figure 1. Device oedometric Geocomp enslaved.

The transition to the next cycle consists in extend the unloading phase, until stabilization of vertical
deformations. The rate of the vertical deformation corresponds to the rate of swelling of the following cycles.
The cycles are repeated several times until stabilization of swelling parameters. The duration of each swelling
test exceed one month.

2.2. Presentation and discussion of results

The study is performed on several samples. However, in this paper only one result is presented (Ech/A).
From Table 1, the studied soil is a highly plastic clay. The cyclic curves are shown in Figure 2. Cycles are shown
in different colors in order to follow the evolution of swelling parameters.

Table 1. Geotechnical characteristics of the sample Ech/A, extracted from Medea site.
Parameters W0 d WL Ip F2µm Pgi Pgf Agi Agf Number
(%) (kN/m3) (%) (%) (%) (kPa) (kPa) (%) (%) of cycles
6.93 18.3 51.6 28.2 53 700 300 8.54 4.38 6
Pgi : Swelling pressure of the first cycle.
Pgf : Swelling pressure of the last cycle.
Agi : Swelling rate of the first cycle.
Agf : Swelling rate of the last cycle.

The rate and the swelling pressure developments are shown in Figures 3 and 4. The form of the oedometric
curve is preserved, despite the large number of cycles. The obtained curves form an hysteresis loop.
Furthermore, the pressure and the rate of swelling of the soil decreased gradually from one cycle to another.

612
During the three first cycles, the difference between the recorded swelling rates is important, the gap between
the swelling rate decreases gradually in the next cycles. The change in swelling pressure, for the same sample,
follow the swelling rate. Comparison of initial and final values shows a significant reduction in both swelling
parameters; the swelling rate increases from 8.54% to 4.38% and the swelling pressure increases from 700-630
(kPa). The initial values of the parameters swelling are reduced by 50% after six cycles of Loading Unloading.
The largest reduction occurred after the first cycle, it is estimated at 26% of the final value for the two
parameters of swelling.
Analysis of the results shows that the swelling parameters are highly dependent on the initial state of the soil.
Geotechnical parameters of state show a low water content and high dry density (Table.1). Which gives the
sample a swelling potential. In the second cycle, the sample is not submitted to the same strain rate, despite the
application of the same series of stress. Because the sample is already saturated, therefore, the moisture content
has probably changed its microstructure, from where the irreversible deformation of the sample.

Figure 2. Curve of Cyclical loading unloading (Ech/A 2,6/3,0 m).

Figure 3. Evolution of the swelling rate under the effect of loading and unloading cycles (Ech/A 2,6 m).

Figure 4. Evolution of the swelling pressure under the effect of loading and unloading cycles (Ech/A 2,6 m)..

613
The swelling index (Cg), also present information on the swelling potential of a soil, it is defined from the
tangent of the unloading curve. This parameter is less affected, firstly, by the loading-unloading cycles,
secondly, by the initial state of the sample, Figure 2.
Figures 3 and 4 present a summary of the evolution of swelling rate and pressure. The decrease of swelling
rate and pressure is gradually during the loading-unloading cycles. The swelling parameters which are registered,
from the fifth cycle, are almost identical.
A similar study on other samples showed that for soils with a high swelling potential, the stabilization of
swelling parameters requires several loading and unloading cycles, and for soils with a low swelling potential,
the stabilization of these parameters occurs during a few cycles. This behavior may be explained by a probable
rearrangement of clay particles to establish a physico-chemical equilibrium within the microstructure of these
particles. The observed strain is generated by the presence of the swelling minerals constituting the studied soil.
A direct relationship is observed between the number of cycles and the stabilization of the rate of strain and
swelling pressure.

3 EFFECT OF SEASONAL CYCLIC SOLICITATION

3.1 Test Procedure

The study was performed on identical samples of intact clay extracted in Haouche Beyzid site located in
Medea (Medjnoun & al 2014). These samples are cut from the same block so as to have a diameter of 63 mm
and a height of 25.4 mm. They are then coated laterally with elastic transparent membranes and low stiffness to
maintain radially samples. The test principle is to expose samples to alternating cycles of shrinking-swelling, in a
climatic chamber equipped with a digital program on which the seasonal characteristics of these two cycles can
be defined.
Drying is carried out in dry air at a temperature of about 41 ∓ 4 ° C and a humidity of 2 ∓ 2% (summer
characteristic), this phase ends when the sample weight becomes constant. Then the program automatically
switches to winter characteristics; a humidity level varying between 92% and 100% and a temperature of about °
C 1∓2. This phase ends after full saturation and total swelling samples, generally after 24 hours, view the small
sample size.
Two modes of saturation samples are used, namely: the saturation by infiltration (Ech/D1) and saturation by
capillarity (Ech/D2). The evolution of the structure is followed during cycles of drying and wetting (shrinking-
swelling) by taking pictures at the end of each cycle.

3.2 Presentation of results

The behavior of the samples during the cycles is shown in Figure 5 to the saturation mode by infiltration and in
Figure 6 for the method of saturation by capillarity. The geotechnical characteristics of these samples are
presented in the following table. The two studied samples (Ech/ D1 and Ech/ D2) have a high dry density and
average water content.

Table 2. Initial geotechnical characteristics of the samples.

Parameters W0 (%) d (kN/m3) WL (%) Ip (%) F 2µm (%)
Ech/D1 and Ech/ D2 11.4 19.91 66.3 39.6 52
5.65m

3.2.1. Saturation by capillarity

When capillary reassembled, the soil sample underwent a macrostructure bursting at the edges, in the phase
of the first wetting cycle. During drying, the shrinkage cracks appeared with different thicknesses. The shrinkage
is observed more in the horizontal direction and the swelling is observed more in the vertical direction. The
sample does not return to its original height, during shrinkage, and is found to have submitted a cumulative
swelling during cycles of shrinking-swelling.
A whitish color gradually covered the surface of the sample (Figure 5), in the swelling-shrinkage cycles,
forming a thin white layer. The latter is trained by an accumulation of water in the capillary reassembled salts.
Chemical analysis showed that the sample has a percentage of calcium carbonate (CaCO3) of about 20% and
traces of sulfate; therefore, the white layer has the salts of calcium carbonate. The thickness of this layer depends
on the percentage of calcium carbonate in the soil.

3.2.2 Saturation by infiltration

During the swelling-shrinkage cycles the evolution of the macrostructure of the sample D2 under the effect of
infiltration by the saturation, shows that the first cycle is characterized by a volumetric shrinkage of the sample

614
after drying, and the appearance of crumbling of aggregates during the first two cycles of shrinking and swelling
(Figure 6). Rearrangements have occurred during the swelling phase of the second, third, fourth and fifth cycle.
These rearrangements have caused the gradual crumbling of the sample and the increase in permeability from the
second cycle.
During the sixth cycle, the temperature decreased to -1 ° C, the water froze in the pores hence increasing their
volume and widening of shrinkage cracks (Figure 6). The sample underwent a major deformation in both vertical
and horizontal directions, because of the freezing water.

3.3. Discussion of Results.

The samples Ech/D1 and Ech/D2 have underwent significant changes in their macro-structures as a result of
water and seasonal solicitation. Ech/D1 underwent bursting of dry upper part under the effect of saturation by
capillarity. Large particles are also crumbled their rounds during cycles of shrinking and swelling. We noticed
that shrinkage cracks are composed of principal cracks and secondary cracks. The principal cracks are cracks
that appear on each drying with a larger thickness, and the secondary cracks are those that appear in a cycle and
disappear to another, their thicknesses are small.
The vertical deformation in the first wetting cycle is important. However, the other cycles were just a
cumulating of vertical deformation. The thickness of the horizontal cracks gradually increases from cycle to
another cycle. Samples acquired a rhythm of shrinkage and swelling more or less stable to the sixth cycle.
The reassembled water capillary generates transport salts dissolved in water to the ground surface. The salts
can be seen after the drying of the soil surface. In the case of marl samples, white color appears on their surface
from the first cycles. Knowing that calcium carbonates are minerals stabilizer swelling clays. Therefore, their
decrease in soil can promote swelling. The effect of the seasons appears in the change of state of water (gas,
liquid and solid) and in the concentration and the state of salts existing in the soil.

(a) Initial state (b) after drying (c) cycle 1 (wet) (d) cycle 1 (dry) (e) cycle 2 (wet)

(f) cycle 2 (dry) (g) cycle 3 (wet) (h) cycle 3 (dry) (i) cycle 4 (wet) (j) cycle 4 (dry)

(k1) cycle 5 (dry) (k2) cycle 5 (vertical show) (l) cycle 6 (dry) (m) cycle 6 (wet) (n) cycle 7 (dry)

Figure 5. Evolution of the clay macrostructure during cycles of shrinkage swelling (imbibition at the base, Ech / D1).

615
(a) initial stat (c) cycle 1 (wet) (d) cycle 1 (dry) (e) cycle 2 (wet) (f) cycle 2 (dry)

(g) Cycle 3 (wet) (h) cycle 3 (dry) (i) cycle 4 (wet) (j) cycle 4 (dry) (k) cycle 5 (wet)

(l) cycle 5 (dry) (m) cycle 6 (wet) (n1) cycle 6 (dry) (n2) cycle 6 (Vertical show)

Figure 6. Evolution of the clay structure during cycles of shrinking and swelling (Ech / D2).

4 EVOLUTION OF MICROSTRUCTURE CLAY AFTER A CYCLE OF SHRINKING AND SWELLING

This work is performed on intact expansive soils samples extracted from the Medea region. The observations
obtained from SEM images shows the evolution of the clay structure after a cycle of shrinking and swelling
(Figure 7). The picture taken on a sample which has underwent a shrinkage-swelling cycle, shows a crumbling of
the clayey micro-particles and the appearance of micro-cracks and micro-pores of different sizes, see Figure 7
(b).
The set of observations obtained after a cycle of shrinking and swelling showed the opening of new cracks
and increased porosity in the samples; which causes the increase of the permeability of swelling clays, after one
swelling-shrinkage cycles. Moreover, the appearance of parallel clay sheets is preserved in the majority of SEM
observations.
Soil underwent movement in vertical and horizontal direction during swelling, with the creation of new
chemical bonds to basic water. However, during the shrinkage, the H-OH connections, evaporate leaving gapes
or discontinuities in the soil structure, these discontinuities are observed as like cracks or pores form in the SEM
figure, figure 7 (b).

Sheets in
parallel Micro-pores

Micro-cracks

X 100 X 100

(a) Intact (b) After a cycle of shrinking and swelling

Figure 7. SEM image of sample extracted of the site Haouch Beyzid, Medea.

616
The macro-structural changes observed in surface samples are only the result of the change in the
macrostructure as a result of shrinking and swelling phenomenon. Macro-cracks accumulate to form shrinkage
cracks distributed in all directions and visible to the human eye.

5 CONCLUSION

This study allowed us to understand that in nature the swelling soils cause a cracking of the building that they
support, when the swelling pressure is greater than the stress applied by the building. And causes a damage
under any applied stress, in the case of sloping land swelling, is recorded motion of land, because, the soil loss
his shear strength, after swelling (Medjnoun, 2014). The procedure of mechanical cyclic loading is very
convenient, it is a passive, economic and environmental clean solution, it allows the soil to adapt to the stress
applied by the structure planned.

REFERENCES

Al-Hamoud, A. S., Basma., A. A., Husein Malkawi A. I. & Al Bashabsheh, M. A. (1995) Cyclic swelling behavior of clays, j.
Geotech. Eng., ASCE, 121 (7): 562-565.
Alshihabi O. (2002) Étude en laboratoire du comportement d’un sol compacté non saturé influence des cycles de séchage-
humidification. Thèse de doctorat, Université des sciences et Technologies, Lille, France, 124.
Basma A. A., Al-Homoud A. S., Malkawi A. I. H., Al-Bashabsheh M. A. (1996) Swelling-Shrinkage behavior of natural
expansive clays. Applied Clay Science, 11, 211-227.
Chen F. H., Ma G. S. (1987) Swelling and shrinkage behaviour of expansive clays. 6th inter. Conf. Expansive soils, New
Delhi, 127-129.
Cuisinier O. (2002) Comportement hydromécanique des sols gonflants compactés. Thése de Doctorat, Institut National
Polytechnique de Lorraine, École Nationale Supérieure de Géologie, Nancy, 176.
Day R. W. (1995) Ultimate Density of a compacted clay subjected to cycles of wetting and drying. Environmental &
Engineering Geoscience, vol. I, n° 2, 229-232.
Day R. W. (1994) Swell-shrink behavior of compacted clay. J. Geotechnical Eng., ASCE, 120(3), 618-623.
Delage P & Howat M. (1993) Etude du comportement hydromécanique d’un matériau argileux gonflant pressenti comme
matériau de remblai de galerie. Rapport de recherche 621 RP CRM 93002, ANDRA, 73 p.
Geremew Z, Audiguier M. & Cojean R. (2009). Analyse du comportement d’un sol argileux sous sollicitations hydriques
cycliques. Bull Eng Geol Environ. 68:421–436
Medjnoun A, (2014) Analyse, caractérisation, prévision et modélisation du comportement des argiles gonflantes. Thèse de
Doctorat, Universisé Mouloud Mammeri, Tizi Ouzou, Algérie.211.
Tripathy S., Subba Rao K. S., Fredlund D. (2002) Water content-void ratio swell-shrink paths of compacted expansive soils.
Can. Geotech. J., 39, 938-959.
Yahia-Aissa M. (1999) Comportement hydromécanique d’une argile gonflante fortement compactée. Thèse de doctorat,
Ecole Nationale des Ponts et Chaussées, CERMES, Paris, 241.

L’evaluation de la susceptibilite au retrait et gonflement des sols argileux du site

d’un projet urbain a Diamniadio à Dakar au Senegal

Assessing susceptibility to withdrawal and swelling clay soil in a site of a urban project in
Diamniadio Dakar Senegal
Amar MBENGUE1, Mouhamadou Moustapha NDIAYE1, Ibrahima K. CISSE2
1
Ingénieur en génie civil diplômé de l’école polytechnique de Thiès, Thiès, Sénégal.
2
Professeur titulaire des Universités à l’Ecole Polytechnique de Thiès.

RESUME - L’hydratation ou la dessiccation de certains sols argileux est parfois la cause de nombreux
dommages dans les ouvrages de génies civils. Ces sols qui peuvent être gonflants, se caractérisent par un
phénomène de gonflement et de retrait élevés. C’est ainsi que dans la zone de Diamniadio au Sénégal des
désordres sur des bâtiments légers ont été observés. Au cours de cette étude, il a été constaté que les sols de ce
site sont des sols gonflants. A cet effet, des solutions constructives des fondations des bâtiments, basées sur une
recherche documentaire ont été proposées.

ABSTRACT - The hydration or the desiccation of certain clay soils is sometimes the cause of many damages in
works of civil engineering infrastructures. These soils which can be swelled are characterized by a phenomenon
of high swelling and withdrawal. Thus in the zone of Diamniadio in Senegal disorders on light buildings were
observed. During this study, it was noted that the soil of this site are expansive soils. For this purpose,
constructive solutions, based on document retrieval were suggested.

1 INTRODUCTION

Les sols gonflants présentent la particularité de voir leur consistance se modifier en fonction de leur teneur en
eau. Durs et cassants lorsqu’ils sont asséchés, un certain degré d’humidité leur fait se transformer en un matériau
plastique et malléable.
Au Sénégal le phénomène de gonflement est observé dans plusieurs zones notamment à Rufisque dans la
région de Dakar où plusieurs cas de désordres, liés au gonflement ont été signalés. Il affecte différents types
d’ouvrages de Génie Civil. Ces désordres peuvent être des fissures, des mouvements de fondations, des
tassements de dallages, des effondrements des murs de soutènement, des dommages au niveau des chaussées
affectant l’état de surface, des engorgements des systèmes d’assainissement autonomes, des ruptures des
conduites d’eau etc.
La zone de Diamniadio, programmée comme futur pôle urbain, qui abrite de nouvelles constructions d’une
grande envergure telles que le centre international de la Francophonie, est concernée par ce phénomène
complexe qui suscite une forte attente de la part des pouvoirs publics. Le besoin de faire une étude minutieuse de
ce phénomène, dans le but de proposer des solutions adaptées aux réalités économiques et environnementales de
notre pays est jugé nécessaire.
Ce présent travail étudie en premier lieu le potentiel de gonflement des sols du site d’un projet urbain à
Diamniadio à partir d’essais géotechniques et de méthodes d’estimation de gonflement existantes. Dans un
deuxième temps des solutions sont proposées pour pérenniser les constructions de ce site qui ont pour assises les
sols gonflants.

2 CARACTERISATION DU GONFLEMENT

Méthodes d’estimation qualitatives du gonflement

2.1.1 Approches suivant différents auteurs
Les limites d’Atterberg, la valeur de bleu et l’activité argileuse ont été utilisés par plusieurs auteurs pour
déterminer le potentiel ou sensibilité de gonflement des sols argileux. Les tableaux ci-dessous nous indiquent
les classifications proposées par six auteurs.

619
Tableau 1. Classification de Dakshanamurthy et al. (1973)

Limite de liquidité Classification Indice de plasticité Sensibilité

0 à 10 non gonflant
< 12 Faible
10 à 35 gonflement faible
35 à 50 gonflement moyen 12 à 15 Moyenne

50 à 70 gonflement élevé 25 à 40 Forte

70 à 90 gonflement très élevé
>= 40 Très forte
Tableau 3. Classification de BRE (1980)

Valeur de bleu Sensibilité

Activité Potentiel de gonflement
< 2.5 Faible
< 0.50 Faible
2.5 à 6 Moyen
0.50 à 1.00 Moyen à fort
6à8 Forte 1.00 à 2.00 Fort à très fort

>8 Très forte > 2.00 Très fort

Tableau 5. Classification de Holtz et Gibbs (1957)

Teneur en particules Indice de plasticité Limite de retrait Potentiel de

Colloïdales (< 1µ) Ip Ws gonflement
>31 >38 <9 Très fort
22 à 31 29 à 38 9 à 10 Fort
14 à 22 17 à 29 10 à 15 Moyen

<14 <17 <15 Faible

2.1.2. Estimation par abaque

Les abaques de Casagrande et de Williams et Donaldson permettent d’estimer le potentiel de gonflement des
sols argileux à partir des limites d’Atterberg (limite de liquidité et de l’indice de plasticité) et de la fraction
argileuse. Ces abaques seront utilisés dans la suite pour situer le potentiel de gonflement.

3 PROPOSITION DE SOLUTIONS

La complexité du phénomène engendre une diversification des solutions qui peuvent être abordées sous
différents angles. Les solutions proposées seront divisées en deux grandes catégories : les solutions préventives
et les solutions curatives.

3.1. Solutions préventives adaptées au sol

3.1.1. Stabilisation
Stabiliser un sol, c’est modifier certaines de ces propriétés en vue d’améliorer ses performances. Il existe
plusieurs techniques de stabilisation.
La stabilisation par ajout de sable consiste à recueillir les données sur le potentiel de gonflement sur un site
donné et de choisir dans ce site le sol qui présente le potentiel de gonflement le plus élevé et de reprendre

620
l’essai de gonflement pour chaque pourcentage de sable ajouté au sol gonflant (par exemple : 15%, 30%, 45%,
55%).
La stabilisation par ajout de chaux consiste à déterminer la quantité optimale de chaux (« lime fixation
point ») qui permet d’avoir un gonflement nul en effectuant des mesures de pressions de gonflement sur des
échantillons obtenus avec différents dosages en chaux (2%, 4%, 6%, 8% et 10% par exemple). Elle permet de
réduire le gonflement et le retrait du sol traité.

3.1.2. Maintien à un état constant de la teneur en eau

Les désordres des constructions résultent notamment des fortes variations de teneur en eau du sol situé sous le
bâtiment. Les dispositifs proposés pour empêcher les variations de teneur en eau sont : l’utilisation de
géomembrane, la conception de système de drainage et évacuation des eaux, l’éloignement de la végétation et la
réalisation d’un trottoir périphérique autour du bâtiment.

3.2. Solutions préventives adaptées à l’interaction sol-structure

3.2.1. Définition de la profondeur minimale des fondations
D’après Mont Carlo, plus la profondeur est élevée, plus le risque de ruine est réduit et plus grande est la fiabilité
des maisons concernées. La probabilité de crise est faible pour les maisons (sans sous-sol) possédant des
fondations à des profondeurs plus basses que - 0.80 m. Il fixe ainsi la profondeur minimale à 0.80 m.
3.2.2. Adaptation de fondation sur sols gonflants
Briaud (2003) a proposé une solution de radier en forme de demi-gaufre, qui confère à la structure une rigidité
importante vis-à-vis de la flexion par rapport à un radier classique.

Figure 1. Radier de fondation en forme de demi-gaufre (Briaud. 2003)

Une solution plus classique consiste à approfondir les fondations jusqu’à une profondeur minimale de 1.5 m.
La solution « fondation profonde » peut également être envisagée si l’épaisseur du sol gonflant est importante et
si celle-ci présente une forte sensibilité au retrait/gonflement. La présence d’un vide sanitaire dans une
construction diminue aussi les effets liés au retrait et gonflement des sols gonflants.

3.3. Solutions préventives adaptées à la structure

Une solution efficace contre le phénomène de retrait et gonflement consiste à rigidifier la structure. Cette
rigidification nécessite la mise en œuvre de chainages horizontaux (haut et bas) et verticaux (poteaux d’angle)
pour les murs porteurs liaisonnés.

3.4. Solutions curatives

La solution curative la plus connue est la reprise en sous-œuvre qui consiste à adopter l’une des méthodes
suivantes : construction de plots de fondation ou réalisation de fondation par micropieux.

621
4. APPLICATION DES SOLUTIONS AU SITE DE DIAMNIADIO

Pour identifier de façon plus précise la nature des sols se trouvant à Diamniadio, des essais géotechniques tels
que la granulométrie, la sédimentométrie, les limites d’Atterberg, l’essai de gonflement à l’œdométrie ainsi que
des essais de teneurs en carbonate ont été réalisés sur 86 échantillons prélevés à 5 m de profondeur par sondage
carotté. Les tableaux qui suivent sont une partie des résultats obtenus à partir des essais réalisés. Les
classifications suivant les méthodes les plus connues dans la littérature (GTR, USCS, LCPC) complètent les
tableaux.

Tableau 6. Synthèse des résultats de quelques essais d’identification (CE 01 – CE 05)

CE
Sondage CE 01 CE 02 CE 04 CE 05
03
0,70 4,30 0,30 0,30 1,00 4,80 1,20
Profondeur (m) - - - - - - -
1,00 4,50 0,50 0,60 1,20 4,90 1,50

Nature de l’échantillon Argile Marne Argile Argile Argile Marne Argile

Teneur en eau (w) en % 9,6 39,2 6,8 7,5 12,8 29,5 20,4
Poids spécifique (kN/m3) 23,7 23,3 23,7 24,4 23,7 23,3 23,3
LL 72 79 43 56 50 78 54
Limites d’Atterberg LP 37,5 39 15,5 23 17 38 21,5
IP 34,5 40 27,5 33 33 40 32,5
Classification NF P11-300 (GTR) A3 A3 A3 A3 A3 A3 A3
OH- OH -
Classification USCS CL CH CH OH -MH CH
MH MH
Classification LCPC LT- OT LT- OT Amp At At LT- OT At

Tableau 7. Résultats de quelques essais de teneurs en carbonates

Sondage CE 02 CE 05 CE 07

Profondeur (m) 3,00 - 3,10 3,40 - 3,50 2,50 - 2,60

Teneur en CaCO3 (%) 38,93 38,27 29,79

Interprétation (tableau
Marne Marne Marne
(fascicule 62 Titre V))

Tableau 8. Résultats de quelques essais de gonflement libre à l’œdomètre

Sondage CE 03 CE 04 CE 06 CE 07 CE 09

Profondeur 4,60 - 4,80 1,70 - 1,90 1,50 - 1,70 1,60 - 1,80 0,30 - 0,50

Nature du sol Marne Argile Argile Argile Argile

Pression de gonflement σg (KPa) 345 460 140 25 68

A partir des classifications utilisées et des résultats des essais de teneur en carbonate, nous constatons que les
sols de Diamniadio sont essentiellement constitués d’argile et marne.

2.4 Estimation qualitative du gonflement suivant différents auteurs

Quelques résultats des potentiels de gonflement des sols de Diamniadio estimés à l’aide des tableaux de
classification des auteurs sont consignés dans les tableaux suivants :

622
Tableau 9. Potentiel de gonflement : (CE 01 – CE 05)
Sondage CE 01 CE 02 CE 03 CE 04 CE 05
0,70 4,30 0,30 0,30 1,00 4,80 1,20
Profondeur (m) - - - - - - -
1,00 4,50 0,50 0,60 1,20 4,90 1,50
Activité argileuse 0.93 0.60 0.69 0.97 0.67 0.50 0.83
Très Très
Dakshanamurthy et Romana Très élevé Moyen Elevé Elevé Elevé
élevé élevé
Prian et al. Fort Très fort Fort Fort Fort Très fort Fort
Williams Fort Moyen Moyen Fort Moyen Moyen Fort
Holtz et Gibbs Fort Très fort Moyen Fort Fort Très fort Fort
Chen Très fort Très fort Fort Fort Fort Très fort Fort

Le potentiel de gonflement varie de « moyen » à « fort ». Sur l’ensemble des points de sondage, il y en a que
deux (02) où le potentiel de gonflement est « faible ». Ces résultats montrent la sensibilité des sols de
Diamniadio au gonflement et constituent un avertissement à toute volonté de construire sur le site sans études
et dispositions préalables.

2.5
stimation du potentiel de gonflement par abaque

Figure 2. Abaque de Casagrande Figure 3. Abaque de Williams et Donaldson

Après avoir placé les points de sondages sur les abaques d’estimation de gonflement, on note une concentration
des points dans la zone où le potentiel de gonflement est estimé entre « moyen » et « grand » par Williams et
Donaldson (voir abaque) et entre « moyen » et « fort » par Casagrande. Ces résultats sont en concordance avec
les autres résultats obtenus par les méthodes précédentes et confirment ainsi le caractère gonflant des sols de
Diamniadio.

4 SOLUTIONS PROPOSEES

Dans les maisons individuelles et les bâtiments collectifs d’habitation, la pression de gonflement est souvent
supérieure à la charge. Il est nécessaire de prendre des dispositions pour limiter au maximum les désordres dus
au gonflement à savoir : le respect de la profondeur minimale de 0.8 m en zone d’aléa moyen à faible et de 1.20
m en zone d’aléa fort, la mise en œuvre des chainages horizontaux et verticaux des murs porteurs liaisonnés
selon les préconisations du DTU 20.1, l’adaptation des dallages sur terre-plein à défaut de la réalisation d’un
plancher sur vide sanitaire, la mise en place de fondations plus profondes à l’aval qu’à l’amont afin de garantir
l’homogénéité de l’ancrage sur les terrains en pente, la réalisation de fondations sur semelles continues, armées
et bétonnées à pleine fouille selon les préconisations du DTU 13.12 (fondations superficielles), la
désolidarisation des parties de constructions fondées différemment au moyen d’un joint de rupture sur toute la
hauteur de la construction, l’éloignement des arbres ou des arbustes à une distance supérieure à la leur hauteur ou
la mise en place d’un écran anti-racines d’une profondeur minimale de 2 m, la mise en place de trottoirs
périphériques et/ou de géomembrane de 1.50 m de largeur pour limiter l’évaporation à proximité immédiate des

623
murs de façades et en fin l’évitement de tout pompage à une nappe superficielle à moins de 10 m de la
construction.
Dans les constructions de grandes envergures il est proposé de faire une étude plus approfondie.

5 CONCLUSION

Les études en laboratoire réalisées sur des échantillons de sol de Diamniadio ont permis de constater que le sol
de la zone est essentiellement constitué de sols marneux et argileux gonflants. L’ampleur des problèmes de
gonflement dans la zone nous a conduits à proposer des dispositions adéquates pour minimiser voire réduire les
désordres liés à ce phénomène. Ces dispositions s’appliquent soit sur le support de l’ouvrage, soit sur les
fondations, soit sur la structure ou bien sur leur combinaison. Ces solutions ne peuvent certes pas résoudre de
façon définitive les problèmes, mais elles œuvrent dans la tenue et la durabilité des constructions sur ces sols.

REMERCIEMENTS

Ces travaux ont été réalisés dans le cadre de notre projet de fin d’étude sous l’encadrement du Professeur
titulaire des Universités, Monsieur Ibrahima K. CISSE à l’Ecole Polytechnique de Thiès et du Laboratoire
Géotechnique, [Link]/DAKAR/SENEGAL.

REFERENCES

[Link] et al. (2009) Etude des phénomènes retrait- gonflement et stabilisation des sols gonflants de la région d’Oran –
19ème congrès français de mécanique – 14 pages.
Abdourahmane DIOP (2005) Contribution à la conception de fondations sur sols gonflants, application aux sols de la
SIPRES CAP DES BICHES – Ecole Polytechnique de THIES – 103 pages.
Benjamin ROTENBERG (2007) Modélisation multi-échelles du comportement de l’eau et des ions dans les argiles –
Université Pierre et Marie Curie – 255 pages
Bulletin des laboratoires des ponts et chaussées (1996) Les essais de gonflement - réf. 4082 - pp. 65-85 65.
Fabrice SALLES (2006) Séquence d’hydratation multi-échelle Détermination des énergies macroscopiques à partir des
propriétés microscopiques – Université Paris VI Pierre et Marie Curie – 327 pages.
Françoise Elsass (2005) Minéralogie des argiles de sols structure, altération, réactivité – Université Louis Pasteur de
Strasbourg – 137 pages.
Gerard Espuña Usan (2010) Analyse du phénomène de retrait gonflement des milieux argileux: impact environnemental et
impact de la sécheresse sur le bâti – Université polytechnique de Catalunya – 49 pages.
Houssine EJJAAOUANI (2008) Interactions des fondations et des sols gonflants : pathologie, calculs et études
expérimentales – Ecole nationale des ponts et chaussées – 243 pages.
Jean Pierre Magnan (2013) Panorama des sols gonflants en géotechnique - Université de Paris Est, IFSTTAR – 20 pages.
Jean Pierre Magnan et al. , SEC (2008 Symposium international sécheresse et constructions – LCPC, 443 pages.
Ministère français de l’écologie et du développement durable (2007) Retrait et gonflement des argiles, comment prévenir les
désordres dans l’habitat individuel – 32 pages.

Problematic soil mechanics in the Algerian arid and semi-arid regions (case of
M’sila expansive clays)
Mécanique des sols problématiques dans les zones arides et semi-arides en Algérie (cas des
argiles expansives de M’sila)
M. Khemissa*1, A. Mahamedi1, L. Mekki1
1
Geomaterials Development Laboratory, Mohamed Boudiaf University, M’sila 28000, Algeria
*
Corresponding Author (khemissa@[Link])

ABSTRACT – This paper presents a state-of-the art review on the behavior of expansive soils in arid and semi-
arid regions in Algeria, where significant damages frequently appear affecting the road infrastructures, roadway
systems and various networks and other small civil and industrial constructions. It deals with characterization
and treatment experimental studies on some Algerian expansive natural soils, including those conducted on
M’sila expansive clays.

RÉSUMÉ – Cet article présente une synthèse sur l’état des connaissances actuelles sur le comportement des sols
expansifs dans les zones arides et semi-arides en Algérie, où d’importants désordres apparaissent régulièrement
affectant les infrastructures routières, voiries et réseaux divers et les constructions civiles et industrielles de
faible hauteur. Il traite des études expérimentales de caractérisation et de traitement effectuées sur quelques sols
naturels expansifs algériens, notamment celles effectuées sur les argiles expansives de M’sila.

1 INTRODUCTION

Problematic soils are a worldwide problem and occur in many parts of the world but particularly in arid and
semi-arid regions. The arid and semi-arid regions cover inter alia a good part of Algeria. These regions are
delimited by the Tellian Atlas in North and the Saharian Atlas in South and extend from East to West until the
bordering Maghreb’s countries. Their meteorology is characterized by weak precipitations and important
temperature variations between winter and summer (cold and wet winters von hot and dry summers). Their
geology comprises clayey formations characterized by a high variation of volume when the conditions of their
equilibrium are modified (natural climatic phenomena due to a prolonged dryness, intense human activity by
modification of the ground water level because of excessive pumping, configuration of constructions in their
environment). These clayey formations were the subject of some characterization studies, which confirmed their
problematic character. The case of expansive, collapsible and salted soils can be in particular cited. Damages
appeared in the road infrastructures, roadway systems and various networks and in other small civil and
industrial constructions compromise the use of problematic soils in their natural state. At dry state, the expansive
soils are very difficult to compact because their consistency varies from hard to very hard. At wet state, they are
sticking, therefore uncompactable. However, their employment can be possibly decided based on specific
treatment with hydraulic binders (LCPC-SETRA, 2000a).
Short-term treatment of fine-grained soils and their long-term stabilization is a current technique in road
construction. This process is mainly used to make compactable the soft soils by reduction of their plasticity and,
consequently, to improve their bearing capacity. Limes mainly calcic (quicklime, extinct lime, lime slurry), road
cements and special binders are the most used treatment products. Action of these products on the hydrous state
of fine-grained soils and on their clayey fraction is highlighted in practice. Treatment studies carried out on some
expansive soils confirm, they also, the cement and lime action on their plasticity and swelling characteristics.
Other treatment products (dune sand, salt, fly ash, bitumen, rice husk ash, stone dust, or their combinations) were
used to stabilize the problematic soils. Obtained results show a certain improvement of geotechnical properties
of studied soils, but effectiveness of tested treatment products is not yet clearly established on the scale of
practice.
This paper presents a state-of-the art review on the significant experimental studies carried out on some
Algerian expansive soils, including those conducted on M’sila expansive clays of which the experimental data
are available.

2 STATE-OF-THE ART REVIEW ON THE BEHAVIOR OF ALGERIAN EXPANSIVE SOILS

625
Expansive, collapsible, liquefiable and salted soils are among the problematic soils most met in the Algerian arid
and semi-arid regions. Their geographical distribution depends on a certain number of geological and/or
geomorphological factors varying from one area to another. Expansive natural soils are most widespread in these
regions. They were the subject of some experimental studies already published. On the other hand, none
geotechnical publication on collapsible, liquefiable or salted natural soils were done. Thus, only the most
significant characterization and stabilization studies of expansive natural soils are discussed hereafter.

2.1 Characterization studies

Significant characterization studies carried out in Algeria concern the following expansive soils:
- In-Aménas expansive clays (Derriche et al., 1989; Derriche and Kebaili, 1998; Hachichi and Fleureau, 1999);
- Oran’s area expansive soils (Hachichi and Fleureau, 1999; Hachichi et al., 2007);
- Tlemcen’s area expansive clays (Djedid et al., 2001);
- M’sila expansive clays (Khemissa et al., 2008; Mahamedi and Khemissa, 2013 and 2014; Khemissa and
Mahamedi, 2014).
These studies confirm the expansive character of tested soils and measure the extent of damage caused to civil
and industrial constructions on which they are founded (roadway systems and various networks, roads, small
buildings).
For M’sila expansive clays, as example, it is concluded that they are classified as low-to-high plastic clays,
overconsolidated, low permeable and very low sensitive to creep. Their overconsolidation is due to the
phenomenon of shrinkage resulting from more-or-less thorough desiccation (table 1). According to French
classification of fine-grained soils and evolutionary rock materials (LCPC-SETRA, 2000b), these clays belong to
subclass A4 (Ip>40 or MBV>8: high plastic clays and marly clays) and are considered as low fragmentary
(FR<7) and low damaged (DG<5). The modifications of water content of these clays, in addition, are
accompanied by shrinkage or swelling. Various classifications used show that these clays are characterized by
high-to-very high swelling potential (figure 1).

Table 1. Geotechnical parameters of M’sila expansive clays.

Parameters Symbols Sidi-Aissa site Sidi-Hadjrès site
Ranges of variation Mean values Ranges of variation Mean values
Depth z (m) 1.5-2.0 1.75 1.30-1.70 1.50
Natural water content wnat (%) 14.63-16.20 15.26 13.21-13.46 13.34
Wet unit weight h (kN/m3) 21.26-21.62 21.43 20.4-24.2 22.3
Dry unit weight d (kN/m3) 18.49-18.68 18.59 18.0-21.4 19.7
Liquid limit wL (%) 40.37-48.39 43.80 81.5-86.7 83.7
Plasticity index IP (%) 21.87-30.25 24.76 50.1-51.9 51.0
Consistency index Ic (%) 1.09-1.24 1.15 1.33-1.47 1.38
Methylene blue value MBV 5.23-5.80 5.56 7.40-9.77 8.31
Clay content C2m (%) 24.2-55.0 37.6 20.5-30.9 25.7
Activity of clay Ac 0.4-1.7 0.9 1.95-2.02 1.98
Optimum water content wopt (%) - 11.1 19.2-19.6 19.51
Maximum dry density d-max - 1.93 1.59-1.61 1.60
Fragmentability coefficient FR - - 2.93-3.51 3.25
Damage coefficient DG - - 2.68-3.50 2.97
In-situ void ratio eo - - 0.60-0.78 0.67
Preconsolidation pressure 'p (kPa) - - 650-1000 700
Overconsolidation ratio OCR - - 8.7-13.5 9.1
Compression index Cc - - 0.16-0.19 0.18
Recompression index Cs - - 0.04-0.06 0.05
Coefficient of permeability kvo (x10-11 m/s) - - 2.1-3.2 3
Creep index Ce - - 0.002-0.011 0.006
Swelling pressure s (kPa) - 195 430-850 600
Secondary rate of swelling Cs - - 0.012-0.132 0.214
Conventional shrinkage limit wR - - 11.6-11.8 11.7
Free swelling fs (%) - 18.44 4.1-68.4 32.7
Effective shrinkage limit wRE - - 17.3-21.8 19.9

2.2 Treatment studies

Some stabilization studies were carried out on Algerian expansive soils. Cement and/or lime (Derriche and
Lazzali, 1997; Afès and Didier, 2000; Hachichi et al., 2009; Mellas et al., 2012; Gueddouda et al., 2013;
Mahamedi and Khemissa, 2013 and 2014; Khemissa and Mahamedi, 2014), sand (Kaoua et al., 1994), salt

626
(Bekkouche et al., 2007), dune sand and salt mixture (Lamara et al., 2007), lime and salt mixture (Gueddouda et
al., 2013) or lime and pozzolana mixture (Harichane et al., 2011) are used like stabilizers. Obtained experimental
results are relatively satisfactory for cement and/or lime treatment. But, effectiveness of other treatment products
is not yet clearly established on the scale of practice.
For M’sila expansive clays, as example, it is concluded that the simple or mix treatment with composed
Portland cement and extinct lime does not involve significant differences on the compaction parameters (figure
2), making it possible to conclude with a good reconstitution of the soil under in-situ as laboratory conditions. It
reduces in addition the plasticity of these clays (figure 3), increases their bearing capacity indexes (figure 4) and
increases their undrained shear strength (figure 5). So, they become unexpansive and compactable, with a
significant reduction of their deformability and a sensitive improvement of their bearing capacity. Swelling
reduction of these treated natural clays gets to them, in addition, certain stability (with respect to the
deformations due to the seasonal variations of water content) and, consequently, a durable behavior (with respect
to the wear of the generating compacted particles of plastic fine particles).

Swelling Potential (Dakshanamurthy and Raman, 1973)

none low medium high very high
60
very high

High plastic clays (CH)

Swelling potential (Chen, 1988)
Plasticity index, Ip (%)

45
40
high

30 Low plastic clays (CL)

medium

20 Sidi-Aissa site
Sidi-Hadjrès site
10
low

0
30 50 70 90
0 20 40 60 80 100 120
Liquid limit, wL (%)
Figure 1. Classification of M’sila expansive clays.

0.2 0.2
Cement
gd-max/wopt

gd-max/wopt=0.085
gd-max/wopt

Lime
0.1
0.1

gd-max/wopt=0.078 0.0

0.0
0 2 4 6 8 10 12
Stabilizer content (%) Cement/Lime content (%)
a/ Cement or lime simple treatment b/ Cement and lime mix treatment

Figure 2. Effect of simple and mix treatment on the normal compaction tests of M’sila expansive clays.

627
Swelling Potential (Dakshanamurthy and Raman, 1973)
none low medium high very high
60

very high
High plastic clays (CH)
(Control sample)

Swelling potential (Chen, 1988)

Plasticity index, Ip (%)

45
40
high
Cement Lime Cement/Lime
30 Low plastic clays (CL) C2 L2 C0L12
medium

C4 L4 C2L10
C6 L6 C4L8
20
C8 L8 C6L6
C10 L10 C8L4
10 C12 L12 C10L2
low

C12L0
0
30 50 70 90
0 20 40 60 80 100 120
Liquid limit, w L (%)

Figure 3. Effect of simple and mix treatment on the plasticity of M’sila expansive clays.
100
California Bearing Ratio, CBR

90 CBR Cement Lime

80 Unsoaked
70 Soaked
60
50
40
30
20
10
0
0 2 4 6 8 10 12
Stabilizer content (%)
California Bearing Ratio, CBR

100
90 Unsoaked CBR
80 Soaked CBR
70
60
50
40
30
20
10
0

Cement/Lime content (%)

a/ Cement or lime simple treatment b/ Cement and lime mix treatment

Figure 4. Effect of simple and mix treatment on the bearing capacity of M’sila expansive clays.

628
250

Undrained cohesion, cu (kPa)

Cement
200 Lime

150

100

0
0 2 4 6 8 10 12
Stabilizer content (%)
250
Undrained cohesion, cu (kPa)

200

150

100

Cement/Lime content (%)

a/ Cement or lime simple treatment b/ Cement and lime mix treatment

Figure 5. Effect of simple and mix treatment on the undrained shear strength of M’sila expansive clays.

3 CONCLUSION

This paper had as a finality to present a state-of-the art review on the behavior of problematic soils in the
Algerian arid and semi-arid regions. Expansive natural soils were the subject of some characterization and
treatment studies already published. On the other hand, none specific geotechnical study of Algerian collapsible,
liquefiable or salted natural soils was published at the present time. Results of available experimental studies
conclude that the expansive natural soils can be used as filling or subgrade materials of roadways and as support
of foundations of geotechnical structures after their specific treatment with hydraulic binders. Indeed, results
presented in this paper show that the simple or mixed treatment with composed Portland cement and extinct lime
improves the mechanical properties of expansive natural clays. However, attempts to improve the mechanical
properties of problematic soils by means of nonconventional treatment products gave mitigated results. Their use
thus does not seem possible for the moment on the scale of practice.

REFERENCES

Afès M., Didier G. (2000). Stabilisation des sols gonflants : cas d’une argile en provenance de Mila (Algérie). Bulletin of
Engineering Geology and the Environment, Springer, 59(1), 75-83, [Link]
Bekkouche A., Azzouz F.Z., Aissa Mamoune S.M. (2007). Stabilisation par les sels des sols gonflants de la région de
Tlemcen. SOMAPRO’07 : Colloque « sols et matériaux à problèmes », ISET Sfax, Tunisie, 19-25.
Chen F.H. (1988). Foundations on expansive soils. Developments in Geotechnical Engineering, Elsevier.

629
Dakshanamurthy V., Raman V. (1973). A simple method of identifying an expansive soil. Soils and Foundations, 13, 97-104.
Derriche Z., Lazzali F. (1997). Analyse des mécanismes de stabilisation d’un sol gonflant par apport de chaux sous
différentes formes. International Proceedings on Engineering Geology and Environment, Springer, 79-84.
Derriche Z., Kebaili M. (1998). Prévision du gonflement des argiles d’In-Aménas. Bulletin des Laboratoires des Ponts et
Chaussées, France, 218, 15-23.
Djedid A., Bekkouche A., Aissa Mamoune S.M. (2001). Identification et prévision de quelques sols de la région de Tlemcen
(Algérie). Bulletin des Laboratoires des Ponts et Chaussées, France, 233, 67-75.
Gueddouda M.K., Goual I., Benabed B. (2013). Effet de chaux, ciment et sel sur le potentiel de gonflement des argiles
gonflantes des régions arides en Algérie. European Journal of Environmental and Civil Engineering, Taylor and
Francis, 17(5), 315-328, [Link]
Hachichi A., Bourokba S.A., Fleureau J.M. (2007). Stabilisation chimique de deux sols gonflants de la région d’Oran. Revue
Française de Géotechnique, 118, 3-11.
Hachichi A., Fleureau J.M. (1999). Caractérisation et stabilisation de quelques sols gonflants d’Algérie. Revue Française de
Géotechnique, 86, 37-51.
Harichane K., Ghrici G., Kenai S., Grine K. (2011). Use of Natural Pozzolana and Lime for Stabilization of Cohesive Soils.
Geotech. Geol. Eng., Springer, 29, 759-769, [Link]
Kaoua F., Derriche Z., Laradi N. (1994). Contribution à l’étude de la stabilisation des sols gonflants par ajouts de sable.
Algérie Equipement : Revue de l’Ecole Nationale des Travaux Publics, 15, 12-15.
Khemissa M., Mekki L., Bakir N. (2008). Comportement oedométrique des argiles expansives de M’sila (Algérie).
SEC’2008: Symposium international sécheresse et constructions, Presses du Laboratoire Central des Ponts et
Chaussées, France, 229-234.
Khemissa M., Mahamedi A. (2014). Cement and lime mixture stabilization of an expansive overconsolidated clay. Applied
Clay Science, Elsevier, 95, 104-110, [Link]
Lamara M., Gueddouda M.K., Benabed B. (2006). Stabilisation physico-chimique des argiles gonflantes (sable de
dunes+sel). Revue Française de Géotechnique, 115, 25-35.
LCPC-SETRA. (2000a). Traitement des sols à la chaux et/ou aux liants hydrauliques : Application à la réalisation des
remblais et des couches de forme. Guide technique.
LCPC-SETRA. (2000b). Guide des terrassements routiers : Réalisation des remblais et des couches de forme. Guide
technique.
Mahamedi A., Khemissa M. (2013). Cement stabilization of compacted expansive clay. TOJSAT: The Online Journal of
Science and Technology, 3(1), 33-38, [Link]
Mahamedi A., Khemissa M. (2014). Stabilization of an expansive overconsolidated clay using hydraulic binders. HBRC
Journal, Elsevier, in-press, [Link]
Mellas M., Hamdane A., Benmeddour D., Mabrouki A. (2012). Improvement of the expansive soils by the lime for their use
in road works. ACE2012: 10th International Congress on Advances in Civil Engineering, 17-19 October 2012, METU,
Ankara, Turkey.

Influence de l’ajout du système chaux ciment sur le potentiel de gonflement et

de la microstructure d’une argile naturelle d’Algérie
Effect of Lime- cement on swelling and on the microstructure of Natural Expansive Clay of
Algeria
Bourokba Mrabent S.A *1, Hachichi A1. , Bengra L. 1 and Boudia M.1
1
Département de Génie Civil, Laboratoire Matériaux sol et Thermique (LMST),Université des Sciences et
Technologie Mohamed Boudiaf Oran, Oran, Algérie

RÉSUMÉ -Le processus de retrait gonflement des sols argileux est à l’origine de dommages considérables tels
que les fissurations et le soulèvement de fond de fouilles. En Algérie durant les trois dernières décennies et plus
particulièrement dans les régions de l’Ouest du pays, ce phénomène à été rencontré Nous nous sommes
intéressés au comportement d’un sol gonflant de la région d’Oran situé au Nord Ouest de l’Algérie. Ce travail est
une continuité de travaux de recherches menées sur l’étude du comportement d’une argile naturelle gonflante de
la région d’Oran ( Mers El Kébir). On a d’une part, apprécié l’influence de l’ajout du système chaux - ciment à
différents pourcentages sur le potentiel de gonflement, la pression de gonflement et les limites de consistance de
ce sol, et d’autre part nous avons étudié l’influence de ces ajouts sur sa microstructure. Ce mode de stabilisation
a conduit à des réductions considérables du gonflement du sol étudié dépassant les 50%. Nous avons complété
cette étude par des observations au microscope électronique à balayage (MEB) afin d’avoir une meilleure
compréhension des traitements au système chaux – ciment et de connaître la disposition des grains dans les
mélanges. Cette étude est sans doute une contribution importante à la compréhension du phénomène du retrait -
gonflement, et un outil complémentaire utile aux ingénieurs du génie civil pour trouver les solutions
appropriées.
Mots clés : argile - gonflement – pression de gonflement – chaux – ciment –Algérie

ABSTRACT –Shrinkage and swelling phenomenon of expansive soils is causing a lot of damages to buildings
founded on them. During the last three decades in Algeria and particularly in the Northwest region this
phenomenon was encountered. We were interested in the behaviour of an expansive soil from the city of Oran in
Northwest part of Algeria. In the first hand we studied the effect of cement in the swelling potential ,the
swelling pressure and the limits of consistence of this soil, in the other hand we studied the effect of cement in
the microstructure of the soil using electronic microscope.. This stabilization method led to significant
reductions in the swelling of the studied soil exceeding 50%. This study was then completed by studying the
microstructure of the treated soil using electronic microscope to obtain information about the disposition of the
pore space with treatment. This study is undoubtedly an important contribution to understanding the
phenomenon of shrinkage - swelling, and a useful complementary tool for civil engineers face this problem
allowing them to find appropriate solutions.
Keywords: clay - swelling –swelling pressure– lime – cement -Algeria

1 INTRODUCTION

Le traitement des sols est souvent utilisé pour augmenter leur résistance, pour réduire ou augmenter leur
perméabilité ainsi que pour diminuer leur compressibilité. Il est aussi utilisé pour minimiser la sensibilité du sol
aux variations de la teneur en eau comme dans le cas des sols expansifs. La stabilisation du gonflement par ajout
de chaux a fait l’objet de plusieurs études (Khattab, 2002), (Cabane, 2004), (Al-Rawas et al., 2005), (Abdo
2008) , (Asgari et al. 2013), et toutes ces études confirment le rôle positif qu’a cet ajout sur la réduction du
gonflement des sols expansifs. Dans cet article, on va d’une part, apprécier l’influence de l’ajout du système
chaux ciment à différents pourcentages sur le potentiel de gonflement, la pression de gonflement et les limites
de consistance de l’argile étudiée, et d’autre part, étudier l’influence de ces ajouts sur la microstructure de cette
argile par des observations au microscope électronique à balayage (MEB). Ces observations nous permettront
d'avoir une meilleure compréhension des traitements et connaître la disposition des grains dans les différents
mélanges.

631
2 LOCALISATION DU SITE ET IDENTIFICATION GÉOTECHNIQUE
Le site objet de cette étude se trouve à Mers El Kébir localisée au Nord Ouest d’Oran. Il s’agit d’une argile
d’environ 3OO m d’épaisseur. Ces argiles marneuses, en gonflant, se soulèvent et peuvent atteindre des
amplitudes importantes, ce qui induit sur place l’apparition de fissures préjudiciables aux fondations des
ouvrages. Le site choisi est localisé dans le fossé miocène de Mers El Kébir. Le matériau de Mers El Kébir a fait
l'objet de recherches antérieures (Hachichi & Fleureau, 1999), (Hachichi et al. 2007), (Bourokba et al. 2010) ,
(Hachichi et al. 2011) et (Bourokba, 2011).
2.1 Caractérisation de l’argile

Les résultats des essais d’identification sont rassemblés dans le Tableau 1. D'après les classifications LCPC1 et
USCS2, la marne de Mers El Kébir est peu plastique (Ap (LCPC) = CL (USCS)). Son activité Ac est élevée
avec un pourcentage d’éléments inférieurs à 2 µm de 26 %. La surface spécifique totale déduite de l’essai au
bleu de méthylène par la formule de Tran Ngoc Lan, 1977 (Bourokba, 2011) est de 189 m2/g. A l’optimum
Proctor normal (OPN), la densité sèche maximale est de 1,63 et la teneur en eau est de 20%.

Tableau 1. Paramètres d’identification de l’argile étudiée.

γs WL WP IP WR F2 Ac VBS SST CaCO3 γd max Wopt
Paramètres
(g/m3) (%) (%) (%) (%) (%) (-) (%) (m2/g) (%) (g/m3) (%)
Valeur du
2,67 49 21,5 27,5 12 26 1,10 9,0 189 18 1,64 20
matériau utilisé

2.2 Caractérisation de la chaux

La chaux hydratée Ca (OH)2 est élaborée à partir d’un calcaire très pur à plus de 95% de CaCO3. Cette chaux est
très fine avec plus de 90% de matières qui passent au tamis de 80 µm et contient un minimum de 50% de chaux
libre et un maximum de 6% de MgO. Le tableau 2 représente l’analyse chimique de la chaux qui montre que le
taux de CaO est élevé, ce qui démontre la pureté de la chaux hydratée testée.

Tableau 2. Analyse chimique de la chaux utilisée. (Laboratoire LTPO3).

Minéraux SiO2 TiO2 Al2O3 Fe2O3 MnO MgO CaO K2O P2O5

Chaux (%) 0.39 0.007 0.22 0.11 0.005 0,428 77.32 - 0.056

2.3. Caractérisation du ciment

Le ciment utilise comme agent stabilisant de l’argile est : Le CPJ-CEM II/A 32.5 de la cimenterie de ZAHANA,
qui a comme ajout cimentaire la pouzzolane. Le tableau 3 représente les caractéristiques chimiques et physiques
du ciment

Tableau 3. Analyse chimique du ciment utilisé. (Laboratoire LTPO).

Paramètres Physiques Chimiques
Poids Surface Temps de
AL2O3 (%)

Perte au feu

(CaO libre)
carbonique
l'anhydrite
Fe2O3 (%)

Insolubles
MgO (%)

Temps de
SiO2 (%)

SO3 (%)
CaO (%)

Dénomination volumique spécifique de début de

CO2

fin de prise
absolu Blaine prise

Notation Mv (g/cm3) SSB (m2/g) t t

Valeur 3.1 0.27 2h 40mn 3h 45mn 22,17 64,05 5,99 2,47 0,27 1,86 2,47 1,6 0,5

3 PRÉPARATION DES ÉCHANTILLONS

Les essais ont été effectués sur des échantillons remaniés en substituant la quantité d’argile par le système chaux
ciment à différents pourcentages. Après avoir broyé finement le sol, la poudre est mélangée à l’eau de façon

1 Laboratoire Centrale des Ponts et Chaussées

2 Unified Soil Classification System
3 Laboratoire des Travaux Publiques de l’Ouest (Algérie)

632
homogène puis compactée sous presse à vitesse lente (1,27 mm/min) dans un moule de 50 mm de diamètre et
100 mm de hauteur. La teneur en eau des échantillons et la densité sèche sont celles obtenues par l’essai Proctor
selon la norme NF P94-093.

4 RESULTATS ET DISCUSSION

4.1. Effet du système ciment chaux sur les limites de consistance

Les limites d’Atterberg sont des paramètres essentiels dans l’estimation du potentiel de gonflement des sols
expansifs. Afin d’étudier l’influence de la stabilisation sur les limites de consistance de l’argile de Mers El
Kébir, des mesures des limites de liquidité et de plasticité ont été effectuées avec le ciment et la chaux à
différents pourcentages suivant la norme ASTM D 4318 (2000). Les résultats de ces essais sont consignés dans
le tableau 4.
Tableau 4 : Limites de consistances mesurées en présence du système ciment- chaux .
CIMENT + CHAUX (%)
Matériau Limites d’Atterberg H2O
1CM+3CH 2CM+2CH 2CM+4CH
WL (%) 49.0 42 46 41
Mers El Kébir WP (%) 21.5 24 22 25
IP (%) 27,5 24 17 16

A partir de ces résultats on constate que :

La limite de liquidité (wL) diminue et la limite de plasticité (wP) augmente avec les pourcentages ajoutés.
L’indice de plasticité (Ip) subit une diminution relativement importante avec le traitement. La limite de liquidité
varie de 49% pour l'argile non traité à 41% pour un ajout de 2% de ciment et 4% de chaux (2CM+4CH). Par
contre la limite de plasticité croit en fonction du pourcentage d'ajout, elle varie de 21% pour l’argile non traitée,
à 25%. Ceci influe directement sur l'indice de plasticité qui diminue de 41,8%. L’addition d’un pourcentage
supplémentaire de chaux a un faible effet sur les limites, wP se stabilise ou diminue légèrement. La diminution
de wL est en accord avec les résultats de la littérature ( Bell, 1994 ; Nalbantoglu 2006; Al Mukhtar, 2010;
Harichane et al, 2010 et 2011 ). Ceci peut être expliqué par la réaction de floculation des grains d’argiles à cause
de la présence de Ca++ et OH- qui jouent un rôle important en diminuant l’épaisseur de la double couche. La
diminution de la limite de liquidité peut être aussi expliquée par la diminution de la surface spécifique
(Bourokba, 2011).

4.2. Effet du système ciment chaux sur le potentiel de gonflement et sa pression.

Les essais utilisés pour mesurer le potentiel et la pression de gonflement ont été réalisés pour différents
pourcentages de ciment et chaux et pour des temps de mûrissements d’une heure, 7 j, 14 j, 28 j, 90j et 180 jours.
Les essais sont effectués sur des échantillons de 20 mm de diamètre et 50 mm de hauteur, compactés à l’OPN
(d = 1.64 Mg/m3 ; wopt = 20%).

4.2.1. Procédure expérimentale.

Afin d’étudier l’effet du système ciment chaux sur le gonflement de l’argile de Mers El Kébir, nous avons
mélangé les échantillons avec les ajouts à différents pourcentages. Ensuite ils ont été soumis à des essais de
gonflement libre et à des pressions de gonflement à volume constant à l’oedomètre selon la norme ASTM D
4546 (1990).
Les réductions du potentiel de gonflement et sa pression en pourcent sont données par les relations :

G G0 (%)  Gajout (%) P P0 (%)  Pajout (%)

 (1)  (2)
G G0 (%) P P0 (%)

Avec : G0 : gonflement de l’argile sans ajout, Gajout : gonflement de l’argile + l'ajout à un certain
pourcentage
P0 : pression de gonflement de l’argile sans ajout, Pajout : pression de gonflement de l’argile + l'ajout

633
4.2.2. Réduction du potentiel de gonflement.
La figure 1 représente la variation du gonflement en fonction du temps, la tendance de cette variation dans le
temps est la même pour tous les mélanges. Nous remarquons d’après les résultats graphiques, que le gonflement
libre présente deux phases distinctes, la première où le gonflement est rapide (gonflement primaire), la
deuxième, où le gonflement est lent (gonflement secondaire).
10
Non traitée
9
2% CM+2%CH

Potentiel de gonflement G (%)

8
1% CM+3%CH
7
2% CM+4%CH
6
5
4
3
2
1
0
1 10 100 1000 10000 100000 1000000

Temps (s)
Figure 1. Variation du potentiel de gonflement en fonction du temps
Les mesures des potentiels de gonflement de l'argile stabilisée sont représentées sur le tableau 5. A partir de ces
résultats, nous remarquons que l’effet de l’ajout est significatif, et que le taux de gonflement le plus faible est
obtenu pour l’argile stabilisée à 2%Ciment +4%Chaux.

Tableau 5. Influence des ajouts sur le potentiel de gonflement.

Ajouts (%) 0 2CM+2CH 1CM+3CH 2CM+4CH
Potentiel de gonflement
9,5 7,56 4,91 3,78
(%)

4.2.3. Réduction de la pression de gonflement.

Les résultats des pressions finales des argiles stabilisées pour un temps de mûrissement de 180 jours sont
présentés par le tableau 6. D'après ces résultats, l'influence des ajouts est très appréciable. Nous remarquons
que la diminution de la pression de gonflement à 1% de ciment + 3 % chaux est significative d’environ 59%. Au-
delà de cette valeur la diminution se stabilise légèrement jusqu'une valeur de 63,5 kPa qui correspond à un
ajout de 2% de ciment + 4 %chaux.

Tableau 6. Influence des ajouts sur la pression de gonflement.

Ajouts (%) 0 2CM+2CH 1CM+3CH 2CM+4CH
Pression de gonflement
180 175 73,5 63,5
(kPa)

4.2.4. Effet du temps de mûrissement sur la pression de gonflement.

Les mesures de la pression de gonflement ont été effectuées après un temps de mûrissement à l’air libre d’1
heure, 7j, 14j, 28j, 90j, et 180 jours suivies par une saturation de l’échantillon à l’eau tout en empêchant la
variation volumique. L’influence du temps de mûrissement a été étudiée pour le pourcentage le plus élevé
(2%CM+4%CH), sur la pression de gonflement et est représentée par la figure 2. Le mûrissement a été réalisé
en laissant les échantillons réagir avec le traitement à l’air libre.

95
90 2% CM+4% CH
85
80
Pg (Kpa)

75
70
65
60
55
50
0 50 100 150 200
Temps (j)

Figure 2. Effet du temps de mûrissement sur la pression de gonflement.

634
En général, on observe une diminution continue de la pression de gonflement avec le temps de mûrissement des
échantillons traités. Ceci confirme la continuité de la réaction chaux+ ciment/argile dans le temps. Une
hypothèse pouvant expliquer cette diminution est que la chaux réagisse avec la silice et l’alumine dans le sol et
forme un gel de C-S-A-H. Ce gel évolue et se cristallise avec le temps, remplissant les vides en réduisant la
possibilité de passages d’eau entre les particules. Cette baisse de la pression pourrait être aussi due à la
précipitation de carbonate de calcium dans les vides à cause de la réaction entre le calcium ionisé et le CO2
dissous dans l’eau (Cabane, 2004 et Lasledj, 2008). Finalement, on constate que la vitesse de diminution de la
pression de gonflement, décroît avec le temps de mûrissement, ce qui traduit un ralentissement du taux de
réaction ciment- chaux /argile.

4.3. Influence des ajouts sur la microstructure

4.3.1 Préparation des échantillons
Les échantillons traités sont compactés à l’OPN, et subissent ensuite des essais de gonflement. Afin que leur
microstructure ne varie pas, on doit procéder à l'élimination de l'eau à l'intérieur des échantillons à analyser tout
en conservant au mieux la texture du matériau. Pour cela, la technique qui semble la mieux adaptée à notre
matériau consiste à réaliser une dessiccation par lyophilisation (lyophilisateur de type Compact Flexi-dry). Les
échantillons, d'un volume de 1 cm3 environ, sont extraits de l'intérieur de l'éprouvette par simple coupe (figure
3). Et ils sont introduits dans de l'azote liquide (Delage et Pellerin 1984). On réalise ensuite une sublimation de
la glace dans le lyophilisateur pour le passage de l'état solide à l'état gazeux.

(1)
1,2, et 3 plans du découpage
(2)
(3)

Avant la congélation Après la congélation

(1)
Découpag Fracturation
e
Plan horizontal

Echantillon pour porosimétrie et MEB

(2 et 3)
Plan vertical

Figure 3. Plan de découpage et d’observation.

4.3.2. Examen au microscope électronique à balayage (MEB)

Le principe du MEB consiste à balayer la surface de l’échantillon avec un pinceau d’électrons, en analysant
les électrons secondaires réémis par le matériau. Cette technique donne des images topographiques de la surface
de l’échantillon avec une grande profondeur de champ. A la fin des essais de gonflement, la méthode consiste à
analyser, à partir d'observations au MEB, l'orientation des particules argileuses dans deux plans de l'éprouvette,
(1) un plan horizontal et (2) un plan vertical. Deux échantillons parallélépipédiques sont prélevés au centre de
l'éprouvette (figure 3) et les plans sont repérés par (1) ou (2). Les plans d'observation (1) et (2) sont obtenus par
fracturation d'un coup sec porté au moyen d'une lame, dans le sens longitudinal des échantillons congelés. Les
observations du plan vertical sont représentées sur la figure 4.

635
Argile non traitée (plan vertical 100 µm) 2% Ciment+ 2% Chaux (plan vertical 100 µm)

2% Ciment+4% Chaux (plan vertical 100 µm) 1% Ciment+ 3% Chaux (plan vertical 100 µm)
Figure 4. Microscopie Electronique à Balayage sur l’argile de Mers El Kébir traitée au système ciment chaux (plan vertical)

A travers cette figure, nous remarquons que le système ciment chaux influe sur la microstructure de l’argile.
Cette influence varie en fonction du pourcentage d’ajout utilisé. A l’état naturel sur le plan vertical la structure
représente un bloc avec des pores. Concernant les argiles traitées à 2% ciment + 2% chaux nous remarquons
qu’il y a une fixation globale et locale. Pour un ajout de 2% ciment + 4% chaux la fixation de la structure
semble aussi importante au niveau local car nous avons utilisé plus de chaux, le fait de l’augmenter donne un
meilleur résultat et enfin pour un ajout de 1% ciment +3% chaux, la fixation est moins importante vu qu’on a
diminué la chaux.
5 CONCLUSION

Lors de cette étude, nous avons essayé de trouver une solution afin d’améliorer les caractéristiques physico
mécanique en procédant à sa stabilisation chimique. Cette stabilisation peut se faire en procédant à la
substitution d’une partie de l’argile par des liants hydrauliques à savoir le ciment - chaux. A travers les essais
effectués sur l’influence de ces ajouts sur l'état de consistance, le potentiel et la pression de gonflement et la
microstructure, on peu conclure ce qui suit :
- La limite de liquidité diminue et la limite de plasticité augmente avec le pourcentage du système ciment
chaux ajouté. En conséquence, l’indice de plasticité (IP) subit une diminution très importante avec le
traitement ce qui diminue la plasticité de notre argile.
- Le système ciment - chaux est un réducteur efficace du potentiel et de la pression de gonflement, plus le
pourcentage d'ajout augmente plus le potentiel de gonflement et sa pression diminuent.
- A partir de l’analyse au microscope électronique à balayage (MEB) de l’argile traitée et pour le
pourcentage d’ajouts le plus élevé à savoir 2% ciment + 4% chaux, nous pouvons remarquer que la
structure reste homogène, les blocs sont collés entre eux et les pores sont remplis par les liants
fixateurs.

636
REMERCIEMENTS
Les auteurs remercient Mme Salima Aggoun du laboratoire L2MGC de l’Université de Cergy Pontoise (France)
de nous avoir aidés à réaliser les observations au microscope électronique à balayage (MEB) du matériau.

REFERENCES
Abdo J. (2008) Traitement des sols en place aux liants hydrauliques. Actes de la journée valorisation des sols médiocres,
Paris, 10 décembre.
Al-Mukhtar M, Lasledj A, Alcover JF (2010) Behaviour and mineralogy changes in lime-treated expansive soil at 20 °C.
Appl Clay Sci 50: 191-198.
Al-Rawas A.A., Hugo A.W, Al-Sarmi H. (2005) Effect of lime,cement and Sarooj (artificial pozzolan) on the swelling
potential of an expansive soil from Oman. Building and Environment. Vol 40, Elsevier, pp. 267- 281.
Asgari M. R, Baghebanzadeh Dezfuli A. & Bayat. M. (2013) Experimental study on stabilization of a low plasticity clayey
soil with cement/lime. Arab J Geosci. DOI 10.1007/s12517-013-1173-1.
ASTM D 4546, (1990) Standard test method for one-dimensional swell or settlement potential of cohesive soils. Annual
Book of ASTM Standards vol. 04.08 American Society for Testing and Materials Easton Philadelphia.
ASTM D-4318 (2000) Standard test method for liquid limit, plastic limit and plasticity index of soils. Annual Book of
ASTM Standards, vol. 04.08 American Society for Testing and Materials Easton Philadelphia.
Bell F.G. (1994) An assessment of cement PFA and lime-PFA used to stabilized clay-size materials. Bulletin of the
Association of Engineering Geology, No.49, pp 25-32, Paris.
Bell F.G. (1996) Lime stabilization of clay minerals and soils. [Link]., Vol.42, pp 223-237.
Bourokba S A, Hachichi A, Fleureau J-M. (2011) Influence du ciment sur le gonflement et la microstructure d’une argile
naturelle d’Algérie. XXIXe Rencontres Universitaires de Génie Civil. Tlemcen Algérie.
Bourokba S A. , Hachichi A., Taibi S., Fleureau J.-M. (2010) Conductivité hydraulique non saturée de l’argile de Mers El
Kébir (Algérie). EJECE. Vol. 14, n° 10.
Bourokba S A. (2011) Etude de la conductivité hydraulique et de la stabilisation par ajouts de liants hydrauliques de l’argile
de Mers El Kébir. Thèse de doctorat Es Sciences, Université des Sciences et de la Technologie Med Boudiaf, Oran, Algérie,
pp.171.
Brandl H. (1981) Alteration of soil parameters by stabilization with lime. Comptes rendus du Xème congrès international de
mécanique des sols et des travaux de fondations, Vol. 3, Stockholm, 15-19 Juin.
Cabane N. ( 2004) Sols traités à la chaux et aux liants hydrauliques : Contribution à l’identification et à l’analyse des
éléments perturbateurs de la stabilisation. Thèse de doctorat, Ecole Nationale Supérieure des Mines de Saint-Étienne , 189
pages.
Delage P., Pellerin M. (1984) Influence de la lyophilisation sur la structure d’une argile sensible du Quebec. Clay Minerals.
Vol. 19, pp. 151-160.
Hachichi A., Bourokba S A., Fleureau J.-M. (2007) Stabilisation chimique des sols gonflants de la région d’Oran. Revue
Française de Géotechnique, n° 118.
Hachichi A., Fleureau J.M. (1999) Caractérisation et stabilisation de quelques sols gonflants d’Algérie. Revue Française de
Géotechnique, n° 86.
Hachichi A, Bourokba S A, Bengraa L Fleureau J-M. (2011) Influence de l’ajout de chaux sur le potentiel gonflement et la
microstructure d’une argile naturelle d’Algérie. INVACO2: Séminaire International, Innovation et valorisation en génie civil
matériaux et construction, Rabat, Moroco
Harichane, K, Ghrici M, Khebizi W, Missoum H. (2010) Effect of the combination of lime and natural pozzolana on the
durability of clayey soils. Elect J Geotech Eng 15 (L):1194-1210.
Harichane K, Ghrici M, Missoum H. (2011) Influence of natural pozzolana and lime additives on the temporal variation of
soil compaction and shear strength. Front Earth Sci 5(2):162-169.
Khattab S. A. (2002) Etude multi-échelles d’un sol argileux plastique traité à la chaux. Thèse de doctorat Université
d’Orléans, 224 pages.
Lasledj A, Al-Mukhtar M (2008) Effect of hydrated lime on the engineering behaviour and the microstructure of highly
expansive clay. The 12th International Conference of International Association for Computer Methods and Advances in
Geomechanics (IACMAG) 1-6 October Goa India.
Nalbantoglu Z. (2006) Lime stabilization of expansive clay In Al Rawas AA. Goosen MFA. Expansive Soils Recent
Advances in Characterization and treatment. Taylor & Francis Group, Balkema, London 341-348.

637
Author Index

Boubaker S. 315
A Boudia M. 631
Abchir Z. 273 Boukadi N. 451
Abdulaziz O.A. 97 Bourgeois E. 147
Abid S. 479 BourokbaMrabent S.A 631
Abou-bekr N. 129, 485 Boussetta S. 399
Abriak N.E. 297 Bruhier J. 379, 501, 509
Ackah P. 553 Burlon S. 121, 273
Ali Ahmed E.M. 225
Alimrina N. 463
Aljallad A. 433 C
Allal M.A. 129 Canou J. 113
Al-Marei A. 433 Chebbi M. 393
Al-Moukdad M. 433 Cherifi F. 517
Alonso E. E. 33 Cisse I.K. 619
Al-Tamimi A.K. 433 Coelho R. 323
Ampadu I.K 553
Ampadu S.I.K. 567
Arab R. 379, 501, 509 D
Assinder P. 379, 501, 509 Daito Y. 339
Attom M.F. 365 Daoud S. 573
Ayeh F. J. F. 553, 567 Debiche F. 241
Della N. 113
Derfouf F.E.M. 129
B Dhahri F. 451
Badenhorst W. 605 Djeraf N. 345
Bahar R. 463, 611 Djerbal L. 463
Bashier S. 561 Droniuc N. 121
Belabed L. 457 Dupla J.C. 113
Belkhatir M. 113
Ben Abdelghani F. 525
Ben Ghawar B.M. 211 E
Benamara F.Z. 457 El Badawy R. 301
Benassi R. 451 El Hamed M. A. 423
Bengra L. 631 Elarabi H. 561
Benmeddour D. 257 Elburas F.S. 211
Bensalem Z. 387 El-Emam M.M. 365, 433
Bheemasetti T. V. 69 El-Gehani G. 189
Boadu F.K. 567 Elhassan A.M. 375
Bouafia A. 249 Elkhawad T. 561
Bouassida M. 359, 387, 399, 407, 573 ElSakhawy N. 301

xii
Elsharief A.M. 97, 197, 375 Khemissa M. 351, 445, 625
Ennour S. 573 Khitas N.E.H. 257
Khouaouci M. 249
Khoudi I. 463
F Kimura M. 339
Fangleu T.R.F. 579 Klai M. 359
FarialaMuchanga V. 471
Fechner T. 219
Frank R. 273 L
Frikha W. 387, 399, 407 Lafhaj Z. 529
Frikha Y. 479 Lamara M. 485
Fukubayashi Y. 339 Larsson S. 307
Lawer A. 567
Le Kouby A. 273
G Leppla S. 17, 81, 89, 233
Goual I. 485 Lingwanda M. I. 307
Goual M.S. 485 Lukanda J. 595
Gress J.C. 181
Grimod A. 537
Gueddouda M.K. 485 M
Guerreros J.C.G. 219 Mabrouki A. 257
Guiras H. 393 Mackens S. 219
Magnan J.P. 121
Mahamedi A. 351, 625
H Maherzi W.
Mahmoud M.H.
525
415
Hachichi A. 631
Hamdi E. 529 Makharamedzha T. 205
Hazzar L. 289 Mamba M. 579
Hcine N. 479 Mansour M.A. 415
Hussien M.N. 289 Marshall D. H. 69
Martin J.R. 265
Matio P.B. 333
J Mbengue A.
Medjnoun A.
619
611
Jamei M. 393
Jjuuko S. 333, 595 Medjo R. 379, 509
Mekki L. 625
Melbouci B. 463
K Mellas M.
Meruti P.
257, 359
155
Kaffel W. 407
Kalumba D. 333, 595 Mirghasemi A. A. 545
Kammoun A. 479 Missaoui A. 529
Kana E. M. 3, 579 Moatlhodi L.J. 155, 161, 205
Karray M. 289 MukokoKalenda G. 587
Katzenbach R. 17, 81, 89, 233 Mutaasa H.K. 595
Kenmogne E. 265
Kettab R.M. 241
Khan Z.H. 365 N
Khatrush S. 189 Naciri O. 493

xiii
Nageeb T. 301 Sakou L. 379, 501, 509
Ndiaye M.M. 619 Samieh A.M. 415
Nechnech A. 345 Schoepf M. 281
Nechnech F.A. 241 Seddiki A. 445
NgoieNsenga D. 587 SeipM. 17
Ngom M.D. 121 Shakshuki M.T. 211
Nhlapho M.T. 205 ShimatuMbuyi J. 471
Nicolas R. 323 Sobolewski J. 501
Niederleithinger E. 219 Stott P. 167, 173, 605
Noaman M. 433
Nyaoro D.L. 307
T
Tabchouche S. 359
O Tabti S. 517
Ojuri O.O. 155 Taibi S. 129, 485
Okonta F.N. 155, 161, 205 Theron E. 167, 173, 605
Omer O. G. 197 Thimus J.F. 587
Osman M.A. 225 Tomásio R. 323
Owusu-Nimo F. 553, 567 TshimangaNzokona H. 471

P V
Pinto A. 323 Vollmert L. 493
Pizette P. 297
Puppala A. J. 69
W
WemeyiE. 265
R
Rabie M. H. 423
Remaud D. 147 Z
Rimoldi P. 537 Zadjaoui A. 137
Zairi M. 479
Zargayouna H. 399
S Zein A.K.M. 105
Safaifard D. 545 Zerhouni M.I. 61
Said I. 529, 573 Zeyeni K. 451

xiv

Common questions

Alimenté par l’IA