N. R. Krishnaswamy,1 K. Rajagopal,1 and G.

Madhavi Latha1

Model Studies on Geocell Supported

Embankments Constructed Over a Soft Clay
Foundation

REFERENCE: Krishnaswamy, N. R., Rajagopal, K., and Mad- al. (1979) identified different possible modes of grid cell failures
havi Latha, G., “Model Studies on Geocell Supported Embank- by conducting laboratory direct shear tests on sand-filled paper grid
ments Constructed Over a Soft Clay Foundation,” Geotechnical cells. Bathurst and Jarrett (1989) have studied the use of geocell re-
Testing Journal, GTJODJ, Vol. 23, No. 1, March 2000, pp.
045–054. inforcement for construction of pavements over extremely soft peat
subgrades. Jenner et al. (1988) have proposed a design methodol-
ABSTRACT: This paper describes the results of laboratory model ogy based on slip line theory for determining the increase in bear-
tests on geocell-supported earth embankments constructed over a ing capacity due to the provision of a geocell layer at the interface
soft clay foundation. The soft clay foundation was prepared in a between the soft clay foundation and embankment. Bathrust and
large test tank to a depth of 600 mm. A single geocell layer was Karpurapu (1993) and Rajagopal et al. (1999) studied the improve-
formed on this clay foundation with different thicknesses, and em-
bankments were constructed above this layer. Four different types
ment in the performance of granular soils by geocell confinement.
of geogrids were used for the formation of the geocell layer. The In this paper, results from a series of load tests on embankments
embankments were subjected to uniform surcharge pressure on the constructed over a soft clay foundation with and without geocell re-
crest until failure. The vertical and horizontal deformations and the inforcement are presented. The results are analyzed to understand
strains developed within the geocell layer were measured during the the influence of geocell geometry in improving the load bearing ca-
test. The influence of various parameters such as tensile stiffness of
geogrids used to fabricate the geocell material, height, and pocket- pacity of soft foundation soil and reducing settlements.
size of the geocell layer, length of the geocell layer, and type of fill
material inside the geocell on the behavior of the embankments was Materials Used in Experiments
investigated through a series of laboratory tests.
The clay soil used for the preparation of the soft clay foundation
KEYWORDS: embankments, soil reinforcement, geocell, geo- can be classified as clay with low plasticity (CL) according to the
synthetics, soft clay, model tests Unified Soil Classification System. The type of soil used for the
construction of the embankment was a poorly graded sand (SP)
with some clay fraction and is referred to as clayey sand in this pa-
The use of geosynthetics for the construction of earth structures per. The geocells were formed using four different types of ge-
is simple and more economical compared with other forms of soil ogrids. Two of these are made of oriented grids, one with a uniax-
improvement techniques such as preconsolidation using vertical ial grid (UX) made of polyethylene and the other with a biaxial grid
drains and stone columns. Geosynthetic reinforcement has suc- (BX) made of polypropylene. The other two geogrids (NP-1 and
cessfully been used to construct embankments over soft clays NP-2) are made of nonoriented high density polyethylene. The
(Rowe et al. 1995). Potential excessive settlements and bearing ca- properties of geogrids were determined from standard wide width
pacity failure of the embankments are arrested by the geosynthetic strip tension tests (ASTM D 4595) and are listed in Table 1. The
reinforcement due to the transfer of stresses from the soil to the re- joints in the geocells were made using 7-mm-wide and 3-mm-thick
inforcement. A more recent advancement of reinforced soil is to plastic strips cut from commercially available bodkins made of
provide three-dimensional confinement to the soil by using geo- low-density polypropylene.
cells (Cowland and Wong 1993; Hendricker et al. 1998). These
geocells are honeycombed interconnected cells that completely en- Laboratory Testing Program
case the soil and provide all-around confinement, thus preventing
the lateral spreading of soil. Because of this confinement, the soil- A steel tank having plan dimensions of 1800 by 800 by 1200-
geocell layer acts as a stiff mat, thus distributing the load over a mm depth was used for performing the model test. The tank was fit-
much larger area of the subgrade soil. This distribution of load over ted with 20-mm-thick Perspex™ (polymethyl methacrylate) sheet
a larger area helps in reducing the total and differential settlements. on one side to monitor the failure surfaces developed in the em-
Several investigators have studied the performance of geocell re- bankment. The other three sides of the tank were made smooth and
inforced soil for construction of foundations and embankments rigid to create plane strain conditions in the tank. All sides of the
over soft soil subgrades. Rea and Mitchell (1978) and Mitchell et test tank were lined with two layers of plastic sheets whose inner
layers are coated with lubricating oil to reduce the effects of side-
1
Professor, associate professor, and Ph.D. scholar, respectively, Department wall friction. The longitudinal sides of the tank were made stiff by
of Civil Engineering, Indian Institute of Technology, Madras, Chennai, India supporting them with rigid braces. The clay soil was initially mixed
600 036. with excessive amounts of water and poked with steel rods to ob-
© 2000 by the American Society for Testing and Materials
45
46 GEOTECHNICAL TESTING JOURNAL

TABLE 1—Properties of geogrids.

Property UX Geogrid BX Geogrid NP-1 Grid NP-2 Grid

Ultimate tensile strength, kN/m 40 20 4.5 7.5

Failure strain, % 28 25 10 55
Initial modulus, kN/m 267 183 75 95
Secant modulus at 5% strain, kN/m 200 160 70 70
Secant modulus at 10% strain, kN/m 95 125 45 50
Aperture opening size, mm 210 ⫻ 16 35 ⫻ 35 50 ⫻ 50 8⫻7

TABLE 2—Properties of foundation clay bed. steady state was achieved within 10 min of the application of load
increment. This surcharge pressure was applied through a precali-
Water content 25%
CBR value 0.55% brated proving ring using a hydraulic jack supported against a re-
Vane shear strength 20 kPa action frame. The uniform distribution of the surcharge pressure
In-situ unit weight 17 kN/m3 was achieved by applying the loads through rigid I-sections run-
In-situ void ratio 0.907 ning over the full width of the test tank. A thick steel plate and an
Degree of saturation 74% expanded polystyrene sheet were placed between the I-sections and
the embankment crest to ensure uniform load distribution as shown
in Fig. 1.
The vertical and horizontal deformations in the model embank-
TABLE 3—Properties of embankment fill. ment were measured using dial gages placed at different locations.
Three dial gages (V1, V2, and V3) were placed on the crest of the
Placement moisture content 12%
Placement unit weight 19 kN/m3 embankment to measure settlement of the embankment. Three dial
Angle of internal friction 30° gages (H1, H2, and H3) were placed along the slope at different
Cohesive strength 10 kPa heights of the embankment to measure lateral deformations. One
CBR value 1.2% dial gage (V4) was placed on top of the geocell layer outside the
body of the embankment to measure the heave of adjacent soil
when the embankment was loaded. The locations of different dial
gages are shown in Fig. 1.
tain uniform mixing and then consolidated under a surcharge pres- The test program also included some tests without the geocell
sure of 10 kPa for one week. After this consolidation, the layer was layer to estimate the performance improvement due to the provi-
trimmed to a 600-mm depth for all tests. This procedure was found sion of geocell layers. The heights of geocell layers (h) were varied
to produce a soft clay bed with uniform properties. Several undis- from 100 to 250 mm in increments of 50 mm. Two different pocket
turbed samples were collected from the bed after consolidation to sizes (d ) of 200 and 400 mm were used in the tests. The aspect ra-
determine the (CBR) value and the vane shear strength of the clay tio (h /d ) of geocell pockets in the tests varied from 0.25 to 0.625.
soil. The unit weight, moisture content, vane shear strength, and Two different patterns, namely diamond and chevron patterns
CBR value of the clay foundation were maintained almost the same (Dean and Lothian 1990; Bush et al. 1990), were used for the for-
for all the model tests by carefully controlling the amount of water mation of the geocell layers (Fig. 2). Four different types of ge-
added and using uniform mixing procedures. The properties of the ogrids were used for the formation of geocell layers. Two different
soft clay bed are listed in Table 2. soils (viz., clayey sand and clay soil) were used to fill the geocell
After leveling the clay bed, a layer of geocells was formed on top openings. In all the tests, the geocell layer was formed over the full
of the clay bed. This was done by cutting the geogrids to required length of the test tank except in one test in which it was truncated
length and height from full rolls and placing them in transverse and at the toe of the embankment.
diagonal directions with bodkins inserted at the connections (Bush
et al. 1990). After the formation of the geocell layer, pockets of the Results and Discussion
geocells were filled with soil. The soil within each cell was com-
pacted uniformly by tamping with a steel rod having an enlarged The load-strain behavior of different geogrids used to form the
base. The unit weight of infill soil was maintained at 17 kN /m3 for geocells is shown in Fig. 3. The uniaxial oriented geogrid (UX) has
all the tests. The compaction quality of this layer was verified by the highest ultimate tensile strength and stiffness. The geogrids
testing undisturbed core samples collected from at least six indi- made of nonoriented polymer (NP-1 and NP-2) have very low
vidual cells. Above the geocell layer, a symmetrical half of the em- strength and stiffness.
bankment was constructed using clayey sand in 50-mm lifts. The In all the embankment model tests, the failure was observed to be
number of blows on each layer was adjusted to achieve a uniform due to the formation of slip circles passing through the soft clay soil
average unit weight of 19 kN/m3 in all the tests. The properties of (Fig. 4). The soil beyond the embankment was observed to heave up
the embankment soil are given in Table 3. as the embankment settled into the soft clay soil. At collapse state,
The physical dimensions of the embankment and the setup used the surcharge pressure on the embankment crest reached a peak, and
to apply uniform surcharge on its crest are shown in Fig. 1. The sur- it was not possible to increase it any further because of large defor-
charge pressure was applied in small increments until the failure. mations taking place below the hydraulic jack. The examination of
Each pressure increment was kept constant until the deformations geocells excavated from the soil after the tests did not reveal any ev-
under that pressure increment reached steady state. Typically, the idence of rupture of the geogrid or joint failures.
 KRISHNASWAMY ET AL. ON GEOCELL SUPPORTED EMBANKMENTS 47

FIG. 1—Geometry and test setup of model embankments.

bankments shown in Figs. 5 and 6. The embankments supported on

geocells made of uniaxial geogrid (UX) showed the least deforma-
tions because of the comparatively higher stiffness of this geogrid.
On the other hand, the embankments supported on geocells made
of nonoriented geogrids (NP-1 and NP-2) showed a softer response
because of their lower stiffness. Although both the nonoriented ge-
ogrids exhibited the same secant stiffness in the tension tests, the
performance of embankments supported on geocells made of NP-2
type geogrid was observed to be better than that with the NP-1 type
FIG. 2—Patterns used for the formation of geocells. geogrid. This may be because of better confinement offered to the
soil by NP-2 geogrids due to their smaller aperture openings.
The slip surfaces formed in the foundation soil for embankments
with three different types of geocell materials are shown in Fig. 4.
The variations of lateral and vertical deformations at different The slip surface observed with the BX type geogrid was found to
surcharge pressures for various model embankments are shown in be very close to that of the UX geogrid and hence is not shown
Figs. 5 and 6. The vertical deformations at all the three locations on here. It may be observed that the slip surface radius is larger for
the crest of the embankment were found to be almost uniform un- geocells made of geogrid with higher stiffness. Hence, the mobi-
til the failure of the embankments. The typical close match between lized shear resistance along the failure surface increases leading to
the results from repeat tests is illustrated in these figures. This close the better performance of embankments supported on geocells
match illustrates the consistency of soil properties achieved in the made of stiffer geogrids.
different trials and the repeatability of the test data. The steep in- The surcharge pressure at which the embankments are deemed to
crease in pressures in Fig. 6 towards the end of the tests may be due have failed is defined as the pressure at which the load-lateral de-
to the increase in the stiffness of clay soil after large settlements formation curve has the highest curvature. This pressure is defined
have taken place. This may have occurred due to the limited depth as the surcharge capacity and was obtained from the intersection
of the soft clay foundation used in the tests. As the embankment is of tangent lines drawn from the origin and through the straight por-
free to move in the lateral direction, the lateral deformations mea- tion at the end of the curve. The surcharge capacity of embank-
sured in the embankment showed expected trends with increasing ments and the corresponding deformations for different tests are re-
deformations at higher surcharge pressures. Hence, for all compar- ported in Table 4. The geocell-supported embankments gave
ative studies, the lateral deformations have been used. higher surcharge capacities and lower deformations compared to
The geocells made of geogrids with higher stiffness values were unreinforced embankments, e.g., embankment supported on 100-
found to be more effective in reducing the deformations. This is ev- mm-thick geocell layer made of UX geogrid developed almost
ident from the pressure-deformation responses of different em- twice the collapse pressure of the unreinforced embankments. The
48 GEOTECHNICAL TESTING JOURNAL

FIG. 3—Load-strain behavior of geogrids used in the test program.

FIG. 4—Failure surfaces observed within the soft clay layer.

 KRISHNASWAMY ET AL. ON GEOCELL SUPPORTED EMBANKMENTS 49

FIG. 5—Influence of stiffness of geocell material on lateral deformation of model embankments.

FIG. 6—Influence of stiffness of geocell material on settlement response of model embankments.

50 GEOTECHNICAL TESTING JOURNAL

TABLE 4—Results from different model embankment tests.

Secant Dimensions of Geocell Maximum Displacements

Modulus Recorded by % Increase
Type of at 5% Pocket Aspect Pattern of Different Dial Gages Surcharge in q from
Geogrid Used to Strain Height, Size, Ratio, Type of Formation Capacity, Unreinforced
Make Geocells (kN/m) h (mm) d (mm) h/d Fill material of Geocells V* H1 H2 H3 V4 q (kPa) Case

Unreinforced … … … … … … 127 3.8 4.5 13 10.3 50 …

UX 200 100 400 0.25 clayey sand diamond 84 0.6 1.5 1.8 2.0 95 90
BX 160 100 400 0.25 clayey sand diamond 97 0.8 1.5 4.3 5.0 75 50
BX 160 150 400 0.375 clayey sand diamond 72 0.5 1.5 2.3 3.6 85 70
BX 160 200 400 0.5 clayey sand diamond 53 0.3 0.3 2.0 2.4 91 82
BX 160 250 400 0.625 clayey sand diamond 42 0.2 0.6 1.0 2.3 95 90
BX 160 100 200 0.5 clayey sand diamond 59 0.4 0.6 2.6 3.2 95 90
BX 160 100 400 0.25 clay diamond 106 0.9 3.0 9.0 4.5 65 30
BX 160 100 400 0.25 clayey sand chevron 89 0.7 2.0 5.7 3.7 80 60
UX 200 100 400 0.25 clayey sand diamond 87 0.4 1.1 3.4 3.6 85 70
(truncated at toe)
NP-1 70 100 400 0.25 clayey sand diamond 107 1.4 3.1 5.3 3.2 65 30
NP-2 70 100 400 0.25 clayey sand diamond 95 0.6 0.9 1.1 2.0 70 40

* V is the average of displacements recorded by V1, V2, and V3.

FIG. 7—Influence of stiffness of geocell material on pressure-heave response of model embankments.

geocell reinforcement also reduced the heave of soil beyond the formations at mid-height of embankments is shown in Fig. 8. The
embankment. This reduction in heave shows that the geocell layers improvement in the performance was observed to be substantial for
distribute embankment loads over a wider area. geocell heights up to 200 mm. When the height was increased to
Figure 7 shows a comparison of the surface heave observed with 250 mm, further improvement in the performance was found to be
100-mm-high geocells of different types. The heaving was ob- relatively small (Table 4). It may be observed that the aspect ratio
served to decrease with an increase in the stiffness of the geogrid. (height to diameter ratio) of geocell pockets plays an important role
The effect of height of geocell (BX geogrid only) on the lateral de- in its performance. For example, when the aspect ratio was in-
 KRISHNASWAMY ET AL. ON GEOCELL SUPPORTED EMBANKMENTS 51

FIG. 8—Influence of aspect ratio of geocell layer on pressure-deformation response of model embankments.

creased from 0.25 to 0.5, the performance of the embankments im- practical difficulties in filling the geocell pockets with clay soil and
proved. When the aspect ratio was increased to 0.625, further im- its compaction.
provement was found to be marginal. For the same aspect ratio of Figure 11 and 12 show the comparison of the performance of the
0.5, the embankments supported on geocell layers (made of BX ge- embankment constructed on a short-length geocell layer extended
ogrid) with pocket dimensions (h ⫻ d) of 100 ⫻ 200 and 200 ⫻ to the embankment toe and one constructed on a full-length geocell
400 mm were observed to give almost similar performance (Table layer. The surcharge capacity was found to be almost the same for
4). both cases. However, the deformations (especially the heave of soil
The performance of geocells formed in both chevron and dia- surface) were found to increase significantly when the geocell layer
mond patterns were found to be very similar (Fig. 9). However, the was truncated at the toe of the embankment as shown in Fig. 12.
chevron pattern involves a larger number of joints (Fig. 2), and In actual field practice, it is common to spread a layer of geogrid
hence the failure of a single joint during the service life may not on the surface of soft clay and attach it to the geocell layer to facil-
lead to serious problems. Further, the joints in the chevron pattern itate the movement of construction equipment on the clay layer.
are easier to form than in the diamond pattern as it involves con- Similarly, one geogrid layer is provided at the top of the geocell
necting only two vertical geogrid surfaces, whereas the diamond layer for ease in construction of fills above the geocell layer. Some
pattern involves connecting three different vertical geogrid sur- tests were repeated with these additional geogrid layers. Because of
faces. the short heights of geocells used in these investigations, further
Figure 10 shows the comparison of lateral deformations at mid- improvement in the performance was not observed. Hence, these
height of embankments supported with geocells filled with clay results are not reported in the paper.
and clayey sand. Two tests with clay soil fill were conducted, and
the average results are shown in the figure. As we can observe from Summary and Conclusions
the figure, even with clay soil fill, significant improvement can be
achieved. From this result we may conclude that the improvement This paper reports the results of model tests used to quantify the
in the performance is due more to the presence of the geocell layer improvement in the performance of embankments constructed on
than the type of soil fill used in the construction. The practical im- soft clays due to the provision of a geocell reinforcement layer at
plication of this observation is that if good quality soils are not the base. The improvement was measured in terms of the increase
available at or near the site, locally available soil fills can be used in surcharge capacity at the crest of the embankment and reduction
in geocells if found suitable from other considerations such as soil in deformations. The results from this investigation have shown the
plasticity and drainage requirements. However, there may be many comparative influence of various parameters on the performance of
 FIG. 9—Influence of pattern of geocell formation on pressure-deformation response of model embankments.

FIG. 10—Influence of type of soil fill in geocell pockets on pressure-deformation response of model embankments.
52
FIG. 11—Influence of truncation of geocell layer on pressure-deformation response of model embankments.

FIG. 12—Influence of truncation of geocell layer on pressure-heave response of model embankments.

53
54 GEOTECHNICAL TESTING JOURNAL

geocell supported embankments constructed on soft foundations. Bathurst, R. J. and Karpurapu, R. G., 1993, “Large-Scale Triaxial
However, because of scale effects, the results from this investiga- Compression Testing of Geocell-Reinforced Granular Soils,”
tion cannot be directly extrapolated to real size embankments. The Geotechnical Testing Journal, Vol. 16, No. 3, pp. 296–303.
following major conclusions are drawn from the results of this Bush, D. I., Jenner, C. G., and Bassett, R. H., 1990, “The Design
investigation. and Construction of Geocell Foundation Mattress Supporting
Embankments Over Soft Ground,” Geotextiles and Geomem-
1. Provision of a layer of geocells at the base of the embankment branes, Vol. 9, pp. 83–98.
improves the load capacity and deformation response of the em- Cowland, J. W. and Wong, S. C. K., 1993, “Performance of a Road
bankment. Embankment on Soft Clay Supported on a Geocell Mattress
2. The tensile stiffness of geogrid used to manufacture the geo- Foundation,” Geotextiles and Geomembranes, Vol. 12, pp.
cell layer has an important influence on the performance of geo- 687–705.
cell-supported embankment. Dean, R. and Lothian, E., 1990, “Embankment Construction Prob-
3. The aspect ratio (height to diameter ratio) of geocell pockets lems Over Deep Variable Soft Deposits Using a Geocell Mat-
was also found to have an important influence on the performance tress,” Performance of Reinforced Soil Structures, British
of geocell-supported embankment models. The optimum aspect ra- Geotechnical Society, London, pp. 443–447.
tio was found to be 0.5. Hendricker, A. T., Fredianelli, K. H., Kavazanjiam, E., Jr., and
4. The geocell layer can be truncated at the toe of the embank- McKelvey, J. A., III, 1998, “Reinforcement Requirements at a
ment without any significant reduction in the load capacity of em- Hazardous Waste Site,” Proceedings of the 8th International
bankments. Conference on Geosynthetics, Atlanta, GA, Vol. 1, pp. 465–
5. The improvement in the performance benefit due to the pro- 468.
vision of geocell layer constructed with both a clayey sand and clay Jenner, C. G., Bush, D. I., and Bassett, R. H., 1988, “The Use of
soil infill was observed to be significant. Hence, even poor quality Slip Line Fields to Assess the Improvement in Bearing Capacity
soils can be used for infill. of Soft Ground Given by a Cellular Foundation Mattress In-
6. The performance of geocells formed in both chevron and di- stalled at the Base of an Embankment,” Proceedings of the In-
amond patterns was found to be very similar, and both can be ternational Geotechnical Symposium on Theory and Practice of
used with the same efficiency. However, the chevron pattern is Earth Reinforcement, Balkema, Rotterdam, pp. 209–214.
more attractive for field applications because of its simplicity in Mitchell, J. K., Kao, T. C., and Kavazanjiam, E., Jr., 1979, “Anal-
joint connection. ysis of Grid Cell Reinforced Pavement Bases,” Technical Report
GL-79-8, U.S. Army Engineer Waterways Experiment Station,
Acknowledgments Vicksburg, VA.
The authors gratefully acknowledge the financial support for this Rajagopal, K., Krishnaswamy, N. R., and Madhavi Latha, G.,
work from the Department of Science and Technology, Govern- 1999, “Behaviour of Sand Confined with Single and Multiple
ment of India, New Delhi under sanction No. III 4(36)/90-ET. They Geocells,” Geotextiles and Geomembranes, Vol. 17, No. 3, pp.
acknowledge also Netlon India, a Division of Parry & Co. Ltd., 171–184.
Vadodara, for both financial support and for supplying different Rea, C. and Mitchell, J. K., 1978, “Sand Reinforcement Using Pa-
geosynthetic products free of cost under an MOU signed with IIT per Grid Cells,” Reprint 3130, Proceedings, ASCE Spring Con-
Madras. vention and Exhibit, Pittsburgh, PA, pp. 24–28.
Rowe, R. K., Gnanendran, C. T., Landva, A. O., and Valsangkar,
References A. J., 1995, “Construction and Performance of a Full-Scale Geo-
Bathurst, R. J. and Jarrett, P. M., 1988, “Large Scale Model Tests textile Reinforced Test Embankment, Sackville, New
of Geocomposite Mattresses Over Peat Subgrades,” Transporta- Brunswick,” Canadian Geotechnical Journal, Vol. 32, No. 3,
tion Research Record, No. 1188, pp. 28–36. pp. 512–534.