American Journal of Civil Engineering and Architecture, 2017, Vol. 5, No.

5, 187-195
Available online at [Link]
©Science and Education Publishing
DOI:10.12691/ajcea-5-5-2

2D and 3D Numerical Simulation of Load-Settlement

Behaviour of Axially Loaded Pile Foundations
Gowthaman S1, Nasvi MCM2,*
1
Department of Engineering Technology, University of Jaffna, Jaffna, Sri Lanka
2
Department of Civil Engineering, University of Peradeniya, Kandy, Sri Lanka
*Corresponding author: nasvimcm@[Link]

Abstract Reliable prediction of settlement behaviour of axially loaded piles is one of the major concerns in
geotechnical engineering. Therefore, this paper focuses on the finite element solutions of load-settlement behaviour
of a single pile and pile group using PLAXIS numerical package. Three different types of analysis were incorporated:
a linear elastic analysis, a complete nonlinear analysis and a combined analysis. The pile case history with settlement
measurements made during field pile load test was considered to validate the single pile load-settlement simulation,
and the same load test result was extended to simulate the load-settlement behaviour of pile group using RATZ
analytical approach. The single pile analysis results suggest that realistic load-settlement predictions can be drawn
by considering complete soil as Mohr-Coulomb model at lower working loads, and incorporation of an interface
zone thickness of two times pile diameter using Hardening-Soil model is required to simulate the load-settlement
behaviour at higher working loads. The group pile analysis results provide a better load-settlement prediction when
incorporating an interface zone thickness of pile dimeter from the pile shat using Hardening-Soil model while
leaving the remaining soil as Linear-Elastic material.
Keywords: Hardening-Soil, Interface zone, Linear-elastic, Mohr-Coulomb, nonlinear, RATZ approach
Cite This Article: Gowthaman S, and Nasvi MCM, “2D and 3D Numerical Simulation of Load-Settlement
Behaviour of Axially Loaded Pile Foundations.” American Journal of Civil Engineering and Architecture, vol. 5,
no. 5 (2017): 187-195. doi: 10.12691/ajcea-5-5-2.

Nowadays, researchers intend to provide immediate and

suitable solutions through numerical analysis for various
1. Introduction field problems and those scenarios can be used for the
same kind of field problem in the future as well. In the
Nowadays, lot of attention has been paid on axially past decade, there was a rapid increment in developing
loaded pile foundations to construct Civil Engineering finite element packages, and it has been concluded that the
infrastructures due to their high load bearing capacity and solutions obtained from finite element analysis are more
applicability in various kinds of geological scenarios. As reliable and accurate [9,15-23]. Zakia et al. [18] have
settlement plays a controlling role in the design of conducted a finite element study on the effect of
pile foundations, number of studies [1-23] have been modelling parameters in settlement predictions using
undertaken by several researchers to capture the exact PLAXIS 2D. Based on their findings, they concluded the
solutions for settlement. All these methods for the following: (i) modulus of interface is very much closer to
settlement analysis of pile foundations can be classified modulus of the soil that in contact with pile, (ii) interface
under three popular categories including (i) Experimental reduction factor provides good agreement when the value
Method, (ii) Analytical methods and (iii) Numerical methods. is between 0.8 and 0.9, and (iii) modelling the soil
Few researchers [1,2,3] have conducted experimental completely as Mohr Coulomb model with consideration of
study on the settlement behaviour of axially loaded piles. interface provides better settlement. Ju [20] has carried out
Analytical method on pile group settlement and concept of settlement analysis using PLAXIS 3D finite element
‘Interaction effect’ in pile groups have been introduced package for pile-group located in sleech strata. Based on
and pioneered by Poulos [4]. Continuously, many the findings, it was concluded that a combination of
researchers [5-13] have conducted analytical study for the nonlinear and linear elastic analysis (using Hardening-Soil
vertical deformation of pile foundations using different model and Linear-Elastic) leads more realistic settlement
techniques such as theoretical load-transfer curves, predictions and a value of half of the pile width as the
discrete layer approach, elastic solutions, theory of interface thickness would be sufficient to capture the load
pile-soil interactions and hybrid approaches. However, in transfer mechanism of a pile group. Jian-lin et al. [21]
this method, problems have often been solved by making have conducted a FEM study on settlement prediction of
number of simplifying assumptions regarding the pile foundation in deep clayey soil deposit. They proposed
geometry and material properties [14]. a useful equation to correct the compression modulus
 American Journal of Civil Engineering and Architecture 188

obtained from laboratory test to simulate deep in-situ soil covers about 15% of the total area of Singapore and
conditions. Alnuiam et al. [19] have developed a finite consists predominantly silty sand with fines content of
element model (FEM) using PLAXIS 3D package to study about 20% to 30% [24]. Site investigation programme
the settlement behaviour of pile in Toyoura sand and including Standard penetration tests was carried out to
they concluded that Mohr-Coulomb model is the best provide the required engineering information and
material model to predict the settlement behaviour of pile description of subsurface soil stratigraphy. The detail of
in Toyoura sand. Fuchun et al. [17] have carried out the sub soil profile in the pile test location is given in
settlement analysis of pile groups using PIGLET finite Table 1. The angle of friction (∅') and effective cohesion
element package and it was concluded that finite element (c') values shown in Table 1 were obtained using the
analysis is a best method to predict the settlement Standard Penetration Test (SPT) results based on the
behavior of group piles with adequate consideration of soil correlations developed for old alluvium [24]. The site
nonlinearity and interaction effect. investigation revealed that the ground water table is at a
However, reliable settlement prediction of pile foundations depth of one meter below the ground surface.
at typical working loads still remains as one of the major The tested single pile was precast circular concrete piles,
geotechnical engineering problems [3,20]. Also, selection 48 m long, driven to a depth of 47.5 m in the ground with
of suitable material models and modelling parameters a free-standing length of 0.5 m above the ground surface.
based on the geological conditions and loading conditions Static load test was undertaken to the reference pile up to
still remains as a huge challenge in numerical simulations 21, 000 kN (300 % of the working load) after three
of settlement behavior of pile foundations. Therefore, months from the installation. In this research study, the
major aim of this study is to analyze the settlement load-settlement behaviour obtained from this static pile
behavior of the axially loaded piles numerically by load test result was numerically simulated.
accounting the realistic nature of the problem for the safe For the numerical simulation of settlement behavior of
and economical design. group pile, a hypothetical 2 × 2 square pile group with a
symmetric arrangement (Figure 1(a)) was considered. The
geometry of individual piles in the pile group and the
2. Adopted Pile Case History stratigraphy (Figure 1(b)) were considered as similar to
that considered in the single pile analysis. Critical design
Results of static pile load test conducted in Woodland, spacing of the pile group to be located in sandy soils is
north western part of Singapore, is adopted for this found to be four times the pile diameter [25] and hence, a
research study. The stratigraphy of Woodland comprises spacing of four times the pile diameter was used in this
mainly the silty sand deposits called Old alluvium. Old study. Material properties of the pile and pile cap are
alluvium is one of the major stratigraphy of Singapore that summarized in Table 2.
Table 1. Summary of Engineering Information of subsurface soils

Dilatancy angle
modulus (MPa)

Poisson’s ratio
kh (×10-8 m/s)
Bulk density
content (%)

Dry density

cʹ (kN/m2)
Moisture

(kN/m3)

(kN/m3)

Elastics
Layer

Φʹ / °

(Ψ)/ °
(vs)
Depth (m) Soil description

1 0 – 2.4 Silty sand 22 20.3 16.6 18.8 5 32 21 0.3 5

2 2.4 – 5.4 Silty sand 22 20.3 16.6 18.8 10 34 22 0.3 5
3 5.4 – 8.4 Medium dense silty sand 22 20.3 16.6 18.8 10 34 28 0.3 5
4 8.4 – 14.4 Medium dense silty sand 18.2 20.7 17.6 6.4 10 34 28 0.3 5
5 14.4 – 26.4 Medium dense silty sand 16.3 20.3 17.6 3.8 10 34 42 0.3 4
6 26.4 – 41.4 Dense silty sand 16.3 20.3 17.6 3.8 15 34 56 0.3 3
7 41.4 – 44.4 Very dense silty sand 16.3 20.3 17.8 3.8 15 34 70 0.3 2
8 44.4 – 47.4 Very dense silty sand 16.3 20.3 17.8 3.8 15 34 63 0.3 2
9 Below 47.4 Siltstone 16.3 20.3 17.8 3.8 15 34 63 0.3 2

Figure 1. (a) Plan-view of 2 × 2 square pile group, and (b) Sectional view X-X
189 American Journal of Civil Engineering and Architecture

Table 2. Properties of pile and pile cap measurement (laboratory level) falls between 100 - 200
Properties Pile Pile cap kPa. The stiffness power in stiffness laws (m) lies between
Unit weight (kN/m3) 24 24 0.5 and 1. Normally consolidated coefficient of earth
pressure at rest (K0) and the Poisson’s ratio for unloading
Elastic Modulus (GPa) 30 30
and reloading (vur) and the failure ratio (Rf) are also
Poisson’s ratio 0.2 0.2
required input parameters for the Hardening Soil model. It
is suggested that, values of vur of 0.2 and Rf of 0.9 are
appropriate for the HS model at drained condition [13].
3. Research Methodology The best estimate parameters of each layers adopted for
the HS model are summarized in Table 3.
3.1. Adopted Finite Element Material Models
Numerical simulation of settlement behavior of axially
3.2. Finite Element Analysis of a Single Pile
loaded single pile and group pile were investigated using Finite Element Analysis of a single pile was performed
PLAXIS 2D and PLAXIS 3D finite element packages using PLAXIS 2D numerical package in axisymmetric
respectively. In both investigations, three primary finite conditions with two degrees of freedom of translation
element material models were incorporated and they are (i) per node. Three different types of FE analyses were
Linear Elastic (LE) model, (ii) Mohr Coulomb (MC) considered in this single pile finite element study: (i) a LE
model and (iii) Hardening Soil (HS) model. Linear Elastic analysis where all the soil was assumed to be linearly
(LE) model is based on Hookes’s Law of isotropic elastic, (ii) a Complete Non-Linear (CNL) analysis where
elasticity. It involves two basic elastic parameters i.e. soil was completely modelled using the HS or the MC
Elastic Modulus (E) and Poisson’s ratio (υ). Although the model, and (iii) a combined analysis (NL-LE) where soil
linear elastic model is not suitable to model the soil, it close to the pile shafts (interface zone) was modelled
may be used to model the stiff volume of soil or stiff using the HS model, while soil in the remaining area was
formulations in the soil. In this study, the piles and pile modelled considering it as either LE or MC material.
cap were modelled using LE material model. Mohr Three different sizes of interface thicknesses were selected
Coulomb (MC) model is one of the nonlinear models for the combined analysis: interface thickness of (i) D/2,
adopted in this research study. It is a simple nonlinear (ii) D and (iii) 2D from the pile shaft, where D is the pile
model which is based on soil parameters that are known in diameter.
most practical situations. This involves five input parameters: The soil was modelled using triangular elements with
Elastic modulus (E) and Poisson’s ratio (υ) for soil elasticity, 15 nodes with an elastoplastic law behaviour obeying the
friction angle (∅) and cohesion (c) for soil plasticity and Mohr-Coulomb failure criterion. The lateral sides of the
dilatancy angle (ψ). However, all the non-linear features computational domain were taken far enough from the pile
of soil behaviour are not included in this model. in order to avoid the boundary effect. Moreover, the
The Hardening Soil model is an advanced nonlinear models of single pile for different type of analysis were
model adopted for the simulation of soil behaviour. It made using the same working area of 70 m × 25 m. At
has been formulated in the framework of classical the bottom level of the model, all movements were
theory of plasticity. The total strains are calculated using restrained, whereas, at the lateral sides, lateral movements
a stress level dependent stiffness with a hyperbolic perpendicular to the boundary were prohibited. Default
stress-strain relationship that defers for virgin load and interface and interface strength reduction factor (Rint) were
unloading/reloading. In Hardening soil model, stiffness is incorporated in the simulation and Rint value of 0.85
described much more accurately by using three additional was adopted [18]. The size of the elements was to be as
stiffness input parameters: triaxial loading stiffness (E50), small as possible close to the pile shaft because of high
triaxial unloading stiffness (Eur) and odometer loading stress gradient, which can capture better pile behaviour.
stiffness (Eoed). It has been concluded that the values The geometry of the Finite element model is shown in
approximately satisfy the following equations: Eur = 3E50 Figure 2(a). The undeformed and deformed FEM mesh are
and Eoed = E50 [20,26]. Hence, these equations were used shown in Figure 2(b) and Figure 2(c) respectively.
to find out the parameters required for HS model. All Prescribed displacement was applied at the pile head and
these stiffness values are specified at a given reference variations of applied load with settlement were obtained.
stress (pref) and are assumed to vary directly with the The computed results from all the three types of analysis
horizontal effective stress raised to the power of m. were compared with the field test results in order to
Generally, the reference stress (pref) at the stiffness identify suitable model for the soil.
Table 3. Hardening Soil (HS) Parameters
Layer Depth (m) E50 (MPa) Eoed (MPa) Eur (MPa) pref (kN) m vur Rf
1 0 – 2.4 15.6 15.6 46.8 100 0.7 0.2 0.9
2 2.4 – 5.4 16.3 16.3 49.0 100 0.7 0.2 0.9
3 5.4 – 8.4 20.8 20.8 62.4 100 0.7 0.2 0.9
4 8.4 – 14.4 20.8 20.8 62.4 100 0.7 0.2 0.9
5 14.4 – 26.4 31.2 31.2 93.6 100 0.7 0.2 0.9
6 26.4 – 41.4 41.6 41.6 124.8 100 0.7 0.2 0.9
7 41.4 – 44.4 52 52 156 100 0.7 0.2 0.9
8 44.4 – 47.4 46.8 46.8 140.4 100 0.7 0.2 0.9
9 Below 47.4 46.8 46.8 140.4 100 0.7 0.2 0.9
 American Journal of Civil Engineering and Architecture 190

Figure 2. (a) Geometry of the Finite Element Model (b) FE mesh and (c) deformed mesh for complete LE analysis

3.3. Finite Element Analysis of a Pile Group The fine mesh analysis was undertaken in all over the
numerical study. The meshed geometry and meshed pile
PLAXIS 3D Numerical package was used to perform foundation are shown in Figure 4. The models of pile
the Finite Element study on settlement behaviour of groups for different type of analysis were performed using
axially loaded pile group. In this 3D analysis, the soil was the same working area of 70 m × 25 m × 25 m. At the
modelled by 15-node wedge elements. Similar to the bottom level of the model all movements were restrained,
single pile analysis, three different types of FE analyses whereas at the lateral sides, lateral movements
were performed in the numerical study including: (i) a LE perpendicular to the boundary were prohibited.
analysis where both the soil adjacent to the pile shafts and
between the piles were assumed to be linear elastic model,
(ii) a complete nonlinear (CNL) analysis where both the
soil adjacent to the pile shafts and between the piles were
modeled using either HS or MC model (iii) a combined
analysis (NL-LE/ NL-NL) where the soil close to the pile
shafts (Zone A in Figure 3) was modelled using HS model,
while the soil in the remaining area (Zone B in Figure 3)
was modelled either as LE or MC material. Three different
size of interface thickness for Zone A were selected for
combined analysis: (i) Zone extended to a distance (d) of
D from the pile shaft, (ii) Zone extended to a distance (d)
of D/2 from the shaft and (iii) Zone extended to a distance
(d) of D/4 from the shaft, where D is pile diameter.

Figure 4. Meshed geometry of group pile foundation

Initially, the results of 3D finite element analysis were

compared with well-established RATZ approach [11], and
based on the outcome, the best finite element model was
selected for any further analysis of settlement behaviour of
pile group. Study on the effect of pile spacing on group
load-settlement behaviour was also considered in this
research study. In addition, settlement analysis was
conducted for following three spacing (centre to centre):
(i) S = 2D, (ii) S = 4D and (iii) S = 8D, where S is the
center to center spacing and D is the pile diameter.
Moreover, group settlement ratios and pile group
efficiencies for the above models were determined and
Figure 3. Pile group layout of combined analysis compared with previous studies.
191 American Journal of Civil Engineering and Architecture

Figure 5. Load-settlement curve of the pile group obtained from RATZ approach

4. Results and Discussions variation between the predicted and measured values is
found to be due to the ignorance of deep-insitu effect in
Modulus, as suggested by Jian-lin [21]. Actually, the
4.1. Load-settlement Behaviour of a Single Modulus obtained from the laboratory tests significantly
Pile vary from the in-situ modulus of deep soil and sometimes
the difference is in the order of several times [21].
Predicted load-settlement plots using complete LE, MC Generally, settlement of shallow foundations using soil
and HS analyses are shown in Figure 6 together with the compression modulus tested under pressure 100−200 kPa
measured field test results. Based on the field test results, does not have too much error. However, calculated
at a typical working load of 7,000 kN, measured settlement values for deep piles are often associated with
settlement of a single pile is 3.5 mm. For the same large differences in measured values [21] as deep in-situ
working load, settlement predicted from the HS model of soil stiffness is always higher than that of samples used to
FEM analysis is 8.2 mm and settlement values predicted obtain the laboratory elastic modulus. Therefore, Equation
from LE model and MC model for the same working load [1] proposed by Jian-lin et al. [21] was used to increase
are almost the same, and it is 6.8 mm. the accuracy of the Elastic modulus in the settlement
analysis.
1/ β
Es, z = Es,0.1−0.2 ( z / h0 ) (1)

where z is the depth of soil layer (m), h0 is reference depth

(generally 1 m), Es,0.1−0.2 is laboratory obtained compression
modulus under the pressure of 100 - 200 kPa, and β is the
plasticity of the soil depending on the liquid limit (LL)
and plasticity index (PI) values from the data, from which
the value of β can be obtained using the specifications of
the BS code. All the FEM analyses were repeated with the
corrected values of modulus and the obtained load-
settlement results are shown in Figure 7.
According to the Figure 7, LE and MC models predict
better prediction than the HS model below the working
Figure 6. Comparison of load-settlement plots obtained from FEA and load of 13, 000 kN. However, beyond the load of 13,000
field test result for single pile analysis
kN, LE and MC models fail to capture the real behaviour
There is a remarkable difference between the calculated of settlement and underestimates the settlement, and it
and the measured settlement for the single pile, indicating might be due to the ignorance of soil nonlinearity. On the
the calculation significantly over predicted the pile head other hand, HS model overestimates the settlement during
settlement or underestimated the pile head stiffness. The the entire analysis due to its advanced nonlinearity.
 American Journal of Civil Engineering and Architecture 192

However, a combined analysis using HS-MC models

provides a better prediction when the load is more than
twice the working load. In this case in which a HS-MC
model has been used, a HS zone of twice the diameter of
the pile shaft with the remaining zone by the MC model,
shows adequate agreement with the field test result than
any of the other models considered.

Figure 7. Comparison of load-settlement plots obtained from FEA and

field test result for a single pile using corrected modulus

Figure 9. Comparison of load-settlement behaviour obtained from HS-

LE analysis and measured Field Test result

Figure 8. Settlement contours of soil obtained when it was modeled as

complete (a) LE model, (b) MC model and (c) HS model

Figure 8 shows the settlement contours obtained for

different analyses. It can be seen that linear settlement
contours appear when soil is modelled as complete LE
(Figure 8a) and this confirms the fact that the LE model
disregards the nonlinearity of soil. When soil is modelled Figure 10. Comparison of load-settlement behaviour obtained from HS-
as complete MC model (Figure 8b), a few nonlinear MC analysis and measured Field Test result
contours are obtained in the settlement zone and this
simple nonlinearity is adequate to predict the settlement 4.2. Load settlement Behavior of Pile Group
behaviour at lower working loads up to about 13,000 KN.
Zakia et al. [18] also have concluded that MC model is Predicted load-settlement plots using complete LE and
good enough to predict the load-settlement behaviour at NL analyses and their comparison with the pile group
lower working loads. On the other hand, when soil is load-settlement curve obtained from RATZ analytical
modelled as complete HS model (Figure 8c), a large zone approach are shown in Figure 11. For comparison
of settlement compared to that obtained from the MC purposes, curves are plotted as the mean pile head load
model (Figure 8b) appears, showing clearly the advanced (defined as the load applied on the rigid cap divided by the
nonlinear contours. Modelling the soil completely using number of piles in the group) versus pile head settlement
HS model leads for the underestimation of the soil for pile group.
stiffness everywhere due to its advanced nonlinearity and
thus, complete HS model simply over predicts the
settlement of pile.
Therefore, it is very clear that, by using a combination
of Nonlinear and Linear (HS-LE) models and Nonlinear
and simple Non-Linear (HS-MC) models, it would be
possible to predict realistic load-settlement behaviour of
settlement single pile. Results obtained from HS-LE
analysis and HS-MC analyses (where soil within the
interface thickness was considered as HS model) are
presented in Figure 9 and Figure 10 respectively.
It can be seen that combined analysis using HS-LE
models fail to provide an accurate prediction and the Figure 11. Comparison of load settlement behaviour of pile group from
results appear to be related only to the average prediction. linear and non-linear FE analysis and RATZ analytical approach
193 American Journal of Civil Engineering and Architecture

Based on the results (Figure 11), it is obvious that there Although the HS model over predicted the settlement or
is a significant difference between the calculated and the under estimated the stiffness, it can capture the nonlinear
RATZ prediction curve for the settlement of pile group, behaviour, which is the precise behaviour of soils.
indicating the Finite Element results significantly over The settlement contours obtained from the Finite
predicted the pile head settlement or underestimated the element analysis using LE, MC and HS model are given in
pile head stiffness. The contrasts between the RATZ the Figure 13. As the nonlinearity affects the interaction of
prediction and Finite Element analysis are because of the the pile group, LE and MC models fail to predict the settlement
ignorance of deep in-situ effect in modulus of soils. behaviour of pile group (Figure 13a and Figure 13b). The
Therefore, Equation proposed by Jian-lin [21] was soil nonlinearity and the interaction effects are clearly
adopted to increase the accuracy of the settlement captured in the HS model as shown in Figure 13c and
prediction as similar to that illustrated in the single pile incorporation of both nonlinear effect leads to real
load-settlement analysis. prediction of the settlement behaviour of group piles.
Therefore, FE analyses were repeated using the However, the calculated Finite Element results provides
corrected modulus calculated using the Equation only a limited level of agreement with the analytical
(Equation [1]) proposed by Jian-lin [21]. Figure 12 shows RATZ results, and based on the observations, it can be
the computed settlement results from all the three types of deduced that combined models can better predict the load-
models (LE, MC and HS) and their comparison with the settlement of pile group. In reality, it is now well-
RATZ prediction. recognized that the modulus of a soil mass decreases with
increasing strain level. For a group of piles, it would be
expected that the strain level will increase as the pile shaft
is approached and hence the stiffness of the soil at this
narrow zone close to the pile shaft is smaller than that
between the piles at some distance from the pile shaft.

Figure 12. Comparison of load settlement behaviour of pile group from

linear and non-linear FE analysis with corrected modulus and RATZ
analytical approach

Based on the comparison between RATZ prediction

and Finite element analysis results (Figure 12), it can be
noticed that the prediction of the pile head settlement is Figure 14. Comparison of load settlement behaviour of pile group from
improved. Although, LE and MC models show better HS-LE analysis and RATZ analytical approach
agreement in settlement of pile group with RATZ
prediction at lower working load (< 12,000 kN), they
failed to predict a better behaviour at higher working load
(> 12,000 kN). MC model is a simple nonlinear model in
which all the nonlinear parameters of the soil are not
incorporated, and hence MC model is only adequate to
predict a limited level of settlement behaviour (up to
12,000 kN) accurately.

Figure 15. Comparison of load settlement behaviour of pile group from

HS-LE analysis and RATZ analytical approach

Therefore, to account this stiffness variation, two types

of combined analysis were considered: (i) HS-LE and (ii)
HS-MC (where soil within the interface zone was
considered as HS model). Results of the settlement of pile
group using HS-LE and HS-MC for different interface
thicknesses are given in Figure 14 and Figure 15
respectively. Based on the results from HS-LE finite
element results (Figure 14), considering a nonlinear
Figure 13. Settlement contours of the considered pile group from (a) LE, interface thickness (d) equal to the pile diameter (D) while
(b) MC and (c) HS analysis
keeping the remaining zone as LE, is sufficient to predict
 American Journal of Civil Engineering and Architecture 194

the load-settlement behaviour of pile group. At the [27] and it has been shown that, at closer spacing, the
average typical working load of 7,000 kN, HS-LE (d = D) block failure mechanism occurs (Figure 17a) with
provides the settlement prediction of 8 mm, while the significant plastic zones being developed below the group
settlement predicted from RATZ approach is 8 mm. and full pile-slip is developed only along the outer piles.
Moreover, at the average ultimate load of 20,000 kN, Based on the results (Figure 17), when the spacing goes to
settlement predicted using HS-LE (d = D) Finite element the value of eight times diameter (8D), group settlement
model is 42 mm, whereas the RATZ analytical approach behaviour reaches closer to the settlement behaviour of
predicts the settlement of 43 mm at the same load. that of single pile. As the pile spacing increases, the
Similarly, based on the observations from HS-MC finite failure mechanism gradually changes to the “single pile”
element results (Figure 15), considering an interface mode (Figure 17b), where full pile-soil slip occurs along
thickness equal to the pile diameter (d = D) from the pile whole individual pile.
shaft for non-linear Hardening-Soil (HS) model while
keeping the remaining zone as Mohr-Coulomb (MC)
model, provides a better agreement with RATZ analytical
approach results. At the average typical working load, the
settlement predicted by the HS-MC (d = D) is 8.5 mm,
while the similar settlement from RATZ prediction is 8
mm. Also, at the average ultimate load of 20,000 kN, the
settlement prediction of the above Finite element model is
42 mm, while the settlement from the RATZ approach is
43 mm. Combined model considering the nonlinear
interface thickness of half the pile diameter (d = D/2)
under predicts the settlement, while using a nonlinear
interface thickness equivalent to two times pile diameter
(d = 2D) over predicts the group pile settlement. Reason
for the under estimation is due to the inadequacy of the
zone considered as HS model, where pile-soil interaction
takes place. The over estimation of settlement is due to the
incorporation of larger zone as HS model, which actually
reduces the stiffness of the soil in a larger area
unnecessarily.

Figure 17. Vertical deformation contours of pile group with the spacing
of (a) 2D and (b) 8D

5. Conclusions
Numerical simulation of load-settlement behavior of
axially loaded single pile and pile group was simulated
using PLAXIS 2D and PLAXIS 3D finite element
packages. Based on the findings of this study, the
following conclusions can be drawn;
Figure 16. Load-settlement relationships of pile group with different
spacing
1. For settlement prediction of a single pile, Mohr
Coulomb (MC) model is the best soil model to
Moreover, a study on the effect of pile spacing on predict the realistic behaviour at lower working
group load-settlement behaviour was conducted and the load (<13,000 kN). At the same time, considering a
results obtained from the analysis are given in Figure 16. combined non-linear model, where hardening soil
The predicted settlements for the average working load of (HS) model is used within an interface thickness of
7,000 KN for S = 2D, S = 4D and S = 8D are 13 mm, 8 two times the pile diameter while keeping the
mm and 4 mm respectively. Based on the results, the remaining soil as MC, will better predict the load-
settlement of the pile group decreases when the spacing settlement behaviour at higher working load
increases. When the spacing is increased by four times, (>13,000 kN).
the settlement decreases to approximately one third. 2. To predict the load-settlement behaviour of pile
From Figure 16, it can be stated that the pile group group, HS model should be used within the
stiffness decreases as the pile spacing decreases. The interface zone with the thickness equals to the pile
stiffness of the pile group with 4D spacing is 1.5 times diameter from the pile shaft while keeping the
higher than the stiffness of pile group with the spacing of remaining zone as LE or MC would be sufficient.
2D. Also, the stiffness of pile group with spacing 8D is 3. When numerical simulations are done to predict
more than 2 times of that of pile group with the spacing of settlements realistically, the understanding and the
2D. This result is also supported by Pressley and Poulos selection of the proper material model is very
195 American Journal of Civil Engineering and Architecture

important. Material model selection will not only [11] Randolph M.F, “Load transfer analysis of axially loaded piles”,
depend on the characteristics of soil, but also will RATZ manual version 4.2. University of Western Australia. 2003.
[12] Cao M, Chen L.Z and Chen S.L, “An Innovative Approach to
depend on the loading conditions of the structure. Evaluate the Behaviour Of Vertically Loaded Pile Groups Based
4. Stiffness of the pile group decreases as the spacing on Elastic Theory”, Journal of Lowland Technology International
of the piles in a pile group decreases. At the closer (1): 1-10. 2007.
spacing, block failure mechanism is developed with [13] Seo H and Prezzi M, “Analytical solutions for a vertically loaded
full pile-soil slip being developed only along the pile in multilayered soil”, International Journal of Geomechanics
and Geoengineering 2 (1): 51-60. 2007.
outer face of the outer piles while widespread [14] Muqtadir A, Desai C.S, “Three-dimensional analysis of a pile-
plastic zones form beneath the entire group. As the group foundation”, International Journal of Numerical and
pile spacing increases, the amount of pile-soil slip Analytical Methods in Geomechanics 10 (1): 41-58. 1986.
along the inner piles increases and the region below [15] Reed L.M, “Load Transfer Criteria of Numerical Analysis of
Axially Loaded Piles in Sand”, Technical report of U. S. Army
the piles in which plasticity develops diminishes. Engineer Division. Part 1, K-84-1, 1984.
Also, at a spacing of about eight times diameter [16] Schanz T and Vermeer P.A, “The Hardening-soil model:
(8D), failure of the pile group is governed by the Formulation and verification”, In Proceedings of the Plaxis
failure of a single pile. Symposium. Beyond 2000 in Computational Geotechnics.
Brinkgreve, R. B. J. (ed.), 281-290. 1999. Rotterdam: Balkema.
[17] Fu­chun XUE, Jian­lin MA and Li­ping YAN, “Three-
dimensional FEM analysis of bridge pile group settlement in soft
References soils”, Geotechnical Special Publication ASCE 220: 135-143. 2011.
[18] Zakia K, Abdelmadjid B and Nazihe T, “Modeling the Behaviour
[1] Sonia, A and Desai C.S., “Load Deformation Response of Axially of Axially loaded and laterally loaded pile with a contact model”,
Loaded Piles”, Journal of Geotechnical Engineering 113(12): Electronic Journal of Geotechnical Engineering 16: 1239-1257.
1483-1500. 1987. 2011.
[2] Horikoshi K, Matsumoto T, Hashizume Y and Watanabe T, [19] Alnuiam A.M.H, Naggar E.I and Naggar H.E.I, “Performance of
“Performance of Piled Raft Foundations Subjected to Dynamic Piled-Raft System under Axial Load”, 18th International
Loading”, International Journal of Physical Modelling in Conference on Soil Mechanics and Geotechnical Engineering:
Geotechnics, 3 (2): 51-62. 2003. 2663-2666. 2013.
[3] McCabe B.A and Lehane B.M, “Behaviour of axially loaded pile [20] Ju J, “Prediction of the Settlement for the Vertically Loaded Pile
groups Driven in clayey silt”, Journal of Geotechnical and Group Using 3D Finite Element Analyses”, Journal of Marine
Environmental Engineering: 401-410. 2006. Georesources & Geotechnology 33 (3): 264-271. 2013.
[4] Poulos H.G, “Analysis of the settlement of pile groups”, [21] Jian-lin M.A, Hussein Y.A and SUChun-hui S.C, “Settlement
Geotechnique 4: 449-471. 1968. Prediction and Behaviour of Pile Foundations in deep clayey soil
[5] Butterfield R and Douglas R.A, “Flexibility coefficients for the deposits”, Central South University Press and Springer-Verlag
design of pile and pile groups and pile groups”, CIRIA technical Berlin Heidelberg 21: 1554-1564. 2014.
note no. 108, London, 1981. [22] Shao-Hung C and Shu-Rong Y, “Numerical Analysis of Small-
[6] Chow Y.K, “Analysis of vertically loaded pile groups”, Scale Model Pile in Unsaturated Clayey Soil”, International
International Journal of Numerical and Analytical Methods in Journal of Civil Engineering: 1-10. 2016.
Geomechanics 10 (1): 59-72. 1986. [23] Mohammad H, Moslem B and Ali F, “Numerical and
[7] Shen W.Y, Chow Y.K, and Yong K.Y, “A Variational Approach Experimental Study of the Optimal Location of Concrete piles in a
for Vertical Deformation Analysis of Pile Group”, International saturated Sandy Slope”, International Journal of Civil
Journal for Numerical and Analytical Methods in Geomechanics Engineering: 1-9. 2017.
21: 741-752. 1997. [24] Orihara K and Khoo K.S, “Engineering properties of Old
[8] Randolph M.F and Wroth C.P, “Analysis and deformation of Alluvium in Singapore and its parameters for bored pile and
vertically loaded piles”, Journal of Geotechnical Engineering excavation design”. Thirteenth Southeast Asian Geotechnical
104(12): 1465-1487. 1978. Conference: 545-550. 1998.
[9] Lee C.Y and Poulos H.G, “Axial response analysis of piles in [25] Craig R.F, Craig’s Soil Mechanics, Seventh edition. 2004.
vertically and horizontally non-homogeneous soils”, Journal of [26] PLAXIS. Material Models Manual. 2015.
Computers and Geotechnics 9: 133-148. 1990. [27] Pressley J.S and Poulos H.G, “Finite element analysis of
[10] Castelli F and Maugeri M, “Simplified non-linear analysis for mechanisms of pile group behavior”, International Journal for
settlement prediction of pile groups”, Journal of Geotechnical and Numerical and Analytical methods in Geomechanics 10: 213-221.
Geo-Environmental Engineering 128 (1): 76-84. 2002. 1986.