See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233559671

Parametric finite element analyses of geocell-supported embankments

Article  in  Canadian Geotechnical Journal · August 2007

DOI: 10.1139/T07-039

CITATIONS READS
41 669

2 authors:

Madhavi Latha Gali Rajagopal Karpurapu

Indian Institute of Science Indian Institute of Technology Madras
120 PUBLICATIONS   1,591 CITATIONS    111 PUBLICATIONS   2,401 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ANALYSIS OF THE BEHAVIOUR OF ORDINARY AND GEOSYNTHETIC ENCASED GRANULAR COLUMNS SUBJECTED TO SHEAR LOADING View project

Laboratory determination of ch View project

All content following this page was uploaded by Madhavi Latha Gali on 16 July 2014.

The user has requested enhancement of the downloaded file.

 917

Parametric finite element analyses of geocell-

supported embankments
G. Madhavi Latha and K. Rajagopal

Abstract: This paper presents the results from parametric finite element analyses of geocell-supported embankments con-
structed on weak foundation soils. A composite model is used to numerically simulate the improvement in the strength
and stiffness of the soil as a result of geocell confinement. The shear strength of the geocell-encased soil is obtained as a
function of the additional confining pressure due to the geocell encasement considering it as a thin cylinder subjected to
internal pressure. The stiffness of the geocell-encased soil is obtained from the stiffness of the unreinforced soil and the
tensile modulus of the geocell material using an empirical equation. The validity of the model is verified by simulating
the laboratory experiments on model geocell-supported embankments. Parametric finite element analyses of the geocell-
supported embankments are carried out by varying the dimensions of the geocell layer, the tensile strength of the material
used for fabricating the geocell layer, the properties of the infill soil, and the depth of the foundation layer. Some impor-
tant guidelines for selecting the geocell reinforcement to support embankments on weak foundation soils are established
through these numerical studies.
Key words: geocell layer, geogrids, finite element analysis, parametric studies, reinforcement, embankment.
Résumé : Cet article présente les résultats d’analyses paramétriques en éléments finis de remblais reposant sur des géocel-
lules construites sur des fondations de sols mous. Un modèle composite est utilisé pour simuler numériquement l’amélio-
ration de la résistance et de la rigidité du sol par le confinement des géocellules. La résistance au cisaillement du sol
contenu dans les géocellules est obtenue en fonction de la pression additionnelle de confinement due à l’encaissement des
géocellules que l’on considère comme des cylindres minces soumis à une pression interne. On obtient la rigidité du sol en-
caissé dans des géocellules à partir de la rigidité du sol non armé et du module en traction du matériau des géocellules
donné par un équation empirique. La validité du modèle est vérifiée en simulant les expériences en laboratoire sur des mo-
dèles de remblais. On réalise des analyses paramétriques en éléments finis de remblais supportés par des géocellules en va-
riant les dimensions de la couche de géocellules, la résistance en traction du matériau utilisé pour fabriquer la couche de
géocellules, les propriétés du sol de remplissage et la profondeur de la couche de fondation. On a interprété au moyen
d’études numériques des directives importantes pour choisir l’armature de géocellules pour supporter les remblais sur des
sols de fondation mous.
Mots-clés : couche de géocellules, géogrille, analyse en éléments finis, études paramétriques, armature, remblai.
[Traduit par la Rédaction]

Introduction reinforcements provide three-dimensional confinement to the

soil. The generic term geocell refers to three-dimensional,
Earthen embankments are important structures because polymeric, honeycomb-like cellular material. A structure of
they form part of the major facilities for transportation, these cells interconnected by joints to form a cellular net-
water storage, and flood control. The embankments con- work and subsequently filled with soil could be used for the
structed on soft clays are prone to excessive settlements and confinement of soil. Horizontal geosynthetic layers contrib-
shear failure due to the high compressibility and low shear ute to the improvement by friction and interlocking between
strength of the foundation soil. The technique of reinforcing the soil and reinforcement, whereas geocells improve the
foundation soils using geosynthetics has emerged as one of strength by friction, interlocking, and all-round confinement
the best solutions over the past three decades for the con- of the soil. A layer of geocell provided at the base of an em-
struction of embankments over soft soils. The geocell rein- bankment improves the bearing capacity of the foundation
forcement is one of the most recent advancements and it is soil and reduces the vertical and lateral deformations of the
particularly suited for construction on soft clays. The geocell embankment. Several researchers have demonstrated the ad-
vantages of geocell reinforcement for the construction of
Received 25 August 2006. Accepted 13 May 2007. Published on embankments on soft soils, for example, Bathurst and Jarrett
the NRC Research Press Web site at cgj.nrc.ca on 11 September
2007.
(1988); Jenner et al. (1988); Bush et al. (1990); Dean and
Lothian (1990); Cowland and Wong (1993); Krishnaswamy
G. Madhavi Latha.1 Department of Civil Engineering, Indian et al. (2000); and Madhavi Latha et al. (2006). Several stud-
Institute of Science, Bangalore, 560 012 India. ies are also available on the application of geocells for sup-
K. Rajagopal. Professor, Department of Civil Engineering, porting foundations, for example, Dash et al. (2001a); Dash
Indian Institute of Technology, Madras, 600 036, India. et al. (2001b); Madhavi Latha et al. (2001); Dash et al.
1Corresponding author (e-mail: [email protected]). (2003); and Dash et al. (2004).

Can. Geotech. J. 44: 917–927 (2007) doi:10.1139/T07-039 # 2007 NRC Canada

918 Can. Geotech. J. Vol. 44, 2007

The response of geocell-supported embankments depends forced and geocell-encased soils, as demonstrated by Bath-
on a number of parameters. Examining the influence of all urst and Karpurapu (1993) and Rajagopal et al. (1999).
of these parameters through laboratory experiments would
consume a lot of time and effort. The finite element techni- Stiffness of the geocell layer
que can be used as a powerful and cost-effective tool for Based on triaxial compression tests on geocell-encased
studying the influence of various parameters through numer- sand, Madhavi Latha (2000) proposed the following empiri-
ical simulations. However, the finite element models should cal equation to express the Young’s modulus of the geocell-
be validated against experimental and field measurements reinforced sand (Eg) in terms of the secant tensile modulus
before applying the results to an actual design. This paper of the geocell material (M) and the Young’s modulus param-
presents the details of parametric finite element analyses eter of the unreinforced sand (Ku):
carried out on geocell-supported embankments to study the
effects of various parameters, such as the dimensions of the ½3
Eg ¼ 4ð
3 Þ0:7 ðKu þ 200M 0:16 Þ
geocell layer, the tensile strength properties of the geocell
material, the properties of the infill soil, and the depth of where Ku is the dimensionless modulus parameter of the un-
soft foundation soil. The validity of the numerical model is reinforced sand, M is the secant modulus of the geocell ma-
initially verified through comparisons with experimental re- terial in kN/m, and
3 is the confining pressure in kPa. The
sults before it is extended to parametric studies. modulus parameter in the above equation corresponds to the
modulus number in the hyperbolic model proposed by Dun-
can and Chang (1970). This value of M should correspond
Composite model for geocell-encased soil to the average strain of 2.5% in the load–elongation re-
A composite model proposed by Madhavi Latha (2000) sponse of the geocell material. The applicability of the
and Rajagopal et al. (2001) for geocell-encased sand is used above procedure is verified through finite element simula-
in finite element simulations presented in this paper. This tions of model embankments tested in the laboratory as de-
model was developed based on triaxial tests on sand encased scribed in subsequent sections.
in single and multiple geocells made of different geosyn-
thetics described by Rajagopal et al. (1999). The model was Numerical model
validated by Rajagopal et al. (2001) against experiments on
All of the finite element analyses reported in this paper
geocell-supported model embankments constructed over a
are performed using the finite element program GEOFEM
soft clay bed. The applicability of this model for various
developed at the Royal Military College of Canada, Kingston
types of geocells was later demonstrated by Madhavi Latha
(Rajagopal and Bathurst 1993). The finite element analyses
and Murthy (2007). A brief description of the model is pre-
of geocell-supported embankments were performed using
sented below for completeness.
plane strain models. The soils were modelled using a non-
linear elastic-plastic constitutive model with Mohr–Coulomb
Shear strength of the geocell layer yield criterion and the nonassociated flow rule. All of the
In these analyses, geocell layers are treated as equivalent analyses were performed using meshes made up of four 3-
foundation soil layers with cohesive strength greater than the node triangles within each rectangle. This mesh arrange-
encased soil and angle of internal friction the same as the ment was reported to be accurate in predicting limit loads
encased soil. The induced cohesion in the soil, cr, is related in plane strain problems (Nagtegaal et al. 1974). Rowe and
to the increase in the confining pressure on the soil due to Soderman (1987) have reported good success in using this
the geocell reinforcement through the following equation: type of mesh arrangement in simulating many reinforced soil


3 pffiffiffiffiffiffi embankment problems. The geocell layers were modelled
½1
cr ¼ Kp as an equivalent composite layer, just like any soil layer.
2
where Kp is the coefficient of passive earth pressure and Model test on geocell-supported


3 is the additional confining pressure due to the mem-
brane stresses. This confining pressure can be calculated
embankment
using the membrane correction theory proposed by Henkel Krishnaswamy et al. (2000) described different model
and Gilbert (1952), which is based on the hoop tension the- tests conducted on geocell-supported embankments in detail.
ory, treating the geocell-encased soil as a thin cylinder sub- In this paper, one of these model embankments is simulated
jected to internal pressure. in finite element analysis to validate the numerical model by
pffiffiffiffiffiffiffiffiffiffiffiffiffi! comparing the performance of the simulated model with the
2M 1  1  "a experimental observations. The model embankment chosen
½2


3 ¼ for finite element analysis is shown in Fig. 1, where V1,
Do 1  "a
V3, and V4 are dial gauges for measuring vertical deforma-
where "a is the axial strain at failure, Do is the initial dia- tions, and H1, H2, and H3 are dial gauges for measuring lat-
meter of the geocell, and M is the secant modulus of the eral deformations.
geocell material at an axial strain of "a. The cohesive The geocell layer in the embankment was made of a poly-
strength (cr) obtained from eq. [1] should be added to the propylene biaxial geogrid (referred to as BX in this paper)
cohesive strength of the unreinforced soil (c) to obtain the having an ultimate tensile strength of 20 kN/m. The load–
cohesive strength of the geocell-encased soil (cg). The angle elongation response for this geogrid in wide-width tension
of internal friction is assumed to be the same for the unrein- tests is presented in Fig. 2. The secant modulus (M) of the
# 2007 NRC Canada
Madhavi Latha and Rajagopal 919

Fig. 1. Experimental embankment chosen for verification of composite model.

Fig. 2. Load–elongation response for the geogrid used in model as 37 kPa using eq. [2]. Hence, the apparent cohesion due
tests. to geocell confinement (cr) is calculated to be 32 kPa. The
geocells are filled with clayey sand having c = 10 kPa and
 = 308 to a unit weight of 19 kN/m3. Therefore, the shear
strength properties of the geocell layer are found to be cg =
42 kPa and g = 308. The Young’s modulus parameter for
the unreinforced clayey sand (Ku) was 382 from the triaxial
tests. Hence the Young’s modulus of the geocell layer was
calculated to be 42 000 kPa from eq. [3].
The experimental model embankment is simulated in the
finite element analysis using triangular elements. The uni-
form settlements applied on the surface of the embankment
through a relatively rigid loading system in the laboratory
tests were simulated in the finite element analyses by apply-
ing equal vertical displacement increments at all of the nodes
on the crest of the embankment. The surcharge pressure was
obtained from the reaction loads generated at these nodes.
The displacements were applied in small increments of
0.10–0.25 mm with several iterations at each load step until
the norms of incremental displacements and out-of-balance
forces were less than 0.5%. The constitutive behaviour of
the soft foundation soil and the soil in the embankment was
simulated using a Mohr–Coulomb elastic – perfectly plastic
yield surface with a nonassociated flow rule. In this model,
a yield surface is defined in terms of the shear strength pa-
geogrid at 2.5% strain is calculated as 160 kN/m from the rameters c and . A plastic potential function is defined in
plot. The geocell pockets are triangular in shape with 0.4 m terms of a dilation angle ( ). The dilation angle in this
base and 0.2 m height in plan. The equivalent diameter of model controls the volume expansion of the soil. The stress
the triangular geocell pockets is calculated as 0.2256 m. For state is corrected back to the yield surface along the flow di-
these values, the all-round confining pressure is calculated rection defined by the plastic potential function. The geocell
# 2007 NRC Canada
920 Can. Geotech. J. Vol. 44, 2007

Fig. 3. Comparison of lateral deformations obtained from finite Fig. 5. Comparison of surface heave obtained from finite element
element analysis with the experimental measurements. analysis with the experimental measurements.

Fig. 4. Comparison of crest settlement from finite element analysis Fig. 6. Profile of the embankment chosen for the parametric finite
with the experimental measurements. element analysis.

be observed from the figure, the numerical model is able to

capture the actual response of the geocell-supported embank-
ment very well at low to moderate surcharge pressures on
the crest. At higher pressures, the simulated models de-
formed more than the experimental models. However, within
the permissible deformation levels of the embankments, the
comparisons are very good.
Comparisons of the average vertical settlement of the
crest and the surface heave measured by the dial gauge at
layer was simulated in the finite element analyses by replac- V4 (shown in Fig. 1) with the finite element results are pre-
ing it with a composite soil layer of equal thickness with sented in Figs. 4 and 5, respectively. From the measured
equivalent shear strength and stiffness properties calculated pressure–settlement behaviour, it can be seen that the stiff-
as above. The Poisson’s ratio () of this soil was assumed to ness of the foundation clay soil increased during the test be-
be 0.30, which is a reasonable value for this type of soil. The cause of the limited size of the model test tank, leading to
values of E and  for the soft clay layer were 200 kPa and strain-hardening behaviour. On the one hand, the clay soil
0.45, respectively. The comparison between the experimental was assumed to be perfectly plastic in the numerical model.
and measured lateral deformations at three different loca- Hence, the stiffness of the clay soil is underestimated, which
tions along the slope of the embankment is shown in Fig. 3, leads to larger deformations. On the other hand, the meas-
where dp represents the pocket size of the geocells. As can ured lateral deformations and heave have shown the usual
# 2007 NRC Canada
Madhavi Latha and Rajagopal 921

Fig. 7. Typical finite element mesh for the embankment chosen for the parametric analysis.

Table 1. Properties of the soils used in the parametric finite element analyses.

Angle of
Cohesive internal Young’s Unit
strength, friction, modulus, weight, Poisson’s
Soil layer c (kPa)  (8) E (kN/m2)  (kN/m3) ratio, 
Foundation soil 10 0 3 000 17 0.45
Embankment soil 10 45 50 000 19 0.30

tendency of increasing deformations at higher pressures as to be 90 kPa, which is the average vertical stress at the mid-
the embankment is free to move in the lateral direction, es- height of the geocell layer multiplied by a lateral pressure
tablishing the validity of the composite model for simulating coefficient of 0.5. The Young’s modulus was then estimated
the geocell-supported embankments with reasonable accu- by substituting the respective secant modulus value in the
racy. Hence, the composite model can be used for the para- equation.
metric analyses of geocell-supported embankments to Some of the parameters considered in the analyses were
understand the influence of each of the parameters on the the aspect ratio of the geocell layer, the stiffness of the geo-
overall response. cell layer, the shear strength of the soil fill in the geocells,
and the depth of the foundation soil (df). Four different geo-
cell materials with different secant moduli (M), viz. 100,
Parametric finite element analyses
200, 500, and 1000 kN/m, were used in the analyses. The
The cross-section of the embankment considered for the aspect ratio (h/D) of the cells was varied from 0.25 to 1.50.
parametric finite element analysis is shown in Fig. 6. Typi- The analyses were carried out with two different fill materi-
cal finite element mesh used for the analysis of the embank- als inside the cells. The shear strength properties of the sand
ment with a 1 m in height geocell layer at the base is shown used to fill the cells were c = 0,  = 458; and for the clay
in Fig. 7. The embankments in all of the parametric analyses fill they were c = 10 kPa and  = 0. The depth of the foun-
were constructed in 1 m thick layers in 10 load steps per dation soil (df) was varied from 5 m to 20 m in increments
layer and 50 iterations at each load step. A total of 100 of 5 m. The equivalent cohesive strength of the geocell layer
load steps were used to construct the full height of the em- (cg) was computed using eq. [1]. The different sets of pa-
bankment. The analysis at each load step was iterated until rameters used in these analyses are listed in Table 2.
the norms of the out-of-balance force and incremental dis-
placements were less than 0.5% or for 50 iterations, which- Effect of aspect ratio
ever happened earlier. The embankment and the foundation In the finite element analyses, the aspect ratio of the cells
soils were treated as elastic – perfectly plastic materials with in a geocell layer was varied by changing the diameter of
Mohr–Coulomb yield criterion. The properties of different the cells while keeping the height of the geocell layer con-
soils used in the analysis are given in Table 1. The geocell stant at 1 m. Some of these analyses were also performed
layer was modelled using the same criterion. The Young’s keeping the diameter constant and varying the height of the
modulus of the geocell layer was estimated using eq. [3]. cells (1 m and 2 m) to achieve different aspect ratios. The
Two different soils, sand, and clay were used to fill the geo- change in diameter of the geocell pocket was simulated in
cells in different tests. The Young’s modulus number Ku is the analyses by using a different cohesive strength as listed
assumed to be 550 for the sand fill and 150 for the clay fill. in Table 2. The secant modulus of the geocell material in
The confining pressure used in the above equation was taken these analyses was taken to be 500 kN/m.
# 2007 NRC Canada
922 Can. Geotech. J. Vol. 44, 2007

Table 2. Parameters varied in finite element analyses.

Cohesive Young’s
Secant modulus Initial strength of modulus of Depth of
of geocell diameter Aspect Type geocell geocell foundation
Simulation material, of cells, ratio of infill layer, layer, soil,
No. M (kN/m) Do (m) of cells soil cg (kPa) Eg (kN/m2) df (m)
1 100 1 1 Sand 6.4 89 038 15
2 200 1 1 Sand 13 93 618 15
3 500 1 1 Sand 32 100 506 15
4 1000 1 1 Sand 64 106 431 15
5 500 4 0.25 Sand 8 100 506 15
6 500 4a 0.5 Sand 8 100 506 15
7 500 2 0.50 Sand 16 100 506 15
8 500 0.67 1.5 Sand 48 100 506 15
9 500 1 1 Clay 23 53 504 15
10 500 1 1 Sand 32 100 506 5
11 500 1 1 Sand 32 100 506 10
12 500 1 1 Sand 32 100 506 20
a
h = 2 m.

Fig. 8. Influence of the aspect ratio of geocells on the settlement Fig. 9. Influence of the aspect ratio of geocells on the net height of
response of the full-scale embankment. the full-scale embankment.

The progressive settlements (at point ‘‘P’’ in Fig. 6) for supported on geocells with higher aspect ratios withstood
different aspect ratios of cells at different embankment loading up to a total height of 10 m (Fig. 9). The perform-
heights are shown in Fig. 8. The same data is presented in ance of the geocell-supported embankments was improved
Fig. 9 as the net height versus the total height of the em- with higher aspect ratios (this was also observed in the labo-
bankment. The net height here is defined as the total height ratory tests). However, this improvement is not significant
minus the surface settlement at that stage. The failure can be beyond an aspect ratio of 1.0. Hence, an aspect ratio of 1.0
defined as the total height at which the net height starts de- (unity) may be taken as the optimum value for best utiliza-
creasing, that is, the settlement is higher than the thickness tion of the geocell layer.
of the fill placed in the embankment (Rowe and Soderman The lateral deformations at the toe of the embankment are
1987). The unreinforced embankment is considered to have shown in Fig. 10. Once again, the lateral deformations have
failed around 6.5 m, while the geocell-supported embank- decreased with increasing aspect ratio and reach the mini-
ments did not even fail at a height of 10 m. However, the mum at an aspect ratio of unity.
embankment supported on geocells with an aspect ratio of The vertical deformations along the base of the embank-
0.25 seems to reach the collapse height at 10 m, as the net ment at the end of construction for different aspect ratios of
height became constant at this fill height. The embankments the cells are shown in Fig. 11. The geocell reinforcement
# 2007 NRC Canada
Madhavi Latha and Rajagopal 923

Fig. 10. Influence of the aspect ratio of geocells on the lateral de- Fig. 12. Surface settlement profiles for full-scale embankment with
formations of the full-scale embankment. the same aspect ratio of geocells (dimensions of cells different). h,
height; D, diameter.

Fig. 11. Surface settlement profiles for different aspect ratios of

geocells for full-scale embankments.

has reduced both the total and differential settlements. Once

again, the effect of aspect ratio is very clear on the perform-
ance improvement. The results from these parametric analy-
ses point towards a conclusive result that the best aspect
ratio (h/D) is unity for the geocells. In cases where the cells
are not circular in shape, the equivalent diameter should be
calculated for the cells, and this value should be used in cal-
culating the aspect ratio. All further analyses were per-
formed with an optimum aspect ratio of unity.
The results from the set of analysis in which the aspect
ratio of the cells was kept constant (varying only the height
and diameter of the cells) are shown in Fig. 12. The aspect
ratio of the cells was 0.5 in both the cases studied, whereas
the height and diameter of the cells are different (Table 2).
The performance of the embankment for both these cases is
similar, as observed from the figures, which leads to the
conclusion that the aspect ratio of the geocells is the primary
factor governing the performance of the geocell layer.

Effect of stiffness of the geocell material

The secant modulus of the material used for making the
geocell was varied for some analyses. A comparison of the
surface settlements obtained with four different secant mod-
uli of geocell material is shown in Fig. 13. In general, the
performance improved slightly with the increase in secant
# 2007 NRC Canada
924 Can. Geotech. J. Vol. 44, 2007

Fig. 13. Surface settlement profiles with different secant moduli of Fig. 15. Influence of the secant modulus of the geocell material on
geocell materials for full-scale embankments. the lateral deformations of the full-scale embankment.

modulus. However, this improvement is not proportional to

the increase in the secant modulus.
The effect of the secant modulus of the geocell material
on the settlements and lateral deformations is shown in
Figs. 14 and 15, respectively. The settlements correspond to
point ‘‘P’’ (from Fig. 6) and the lateral deformations corre-
spond to the toe of the embankment. From these figures it
is also evident that although the deformations are reduced
with increase in the secant modulus of the geocell material,
the magnitude of this reduction is not proportional to the in-
crease in the secant modulus. The net height versus the total
Fig. 14. Influence of the secant modulus of the geocell material on height of the embankment for the different cases is shown in
the settlements of the full-scale embankment. Fig. 16. It can be observed from Fig. 16 that as the unrein-
forced embankment failed after reaching a height of 6.5 m,
the embankments supported on geocells even of lower stiff-
ness (M = 100 kN/m) were safe up to a height of 9 m. An
increase in the secant modulus of the geocell material above
200 kN/m did not result in a significant increase in the net
height of the embankment. Hence, it can be concluded that
the geocells do not need to be made of very stiff geogrids.

Effect of type of infill soil

All of the previous analyses were performed with granular
soil fill in the geocell pockets. One analysis was performed
with cohesive soil infill to examine the influence of the infill
soil on the overall performance. The comparison of average
settlements and lateral deformations of the embankments ob-
tained with granular and cohesive soil fills in the geocells is
shown in Figs. 17 and 18, respectively. At any given sur-
charge pressure, the lateral deformations in the embankment
on the geocells with granular soil fill are 20% less than the
embankment on the geocells with cohesive soil fill, while
the vertical deformations are not influenced very much by
the type of fill. Because of the higher shear strength proper-
ties of the granular soil filled geocells, higher restraining
forces in the lateral direction may have been mobilized as
# 2007 NRC Canada
Madhavi Latha and Rajagopal 925

Fig. 16. Influence of the secant modulus of the geocell material on Fig. 18. Influence of shear strength of the infill soil on the lateral
the net height of the full-scale embankment. deformations of the full-scale embankment.

Fig. 17. Influence of shear strength of the infill soil on the settle- Fig. 19. Influence of the depth of the foundation soil on the settle-
ment response of the full-scale embankment. ments of the full-scale geocell-supported embankment.

compared to the cohesive soil fill case. This could have re- Effect of depth of foundation soil
sulted in slightly better performance with the granular soil The depth of the soft foundation soil (df) overlying the rigid
fill. These results once again confirm that even with clay foundation layer was varied from 5 m to 20 m in increments of
fill inside the geocells, significant improvement in the per- 5 m. This set of analyses was carried out on embankments
formance could be achieved. Hence, if there is an absence supported on geocell layers having a height of 1 m, a geocell
of good granular soils for infill in the vicinity, local soils material secant modulus of 500 kN/m, and an aspect ratio of
can also be used to fill the geocells and still achieve consid- unity. The effect of the depth of the foundation soil on the ver-
erable improvement in performance. However, it should be tical and lateral deformations of the embankment is presented
noted that compaction of clay inside the geocells may pose in Figs. 19 and 20, respectively. The variation of the net
practical difficulties in the field. height of the embankment with the total height for different
# 2007 NRC Canada
926 Can. Geotech. J. Vol. 44, 2007

Fig. 20. Influence of the depth of the foundation soil on the lateral be plastic failure within the clay layer, leading to increased
deformations of the full-scale geocell-supported embankment. deformations in the embankment. Earlier studies by Rowe
and Soderman (1987) also demonstrated a similar mechanism.
When the lateral deformations were compared for the em-
bankments with different depths of soft foundation layer, it
was observed that the depth of the foundation affected the
performance of the embankments. The embankments con-
structed on deeper foundation layers deformed more than
the embankments on thinner foundation layers, but this in-
fluence of increased depth on lateral deformations was
found to cease at a depth of 20 m, which is half the base
width of the embankment. Therefore, when studying the de-
formation behaviour of embankments constructed on deeper
soft foundation layers through model studies, the depth of
the foundation layer (df) should be at least equal to half the
base width of the embankment in the model.

Conclusions
The following specific conclusions can be drawn from
this study.
(1) The performance of geocell-supported embankments im-
proved with higher aspect ratios, the improvement being
insignificant beyond an aspect ratio of 1.0. Hence, an as-
pect ratio of 1.0 is recommended for the geocells to obtain
Fig. 21. Influence of the depth of the foundation soil on the net the best performance in terms of reduction in deformations
height of the full-scale geocell-supported embankment. and increase in surcharge-carrying capacity.
(2) Granular soils are preferred for fill inside the geocells
because the confinement effect is more pronounced in
these soils, leading to a greater reduction in overall de-
formations. However, in the absence of granular fill, the
geocell pockets can be filled with locally available mate-
rials without compromising the performance, as evi-
denced from the results of this study.
(3) The performance of the geocells made with geogrids
having secant modulus more than 200 kN/m was not ob-
served to increase significantly. Hence, geogrids having
moderate secant moduli (in the range of 100–200 kN/m)
may be used to form the geocells.
(4) The performance of the geocell-supported embankment
is adversely affected with increase in the depth of the
foundation soil. The reason for the drop in overall perfor-
mance could be attributed to the increased deformations
because of the plastic failure within the soft foundation
with the increase in the depth of the foundation layer.

References
Bathurst, R.J., and Jarrett, P.M. 1988. Large scale model tests of
geocomposite mattresses over peat subgrades. In Transportation
Research Record No. 1188. Transportation Research Board, Wa-
shington, D.C. pp. 28–36.
cases is presented in Fig. 21. It can be seen from these figures
Bathurst, R.J., and Karpurapu, R.G. 1993. Large-scale triaxial com-
that as the depth of the soft foundation layer (df) is increased,
pression testing of geocell-reinforced granular soils. Geotechni-
the settlements were increased. When the depth of the soft cal Testing Journal, 16: 296–303.
foundation soil was 20 m, even with geocell reinforcement, Bush, D.I., Jenner, C.G., and Bassett, R.H. 1990. The design and
the net height of the embankment was 7.5 m at a total height construction of geocell foundation mattress supporting embank-
of 10 m. With the same amount of reinforcement, when the ments over soft ground. Geotextiles and Geomembranes, 9: 83–
depth of the foundation soil was 5 m, the net height and total 98. doi:10.1016/0266-1144(90)90006-X.
height were equal until the full height was reached. When the Cowland, J.W., and Wong, S.C.K. 1993. Performance of a road
foundation layer is thin, the soft soil exhibits elastic behav- embankment on soft clay supported on a geocell mattress foun-
iour, and hence the deformations in the embankment are dation. Geotextiles and Geomembranes, 12: 687–705. doi:10.
small. As the depth of the foundation soil increases, there will 1016/0266-1144(93)90046-Q.
# 2007 NRC Canada
Madhavi Latha and Rajagopal 927

Dash, S.K., Krishnaswamy, N.R., and Rajagopal, K. 2001a. Bear- Madhavi Latha, G. 2000. Investigations on the behaviour of geocell
ing capacity of strip footings supported on geocell-reinforced supported embankments. Ph.D. thesis, Department of Civil Engi-
sand. Geotextiles and Geomembranes, 19: 235–256. doi:10. neering, Indian Institute of Technology Madras, Chennai.
1016/S0266-1144(01)00006-1. Madhavi Latha, G., and Murthy, V.S. 2007. Effects of reinforce-
Dash, S.K., Rajagopal, K., and Krishnaswamy, N.R. 2001b. Strip ment form on the behaviour of geosynthetic reinforced sand.
footing on geocell reinforced sand beds with additional planar Geotextiles and Geomembranes, 25: 23–32. doi:10.1016/j.
reinforcements. Geotextiles and Geomembranes, 19: 529–538. geotexmem.2006.09.002.
doi:10.1016/S0266-1144(01)00022-X. Madhavi Latha, G., Dash, S.K., Rajagopal, K., and Krishnaswamy,
Dash, S.K., Sireesh, S., and Sitharam, T.G. 2003. Bearing capacity N.R. 2001. Finite element analysis of strip footing on geocell re-
of a circular footing on geocell reinforced sand underlain by soft inforced sand beds. Indian Geotechnical Journal, 31: 454–478.
clay. Geotextiles and Geomembranes, 21: 197–219. Madhavi Latha, G., Rajagopal, K., and Krishnaswamy, N.R. 2006.
Dash, S.K., Rajagopal, K., and Krishnaswamy, N.R. 2004. Perfor- Experimental and theoretical investigations on geocell-supported
mance of different geosynthetic reinforcement materials in sand embankments. International Journal of Geomechanics, 6: 30–35.
foundations. Geosynthetics International, 11: 35–42. doi:10. doi:10.1061/(ASCE)1532-3641(2006)6:1(30).
1680/gein.11.1.35.36317. Nagtegaal, J.C., Parks, D.M., and Rice, J.R. 1974. On numerically
Dean, R., and Lothian, E. 1990. Embankment construction pro- accurate finite element solutions in the fully plastic range. Com-
blems over deep variable soft deposits using a geocell mattress. puter Methods in Applied Mechanics and Engineering, 4: 153–
In Performance of reinforced soil structures. Edited by A. 177. doi:10.1016/0045-7825(74)90032-2.
McGown, K.C. Yeo, and K.Z. Andrawes. British Geotechnical Rajagopal, K., and Bathurst, R.J. 1993. Users manual for geotech-
Society, Thomas Telford Ltd., London. pp. 443–447. nical finite element modelling GEOFEM, Department of Civil
Duncan, J.M., and Chang, C.Y. 1970. Non-linear analysis of stress Engineering, Royal Military College of Canada, Kingston, Ont.
and strain in soils. Journal of the Soil Mechanics and Founda- Rajagopal, K., Krishnaswamy, N.R., and Madhavi Latha, G. 1999.
tions Division, 96: 1629–1653. Behavior of sand confined in single and multiple geocells. Geo-
Henkel, D.J., and Gilbert, G.C. 1952. The effect of rubber mem- textiles and Geomembranes, 17: 171–184. doi:10.1016/S0266-
branes on the measured triaxial compression strength of clay 1144(98)00034-X.
samples. Géotechnique, 3: 20–29. Rajagopal, K., Krishnaswamy, N.R., and Madhavi Latha, G. 2001.
Jenner, C.G., Bush, D.I., and Bassett, R.H. 1988. The use of slip line Finite element analysis of embankments supported on geocell
fields to assess the improvement in bearing capacity of soft layer using composite model. In Proceedings of 10th Interna-
ground given by a cellular foundation mattress installed at the tional Conference on Computer Methods and Advances in Geo-
base of an embankment. In Proceedings of the International mechanics, Tuscon, Arizona, January 2001. Edited by C.S.
Geotechnical Symposium on Theory and Practice of Earth Desai, T. Kundu, S. Harpalani, D. Contractor, and J. Kemeny.
Reinforcement, Fukuoka, Kyushu, Japan, October 1988. Edited Taylor & Francis, pp. 1251–1254.
by T.?Yamanouchi, N. Miura, and H. Ochiai. A.A. Balkema, Rowe, R.K., and Soderman, K.L. 1987. Stabilization of very soft
Rotterdam. pp. 209–214. soils using high strength geosynthetics: the role of finite element
Krishnaswamy, N.R., Rajagopal, K., and Madhavi Latha, G. 2000. analyses. Geotextiles and Geomembranes, 6: 53–80. doi:10.
Model studies on geocell supported embankments constructed 1016/0266-1144(87)90057-4.
over a soft clay foundation. Geotechnical Testing Journal,
ASTM, 23: 45–54.

# 2007 NRC Canada

View publication stats