KSCE Journal of Civil Engineering (2024) 28(7):2613-2624 pISSN 1226-7988, eISSN 1976-3808

DOI 10.1007/s12205-024-2458-5 www.springer.com/12205

Geotechnical Engineering

Numerical Study on Shear Behavior of Geocell-reinforced Layer Based on

Large-scale Direct Shear Tests
a a,b
Yang Zhao , Zheng Lu , Jie Liuc, and Hailin Yaoa

a
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences,
Wuhan 430071, China
b
Hubei Key Laboratory of Geo-Environmental Engineering, Wuhan 430071, China
c
Xinjiang Transportation Planning Survey and Design Institute Co., Ltd., Urumqi 830006, China

ARTICLE HISTORY ABSTRACT

Received 16 August 2022 Although geocells have been extensively used to improve the performance of road embankments
Revised 9 March 2023 and slopes, their positive effect on shear strength under various shear failure scenarios has not
Accepted 29 January 2024 been sufficiently emphasized. This paper attempts to fill this research gap through a series of
Published Online 20 March 2024 numerical studies and presents ongoing research. First, FLAC3D is adopted to simulate two
large-scale direct shear tests, and the outcomes are compared with laboratory test results
KEYWORDS reported in the literature. Then, the layout of the geocell in the shear box, the angle between
the geocell layer and the shear plane, and the aspect ratio of the geocells are investigated to
Geosynthetics analyze the effects of each factor on the shear strength of the geocell-reinforced layer. The
Geocell findings confirm that the geocell-reinforced layer exhibits substantial anisotropy, which is
Large-scale direct shear test caused by variations in the shear failure plane. When the shear plane is near the center of the
FLAC3D geocell, or the angle between the shear plane and the horizontal plane is less than 45 degrees,
Geocell reinforced layer the geocells can provide higher shear strength. The shear strength of the geocell-reinforced
layer can also be improved by increasing the aspect ratio or geocell height, where the
apparent cohesiveness and friction angle are increased by 156% and 13%, respectively, over a
range of aspect ratios from 0.394 to 1.575.

1. Introduction (1993) first carried out several triaxial compression tests. The test
results indicated that the frictional resistance of the composite
Due to the exceptional performance and affordability, geocells, a structure matched the friction angle of the soil infill. The lateral
specialized three-dimensional geosynthetic material has been confinement mechanism can make the composite structure exhibit
widely utilized in geotechnical engineering (Yang et al., 2013; stronger cohesion. Further large-scale triaxial compression
Hegde and Sitharam, 2017; Song et al., 2018; Ardakani and experiments (Song et al., 2019; Song et al., 2021) confirmed similar
Namaei, 2021; Gedela and Karpurapu, 2021b). In the early 1970s, findings. However, these studies were performed using cylindrical
the US Army Corps of Engineers employed geocells to build isolated geocell-soil composite specimens that were incompatible
roadways over poor soils for faster and more flexible vehicle with honeycomb- or diamond-shaped geocells and did not
movement. Compared to other planar geosynthetics like geotextiles accurately represent the interaction effects of neighboring cells.
and geogrids, the greatest advantage of geocells is the increase in A number of conventional triaxial compression tests taking into
the shear strength of granular infill materials provided by lateral account multiple geocell pockets, roundness, rectangle, and hexagon
confinement of the vertical geocell walls (Bathurst and Karpurapu, shapes (Chen et al., 2013), as well as different raw materials (Latha
1993). and Murthy, 2007; Wang et al., 2021) were performed (Rajagopal et
The direct shear test and triaxial compression test are most al., 1999; Wesseloo et al., 2009). In addition, the apparent cohesion
frequently utilized to determine the shear strength parameters. To increased by the equivalent composite layer can be determined
gauge the advancement of the geocell, Bathurst and Karpurapu from the results of triaxial compression tests using the equations

CORRESPONDENCE Zheng Lu [email protected] State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese
Academy of Sciences, Wuhan 430071, China
ⓒ 2024 Korean Society of Civil Engineers
2614 Y. Zhao et al.

provided by Rajagopal et al. (1999). Hence, the geocells and infill of triaxial compression tests to fully evaluate the improvement of
materials could be regarded as a homogeneous flexible slab, which is shear strength (i.e., cohesion and internal friction) of reinforced
known as the Equivalent Composite Approach (ECA) and is soils is inappropriate. This is because, in some cases, such as
widely used in numerical calculations (Hegde and Sitharam, geocell-reinforced retaining walls or geocell-reinforced roadbed
2013; Hegde and Sitharam, 2015a; Hegde and Sitharam, 2015b; slopes (see Fig. 1), the stress state of the geocell-reinforced layer
Hegde, 2017; Ujjawal et al., 2019; Kabiri Kouchaksaraei and is not always compressive. Instead, the angles between the
Bagherzadeh Khalkhali, 2020). sliding plane of the potential failure and the geocell layers may
Direct shear test is advantageous over triaxial compression vary in the geocell-reinforced subgrade slope at the site. Also,
test because the shear plane can be adjusted to model various the shear plane may pass through the geocells at different wall
shear failure states of the geocell-reinforced layer. Also, the angle of height positions. Therefore, compared to the triaxial compression
the shear plane in the triaxial compression tests is always equal tests, the direct shear tests, especially for the direct shear numerical
to the 45 degrees plus the friction angle. Hence, the shear calculations, have the specific advantage of being able to adjust
strength parameters from direct shear tests were different from the position and angle of the shear plane.
those from triaxial compression tests (Cornforth, 1964; Lee, In this study, the geocell-reinforced layer's direct shear behavior
1970; Lini Dev et al., 2016; Medzvieckas et al., 2017). To explore is investigated considering the above factors. Also, geogrid structural
the properties at the geocell-soil interface, Tavakoli Mehrjardi elements are used to accurately model the three-dimensional
and Motarjemi (2018); Kabiri Kouchaksaraei and Bagherzadeh honeycomb-shaped geocells to represent the reinforced scenario.
Khalkhali (2020); Altay et al. (2021); Li et al. (2021) conducted a Firstly, the outcomes of two large direct shear tests are cited to
series of large-scale direct shear experiments. In this case, the verify the initial numerical models. The baseline model is developed
shear plane is kept parallel to the top of the geocell layer. The by extending a basic model. The influence of a single reinforcement
interlocking effect caused by the increase in medium grain size parameter on the shear behavior of the geocell-reinforced layer is
and relative density of soil resulted in an enhancement of the then investigated using a series of parametric calculations that
geocell-soil interface. In these investigations when the situation consider the geocell layouts, the angle between the geocell and
was more comparable to the pullout behavior, the interface the shear plane, and the aspect ratio. The study's results can guide
between the geocell-reinforced layer and soil layer was analyzed designers and researchers to understand the geocell-reinforced
rather than the interaction between the geocell walls and infill mechanism under different stress and shear failure states.
soil. Nowadays, few researchers concentrate on the geocell-
reinforced performance under shear failure conditions (Kabiri 2. Validation of Numerical Models
Kouchaksaraei and Bagherzadeh Khalkhali, 2020). There is a wide
gap between research studies evaluating geocell-reinforced Three-dimensional numerical simulations of two significant
performance as well as the stress and failure state of the geocell- large-scale direct shear tests are performed using FLAC3D software
reinforced layer in reinforced structures. Only Kabiri Kouchaksaraei to investigate the shear behaviors of unreinforced and geocell-
and Bagherzadeh Khalkhali (2020) considered the layout location of reinforced layers. Plastic collapse and flow caused by significant
geocell with the shear plane in their finite element analysis using shear deformations can be accurately modeled using the Lagrangian
the ECA method. Numerical results confirmed that the closer the formulas.
middle height of the geocell was to the shear surface, the more
effective the shear strength could be increased. 2.1 Model Description
Triaxial compression tests can reflect the true state of stress, in Two large-scale direct shear tests conducted by Kabiri
which the angle of the shear plane is always equal to 45 degrees Kouchaksaraei and Bagherzadeh Khalkhali (2020) and Tavakoli
plus the friction angle. However, the literature review indicates Mehrjardi and Motarjemi (2018) were adopted to validate the
that due to the lack of consideration of the real honeycomb form, numerical models. The tests are named DST-A and DST-B. The
shear failure plane, and multi-geocell pocket interactions, the use experimental results of the direct shear test with a 196 kPa
overburden are employed to simulate the DST-A. Under three
different overburdens of 100 kPa, 200 kPa, and 300 kPa, the
interface shear behavior of geocell unreinforced and reinforced
sand (S3) with a relative density of 50% is simulated using the
DST-B model. By neglecting the rigid direct shear box, the
dimensions of the two models (Figs. 2 and 3) are kept constant
with the dimensions of the samples inside the large direct shear
box. The wall height and size of the geocells are likewise kept
constant with the experimental tests. It is worth noting that the
paper by Kabiri Kouchaksaraei and Bagherzadeh Khalkhali
Fig. 1. The Possible Forms of Shearing between Geocells and Soil (2020) provided comprehensive information about the geocells.
(Modified from Palmeria and Milligan (1989)) In DST-A, geocells with 25-mm pocket openings (equivalent pocket
 KSCE Journal of Civil Engineering 2615

expanded honeycomb shape. To make the geocells resemble an

enlarged honeycomb form, the opening size of geocell is changed in
the numerical validation. The length of the numerical model in
the Y-direction is 3.5 times the length of the cell. The cell width
in the X direction is also extended to simulate the fully opening
geocell forms. The modified geocells maintain the same wall
height as the direct shear tests and have a pocket opening
diameter of 63.5 mm. The top and center of the geocell mattress
are placed horizontally to the shear plane in both DST-A and
DST-B models, respectively.

2.2 Constitutive Model and Boundary Conditions

Fig. 2. Isometric View of the Model Simulating DST-A 2.2.1 Constitutive Model
The selection of an appropriate constitutive model for modeling
the behavior of geomaterials is essential for numerical research.
The granular soils of DST-B exhibit a typical strain-softening
behavior, while the silty sand of DST-A exhibits a strain-hardening
behavior beyond the yield threshold. The direct shear tests performed
by Mohapatra et al. (2016); Ensani et al. (2021); Grabowski et al.
(2021); Lashkari and Jamali (2021) all revealed a similar behavior
pattern. Several researchers used the Mohr-Coulomb model (Kabiri
Kouchaksaraei and Bagherzadeh Khalkhali, 2020; Rashma et al.,
2021) to simulate the shear behavior of soil. It is well known that
the Mohr-Coulomb yield criterion, which is based on elasticity-
perfect plasticity, does not fully match the strain-softening and
Fig. 3. Isometric View of the Model Simulating DST-B strain-hardening behaviors. To simulate the behavior of cohesionless
soil, the strain-softening/hardening model is employed in this
paper instead of the Mohr-Coulomb model. Based on predefined
modifications of the Mohr-Coulomb model (such as cohesion,
friction, dilatation, and tensile strength as functions of the deviatoric
plastic strain), this constitutive model is capable of describing
nonlinear material behavior, such as softening or hardening. It is
assumed that the total strain () before the yield point of the soil
is purely elastic strain, i.e.,  =  e, while the total strain is the
elastic and plastic strains (  =  e +  p ) after the yield point.
Compared to the Mohr-Coulomb model, this model can modify
the cohesion, friction, and dilation according to the accumulated
plastic strain. Users can redefine these values as a piecewise
linear function related to plastic shear strain. The comprehensive
explanation of the strain-softening/hardening model is illustrated
in Fig. 5. In addition, the difference between the peak friction
Fig. 4. Detailed Geocell Size and Arrangement in the Shear Box (Modified angle and the critical state friction angle is used to characterize
from Tavakoli Mehrjardi and Motarjemi (2018)) the expansion angle in this study (Bolton, 1986; Gedela and
Karpurapu, 2021a). After fitting computational results to the
experimental data without geocells, the elastic modulus and
opening diameter of 28.2 mm) and 25-mm wall height are used. shear strength parameters of the soil are summarized in Tables 1
Although Tavakoli Mehrjardi and Motarjemi (2018) stated that and 2, respectively. It is worth noting that the elastic modulus of
geocells had a cell area of 55 mm2 and a wall height of 50 mm, sand in DST-B is correlated with the overburdened load (normal
respectively, there was a significant discrepancy between these stress) that follows the gradient law of the following Eq. (1) (Shi
figures and the actual geocell size described in their study (see et al., 2020).
Fig. 4). The honeycomb-shaped geocells were not fully expanded in n
P
DST-B. Hence, in this numerical validation, the geocell opening E = E0 ⎛ -----⎞ , (1)
⎝ P0⎠
size was adjusted to make the geocell mattress more like the
2616 Y. Zhao et al.

Fig. 5. Schematic Representation of Strain-Softening/Hardening Model: (a) Stress-Strain Curve, (b) Variation of ,  with  p, (c) Piecewise Linear
Variation of ,  with  p

Table 1. Properties of Soil and Geocell in the Validation Modeling Table 2. Variation of Shear Strength Parameters with Plastic Strain
Parameters Value Plastic strain, Plastic strain,
DST-A DST-B
 p (%)  p (%)
Silty sand (DST-A)
Young modulus, M (MPa) 25 Cohesion Friction Friction Dilation
(kPa) angle (°) angle (°) angle (°)
Poisson's ratio, ϑ 0.3
Cohesion, c (kPa) 6.2 0 6.2 30 0 53 8.7
Friction,  (°) 30 0.87 6.6 32.5 1 52.5 8.2
Dilation,  (°) 3 2.5 7.2 36.87 2 52.0 7.7
Density,  (kg/m3) 2090 4.57 7.3 37.1 3 51.5 7.2
5.23 7.3 37.1 4 47.0 2.7
Sand (DST-B)
− − − 5 46.5 2.2
Young modulus, M (MPa) z 12.5 (100 kPa)
23.1 (200 kPa) − − − 6 45.5 1.2
33.2 (300 kPa) − − − 7 44.3 0
Poisson's ratio, ϑ 0.3 − − − 8 44.3 0
Cohesion, c (kPa) 0
Friction,  (°) 53
Dilation,  (°) 6.7
geocells and the infill materials is considered to be linear (Hegde
Density,  (kg/m3) 2000 and Sitharam, 2015c). The parameter of the interface shear modulus
Geocells is calculated with reference to the study of Yang et al. (2010).
Young modulus, Mg (MPa) 250 Also, the interface cohesion and friction angle can be obtained
Poisson's ratio, ϑ 0.45 by discounting the corresponding values of infill soil of geocell
Interface shear modulus ki (MPa/m) 19.7 pockets. In this study, they are taken by referring the outcomes
Interface cohesion, ci (kPa) 4.96 (DST-A) from Oliaei and Kouzegaran (2017).
0 (DST-B)
Interface friction angle = atan (0.8 × tan(ϕ)), (2)
Interface friction, i (°) 24.8 (DST-A)
46.7 (DST-B) Interface cohesion = 0.8 × c, (3)
Thickness of geocell, t (mm) 1 where  and c are the friction angle and cohesion of soil infill,
respectively. It is essential to note that, due to the use of the
strain-softening/hardening model, cohesion and friction angle
where E0 is the fitting parameter to adjust the elastic modulus for are plastic shear strain functions. However, in this numerical
a given normal stress. Herein, the back-calculated elastic modulus study, the initial values of the interface cohesion and friction angle
value is 12.5 MPa at the normal stress of 100 kPa. P and P0 are determined by ignoring the parameter changes. Detailed
represent the normal stress and the atmospheric pressure, parameters of geocell in numerical analysis are displayed in Table 1.
respectively. The exponent n depends on the soil type. In this
study, it is set to 0.9 by considering the soil type and back- 2.2.2 Boundary Conditions
calculating from the experimental results. The numerical boundary conditions for the DST-A and DST-B
The diamond and honeycomb geocells are simulated using models are the same. Equal horizontal displacements are applied
the geogrid structure element. It is noted that the important to the grid points on the bottom and lateral limits of the lower
mechanism offered by the geogrid structure element in FLAC3D shear box in the X-direction at a rate of 5e-7 m/step to achieve
is the interface shear behavior. According to the Mohr-Coulomb the desired horizontal displacement. Some corresponding lower
failure criterion, the interface shear relationship between the shear box surfaces are fixed to prevent lateral movement to
 KSCE Journal of Civil Engineering 2617

simulate the rigid shear box. To maintain the fixity of the upper
shear box, all grid points on the top and vertical boundaries are
also constrained in the X-direction. Mohapatra et al. (2017)
employed identical boundary conditions to simulate the direct shear
test.
Several FISH codes are compiled to determine the X-direction
unbalanced force generated by the bottom shear box, which is
then divided by the area of the shear plane to obtain the shear
stress. The matched Z-direction continuous stress at the top of
the models is used as the normal stress of the direct shear tests. A
dial gauge is used in the DST-B test to track the volume change
caused by the shear behavior. To simulate this phenomenon,
some FISH programs are also employed to determine the average
Z-direction displacement at the top surface.

2.3 Validation
The variations of shear stress with shear displacement for DST-A Fig. 8. Variation of Top-surface Normal Displacement with Shear
Displacement for DST-B under the Normal Stress of 200 kPa
and DST-B are presented in Figs. 6 and 7, respectively. Fig. 8

shows the variations of top-surface normal displacement and

shear displacement for DST-B at a normal load of 200 kPa. It can
be concluded that for the unreinforced and geocell-reinforced
cases, the numerical results are in very good agreement with the
direct shear test data, with very small deviations. The reasons for
this small difference can be summarized as follows. In the direct
shear test, FLAC3D can play the role of the interlock of sand
particles, but it cannot simulate it. The present study will use the
discrete element method to avoid the deficiency. In addition, the
calculated parameters are obtained by the back calculating and
trial-and-error method in this study. The unavailability of actual
soil mechanical parameters may the another reason.
Additionally, this study evaluates the apparent cohesion and
friction angle of the DST-B based on the experimental results
and numerical analysis. The apparent cohesion and friction angle
Fig. 6. Variation of Shear Stress with Shear Displacement for DST-A
for the unreinforced and reinforced cases in the tests are 0.0 kPa,
(FLAC3D versus Test) 42.3°, and 1.9 kPa, 42.2°, respectively; while those in the
numerical simulation are 5.6 kPa, 42.1°, and 6.7 kPa, 40.8°,
respectively. Furthermore, it should be noted that the outcomes
from direct shear tests differ from the input parameters displayed
in Table 2 in terms of the friction angle. This is because the
input-mechanical parameters in FLAC3D represent the values
obtained from the true triaxial compression tests. However, the
cohesion and friction angle mentioned above are obtained from
the numerical calculations of direct shear tests. According to
numerous studies (Cornforth, 1964; Lee, 1970; Lini Dev et al.,
2016; Medzvieckas et al., 2017), the internal friction angle of
sands in direct shear tests was 2 to 8 degrees higher than that in
triaxial compression tests.

3. Numerical Analysis of Shear Behavior of

Geocell-Reinforced Layer

Fig. 7. Variation of Shear Stress with Shear Displacement for DST-B Based on the DST-B numerical model, the shear behavior of the
(FLAC3D vesus Test) geocell-reinforced layer is investigated in this section. Geocells
2618 Y. Zhao et al.

Table 3. Details of Parametric Study

Test series Details Variable parameter Constant parameters
0 Baseline model None h = 50 mm
S = 25 mm
 = 0o
1 Effect of geocell layout S = 0 5 10 15 20 25 mm h = 50 mm
=0
2 Effect of angle between geocell and shear plane  = 0o 30o 45o 60o 90o h = 50 mm
S = 25 mm
3 Effect of aspect ratio h = 25 50 75 100 mm S = 0.5 h
(aspect ratio: h/d = 0.394 0.787 1.181 1.575)  = 0o

Fig. 9. Schematic Diagram of the Geocell-Reinforced Details: (a) Layout of Geocells, (b) Angle between Geocell and Shear Plane

in the shape of honeycombs and infill sand have properties more Table 3 provides more details of the baseline model. A series of
similar to those of field reinforcement works. On the other hand, parametric simulations also investigate the effect of a single
the DST-B geocell layer is located on the lower shear box, and reinforced parameter on the shear behavior of a geocell-reinforced
the top of the layer is kept horizontal and aligned with the shear layer. Separate changes are made to the geocell layout, the angle
plane in order to study the interfacial properties between the between the geocell and shear plane, and the geocell aspect ratio
geocell and the soil rather than the shear characteristics of the (see Table 3). The schematic representation of the geocell placement
geocell-reinforced layer. Therefore, in the present numerical and angle to the shear plane is shown in Fig. 9.
analysis, the baseline models are developed by varying the locations
of the geocell-reinforced layer. The model dimensions, loading 4. Results and Discussions
condition, boundary condition, constitutive model, parameters,
applied normal stress, etc., are kept the same as the DST-B numerical 4.1 Effect of Geocell Layout
model, except for the location of the geocell-reinforced layer. The variation of shear stress with shear displacement for various

Fig. 10. Variation of Shear Stress with Shear Displacement: (a) Different Layout under the Normal Stress of 200 kPa, (b) Different Normal Stress for
the Case of S = 25 mm
 KSCE Journal of Civil Engineering 2619

geocell layouts is shown in Fig. 10. Since the shear behavior is from the strain-softening behavior curves. According to the
similar for each scenario, this study only reports results for the Chinese standard (Test Methods of Geosynthetics for Highway
case of 200 kPa normal stress and S = 25 mm. Unreinforced soil Engineering, JTG E50-2006), the peak shear stress for the strain-
exhibits strain softening behavior in Fig. 10(a), whereas geocell hardening behavior curves is also considered to be the shear
reinforcement results in strain hardening. The shear strength of stress corresponding to 10% of the length of the shear box. The
the geocell-reinforced layer increases somewhat with an increase variation of the geocell layout and interface shear strength
in the distance between the geocell surface and the shear plane coefficient under various normal stresses is shown in Fig. 11.
(i.e., the value of S), especially when the value of S is more than The coefficient becomes stabilized when S is larger than 15 mm,
15 mm. Furthermore, it can be concluded that the geocell- as indicated by the data. Under different normal stresses, geocell
reinforced layer exhibits anisotropy in terms of shear strength can increase the interface shear strength by a maximum of 1.75
(Latha and Somwanshi, 2009; Hegde, 2017; Song et al., 2019). to 2.75 compared to unreinforced sand. It is worth noting that
Two curves of the unreinforced and S = 0 cases almost overlap normal stress negatively influences the advancement of geocells,
each other. Experimental findings (Tavakoli Mehrjardi and which yields a reduction in the beneficial effect of improving the
Motarjemi, 2018), confirmed that the shear strength at the shear strength of the geocell-reinforced layer. Tavakoli Mehrjardi
geocell-soil interface slightly decreased relative to its initial shear and Motarjemi (2018) obtained the same results. The increment
strength when the shear displacement exceeded the yield point. of shear stress provided by geocells remains relatively constant.
Two factors contribute to this. 1) 3.0 mm medium-grained sand As demonstrated in the calculated equation of interface shear-
is used in the numerical analysis. The interlocking effect cannot strength coefficient, the Soil internal is relative smaller in the low
be facilitated by small particle sizes. 2) Grain sliding alongside normal stress. Therefore, the smaller the A in the denominator,
the geocell walls leads to a weakening of the shear strength of the the higher value of . Or, a greater shear stress can be obtained at
geocell-soil contact, deforming the geocell walls (Tavakoli Mehrjardi a high normal stress. In this case, the proportion of the improved
and Motarjemi, 2018). According to Fig. 10(b), the direct shear shear strength of the geocell is relatively small compared to the
tests exhibit an increase in shear stress due to an increase in shear strength of the unreinforced soil.
normal stress, which is consistent with the prediction. The linear Mohr failure envelopes of the shear plane for
The interface shear-strength coefficient (α), a nondimensional various geocell layouts are shown in Fig. 12. The Mohr envelopes
parameter, was utilized by Liu et al. (2009); Makkar et al. (2019) move upward as the distance between the geocell surface and the
to compare the interface shear strengths of soil and geosynthetics. shear plane increases. The shear strength parameters of the geocell-
This study employs this coefficient to evaluate the shear strength reinforced layer, i.e., apparent cohesion and friction angle, can be
of the geocell-reinforced layer. determined, as illustrated in Fig. 13. Many researchers have
claimed that the peak angles of composite structure is the same
Soil – geocell
 = ----------------------
-, (4) as that of soil fill. The confining system only enhanced apparent
Soil internal
cohesion (Bathurst and Karpurapu, 1993; Hegde, 2017; Song et
where Soil – geocell is the peak shear stress at the shear plane of the al., 2019). However, the numerical analysis of the direct shear
geocell-reinforced layer; and Soil internal is the peak shear stress test suggests that the two parameters both increase due to the
of unreinforced soil. The peak shear stress can be easily obtained cellular confining system, which is in accordance with the
experimental results of Li et al. (2021). The angle of shear plane

Fig. 11. Variation of Interface Shear-Strength Coefficient with Geocell

Layout Fig. 12. Linear Mohr Failure Envelopes for Various Reinforced Cases
2620 Y. Zhao et al.

Fig. 13. Variation of Shear Strength Parameters with Geocell Layout

in the triaxial test is 45° + /2. Since the geocells are located
outside the soil sample, soil particles do not move on the shear
surface. However, the shear plane in the present study is set to
run horizontally. Increased S causes more soil particles close to
the shear surface to move along the geocell walls. The geocell
walls can offer lateral confinement and vertical frictional resistance,
Fig. 15. Contours of X Direction in the Middle of Shear Box: (a) S = 5 mm,
which is how geocell reinforcement works. These two points are
(b) S = 25 mm
directly identified in this subsection. Furthermore, the increased
magnitude of the friction angle is smaller than the increased
amplitude of apparent cohesion. The geocell-soil interface has significant normal displacement, as shown in Fig. 14(a). It is
much less shear strength than a geocell mattress. When S/H further demonstrated by the X-direction contours at the center of
varies from 0 to 0.5, the apparent cohesion increases by around the shear box (see Fig. 15). It should be noted that the contours in
140 kPa, whereas the friction angle increases from 40.8° to 52.6°. the case of S = 25 mm might mobilize more soils than those in
The variation of normal displacement with shear displacement the case of S = 5 mm. In addition, as indicated in Fig. 14(b), it is
for different geocell layouts is shown in Fig. 14. In this study, similar to the curves trend presented in Fig. 11, normal stress
only the results under normal stress of 200 kPa and S = 25 mm may influence this behavior. They all reveal that the development of
are reported. The increased S value suggests that geocells mobilize normal displacement is negatively affected by normal stress
more soil zones in the shear boxes, which contributes to the (Tavakoli Mehrjardi and Motarjemi, 2018). In other words,

Fig. 14. Variation of Normal Displacement with Shear Displacement: (a) Different Layout under the Normal Stress of 200 kPa, (b) Different Normal
Stress for the Case of S = 25 mm
 KSCE Journal of Civil Engineering 2621

Fig. 16. Variation of Shear Stress with Shear Displacement: (a) Different Angles under the Normal Stress of 200 kPa, (b) Different Normal Stress for
the Case of  = 45°

Fig. 17. Variation of Interface Shear-strength Coefficient with Different Fig. 18. Linear Mohr Failure Envelopes for Reinforced Cases of Different
Angles Angles

larger normal stresses may prevent the movement of soil reinforced layer can only be increased by 1.1 to 1.35 times
particles. compared to the unreinforced soil.
The apparent cohesion and friction angle are important factors
4.2 Effect of the Angle between Geocell and Shear that need to be considered when assessing the performance of
Plane reinforced structures at various angles. The linear Mohr failure
The variation of shear stress with shear displacement at different envelopes (see Fig. 18) are used to derive these two parameters.
angles between the geocell and the shear plane is displayed in It can be observed from Fig. 19 that the diminishing tendency of
Fig. 16, the results at different angles for a normal stress of the two parameters is the same. The apparent cohesion and friction
200 kPa and  = 45° are demonstrated. As with the results in angle decrease with the increasing angle between the geocell and
Fig. 10, the stress-strain behavior of reinforced layer exhibits the shear plane. In addition, the initial rotation of geocell mattress
strain hardening, with higher normal stress increasing the shear significantly affects this behavior, particularly for cohesion. The
stress. The interface shear strength coefficient can be used to numerical results imply that geocells cannot considerably enhance
investigate the improvement of the geocell under various shear strength when the angle between the geocell and the shear
orientations. As indicated in Fig. 17, the shear strength is plane is greater than 45°. Therefore, it is recommended to use the
noticeably reduced by the first rotation of the geocell layer, actual geocell and soil models to substitute the Equivalent
particularly at low normal stress cases. When the angle between the Composite Approach for assessing slope stability in numerical
geocell and shear plane reaches 45°, the peak shear stress of the simulations.
2622 Y. Zhao et al.

Fig. 19. Variation of Shear Strength Parameters with Different Angles

Fig. 21. Variation of Interface Shear-strength Coefficient with Different
Aspect Ratios
4.3 Effect of the Aspect Ratio of Geocells
Based on the pre-calculated numerical analysis, the horizontal
distance between the geocell and the edge of the shear boxes
affects the outcomes. As a result, the current study uses an
increase in height to adjust the aspect ratio. Table 3 presents
detailed aspect ratios. Furthermore, in this section, a wall height
legend is used for all graphs utilize instead of an aspect ratio
legend. The results in Sections 4.1 and 4.2 are consistent with the
shear stress pattern with shear displacement depicted in Fig. 20.
The various interface shear strength coefficient with wall height
is displayed in Fig. 21 for comparison purposes. The peak shear
stress increases approximately linearly with wall height (aspect
ratio). According to Kabiri Kouchaksaraei and Bagherzadeh
Khalkhali (2020), the interfacial shear strength was not improved
by increasing the aspect ratio. However, the above-mentioned
results are not applicable to increasing the shear strength of the
geocell-reinforced layer. Peak shear stress increases by 2.0 to 3.5
times at low normal stress (100 kPa), but there is no appreciable Fig. 22. Linear Mohr Failure Envelopes for Reinforced Cases of Different
improvement at high normal stress (300 kPa). It is important to Aspect Ratios

Fig. 20. Variation of Shear Stress with Shear Displacement: (a) Different Aspect Ratios under the Normal Stress of 200 kPa, (b) Different Normal
Stress for the Case of H = 75 mm
 KSCE Journal of Civil Engineering 2623

increases from 40.8° to 52.6°. The initial rotation significantly

reduces the shear strength of the geocell-reinforced layer with a
minimum value of 45 degrees. When the angle between the
geocell and shear plane reaches 45°, the peak shear stress of the
reinforced layer is increased only by 1.1 to 1.35 times compared
to the unreinforced soil. In addition, the rotation of geocells yields a
greater effect on the cohesion than the friction of the reinforced
layer. Further, the shear strength of the geocell-reinforced layer
may be greatly improved by increasing the aspect ratio. The
apparent cohesion and friction angle increase by 156% and 13%,
respectively in case of the aspect ratio varying from 0.394 to
1.575.

Acknowledgments
Fig. 23. Variation of Shear Strength Parameters with Different Aspect The research described in this paper was financially supported
Ratios by the National Natural Science Foundation of China (Nos.
42077262, 42077261, and 41972294), the Sichuan Transportation
note that the wall height is changed in this study to represent a Science and Technology Project (No. KJFZ-2022Y-022), the
change in aspect ratio from 0.394 to 1.575. The results of this 2022 Annual Transportation Industry Science and Technology
study may differ somewhat from the actual situation or, in the Project (Nos. 2022-ZD-017, and 2022-ZD-018), and the Research
case of the constant height but varying pocket size, this still Fund Project of Xinjiang Transportation Planning Survey and Design
needs to be supported by the follow-up investigations. Institute Co., Ltd. (Nos. KY2022042504, and KY2022042501). In
Based on the linear Mohr failure envelopes displayed in Fig. 22, addition, the first author, Yang Zhao wants to thank Juan Li and
plots are developed to demonstrate how the apparent cohesion Guan-lin Zhao for the great support of his scientific studies during
and friction angle vary with geocell height. As shown in Fig. 23, those dark days.
both shear strength values increase as the wall height increases.
From an aspect ratio of 0.394 to 1.575, the apparent cohesion and ORCID
friction angle increase by 156% and 13%, respectively, indicating
that increasing the aspect ratio is more favorable for increasing Yang Zhao https://orcid.org/0000-0002-0521-0651
the apparent cohesion. However, the friction resistance may only Zheng Lu https://orcid.org/0000-0003-3226-6535
be somewhat enhanced, which is comparable to the results of
conventional triaxial compression tests. References

5. Conclusions Altay G, Kayadelen C, Taskiran T, Bagriacik B, Toprak O (2021) Frictional

properties between geocells filled with granular material. Revista de
In this paper, the direct shear behaviors of geocell-reinforced la Construcción 20(2):332-45, DOI: 10.7764/rdlc.20.2.332
layers with various reinforced scenarios are investigated through Ardakani A, Namaei A (2021) Numerical investigation of geocell reinforced
a series of numerical calculations. Initially, the direct shear tests slopes behavior by considering geocell geometry effect. Geomechanics
verify the accuracy of the numerical calculations, and the results and Engineering 24(6):589-97, DOI: 10.12989/gae.2021.24.6.589
Bathurst RJ, Karpurapu R (1993) Large-scale triaxial compression testing
agree well with the numerical calculations. Then, based on the
of geocell-reinforced granular soils. Geotechnical Testing Journal
baseline model, the influencing factors including geocell layout, 16(3):296-303, DOI: 10.1520/GTJ10050J
the angle between the geocell layer and shear plane, and aspect Bolton MD (1986) The strength and dilatancy of sands. Géotechnique
ratio are adjusted to compare their effects on the reinforced 36(1):65-78, DOI: 10.1680/geot.1986.36.1.65
performance. The results confirm that the stress-strain curve of Chen R-H, Huang Y-W, Huang F-C (2013) Confinement effect of
the unreinforced case displays a strain-softening behavior, whereas geocells on sand samples under triaxial compression. Geotextiles
all geocell-reinforced cases exhibit a strain-hardening behavior. and Geomembranes 37(3):35-44, DOI: 10.1016/j.geotexmem.2013.
01.004
The initial geocell layout has an important effect on the interface
Cornforth DH (1964) Some experiments on the influence of strain
shear strength coefficient. When the middle of the geocell walls conditions on the strength of sand. Géotechnique 14(2):143-67,
is alignment with the horizontal shear plane, the interfacial shear DOI: 10.1680/geot.1964.14.2.143
stress reaches its maximum. The maximum interface shear strength Ensani A, Razeghi HR, Mamaghanian J (2021) Application of hyperbolic
of geocell-reinforced layer increases from 1.75 to 2.75 at different model for mechanical behaviour of marginal soil-geosynthetic interface.
normal stresses. Increasing S/H from 0 to 0.5 increases the Geosynthetics International, 1-60, DOI: 10.1680/jgein.21.00039
apparent cohesion by around 140 kPa, whereas the friction angle Gedela R, Karpurapu R (2021a) Influence of pocket shape on numerical
2624 Y. Zhao et al.

response of geocell reinforced foundation systems. Geosynthetics (ASCE)GT.1943-5606.0000150

International 28(3):327-37, DOI: 10.1680/jgein.20.00042 Makkar FM, Chandrakaran S, Sankar N (2019) Experimental investigation
Gedela R, Karpurapu R (2021b) Laboratory and numerical studies on of response of different granular Soil–3D geogrid interfaces using
the performance of geocell reinforced base layer overlying soft large-scale direct shear tests. Journal of Materials in Civil Engineering
subgrade. International Journal of Geosynthetics and Ground 31(4):04019012, DOI: 10.1061/(asce)mt.1943-5533.0002645
Engineering 7(1):1-18, DOI: 10.1007/s40891-020-00249-4 Medzvieckas J, Dirgėlienė N, Skuodis Š (2017) Stress-strain states
Grabowski A, Nitka M, Tejchman J (2021) Comparative 3D DEM differences in specimens during triaxial compression and direct
simulations of sand–structure interfaces with similarly shaped clumps shear tests. Procedia Engineering 172:739-45, DOI: 10.1016/j.proeng.
versus spheres with contact moments. Acta Geotechnica 3533-3554, 2017.02.094
DOI: 10.1007/s11440-021-01255-0 Mohapatra SR, Rajagopal K, Sharma J (2016) Direct shear tests on
Hegde A (2017) Geocell reinforced foundation beds-past findings, present geosynthetic-encased granular columns. Geotextiles and Geomembranes
trends and future prospects: A state-of-the-art review. Construction 44(3):396-405, DOI: 10.1016/j.geotexmem.2016.01.002
and Building Materials 154(15):658-74, DOI: 10.1016/j.conbuildmat. Oliaei M, Kouzegaran S (2017) Efficiency of cellular geosynthetics for
2017.07.230 foundation reinforcement. Geotextiles and Geomembranes 45(2):11-
Hegde A, Sitharam TG (2013) Experimental and numerical studies 22, DOI: 10.1016/j.geotexmem.2016.11.001
on footings supported on geocell reinforced sand and clay beds. Palmeria EM, Milligan GWE (1989) Scale and other factors affecting
International Journal of Geotechnical Engineering 7(4):346-54, the results of pull-out tests of grids buried in sand. Géotechnique
DOI: 10.1179/1938636213Z.00000000043 39(3):511-42, DOI: 10.1680/geot.1989.39.3.511
Hegde A, Sitharam TG (2015a) 3-Dimensional numerical modelling of Rajagopal K, Krishnaswamy N, Latha GM (1999) Behaviour of sand
geocell reinforced sand beds. Geotextiles and Geomembranes 43(2): confined with single and multiple geocells. Geotextiles and
171-81, DOI: 10.1016/j.geotexmem.2014.11.009 Geomembranes 17(3):171-84, DOI: 10.1016/S0266-1144(98)00034-X
Hegde A, Sitharam TG (2017) Experiment and 3D-numerical studies on Rashma RSV, Jayalekshmi BR, Shivashankar R (2021) Shear strength
soft clay bed reinforced with different types of cellular confinement behaviour of pervious concrete column improved soft clay bed: A
systems. Transportation Geotechnics 10:73-84, DOI: 10.1016/j.trgeo. numerical study. Transportation Infrastructure Geotechnology, DOI:
2017.01.001 10.1007/s40515-021-00179-2
Hegde AM, Sitharam TG (2015b) Effect of infill materials on the Shi C, Zhao C, Zhang X, Guo Y (2020) Coupled discrete-continuum
performance of geocell reinforced soft clay beds. Geomechanics and approach for railway ballast track and subgrade macro-meso analysis.
Geoengineering 10(3):163-73, DOI: 10.1080/17486025.2014.921334 International Journal of Pavement Engineering, 1-16, DOI: 10.1080/
Hegde AM, Sitharam TG (2015c) Three-dimensional numerical analysis of 10298436.2020.1721498
geocell-reinforced soft clay beds by considering the actual geometry Song F, Chen W, Nie Y, Ma L (2021) Evaluation of required stiffness
of geocell pockets. Canadian Geotechnical Journal 52(9):1396-407, and strength of cellular geosynthetics. Geosynthetics International,
DOI: 10.1139/cgj-2014-0387 1-36, DOI: 10.1680/jgein.21.00032
Kabiri Kouchaksaraei M, Bagherzadeh Khalkhali A (2020) The effect Song F, Liu H, Ma L, Hu H (2018) Numerical analysis of geocell-
of geocell dimensions and layout on the strength properties of reinforced retaining wall failure modes. Geotextiles and Geomembranes
reinforced soil. SN Applied Sciences 2(10):1701, DOI: 10.1007/ 46(3):284-96, DOI: 10.1016/j.geotexmem.2018.01.004
s42452-020-03480-w Song F, Liu H, Yang B, Zhao J (2019) Large-scale triaxial compression
Lashkari A, Jamali V (2021) Global and local sand–geosynthetic interface tests of geocell-reinforced sand. Geosynthetics International 26(4):388-
behaviour. Géotechnique 71(4):346-67, DOI: 10.1680/jgeot.19.P.109 95, DOI: 10.1680/jgein.19.00019
Latha GM, Murthy VS (2007) Effects of reinforcement form on the behavior Tavakoli Mehrjardi G, Motarjemi F (2018) Interfacial properties of
of geosynthetic reinforced sand. Geotextiles and Geomembranes geocell-reinforced granular soils. Geotextiles and Geomembranes
25(1):23-32, DOI: 10.1016/j.geotexmem.2006.09.002 46(4):384-95, DOI: 10.1016/j.geotexmem.2018.03.002
Latha GM, Somwanshi A (2009) Effect of reinforcement form on Ujjawal KN, Venkateswarlu H, Hegde A (2019) Vibration isolation
the bearing capacity of square footings on sand. Geotextiles and using 3D cellular confinement system: A numerical investigation.
Geomembranes 27(6):409-22, DOI: 10.1016/j.geotexmem.2009.03. Soil Dynamics and Earthquake Engineering 119:220-34, DOI:
005 10.1016/j.soildyn.2018.12.021
Lee KL (1970) Comparison of plane strain and triaxial tests on sand. Wang J-Q, Hou S-L, Xue J-F, Lin Z-N, Tang Y, Zhou J-w (2021) Triaxial
Journal of the Soil Mechanics and Foundations Division 96(3):901- shear behavior of a gravelly sand with different forms of reinforcement.
23, DOI: 10.1061/JSFEAQ.0001425 Advances in Civil Engineering 2021:1-11, DOI: 10.1155/2021/8891400
Li H, Yang X, Zeng H, Zhao X (2021) Shear plane characteristics of Wesseloo J, Visser AT, Rust E (2009) The stress–strain behaviour of
geocell-reinforced fine sand through direct shear test. Arabian Journal multiple cell geocell packs. Geotextiles and Geomembranes 27(1):31-
of Geosciences 14(21):2226, DOI: 10.1007/s12517-021-08604-z 38, DOI: 10.1016/j.geotexmem.2008.05.009
Lini Dev K, Pillai RJ, Robinson RG (2016) Drained angle of internal Yang X, Han J, Leshchinsky D, Parsons RL (2013) A three-dimensional
friction from direct shear and triaxial compression tests. International mechanistic-empirical model for geocell-reinforced unpaved roads.
Journal of Geotechnical Engineering 10(3):283-87, DOI: 10.1080/ Acta Geotechnica 8(2):201-13, DOI: 10.1007/s11440-012-0183-6
19386362.2015.1133754 Yang X, Han J, Parsons RL, Leshchinsky D (2010) Three-dimensional
Liu C-N, Zornberg JG, Chen T-C, Ho Y-H, Lin B-H (2009) Behavior of numerical modeling of single geocell-reinforced sand. Frontiers of
geogrid-sand interface in direct shear mode. Journal of Geotechnical Architecture and Civil Engineering in China 4(2):233-40, DOI: 10.1007/
and Geoenvironmental Engineering 135(12):1863-71, DOI: 10.1061/ s11709-010-0020-7