Geotextiles and Geomembranes 44 (2016) 332e343

Contents lists available at ScienceDirect

Geotextiles and Geomembranes

journal homepage: www.elsevier.com/locate/geotexmem

Modified unit cell approach for modelling geosynthetic-reinforced

column-supported embankments
Yan Yu a, *, Richard J. Bathurst a, 1, Ivan P. Damians b, 2
a
GeoEngineering Centre at Queen's-RMC, Department of Civil Engineering, Royal Military College of Canada, Kingston, Ontario K7K 7B4, Canada
b
Department of Civil and Environmental Engineering, School of Civil Engineering, Universitat Polit

ecnica de Catalunya-BarcelonaTech (UPC), Spain

a r t i c l e i n f o a b s t r a c t

Article history: Geosynthetic-reinforced and column-supported (GRCS) embankments have proven to be an effective
Received 3 September 2015 construction technique for fills on soft foundations. The paper introduces a modified unit cell approach to
Received in revised form model GRCS embankments supported by deep mixed column walls. The modified unit cells include linear
4 January 2016
elastic springs at one or both vertical boundaries to simulate lateral displacements of the embankment
Accepted 14 January 2016
Available online xxx
fill and foundation soil. The finite difference program FLAC is used to compare numerical outcomes using
the modified unit cells with those using the typical unit cell arrangement with lateral rigid side
boundaries. Numerical results demonstrate good agreement between simulations using small-strain and
Keywords:
Geosynthetic reinforcement
large-strain modes in some cases and large differences in other cases. Lateral displacements of the
Column-supported embankments embankment fill and foundation soil using the modified unit cells are shown to have large influence on
Numerical modelling reinforcement loads. Finally the paper demonstrates that calculated reinforcement loads are sensitive to
Unit cell choice of small-strain or large-strain mode when using program FLAC.
Deep mixing © 2016 Elsevier Ltd. All rights reserved.
FLAC

1. Introduction
Embankments over soft foundations must be designed to avoid
bearing capacity failure of the foundation, unacceptable lateral
spreading of the embankment fill, and damage to adjacent struc-
tures due to large differential settlements. An effective technique to
overcome these challenges is to use geosynthetic-reinforced and
column-supported (GRCS) embankments (Fig. 1). The addition of
geosynthetic reinforcement improves the performance of column-
supported embankments that predate the use of GRCS embank-
ments. Common support types for GRCS embankments are
cementesoil deep mixing (DM) columns (e.g., Bergado et al., 1999;
Borges and Marques, 2011; Bruce et al., 2013; Chai et al., 2015;
Forsman et al., 1999; Han et al., 2007; Huang and Han, 2009,
2010; Huang et al., 2009; Lai et al., 2006; Liu and Rowe, 2015,
2016; Yapage and Liyanapathirana, 2014) and geosynthetic-
encased stone columns (e.g., Hosseinpour et al., 2015; Khabbazian
et al., 2015; Yoo, 2010; Ali et al., 2014; Gu et al., 2016). Concrete
* Corresponding author. Tel.: þ1 613 541 6000x6347; fax: þ1 613 541 6218.
E-mail addresses: yan.yu@queensu.ca (Y. Yu), bathurst-r@rmc.ca (R.J. Bathurst),
ivan.puig@upc.edu (I.P. Damians).
1
Tel.: þ1 613 541 6000x6479, fax: þ1 613 541 6218.
2
Tel.: þ34 93 401 1695; fax: þ34 93 401 7251.
and timber piles with and without pile caps and basal reinforce-
ment have also been used to increase the construction rate and to
improve load transfer from the soft soil to the stiffer piles (e.g.,
Briançon and Simon, 2012; Liu et al., 2007; Nunez et al., 2013; Rowe
and Liu, 2015; Zhang et al., 2013; Blanc et al., 2014; Xing et al., 2014;
Bhasi and Rajagopal, 2015a, 2015b). The basic working mechanisms
for DM column/pile-supported embankments with geosynthetic
reinforcement are soil arching and tensioned membrane effects
resulting in load transfer from the embankment fill self-weight
(plus any surcharge) to the DM columns/piles. The load on the
DM columns/piles is then transferred to the deeper and stiffer soil
stratum (Fig. 1).
GRCS embankments can be designed using closed-form solu-
tions that take advantage of soil arching and tensioned membrane
load transfer mechanisms within the GRCS embankment system
(e.g., Hewlett and Randolph, 1988; Low et al., 1994; Love and
Milligan, 2003; Kempfert et al., 2004; BS8006, 2010; EBGEO,
2011; Van Eekelen et al., 2011, 2013, 2015). Advanced numerical
models for complex soilestructure interaction problems using the
finite element method (FEM) and finite difference method (FDM)
are becoming more common as a research tool to improve under-
standing of the behaviour of GRCS embankments (e.g., Liu and
Rowe, 2015; Han et al., 2007). The advantage of using a full-width
numerical model of a GRCS embankment is that lateral

http://dx.doi.org/10.1016/j.geotexmem.2016.01.003
0266-1144/© 2016 Elsevier Ltd. All rights reserved.
 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
Fig. 1. Schematic showing a full-width model of a GRCS embankment with DM column walls.
deformations that vary across the width and depth of the
embankment fill and foundation are predicted. Of course, the ac-
curacy of numerical predictions will depend on mesh refinement
and the complexity of the constitutive models used for the
component materials and their interfaces. However, parametric
analyses at the design stage using a full-width model can be very
time consuming and may not adequately capture local load transfer
mechanisms particularly if a coarse numerical mesh is used in the
simulations. A strategy to overcome this shortcoming is the unit
cell approach (e.g., Han and Gabr, 2002; Smith and Filz, 2007;
Zhuang et al., 2010; Khabbazian et al., 2015). The location of an
example unit cell in a GRCS embankment is shown in Fig. 1. How-
ever, there are also limitations associated with typical unit cells
including the use of fixed lateral boundaries (Fig. 2a). Khabbazian
et al. (2015) showed that the tensile loads in the geosynthetic
reinforcement using a full-width GRCS embankment model were
much greater than those using the unit cell approach for the same
structure. They attributed this discrepancy to lateral spreading of
the embankment fill and foundation soil in the full-width model
that was not captured by the unit cell. Regardless of which
approach is used to model GRCS embankment performance, nu-
merical results can also depend on how geometric nonlinearity of
the soil and reinforcement is modelled using the small-strain and
large-strain options in the FDM program FLAC (Itasca, 2011) or with
and without mesh updating in FEM software programs.
The objectives of this paper are to demonstrate a new modelling
technique using a modified unit cell approach to simulate the
lateral spreading of the embankment fill and foundation soil, and to
examine the influence of large-strain and small-strain model op-
tions in program FLAC on numerical outcomes (i.e., with and
without mesh updating during calculation steps). Numerical results
using (conventional) unit cells with lateral rigid boundaries and
units cells with one or both vertical boundaries supported by
horizontal linear elastic springs are presented and compared. The
effect of lateral spring stiffness values on lateral spreading of the
embankment fill and foundation soil, and reinforcement loads are
demonstrated.
333
2. Small- and large-strain mode in FLAC
Numerical analyses using FLAC models (Itasca, 2011) can be
executed in either large-strain mode (based on the Lagrangian
formulation) or small-strain mode (based on the Eulerian formu-
lation). For the Lagrangian formulation, the numerical grid co-
ordinates at the end of each calculation step (or specified steps) are
updated by adding the grid incremental displacements to grid co-
ordinates before the next step. Hence, stresses and displacements
at the current calculation step are calculated based on the updated
grid representing the deformed material zones. However for the
Eulerian formulation, the grid is fixed to the original geometry and
material zones. The calculation of stresses and displacements is
based on the fixed grid even though the material zones move and
deform during subsequent calculation steps. The reader is directed
to the FLAC manual (Itasca, 2011) for details regarding small- and
large-strain options in the program.
3. Problem definition and parameter values
Fig. 1 shows a GRCS embankment where the soft foundation soil
is improved by the cementesoil DM column walls. The numerical
simulations carried out in this paper are for two-dimensional cases
because of the plane-strain condition associated with Fig. 1. How-
ever, the general approach presented in this paper can be extended
to model three-dimensional GRCS embankment cases. This paper
uses the example of GRCS embankments with 10-m thick soft
foundation soil and 1e5-m thick embankment fills. Above the soft
foundation soil is a working platform fill with a geosynthetic layer
placed 0.3 m above the foundation soil surface. The spacing of 0.7-
m thick column walls was 2.8 m (e.g., Forsman et al., 1999; Han
et al., 2007; Huang and Han, 2010). The area replacement ratio in
the numerical examples in this paper is 25%. The column walls are
founded on bedrock.
The location of an example unit cell in this study is shown in
Fig. 1. Fig. 2a shows a unit cell (Case 1) with typical boundary
conditions (e.g., fixed y-direction at bottom of the cell, and fixed x-
334 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343

Fig. 2. Unit cell (with half of DM column wall on each side of the cell) with: (a) fixed x-displacement boundary condition on both left and right sides of the cell (Case 1), (b) fixed x-
displacement boundary condition on left side and springs on right side of the cell (Case 2), and (c) springs on both left and right sides of the cell (Case 3).

displacement at both left and right sides of the cell). The width of Table 1
Material property values in FLAC.
the unit cell is 2.8 m with half of the DM column wall (i.e., 0.35-m
thick) on both right and left sides of the unit cell. To allow for lateral Properties Value
spreading of the embankment fill and foundation soil, the Case 1 Embankment and working platform fill
unit cell is first modified using a column of horizontal springs (with Unit weight, gef (kN/m3) 20
axial stiffness k) at the right side boundary of both the embankment Young's modulus, Eef (MPa) 40
Poisson's ratio, nef (e) 0.3
fill and foundation soil (Case 2 e Fig. 2b) and then both left and
Friction angle, fef (degree) 35
right side boundaries (Case 3 e Fig. 2c). The spring stiffness values Dilation angle, jef (degree) 5
can be different on left and right sides of the unit cell and different Cohesion, cef (kPa) 1
for the embankment fill and soft foundation soil. These cases are Soft foundation soil
also examined in the paper. Unit weight, gfs (kN/m3) 16
For a 5-m high embankment fill, the unit cell was modelled Young's modulus of soft clay 1, Efs,1 (MPa) 0.5
using 8400 zones and 112 cable elements for the geosynthetic Young's modulus of soft clay 2, Efs,2 (MPa) 1.0
Young's modulus of soft clay 3, Efs,3 (MPa) 1.5
reinforcement. A linear elastic model with MohreCoulomb failure
Young's modulus of soft clay 4, Efs,4 (MPa) 2.0
criterion was selected to model both the embankment fill and soft Young's modulus of soft clay 5, Efs,5 (MPa) 2.5
foundation soil. More advanced constitutive models for the soils Poisson's ratio, nfs () 0.25
and reinforcement were not used in this study in order to focus on Friction angle, ffs (degree) 15
the modified unit cell concept. Parameter values are summarized in Dilation angle, jfs (degree) 0
Cohesion, cfs (kPa) 5
Table 1 and were selected based on ranges reported in the literature
(Budhu, 2010; Huang and Han, 2010; Huang et al., 2009). The ma- DM column
Unit weight, gc (kN/m3) 20
terial properties for the working platform fill were taken to be the
Young's modulus, Ec (MPa) 150
same as those for the embankment fill (Huang and Han, 2010). A Poisson's ratio, nc () 0.3
depth-dependent Young's modulus for the foundation soil was
used as follows: Efs,1 ¼ 0.5 MPa for foundation soil depth between
0 and 2 m, Efs,2 ¼ 1 MPa between 2 and 4 m, Efs,3 ¼ 1.5 MPa between Table 2
4 and 6 m, Efs,4 ¼ 2 MPa between 6 and 8 m, and Efs,5 ¼ 2.5 MPa Interfaces and corresponding parameter values in FLAC.
between 8 and 10 m. A linear elastic model was used for the DM
Interface and parameters Value
column walls with Young's modulus of 150 MPa and Poisson's ratio
of 0.3 taken from Huang and Han (2010). Soft clay-DM column and soft clay-platform fill
Friction angle, fi,fs (degree) 10
The load transfer between the dissimilar materials within the
Dilation angle, ji,fs (degree) 0
unit cell was modelled using interfaces with the normal stiffness, Adhesion, ci,fs (kPa) 3.3
shear stiffness, friction angle, dilation angle and cohesion values Normal stiffness, kn,fs (MPa/m) 10
shown in Table 2 together with a strength reduction factor of Ri ¼ 2/ Shear stiffness, ks,fs (MPa/m) 1
3. The method to calculate the interface parameter values based on Platform fill-DM column
the surrounding soil properties is described by Yu et al. (2015b). For Friction angle, fi,pf (degree) 25
example, the friction angle (fi,fs) and cohesion (ci,fs) for the inter- Dilation angle, ji,pf (degree) 0
Cohesion, ci,pf (kPa) 0.67
face between the foundation soil and DM column is computed by
Normal stiffness, kn,pf (MPa/m) 100
applying a reduction factor Ri ¼ 2/3 to the friction angle (ffs) and Shear stiffness, ks,pf (MPa/m) 10
cohesion (cfs) of the foundation soil resulting in
 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
fi,fs ¼ tan1[Ri  tan(ffs)] ¼ tan1[2/3  tan(15 )] ¼ 10 and
ci,fs ¼ Ri  cfs ¼ 2/3  5.0 ¼ 3.3 kPa.
The lateral springs shown in Fig. 2b and c were modelled using
special interfaces with only normal stiffness values (the shear
stiffness values were set to zero) using the ‘glued’ condition in FLAC.
Thus for these interfaces, friction angle, dilation angle, and cohe-
sion values were not required. The concept of using horizontal
springs to simulate the soil lateral response in this paper is similar
to that used in soilestructure interaction design of supported
excavation walls and analyses for laterally loaded piles and verti-
cally loaded beams/plates (e.g., Selvadurai, 1979; Canadian
Geotechnical Society, 2006; Prendergast and Gavin, 2016). An
initial estimation of the spring stiffness k was based on an empirical
method from Barden (1962) using k ¼ 0.65Efs/B/(1-v2fs) where Efs
and vfs are the Young's modulus and Poisson's ratio of the foun-
dation soil, and B is the foundation width. A constant value of
k ¼ 1 MPa/m was used as the reference value in this paper and is
close to the average of k ¼ 0.35e1.73 MPa/m computed using
Efs ¼ 0.5e2.5 MPa, vfs ¼ 0.25 and foundation width B ¼ 1 m.
However, a sensitivity analysis using larger spring stiffness values
including infinite stiffness was also carried out.
The cable nodes for the geosynthetic reinforcement on the left
and right side of the unit cell were defined using grid numbers and
the remainder were defined using x- and y-coordinates. The geo-
synthetic reinforcement was assigned an axial stiffness of
J ¼ 2000 kN/m. This value falls within measured ranges reported by
Forsman et al. (1999) and is similar to the value used in numerical
models by Han et al. (2007) and Huang and Han (2010). The cable
elements have a cross-section area of Ag ¼ 0.002 m2 per metre
running length of column and perimeter of Pg ¼ 2 m (with an out-
of-plane width of 1 m). Thus the Young's modulus of each cable
element is calculated as Eg ¼ J/Ag ¼ 2000/0.002 ¼ 1000 MPa. For the
cable grout properties, the shear stiffness is
Ks,g ¼ Pg  ks,pf ¼ 2  10 ¼ 20 MN/m/m, the adhesion is
Cg ¼ Pg  ci,pf ¼ 2  0.67 ¼ 1.33 kPa, and the friction angle is
fg ¼ fi,pf ¼ 25 , where ks,pf, ci,pf and fi,pf are the shear stiffness,
cohesion and fiction angle of the interface between the platform fill
and DM column, respectively (Table 2). The shear stiffness was set
to be ks,pf ¼ 10 MPa/m (Yu et al., 2015a) and the adhesion was taken
as ci,pf ¼ Ri  cef ¼ 2/3  1.0 ¼ 0.67 kPa (where Ri is the strength
reduction factor introduced earlier and cef is the cohesion of
embankment fill/working platform fill). Table 3 summarizes all
properties related to the cable elements.
The modelling started by setting the initial ground stresses us-
ing K0 ¼ 1  sin(ffs) ¼ 0.741 for the soft foundation soil without the
DM column walls (e.g., by assuming that the DM column walls have
the same material properties as the soft foundation soil). To
simulate the DM column walls, the true wall material properties
were then applied and the model was solved to equilibrium. The
influence of installing the DM column wall on the stress distribu-
tion within the soft foundation soil was not considered in this pa-
per. The numerical grid displacements and velocities were then set
to zero. Small-strain mode was used in all simulations prior to
placement of the fill. Thereafter, during embankment construction,
Table 3
Parameter values of cable elements in FLAC.
Cable element parameters Value
Young's modulus, Eg (MPa) 1000
Cross-sectional area, Ag (m2/m) 2  103
Exposed perimeter, Pg (m) 2 intersections between the DM column walls and foundation soil
Grout stiffness, Ks,g (MN/m/m) 20
Grout cohesion, Cg (kN/m) 1.33
Grout frictional resistance, fg (degree) 25
335
small-strain or large-strain mode was selected to investigate the
influence of strain mode on numerical outcomes. To simulate the
embankment construction, 0.2-m thick embankment fill lifts were
activated in sequence and the model was solved to equilibrium for
each embankment fill lift until reaching the final embankment
height of 5 m. The activation of the second embankment fill lift was
accompanied by adding the cable elements to the model. It should
be noted that only idealized (fully drained) conditions for the
foundation soil were examined in this study; hence the generation
and dissipation of pore-water pressures that may occur in field
cases was not considered.
All numerical simulations were run using a desk top computer
with two Intel Xeon X5670 Six-Core 2.93 GHz central processor
units. The computer solved the numerical models in about 1e2 h
depending on the boundary conditions and the choice of large- and
small-strain modes. For example, it took longer to solve Case 3
models than Cases 2 and 1 with the same material properties and
large/small-strain mode, and models using large-strain mode took
longer to solve than those using small-strain mode for the same
boundary conditions and material properties.
4. Results
4.1. Unit cell for Case 1
Fig. 3a shows the calculated maximum vertical stresses at the
foundation surface of an embankment supported by DM column
walls and geosynthetic reinforcement using a unit cell with fixed
x-displacement at both right and left sides of the unit cell (Case 1;
Fig. 2a). The numerical results show that increasing the
embankment fill height increased the maximum vertical stress at
the foundation surface. For example, the maximum foundation
surface vertical stress was about 14 kPa for 1-m high embankment
fill and increased to about 22 kPa for 5-m high embankment fill
when using the large-strain mode. The differences in calculated
maximum vertical stresses at the foundation surface using both
large- and small-strain modes are negligible. However, for the
embankment fill higher than 0.5 m, the calculated maximum
vertical stresses at the foundation surface were much less than
those calculated using only soil self-weight (gefh) because of the
soil arching effect. The greater vertical stress at the foundation
surface with increasing embankment height resulted in larger
vertical settlement as shown in Fig. 3b. The maximum settlement
at the foundation surface was about 44 mm at 1-m high
embankment fill and increased to about 85 mm at 5-m high
embankment fill when using large-strain mode. The differences in
vertical settlements at the foundation surface using large- and
small-strain modes were less than 3 mm for the Case 1 unit cell
examined in this paper.
The load transfer from the embankment fill to the DM column
walls and soft foundation soil was modelled using interfaces. Fig. 4a
shows the interface normal stresses acting on the top of the DM
column wall and foundation surface at 5-m high embankment fill
for the Case 1 unit cell (Fig. 2a). The vertical stresses on the soft
foundation surface (i.e., on the centre part of the unit cell) were
much lower than those at the top of the DM column walls (i.e., left
and right sides of the unit cell) because the soil self-weight load
above the soft foundation is transferred to the DM column walls
due to the soil arching effect. For example, the vertical stresses at
the centres of the DM column walls (i.e., left and right edges of the
unit cell) were about 332 kPa and increased to about 388 kPa at the
when using large-strain mode and are much greater than those
computed based on soil self-weight alone (i.e., 100 kPa). The ver-
tical stresses on the foundation surface were generally less than
336 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
Fig. 3. Influence of large- and small-strain modes for Case 1 unit cell on (a) the
maximum vertical stresses at the foundation surface, and (b) the maximum settle-
ments at the foundation surface.
24 kPa which are much lower than stresses due to soil self-weight.
The choice of large- and small-strain modes had negligible influ-
ence on the vertical stresses at the foundation surface and the top
of the DM column walls using the Case 1 unit cell. The reason for
zero stress at the column edges is that the foundation surface
settled more than the bottom of the embankment backfill resulting
in separation between the foundation surface and backfill near the
edges of the columns.
The soil arching effect can also be appreciated from the vertical
stresses at the centre of the unit cell within the 5-m high
embankment fill shown in Fig. 4b. The calculated vertical stresses
within the embankment fill were equal to those from soil self-
weight over the first 2 m below the embankment surface. There-
after, the calculated vertical stress plots depart from the soil self-
weight line. This is because of the stress redistribution within the
embankment fill due to the soil arching mechanism which redirects
the embankment fill self-weight to the DM column walls and then
to the stiffer stratum at depth.
Fig. 5a shows the reinforcement tensile loads for the 5-m high
embankment fill and Case 1 unit cell (Fig. 2a) using both small-
and large-strain modes in FLAC. The use of large-strain mode
resulted in reinforcement tensile loads of about 8 kN/m at x ¼ 0
and 2.8 m, maximum values of 11 kN/m at x ¼ 0.35 and 2.45 m,
Fig. 4. Influence of large- and small-strain modes for Case 1 unit cell with 5-m high
embankment fill on (a) the interface normal stresses at the foundation surface
elevation, and (b) the vertical stress above the centre of the unit cell.
and minimum value of 1.6 kN/m at x ¼ 1.4 m. However, when
using the small-strain mode the reinforcement tensile loads were
much lower (e.g., the maximum reinforcement load was about
3 kN/m at x ¼ 0 and 2.8 m for the small-strain mode versus
11 kN/m at x ¼ 0.35 and 2.45 m noted above for the large-strain
mode). It should be noted that the reinforcement loads for Case 1
using small-strain mode were positive between x ¼ 0 and 2.8 m
and near zero between x ¼ 0.8 and 2.0 m. The influence of choice
of small- and large-strain mode in FLAC on the magnitude of
reinforcement tensile loads is judged to be significant for the
Case 1 unit cell conditions. The differences are because the
reinforcement tensile loads using the large-strain mode (i.e., with
mesh updating) were generated from both the horizontal and
vertical differential displacements between the soil and the
reinforcement, while those using small-strain mode (i.e., without
mesh updating) were generated only from the horizontal differ-
ential displacements between the soil and the reinforcement.
Thus the results from large-strain mode are judged to be more
accurate and representative of field performance than those us-
ing the small-strain mode especially for a soft foundation with
horizontal geosynthetic reinforcement; this is because the
tensioned membrane effect can be modelled only using the large-
strain mode.
 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
Fig. 5. Influence of large- and small-strain modes for Case 1 unit cell on (a) the
reinforcement tensile loads, and (b) the reinforcement vertical displacements for 5-m
high embankment fill.
The influence of the large- and small-strain modes on the
reinforcement vertical displacement is shown in Fig. 5b. The use of
the small-strain mode generally resulted in larger reinforcement
vertical displacements above the soft foundation compared to the
large-strain mode because the tensioned membrane effect was not
generated using small-strain mode.
4.2. Unit cell for Case 2
The results presented thus far are for the Case 1 unit cell (Fig. 2a)
with typical boundary conditions which do not account for lateral
spreading of embankment fill and soft foundation soil. This section
examines results for the first of two modified unit cells proposed in
this paper where the left side of the cell is fixed in x-direction and
the right side of the unit cell is supported by a series of horizontal
springs (Case 2 in Fig. 2b). To simplify the numerical analysis, only
linear elastic springs are considered.
Fig. 6a shows the maximum vertical stresses at the foundation
surface for the Case 2 unit cell with a spring stiffness value of
k ¼ 1 MPa/m using both large- and small-strain modes. For this
example the spring stiffness values for the embankment fill and the
soft foundation soil were the same. The modelling results show that
the maximum vertical stresses at the foundation surface were
337
Fig. 6. Influence of large- and small-strain modes for Case 2 unit cell with spring
stiffness k ¼ 1 MPa/m on (a) the maximum vertical stresses at the foundation surface,
and (b) the maximum settlements at the foundation surface.
higher than those for the Case 1 unit cell example in Fig. 3a when
the embankment fill is higher than 2 m and other conditions being
equal. For example, the use of large-strain mode for Case 2 unit cell
resulted in the maximum vertical stress at the foundation surface of
about 14 kPa at 1-m high embankment fill (same as 14 kPa in
Fig. 3a), and about 32 kPa at 5-m high embankment fill (compared
to 22 kPa in Fig. 3a). The maximum difference between the large-
and small-strain modes for the maximum vertical stresses at the
foundation surface with Case 2 unit cell was about 4 kPa (i.e.,
relative difference about 12%) at 5-m high embankment fill
compared to less than 1 kPa (i.e., relative difference about 2%) with
Case 1 unit cell. Fig. 6b shows the maximum settlements at the
foundation surface for Case 2 unit cell with a spring stiffness value
of k ¼ 1 MPa/m using both large- and small-strain modes. The re-
sults show that the use of horizontal springs at right side of the unit
cell (Fig. 2b) resulted in larger maximum settlements at the foun-
dation surface compared to those in Fig. 3b for embankment fills
higher than 1 m and other conditions being equal. The maximum
difference between the large- and small-strain modes for the ver-
tical settlement at 5-m high embankment fill was larger for Case 2
unit cell (about 23 mm) than for Case 1 unit cell (about 3 mm).
The normal stresses on the top of the DM column wall and
foundation surface for Case 2 unit cell with spring stiffness of
338 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343

k ¼ 1 MPa/m are shown in Fig. 7a. When using the large-strain

mode, the normal stress for Case 2 unit cell was about 309 kPa at
x ¼ 0 (compared to 332 kPa for Case 1 unit cell in Fig. 4a), about
373 kPa at x ¼ 0.35 m (compared to 388 kPa in Fig. 4a), less than
33 kPa between x ¼ 0.35 and 2.45 m (compared to less than 24 kPa
in Fig. 4a), about 480 kPa at x ¼ 2.45 m (compared to 388 kPa in
Fig. 4a), and about 164 kPa at x ¼ 2.8 m (compared to 332 kPa for
Case 1 unit cell in Fig. 4a). Fig. 7b indicates that lateral spreading of
embankment fill and foundation soil can also influence the vertical
stress distribution within the embankment fill. This can be seen by
comparing Fig. 7b for Case 2 unit cell with Fig. 4b for Case 1 unit
cell.
Fig. 8a shows the reinforcement tensile loads for Case 2 unit cell
with spring stiffness k ¼ 1 MPa/m. The results show that lateral
spreading of the embankment fill and foundation soil can have
significant influence on the reinforcement loads. For example, the
use of Case 2 unit cell with spring stiffness k ¼ 1 MPa/m and large-
stain mode resulted in the reinforcement load of about 31 kN/m at
x ¼ 0 (compared to 8 kN/m for Case 1 unit cell in Fig. 5a), about
40 kN/m at x ¼ 0.35 m (compared to 11 kN/m in Fig. 5a), about
15 kN/m at x ¼ 1.4 m (compared to 1.6 kN/m in Fig. 5a), about
38 kN/m at x ¼ 2.45 m (compared to 11 kN/m in Fig. 5a), and about
17 kN/m at x ¼ 2.8 m (compared to 8 kN/m in Fig. 5a). Using the

Fig. 8. Influence of large- and small-strain modes for Case 2 unit cell with spring
stiffness k ¼ 1 MPa/m on (a) the reinforcement tensile loads, and (b) the reinforcement
vertical displacements for 5-m high embankment fill.

small-strain mode, the maximum reinforcement load was 31 kN/m

at x ¼ 2.45 m in Fig. 8a compared to 3 kN/m at x ¼ 0 and 2.8 m for
Case 1 in Fig. 5a. It is the lateral spreading of embankment fill and
foundation soil that occurs using the Case 2 unit cell that results in
greater axial extension of the reinforcement and thus larger rein-
forcement loads for the Case 2 unit cell compared to the Case 1 unit
cell. The reinforcement vertical displacements above the soft
foundation for Case 2 with small-strain mode were generally larger
than those with large-strain mode (Fig. 8b).
The vertical stresses above and below the reinforcement are
shown in Fig. 9. The numerical results indicate that the net
vertical stresses (i.e., the differences between the vertical stresses
above and below the reinforcement) on the reinforcement be-
tween the DM column walls deviated from the uniform load
distribution assumed in BS8006 (2010) and the triangular load
distribution assumed in EBGEO (2011). This net vertical stress
distribution on the reinforcement was closer to the load distri-
bution proposed by Van Eekelen et al. (2013) using an inverted
triangular shape.

4.3. Influence of spring stiffness on reinforcement loads for Case 2

Fig. 7. Influence of large- and small-strain modes for Case 2 unit cell with spring
stiffness k ¼ 1 MPa/m on (a) the interface normal stresses at the foundation surface
elevation and 5-m high embankment fill, and (b) the vertical stress at the centre of unit The choice of magnitude of the spring stiffness can affect the
cell within the embankment fill. lateral displacements generated within the embankment fill and
 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343 339

displacement between the reinforcement and platform fill. For

example, the maximum reinforcement load was about 44 kN/m for
the foundation spring stiffness of kf ¼ 0.1 MPa/m and decreased to
about 40 and 35 kN/m for kf ¼ 1 and 10 MPa/m, respectively.

4.4. Unit cell for Case 3

In this section, the lateral spreading of the embankment fill

and foundation soil on both sides of the cell is considered using
springs on both sides of the unit cell (Case 3 unit cell in Fig. 2c).
Fig. 12 shows the maximum vertical stresses and settlements at
the foundation surface for Case 3 unit cell with spring stiffness
values of kl ¼ kr ¼ 1 MPa/m using both small- and large-strain
modes. The results show that the use of lateral springs on both
sides of the unit cell resulted in maximum vertical stresses of 42
and 36 kPa (compared to 36 and 32 kPa for Case 2 unit cell) and
maximum settlements of 192 and 151 mm (compared to 149 and
126 mm for Case 2 unit cell) for the 5-m high embankment fill
Fig. 9. Vertical stresses above and below the reinforcement for Case 2 unit cell with
simulations using the small- and large-strain modes,
large-strain mode.
respectively.
Fig. 13 shows the interface normal stresses at the foundation
foundation soil. These displacements will also depend on where the
surface and vertical stresses at the centre of the unit cell for the 5-m
Case 2 unit cell is taken along the full-width of the GRCS
high embankment fill and Case 3 unit cell with spring stiffness
embankment (e.g., near the centre line or toe of the embankment
values of kl ¼ kr ¼ 1 MPa/m. The maximum interface normal stress
due to different fill depths). The influence of different spring stiff-
was about 462 and 538 kPa at the intersection between the DM
ness values on the reinforcement loads with large-strain mode and
column wall and soft foundation soil for simulations in small- and
5-m high embankment fill is shown in Fig. 10 (both the embank-
large-stain mode, respectively (Fig. 13a). These values are greater
ment fill and the soft foundation soil were assumed to have the
than those for Case 2 unit cell (i.e., 421 and 480 kPa in Fig. 7a for the
same spring stiffness in this example). The numerical results show
small- and large-strain modes, respectively). The differences in
that the reinforcement loads decreased with increasing soil spring
vertical stresses at the centre of the unit cell between Case 3 and
stiffness because of less lateral spreading of embankment fill and
Case 2 unit cells can also be observed by comparing results from
foundation soil associated with higher spring stiffness. For
Fig. 13b with those from Fig. 7b.
example, the maximum reinforcement load for Case 2 unit cell was
Fig. 14a shows the reinforcement loads for Case 3 unit cell at 5-m
about 40 kN/m when using spring stiffness k ¼ 1 MPa/m, but
high embankment fill using both small- and large-strain modes and
decreased to about 19, 11 and 11 kN/m when increased to
spring stiffness values of kl ¼ kr ¼ 1 MPa/m. The maximum rein-
k ¼ 10 MPa, 100 MPa and infinite stiffness, respectively.
forcement loads for Case 3 unit cell were about 34 and 44 kN/m
The spring stiffness of the soft foundation soil can be different
using small- and large-strain modes, respectively. The maximum
from that of the embankment fill. The influence of different foun-
difference in reinforcement load using large- and small-strain
dation soil spring stiffness values on reinforcement loads below a 5-
modes was about 13 kN/m and occurred at the centre of the unit
m high embankment fill using large-strain mode is shown in Fig. 11.
cell. Fig. 14b shows the reinforcement vertical displacements for
The results show that increasing the spring stiffness value of the
Case 3 unit cell at 5-m high embankment fill where small-strain
foundation soil reduced the reinforcement loads because less
mode resulted in larger reinforcement vertical displacements
lateral spreading of the foundation soil resulted in less differential
compared to results using large-strain mode.

Fig. 10. Influence of soil spring stiffness values for Case 2 unit cell on the reinforce-
ment tensile loads with large-strain mode at 5-m high embankment fill using the same Fig. 11. Influence of spring stiffness values of the foundation soil for Case 2 unit cell on
spring stiffness values for the embankment fill and foundation soil. the reinforcement tensile loads with large-strain mode at 5-m high embankment fill.
340 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
Fig. 12. Influence of large- and small-strain modes for Case 3 unit cell with spring
stiffness values kl ¼ kr ¼ 1 MPa/m on (a) the maximum vertical stresses at the foun-
dation surface, and (b) the maximum settlements at the foundation surface.
4.5. Influence of spring stiffness on reinforcement loads for Case 3
The influence of different spring stiffness values at left side and
right of the unit cell (Case 3) on the reinforcement loads is shown in
Fig. 15a using large-strain mode. In this example the embankment
fill is 5 m high and the spring stiffness at the right side of the unit
cell is fixed at kr ¼ 1 MPa/m. Increasing the spring stiffness at left
side of the unit cell increased the reinforcement load at x ¼ 0 but
had negligible effect on the reinforcement load at x ¼ 2.8 m. Using
spring stiffness greater than kl ¼ 10 MPa/m at left side of the unit
cell had minor influence on the reinforcement loads between
x ¼ 0.35 and 2.8 m. Fig. 15b shows the influence of different spring
stiffness values at left side of the unit cell (Case 3) on the rein-
forcement vertical displacements. The numerical results show that
increasing the spring stiffness at left side of the unit cell decreased
the reinforcement vertical displacements. However, spring stiffness
greater than kl ¼ 10 MPa/m at left side of the unit cell had minor
influence on the reinforcement vertical displacements.
4.6. Influence of foundation modulus on reinforcement loads for
Case 3
Fig. 16 shows the influence of foundation modulus on the
reinforcement loads for Case 3 with large-strain mode. Increasing
Fig. 13. Influence of large- and small-strain modes for Case 3 unit cell with spring
stiffness values kl ¼ kr ¼ 1 MPa/m on (a) the interface normal stresses at the foun-
dation surface elevation and 5-m high embankment fill, and (b) the vertical stress at
the centre of unit cell within the embankment fill.
the foundation Young's modulus from 0.5 to 2.5 MPa decreased the
reinforcement loads because of less settlement of the foundation
surface (thus less reinforcement vertical displacements). However,
only increasing the Young's modulus for the foundation soil 2 m
below the foundation surface had negligible influence on the
reinforcement loads.
4.7. Influence of reinforcement stiffness on reinforcement loads for
Case 3
The influence of reinforcement stiffness on the reinforcement
loads for Case 3 with large-strain mode is shown in Fig. 17. The
numerical results show that the reinforcement loads increased
with increasing reinforcement stiffness. For example, the
maximum reinforcement load was about 30 kN/m when the rein-
forcement stiffness was 500 kN/m and increased to about 44 and
51 kN/m for J ¼ 2000 and 5000 kN/m, respectively.
4.8. Stress reduction ratio
The influence of different unit cells on load transfer can also be
examined using the stress reduction ratio (SRR) which is defined as
the ratio of the stress acting on the soft foundation surface between
 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343 341

Fig. 14. Influence of large- and small-strain modes for Case 3 unit cell with spring Fig. 15. Influence of spring stiffness values at left side of Case 3 unit cell on (a) the
stiffness values kl ¼ kr ¼ 1 MPa/m on (a) the reinforcement tensile loads, and (b) the reinforcement tensile loads, and (b) the reinforcement vertical displacements with
reinforcement vertical displacements for 5-m high embankment fill. large-strain mode at 5-m high embankment fill using the same spring stiffness values
for the embankment fill and foundation soil.

the DM columns to the embankment fill self-weight at the foun-

dation surface (no surcharge pressure was considered in this pa-
per). Fig. 18 shows the calculated SRR for the three different unit
cell cases with large-strain mode. The results show that increasing
the embankment fill thickness decreased the SRR when other
conditions were equal. For the same embankment height, the SRR
was lower for Case 1 than for Case 2 which was also lower than that
for Case 3. For example, for 5-m high embankment the SRR was
about 0.218 for Case 1 and increased to about 0.327 for Case 2 and
0.360 for Case 3.

5. Conclusions

An effective technique to improve the performance of em-

bankments over soft foundations is to use geosynthetic-
reinforced and column-supported (GRCS) embankments. Both
full-width model and unit cell approaches have been used by
researchers to study the behaviour of these structures. However,
the typical unit cell with fixed horizontal displacements at side
boundaries is unable to capture the lateral spreading of the
embankment fill and foundation soil. Khabbazian et al. (2015)
showed that this can lead to large differences in numerical out- Fig. 16. Influence of foundation modulus on the reinforcement load for Case 3 with
comes between a full-width model and the simulation of the large-strain mode.
342 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
Fig. 17. Influence of reinforcement stiffness on the reinforcement load for Case 3 with
large-strain mode.
1.0
Large-strain mode
Stress reduction ratio, SRR (-)
0.8
0.6
0.4
Unit cell boundary conditions
0.2 Case 1: Fixed at left and right sides
Case 2: Fixed at left side and spring k = 1 MPa/m at right side
Case 3: Spring kl = kr = 1 MPa/m at left and right sides
0.0
0 1 2 3
Embankment fill height, h (m)
Fig. 18. Calculated stress reduction ratio for Cases 1, 2, and 3 with large-strain mode.
same structures using the typical unit cell approach. The current
investigation presents a modified unit cell approach to simulate
the lateral spreading of the embankment fill and foundation soil
using horizontal linear elastic springs at the vertical boundaries of
the unit cell. Program FLAC is used to examine the influence of
large- and small-strain modes on conventional and modified unit
cell behaviour. Based on the numerical results using the three
different unit cells examined in this paper, the following con-
clusions can be summarized:
 For a typical unit cell with fixed lateral horizontal displacements
at the side boundaries (Case 1 unit cell in Fig. 2a), the use of
large- and small-strain modes has minor influence on the
maximum vertical stresses and settlements, and interface
normal stresses at the foundation surface. However, when the
numerical model was run in small-strain mode, the calculated
reinforcement loads were much lower compared to large-strain
mode.
 For the modified unit cells with lateral springs (Case 2 unit cell
in Fig. 2b and Case 3 unit cell in Fig. 2c), the lateral spreading of
the embankment fill and foundation soil increased the
maximum vertical stresses and settlements at the foundation
surface, and the reinforcement loads when compared to
numerical results using fixed side boundaries (Case 1 unit cell)
and other conditions being equal.
 The magnitude of calculated reinforcement loads for Cases 2 and
3 unit cell examples depended on the choice of lateral spring
stiffness value and whether or not the same stiffness value was
assigned to the embankment fill and foundation soil.
This paper has demonstrated that lateral spreading of the
embankment fill and foundation soil can be modelled using a
modified unit cell with a series of lateral springs supporting one or
both side boundaries. However, numerical outcomes depended on
the choice of small-strain or large-strain mode when using pro-
gram FLAC. The computed reinforcement loads were particularly
sensitive to this choice.
The influence of equivalent large-strain mode in FEM programs
(i.e., mesh updating option) was not investigated in this study.
However, similar sensitivity to choice of fixed or updated mesh
generation on reinforcement loads using these codes is expected.
Acknowledgement
The work reported in this paper was supported by a grant from
the Natural Sciences and Engineering Research Council of Canada
Case 1
Case 2 (NSERC; Grant number: RGPAS-446148-2013).
Case 3
Notation
Basic SI units are given in parentheses
K0 coefficient of lateral earth pressure at rest
(dimensionless)
k spring stiffness at right side of unit cell for Case 2 (Pa/m)
kb spring stiffness for embankment fill at right side of unit
cell for Case 2 (Pa/m)
kf spring stiffness for foundation soil at right side of unit cell
for Case 2 (Pa/m)
kl spring stiffness at left side of unit cell for Case 3 (Pa/m)
kr spring stiffness at right side of unit cell for Case 3 (Pa/m)
4 5 h thickness of embankment fill (m)
J axial stiffness of geosynthetic (N/m)
Ri strength reduction factor (dimensionless)
x, y Cartesian co-ordinates with origin at base of unit cell (m)
Other notations appear in the tables.
References
Ali, K., Shahu, J.T., Sharma, K.G., 2014. Model tests on single and groups of stone
columns with different geosynthetic reinforcement arrangement. Geosynth. Int.
21 (2), 103e118.
Barden, L., 1962. Distribution of contact pressure under foundations. Geotechnique
12 (3), 181e198.
Bergado, D.T., Ruenkrairergsa, T., Taesiri, Y., Balasubramaniam, A.S., 1999. Deep soil
mixing used to reduce embankment settlement. Ground Improv. 3, 145e162.
Bhasi, A., Rajagopal, K., 2015a. Geosynthetic-reinforced piled embankments: com-
parison of numerical and analytical methods. ASCE Int. J. Geomech. 15 (5),
04014074.
Bhasi, A., Rajagopal, K., 2015b. Numerical study of basal reinforced embankments
supported on floating/end bearing piles considering pileesoil interaction.
Geotext. Geomembr. 43 (6), 524e536.
Blanc, M., Thorel, L., Girout, R., Almeida, M., 2014. Geosynthetic reinforcement of a
granular load transfer platform above rigid inclusions: comparison between
centrifuge testing and analytical modelling. Geosynth. Int. 21 (1), 37e52.
Borges, J.L., Marques, D.O., 2011. Geosynthetic-reinforced and jet grout column-
supported embankments on soft soils: numerical analysis and parametric
study. Comput. Geotechnics 38, 883e896.
Briançon, L., Simon, B., 2012. Performance of pile-supported embankment over soft
soil: full-scale experiment. ASCE J. Geotech. Geoenviron. Eng. 138 (4), 551e561.
Bruce, M.E.C., Berg, R.R., Collin, J.G., Filz, G.M., Terashi, M., Yang, D.S., 2013. Federal
Highway Administration Design Manual: Deep Mixing for Embankment and
Foundation Support. U.S. Department of Transportation, USA, 248 pp.
BS8006, 2010. Code of Practice for Strengthened/Reinforced Soil and Other Fills.
British Standards Institution (BSI), Milton Keynes, UK.
 Y. Yu et al. / Geotextiles and Geomembranes 44 (2016) 332e343
Budhu, M., 2010. Soil Mechanics and Foundations, third ed. John Wiley & Sons, Inc.
Canadian Geotechnical Society, 2006. Canadian Foundation Engineering Manual,
fourth ed. BiTech Publishers Ltd., Richmond, B.C.
Chai, J.C., Shrestha, S., Hino, T., Ding, W.Q., Kamo, Y., Carter, J., 2015. 2D and 3D
analyses of an embankment on clay improved by soilecement columns. Com-
put. Geotechnics 68, 28e37.
EBGEO, 2011. Recommendations for Design and Analysis of Earth Structures Using
Geosynthetic Reinforcements e EBGEO. German Geotechnical Society, ISBN
978-3-433-02983-1. Digital Version in English, ISBN: 978-3-433-60093-1.
Forsman, J., Honkala, A., Smura, M., 1999. Hertsby case: a column stabilized and
geotextile reinforced road embankment on soft subsoil. In: Dry Mix Method for
Deep Soil Stabilization. Balkema, Rotterdam, The Netherlands, pp. 263e268.
Gu, M., Zhao, M., Zhang, L., Han, J., 2016. Effects of geogrid encasement on lateral
and vertical deformations of stone columns in model tests. Geosynth. Int.
http://dx.doi.org/10.1680/jgein.15.0003.
Han, J., Gabr, M.A., 2002. Numerical analysis of geosynthetic reinforced and pile-
supported earth platforms over soft soil. ASCE J. Geotech. Geoenviron. Eng.
128 (1), 44e53.
Han, J., Oztoprak, S., Parsons, R.L., Huang, J., 2007. Numerical analysis of foundation
columns to support widening of embankments. Comput. Geotechnics 34 (6),
435e448.
Hewlett, W.J., Randolph, M.F., 1988. Analysis of piled embankments. Ground Eng. 21
(3), 12e18.
Hosseinpour, I., Almeida, M.S.S., Riccio, M., 2015. Full-scale load test and finite-
element analysis of soft ground improved by geotextile-encased granular col-
umns. Geosynth. Int. 22 (6), 428e438.
Huang, J., Han, J., 2009. 3D coupled mechanical and hydraulic modeling of a
geosynthetic-reinforced deep mixed column-supported embankment. Geotext.
Geomembr. 27, 272e280.
Huang, J., Han, J., 2010. Two-dimensional parametric study of geosynthetic-
reinforced column-supported embankments by coupled hydraulic and me-
chanical modeling. Comput. Geotechnics 37, 638e648.
Huang, J., Han, J., Oztoprak, S., 2009. Coupled mechanical and hydraulic modeling of
geosynthetic-reinforced column-supported embankments. ASCE J. Geotech.
Geoenviron. Eng. 135, 1011e1021.
Itasca, 2011. FLAC: Fast Lagrangian Analysis of Continua. Version 7.0 (computer
program). Itasca Consulting Group, Inc., Minneapolis, MN.
Kempfert, H.G., Go € bel, C., Alexiew, D., Heitz, C., 2004. German recommendations for
the reinforced embankments on pile-similar elements. In: Proceedings of the
3rd European Conference on Geosynthetics, Munich, Germany, pp. 279e284.
Khabbazian, M., Kaliakin, V.N., Meehan, C.L., 2015. Column supported embankments
with geosynthetic encased columns: validity of the unit cell concept. Geotech.
Geol. Eng. 33, 425e442.
Lai, Y.P., Bergado, D.T., Lorenzo, G.A., Duangchan, T., 2006. Full-scale reinforced
embankment on deep jet mixing improved ground. Ground Improv. 10 (4),
153e164.
Liu, H.L., Ng, C.W.W., Fei, K., 2007. Performance of geogrid-reinforced and pile-
supported highway embankment over soft clay: case study. ASCE J. Geotech.
Geoenviron. Eng. 133, 1483e1493.
343
Liu, K.W., Rowe, R.K., 2015. Numerical study of the effects of geosynthetic rein-
forcement viscosity on behaviour of embankments supported by deep-mixing
method columns. Geotext. Geomembr. 43 (6), 567e578.
Liu, K.W., Rowe, R.K., 2016. Performance of reinforced, DMM column-supported
embankment considering reinforcement viscosity and subsoil's decreasing
hydraulic conductivity. Comput. Geotechnics 71, 147e158.
Love, J., Milligan, G., 2003. Design methods for basally reinforced pile-supported
embankments over soft ground. Ground Eng. 39e43.
Low, B.K., Tang, S.K., Choa, V., 1994. Arching in piled embankments. ASCE J. Geotech.
Eng. 120 (11), 1917e1938.
Nunez, M.A., Briancon, L., Dias, D., 2013. Analyses of a pile-supported embankment
over soft clay: full-scale experiment, analytical and numerical approaches. Eng.
Geol. 153, 53e67.
Prendergast, L.J., Gavin, K., 2016. A comparison of initial stiffness formulations for
small-strain soil-pile dynamic Winkler modelling. Soil Dyn. Earthq. Eng. 81,
27e41.
Rowe, R.K., Liu, K.W., 2015. 3D finite element modeling of a full-scale geosynthetic-
reinforced, pile-supported embankment. Can. Geotech. J. 52 (12), 2041e2054.
Selvadurai, A.P.S., 1979. Elastic Analysis of Soilefoundation Interaction. Elsevier,
Amsterdam.
Smith, M., Filz, G., 2007. Axisymmetric numerical modeling of a unit cell in
geosynthetic-reinforced, column-supported embankments. Geosynth. Int. 14
(1), 13e22.
Van Eekelen, S.J.M., Bezuijen, A., Van Tol, A.F., 2011. Analysis and modification of the
British Standard BS8006 for the design of piled embankments. Geotext. Geo-
membr. 29 (3), 345e359.
Van Eekelen, S.J.M., Bezuijen, A., Van Tol, A.F., 2013. An analytical model for arching
in piled embankments. Geotext. Geomembr. 39, 78e102.
Van Eekelen, S.J.M., Bezuijen, A., Van Tol, A.F., 2015. Validation of analytical models
for the design of basal reinforced piled embankments. Geotext. Geomembr. 43
(1), 56e81.
Xing, H., Zhang, Z., Liu, H., Wei, H., 2014. Large-scale tests of pile-supported earth
platform with and without geogrid. Geotext. Geomembr. 42 (6), 586e598.
Yapage, N.N.S., Liyanapathirana, D.S., 2014. A parametric study of geosynthetic-
reinforced column-supported embankments. Geosynth. Int. 21 (3),
213e232.
Yoo, C., 2010. Performance of geosynthetic-encased stone columns in embankment
construction: numerical investigation. ASCE J. Geotech. Geoenviron. Eng. 136
(8), 1148e1160.
Yu, Y., Bathurst, R.J., Miyata, Y., 2015a. Numerical analysis of a mechanically stabi-
lized earth wall reinforced with steel strips. Soils Found. 55 (3), 536e547.
Yu, Y., Damians, I.P., Bathurst, R.J., 2015b. Influence of choice of FLAC and PLAXIS
interface models on reinforced soilestructure interactions. Comput. Geo-
technics 65, 164e174.
Zhang, J., Zheng, J.J., Chen, B.G., Yin, J.H., 2013. Coupled mechanical and hydraulic
modeling of a geosynthetic-reinforced and pile-supported embankment.
Comput. Geotechnics 52, 28e37.
Zhuang, Y., Ellis, E.A., Yu, H.S., 2010. Plane strain FE analysis of arching in a piled
embankment. Ground Improv. 163 (4), 207e215.