You are on page 1of 20

Engineering Structures 242 (2021) 112337

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Geosynthetic-reinforced pile-supported embankments − 3D discrete


numerical analyses of the interaction and mobilization mechanisms
Tuan A. Pham a, b, *, Quoc-Anh Tran c, Pascal Villard d, Daniel Dias d, e, f
a
School of Energy, Geoscience, Infrastructure and Society, Heriot-Watt University, Edinburgh, UK
b
Department of Civil Engineering, The University of Tokyo, Hongo-Bunkyo, Tokyo, Japan
c
Department of Civil and Environmental Engineering, Norwegian University of Science and Technology, Trondheim, Norway
d
Univ. Grenoble Alpes, CNRS, Grenoble INP, 3SR, 38000 Grenoble, France
e
Antea Group, Antony, France
f
School of Automotive and Transportation Engineering, Hefei University of Technology, Hefei, China

A R T I C L E I N F O A B S T R A C T

Keywords: Three-dimensional numerical analyses using the discrete element method are conducted to investigate several
Piled embankment fundamental aspects related to soil-structure interaction and mobilization mechanisms in the geosynthetic-
Geosynthetic reinforced and pile-supported embankments. The contributions of the soil arching, tensioned membrane ef­
Numerical analysis
fect, friction interaction, subsoil support, and punching failure are investigated. The results indicated that the
Soil arching
inclusion of the geosynthetic enhances the stress transfer from the subsoil to piles due to the tensioned membrane
Soil-structure interaction
Load transfer mechanism action, and the stress distribution is more uniform as compared to piled embankment without geosynthetic.
Discrete element model However, the tension distribution in geosynthetic is not uniform and the maximum tension occurs near the pile
Design methods edge. Numerical results also proved that the subsoil provides substantial support and reduces the reinforcement
tension while shear stresses are mobilized along the upper and lower sides of soil-geosynthetic interfaces. These
mechanisms should be considered in theoretical models to produce a more realistic approach. Finally, ten
available design methods are reviewed and compared to the numerical results to assess the performance of
analytical models. The results showed that the design method of Pham, CUR 226 design guideline, and EBGEO
design standard agree well with the numerical results and are generally better than the results of all other
methods.

presented a semi-spherical arching model to describe the load transfer


mechanism based on their three-dimensional model tests. This model
1. Introduction was adopted into the BS 8006 [12]. Low et al. [13] improved the method
of Hewlett and Randolph by considering a body force into the plane-
Geosynthetic reinforced and pile-supported (GRPS) embankments strain differential equation of equilibrium. Miki and Nozu [14] pre­
are increasingly used in highway construction on soft soil to reduce sented a model to predict the arching soil in GRPS embankments based
settlements, minimize soil yield above pile caps, and enhance the load on laboratory model test results. The model considered the arching
transfer efficiency (Fig. 1). Many case histories were reported in the shape change by using a variable arching angle in calculations. It was
literature over recent years [1–8]. implemented in PWRC [15]. Kempfert et al. [16] presented a new design
Because the use of the GRPS solution is increasing rapidly, some method derived from 1:3 laboratory models of piled embankment
researchers proposed analytical models to investigate the GRPS problems. Before estimating the reinforcement tension necessary to
embankment behavior [9–18]. Terzaghi [9] presented an arching carry the embankment load, the method firstly estimates the soft soil
analysis to account for the three-dimensional shape of the settling soil load magnitude with no reinforcement included. This analysis method
mass in the embankment above the foundation soil between columns was adopted into EBGEO [17]. Van Eekelen et al. [18] presented a new
and proposed a friction arching model. Guido et al. [10] showed that the model, namely, the concentric arches model. This model consists of a set
load spread angle through the geogrid-reinforced cohesionless soil is at of concentric hemispheres and arches. It is used by CUR 226 [19]. Pham
least equal to 45 degrees from the vertical. Hewlett and Randolph [11]

* Corresponding author.
E-mail address: ta.pham@hw.ac.uk (T.A. Pham).

https://doi.org/10.1016/j.engstruct.2021.112337
Received 28 November 2020; Received in revised form 14 March 2021; Accepted 30 March 2021
Available online 30 May 2021
0141-0296/© 2021 Elsevier Ltd. All rights reserved.
T.A. Pham et al. Engineering Structures 242 (2021) 112337

Nomenclature (kN)
Ps load part resting on the geosynthetic and subsoil (kN),
A tributary area of one column cap (m2) q surcharge (kPa)
As area covered by the soft ground (m2) SRRa stress reduction ratio induced by arching (dimensionless)
a equivalent width of square pile cap (m) s center-to-center pile spacing (m)
D thickness of soft subsoil (m) sx , s y center-to-center pile spacing in directions x and y (m)
dc diameter of circular pile cap (m) s′ clear spacing (m)
E total efficacy (dimensionless) sd diagonal pile spacing (m)
Ea efficacy component from arching effect (dimensionless) T maximum tension in geosynthetic (kN/m)
Em efficacy component from the membrane effect y maximum deflection of geosynthetic (m)
(dimensionless) γ unit weight of embankment soils (kN/m3)
Es modulus of soft subsoil (kPa) σc total vertical stress acting on the pile cap by both arching
FGRsquare total vertical load exerted by the 3D hemisphere on their and membrane effect (kPa)
square subsurface (kN) σ as vertical stress acting on the geosynthetic and subsoil by
FGRstrip total vertical load exerted by the 2D arches on the GR strips arching effect (kPa)
(kN) θ arching angle (degree)
hg arching height (m) ϕ friction angle of embankment soils (degree)
H height of embankment (m) ε strain of geosynthetic (dimensionless)
JGR tensile stiffness of geosynthetic (kN/m) εf strain of geosynthetic due to skin friction (dimensionless)
KP passive earth pressure coefficient (dimensionless) δ deflected angle (degree)
K earth pressure coefficient at rest (dimensionless)
Pc load component distributed on pile cap by arching effect

[20–22] presented a simplified method for GRPS embankment analyses the precise mechanism by which the load is transferred to piles, how­
considering a combination of the fill soils arching theory, the tensioned ever, remains poorly understood.
geosynthetic membrane theory, and soil-structure interaction. However, In the past decade, various modeling studies were reported to
all design procedures still contain some simplifying assumptions. improve the understanding of the behavior of geosynthetic-reinforced
The geosynthetic-reinforced and pile-supported embankment are and pile-supported embankments. When compared with small-scale,
integrated systems combining various elements such as embankment full-scale, or large-scale experimental studies, numerical studies are
soils, geosynthetic, piles, platform, and, subsoil. So, the response of economic, time-efficient, and flexible. They are also a very powerful tool
GRPS embankments is affected by the interactions between these ele­ to investigate and get insights into the behavior of GRPS embankments.
ments. Pham [21] noted that the load transfer mechanism in GRPS is a The first approach often used is the finite element method (FEM).
combination of the following five phenomena: (1) stress transfer from Numerous numerical studies were conducted with the finite element
the soft soil to the piles due to their stiffness difference or so-called soil method [23–40]. Ariyarathne and Liyanapathirana [23] conducted nu­
arching, (2) geosynthetic stretching to transfer the load from the subsoil merical modeling with the finite element program ABAQUS under three-
to the pile cap, so-called tensioned geosynthetic membrane effect, (3) dimensional and two-dimensional plane-strain conditions. The main
support of the subsoil that reduces the load carried by the geosynthetic, finding was that the formation of soil arches is visible in the three-
(4) frictional interaction along with the upper and lower sides of dimensional model as a dome which is supported by piles, and in the
geosynthetic-soil interfaces, and (5) interaction between the piles and two-dimensional model as arches that span along the pile walls. Lo et al.
soils. The complexity of the load transfer mechanism in GRPS em­ [24] performed a numerical study of a GRPS embankment considering
bankments is shown in Fig. 2. Although this technique is widely used, the time-dependent interaction between a stone column and the

Fig. 1. Geosynthetic reinforced and pile-supported (GRPS) embankment.

2
T.A. Pham et al. Engineering Structures 242 (2021) 112337

a Uniform surcharge, q b

Embankment

r
+

H
a
s/2
s T T

Pile cap Pile cap Pile cap Pile cap


Soft subsoil
deflected
Pile geosynthetic
Pile Pile

Vertical stress acting on upper


c surface of reinforcement
d

+ + T T
Geosynthetic reinforcement

top
Pile cap Pile cap
bottom

Pile
Ground reaction

(a) soil arching, (b) tensioned membrane effect, (c) subsoil support, (d) friction interaction
Fig. 2. Sketch of load transfer mechanism in GRPS embankment.

surrounding soft clay. They found that the stone columns’ role in sup­ finite element method with an elastoplastic constitutive model. Jiang
porting the fill loading evolves with time. Zhuang and Ellis [25] con­ et al. [33] presented a numerical analysis of a monitored pile–slab-
ducted a finite element analysis of a piled embankment considering the supported embankment for a high-speed railway. A coupled two-
subsoil contribution. As expected, the subsoil support effect on the dimensional mechanical and hydraulic numerical model was used.
reinforcement tension reduction was showed. Wu et al. [26] performed a Meena et al. [34] studied the soil arching phenomenon in a granular
three-dimensional numerical model to analyze the performance of a embankment subjected to train loadings using two-dimensional finite
piled embankment using Prefabricated high-strength concrete (PHC) element approaches. A 2D plane strain idealization was used to convert a
piles. Ariyarathne et al. [27] suggested that the geosynthetic layer ten­ real three-dimensional case into 2D. Yoo [35] presented a numerical
sion was close to the 3D one when truss elements were used. Therefore, investigation of geosynthetic-encased stone columns performance. A
the equivalent area method used with the truss element for the geo­ three-dimensional finite-element model was employed to carry out a
synthetic reinforcement layer can be considered as the most appropriate parametric study on some governing factors. Potts and Zdravkovic [36]
2D idealization method. Borges and Marques [28] conducted a numer­ presented finite-element analyses of a geosynthetic-reinforced pile-
ical analysis to study the time-dependent behavior of geosynthetic- supported embankment. An updated Lagrangian large displacement
reinforced and jet grout column-supported embankments on soft soils. formulation was utilized to mobilize the geosynthetic membrane above
The cylindrical unit cell formulation was used and, the numerical model the void.
incorporated the consolidation theory and critical state constitutive re­ The second approach usually used is the Finite different method
lations. Jamsawang et al. [29] presented a three-dimensional numerical (FDM). The finite element method has the same features as the finite
analysis of a column-supported embankment for a highway, in which the difference one as both are methods to solve differential equations. A
consolidation behavior was simulated by coupling a three-dimensional three-dimensional numerical analysis was done by Das and Deb [41] to
mechanical and hydraulic model. Oliveira et al. [30] analyzed the investigate the time-dependent behavior of an end-bearing stone col­
behavior of an embankment constructed on deep mixing columns (DMC) umn–supported embankment. Indraratna et al. [42] conducted a nu­
using a couple soil–water formulations. Due to the arching effect, the merical analysis with a finite-difference model to analyze the behavior
load applied by the embankment was concentrated on the DMC, the of a stone column–supported embankment, adopting the free strain
effective stresses increase on the soil being negligible. Rowe and Liu [31] approach, and considering soil arching, clogging of the stone column.
conducted a three-dimensional finite element modeling of a full-scale Jenck et al. [43] performed a 3D finite-difference analysis of a piled
geosynthetic-reinforced, in which a fully 3D coupled model was used embankment. However, no relative displacements were considered at
with a consolidation analysis and the Drucker–Prager cap material the interface between the piles and the soft soil, which did not permit to
behavior. Yadav et al. [32] presented a comparative study on floating study of the positive and negative skin friction along with the piles.
and fixed sand compaction piles (SCPs), in which the settlement and Huang et al. [44] presented three-dimensional numerical analyses of a
pore water pressure behaviors were studied using a soil–water coupled column-supported embankment using the finite difference method.

3
T.A. Pham et al. Engineering Structures 242 (2021) 112337

Numerical calculations were performed for a unit cell and a half- investigated the soil-arching phenomenon between piles by adopting
embankment model for the lateral spreading case. Yu and Bathurst two different numerical methods: the discrete-element method and the
[45] simulated a GRPS embankment using a full-width model consid­ finite-difference method. A finding was that for high embankments
ering small- and large-strain modes. Numerical analyses showed that the subjected to small deformations, the FDM numerical results were
reinforcement tensile loads from the modified unit cell model are in equivalent to those of the DEM. Indraratna et al. [56], and Gholaminejad
good agreement with those from the full-width model. Han and Garb et al. [57] presented a coupled model of a discrete and finite difference
[46] conducted a numerical study by using a finite difference method to method to study the deformation of a single stone column. This model is
investigate the pile-soil-geosynthetic interactions. interesting as it combines the advantages of the discrete and continuum
The finite element method or finite difference method achieved great methods and facilitates the exchange of interface contact forces and
success in obtaining the soil arching in piled-supported embankments. displacements between the discrete and continuum boundaries. The
Nevertheless, FEM or FDM methods are based on the framework of contact forces and moments at the interface were obtained from the
continuum mechanics. Therefore, DEM modeling is an alternative discrete domain and transferred to the continuum zone as nodal forces,
method to numerical simulations of granular soils as it can consider the while the interface displacements were obtained from the continuum
interaction of the particles, allowing the soil response to be analyzed at boundaries and transferred to the discrete zone. However, the number of
the particle level in detail. FEM of FDM cannot be employed to inves­ numerical studies using the coupled model was still limited. Most of the
tigate the particle shape effect. The discrete element method (DEM) is an presented studies focused on the stress transfer between the soil and the
alternative for the investigation of fundamental soil behavior. It has piles while the interaction between piles, subsoil, and the geosynthetic,
made significant contributions to the study of granular materials over the subsoil support, frictional behavior, and tensioned geosynthetic
the last decades. Several researchers used DEM to explore the behavior membrane mechanism was not been well investigated. They should be
of soil arching in pile-supported embankments [47–53]. Badakhshan considered for a better understanding of the complex behavior and load
et al. [47] presented 3D discrete element simulations to explore the transfer mechanism in geosynthetic-reinforced pile-supported
arching mechanism in piled embankments. Resulted showed that embankments.
considering the arching process kinematic, the discrete nature of the Analytical methods are available. They usually allow designing
utilized numerical model for soil allows a good description of the load geosynthetic-reinforced pile-supported embankments using two sepa­
recovery evolution and contact forces within the granular material. rate calculation steps. In step 1, the stress redistribution with a load on
Chen et al. [48] conducted numerical simulations using DEM to inves­ the reinforcement is calculated, and in step 2, the membrane action of
tigate the particle shape effects on soil arching. The particle shape can the reinforcement is defined. No interaction of the soil and the tensioned
significantly affect the development of soil arching, while it less has an membrane is considered. So, a better understanding of the complete
impact on the ultimate soil arching. Nevertheless, the particle shape has behavior is necessary. In this paper, a DEM numerical study is conducted
a slight influence on the ultimate soil arching. Lai et al. [49] conducted showing the fundamental aspects related to the behavior and interaction
discrete element method models for trapdoor tests in a total of 131 cases between the soil-structure elements of geosynthetic-reinforced pile-
to present the soil arching effect, stress state, and deformation behaviors supported embankments. The contributions of the soil arching,
of piled embankments. In-depth discussions on the composition and tensioned membrane effect, friction interaction, subsoil, and punching
mobilization mechanisms of soil the arching were analyzed. Rui et al. failure are investigated. Results are presented as a discussion related to
[50] simulated a series of 2D multiple trapdoor systems using the stress, settlement, efficacy, and tensile forces. The analysis is extended
discrete element method. The contact forces between the particles with a comparison between the available design methods for different
formed a strong chain force network enveloping a triangular and tower- values of embankment heights. Thus, the results of this study are
shaped deformation region. Force chains within the deformation region intended to provide a better understanding of the interaction behavior of
supported the load coming from outside of this region. Han et al. [51] systems composed of an embankment, a geosynthetic, soft soil, and piles
also conducted numerical simulations of unreinforced and reinforced and to bring insight into the mobilization mechanisms in such systems.
pile-supported embankments using the discrete element method. The
mechanical behavior of a platform over piles in soft soil was investigated 2. Load transfer mechanisms
through DEM modeling and a numerical continuum model by Jenck
et al. [52]. They summarized that the particle contact friction coefficient Under the embankment weight and surcharge, the geosynthetic
and the particle contact bond of the fill had a preponderant influence on
the pile load transfer. Chareyre and Villard [53] investigated the
behavior of soil-inclusion systems using the discrete element method.
The movement of the inclusion was considered as a dynamic process and
computed step by step in the same way as in the DEM.
It should be noted that the discrete element method based on a
Newtonian approach which uses rigid bodies, can consider the interac­
tion between soil particles, thus allowing transfer load mechanisms of a
particle to another. This model is therefore well adapted to simulate the
behavior of granular materials. However, in DEM modeling it is not so
easy to simulate the tensioned geosynthetic membrane mechanism and
the soil-structure interface behavior. In contrast, the continuous
approach is effective in describing the geosynthetic sheet structure and
its interface behavior. Thus, a coupled model between a continuous
approach and a discrete approach has to be considered to overcome the
DEM numerical problems. Le Hello and Villard [54] used a numerical
model based on the coupling between finite and discrete element
methods to simulate the behavior of GRPS embankments. The compar­
ison with experimental results based on a true scale experiment showed
that the numerical model can consider the geosynthetic sheet membrane
behavior considering a finite element and the mechanisms acting in the
soil embankment by the use of discrete elements. Anh Tran et al. [55] Fig. 3. Interaction mechanism between pile-soil-geosynthetic.

4
T.A. Pham et al. Engineering Structures 242 (2021) 112337

reinforcement deflects. The deflection has two effects as shown in Fig. 3. spacing of 3 m. The width of the square pile cap is 0.6 m which yields an
The first effect is the soil layer bending which generates arching inside improvement area ratio equal to 4%, as shown in Fig. 5.
the soil and induces a redistribution of stresses on the stiffer elements. The embankment is modeled by discrete elements. The number of
This phenomenon is known as “soil arching”. Another effect is the particles varies within the range of 16,000–54,000 which corresponds
geosynthetic material stretching which induces mobilization of a respectively to a minimum embankment height of 0.75 m and a
portion of its strength. It is known as the “tensioned membrane effect”. maximum embankment height of 3 m. Kozicki et al. [68] showed that
In any case, the settlement between piles is larger than on the top of clustering particles can better replicate the mechanical behavior of
them, this generates tensile forces in the horizontal geosynthetic rein­ granular materials. Therefore, each particle is created by two clumped
forcement. Therefore, the geosynthetic layer is clamped like a mem­ spheres (distance between the centroid of two clumped spheres equal to
brane over the piles. The deflection of the geosynthetic layer will then 0.8D where D represents the single sphere diameter) with sphere di­
increase the load transference to the piles. Only small residual stresses ameters distributed from 0.01 m to 0.04 m. The micro-parameters of the
will be present in the soil between the piles. The membrane support DEM particles are shown in Table 1 for three different soil states
effect was confirmed by experimental investigations [58–66]. However, including loose, medium, and dense. Through numerical simulations
the basal reinforcement effect on the settlement improvement was not triaxial tests, the macro-parameters are calibrated with the DEM pa­
studied. rameters as shown in Table 2.
In the case of very soft soil with high organic content, additional The compressible subsoil under the geosynthetic sheet is assumed to
loadings can cause large primary and secondary consolidation dis­ be compressible. To account for the ground reaction, the action of the
placements. Providing that the contact is maintained between the geo­ underlying soil is modeled using spheres arranged regularly according to
synthetic layer and the subsoil, the subsoil will carry some small loading a plan and Winkler’s spring model, in which spheres and springs are
amount. Assessing the magnitude of this support is complex due to the positioned under the geosynthetic sheet. Spheres are used to allow
soil-structure interaction. The subsoil compressibility can also vary friction between the soft soil and the geosynthetic. Similarly, the pile
locally. The magnitude of the support will be time-dependent in short elements are modeled using elastic springs with a subgrade reaction
term due to the excess pore pressure dissipation and in the long term due modulus which is significantly higher than the subsoil one.
to the secondary creep. Several 2D analytical methods were developed, The considered geosynthetic sheet is a non-woven geotextile with a
and some were extended to three-dimensional cases. Most of these tensile stiffness of 3000 kN/m reinforced in the two perpendicular di­
methods were developed for embankments over soft soils where the rections. Three-node structural elements are used to define the geo­
upwards ground counter-pressure on the geosynthetic is negligible. synthetic sheets. To guarantee the regularity and the continuity of the
However, in the case of a medium compressibility subsoil, ignoring this contact surface between the triangular elements and the soil particles
counter-pressure may result in a significant overestimation of the during the sheet deformation, cylinders, and spheres are respectively
required geosynthetic tensile strength. placed on the side and nodes of each triangle, as shown in Fig. 6.
On the other hand, the friction interactions between the geo­ To consider the interaction between the soil-structure elements,
synthetic, soil, and, piles are still a matter of discussion. Due to the specific interaction laws are used to characterize the interface behavior
stresses transfer from the soil to the piles, the soil stresses decrease along between the soil particles and the sheet elements. Hence, a microscopic
with the pile depth based on the mobilized shear stresses acting. As the contact friction angle is modelled to consider a relative roughness be­
geosynthetic deflects, a relative movement between the geosynthetic tween the sheet elements and the soil particles. The boundary conditions
and the surrounded soils will occur and the shear stresses will therefore include four frictionless vertical rigid walls to fix the horizontal dis­
be mobilized, as illustrated in Fig. 4. There is still no evidence provided placements because of the symmetric condition. Fig. 7 shows a sche­
to show this interaction aspect. The main reason may come from the matic diagram of the DEM model with the interaction description
difficulty in describing the interaction behavior at the geosynthetic-soil between the pile-soil-geosynthetic elements.
interfaces.
4. Result discussion
3. Numerical analysis
The results obtained by the discrete numerical study are presented in
A geosynthetic-reinforced and pile-supported embankment system the following sections. It should be noted that the main goal of this
consisted of piles, embankment, platform layer, geosynthetic, and sub­ section is to present the important findings of the soil-structure inter­
soils is considered. The main objective of this study is to consider the action and mechanisms for mobilizing geosynthetic reinforcement. So
influence of the soil-structure interaction on the load transfer mecha­ that, a representative case of geosynthetic-reinforced and pile-supported
nisms, and therefore only a quarter of the square mesh is modeled. The embankments was therefore selected for discussion. The fundamental
GRPS model using DEM was validated with the full-scale experiment parameters of this reference case are listed in Table 3.
[54,59,66,67]. A square mesh is modeled with a center-to-center pile

T T
Original surface
upper
Pile cap Deformed geosynthetic Pile cap

lower

Interface friction

Pile Pile

s
Fig. 4. Friction mobilization mechanism along interface of deformed geosynthetic.

5
T.A. Pham et al. Engineering Structures 242 (2021) 112337

Fig. 5. A quarter of the mesh for simulation.

Table 1
Micro-parameters in the DEM model for granular materials.
Samples Numerical Porosity Kn (MN/m2) kn/kt ϕ

Loose (L) 0.41 10 1 40o


Medium (M) 0.38 10 1 40o
Dense (D) 0.34 10 1 40o Granular
embankment

Upper Interface
Table 2
Interpretation of macro-parameters from the DEM model.
Sample ϕp (degrees) ϕcs (degrees) m Lower Interface
Geosynthetic
o o
Loose (L) 34 26 0.36 Soft soil
Medium (M) 40o 26o 0.36
Dense (D) 46o 26o 0.36 Soft spring
Rigid pile
Rigid spring

Fig. 7. Scheme of the numerical model of a reinforced piled embankment.

Table 3
Parameters for Control case.
Height of embankment H = 0.75–3.0 m
Unit weight of embankment fill γ = 18.5 kN/m3
Soft soil stiffness Ks = 0.2 MPa/m
Friction angle of embankment fill 40◦
Pile center-to-center spacing 3.0 m
Width of pile cap or diameter 0.6 m
Young’s modulus of pile 2000 MPa
Stiffness of geosynthetic 3000 kN/m
The friction angle of the interface soil/geosynthetic 26◦
Fig. 6. Interaction simulation between soil and geosynthetics (Le Hello and
Villard, 2009).
4.1. Vertical stress distribution

Concerning the load transfer mechanisms to the piles, there is a


major difference between piled embankments with geosynthetic (rein­
forced case) and without geosynthetic (unreinforced). Fig. 8 compares

6
T.A. Pham et al. Engineering Structures 242 (2021) 112337

1000
Above geosynthetic

Vertical stress distribution (kPa)


900 Peak point
Pile cap Below geosynthetic
800 Unreinforced case
700
600
500
400
300
200
100
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Horizontal distance away from the centerline of pile (m)

Fig. 8. Vertical stress distribution for the reinforced and unreinforced cases.

the vertical stress distribution obtained over the piles considering rein­ by soil arching, while the stress distribution below the geosynthetic is
forced and unreinforced cases. As it can be seen, the direct load transfer influenced not only by the soil arching but also by the tensioned geo­
mechanisms within the granular embankment are disturbed by the synthetic membrane effect.
presence of the reinforcement: the vertical stress above the geosynthetic
in the reinforced case is significantly lower than the unreinforced case
one within the pile head range. However, in the reinforced case, the 4.2. Punching shear failure
vertical stress below the geosynthetic increases and the total vertical
stress distributed on the pile cap is, therefore, higher than the unrein­ The punching failure can be categorized into two types, namely, the
forced one. Besides, due to the action of the geosynthetic at the pile local punching failure and overall punching failure. Due to the stiffness
edge, the load distribution acting on the pile head is strongly affected difference between the pile and subsoil, the soft subsoil between piles
leading to the higher stress concentration at the periphery of the piles. In tends to settle more than piles. Hence, pile heads can penetrate the
fact, the geosynthetic reinforcement induces a more efficient load embankment layers, which is known as the local punching shear failure.
transfer mechanism to the piles. This phenomenon can be explained by In the case where overall punching shear failure occurs (high surface
the geosynthetic stretching mechanism. Under embankment loads, the differential settlements), it can be visible at the embankment crest. An
geosynthetic deflects to mobilize a portion of the geosynthetic’s strength irregular surface with “humps” at the pile locations and “depressions”
and acts as a clamping membrane to attract loads onto the pile cap in-between piles could then be observed. Punching failure can be
[20,21]. identified from excessive shear strains above the columns, and excessive
It should be noted that due to the use of a geosynthetic, the vertical differential settlements at the embankment base in numerical modeling.
stress distributions above and below the geosynthetic are different. The overall punching failure is critical for low embankments while the
Fig. 9 shows the vertical stress distribution considering a three- local punching failure is often crucial for high embankments.
dimensional view. The vertical stresses above the geosynthetic are If the vertical stress acting on the pile cap due to arching is too large,
uniformly distributed along the pile cap side edge while the vertical the high reaction pressure from the pile can cause punching shear fail­
stresses below the geosynthetic are significantly concentrated at the pile ure. Compared with an unreinforced case, the vertical stresses above the
edge along the diagonal line. Furthermore, the vertical stress below the geosynthetic within the pile cap range are lower than in the unrein­
geosynthetic is higher than above it. The difference can be explained by forced case (Fig. 8). The benefit of reducing the vertical stresses above
the fact that the vertical stress above the geosynthetic is mainly induced the piles induces also a reduction of the risk by the soil yielding or
punching failure into the embankment fill. Meanwhile, increasing the

Fig. 9. 3D Surface of vertical stress distribution: (a) above geosynthetic, (b) below geosynthetic.

7
T.A. Pham et al. Engineering Structures 242 (2021) 112337

vertical stresses below the geosynthetic induces a tensioned membrane only a part of the subsoil under the GR strips area is mobilized while
effect increases, more loads are then transferred to the piles. The nu­ CUR226 suggests that the subsoil below the entire GR area should be
merical results suggest that placing a stiffer geotextile immediately at considered.
the pile top can mitigate the local punching failure or yielding soil. The
overall punching failure can be minimized by increasing the efficacy of
piles, improving the piles covering ratio, placing geosynthetics, and 4.5. Differential settlements
adjusting the thickness of the load transfer platform by adding more
layers of geosynthetics. In any case, the subsoil settlement between piles is larger than pile
settlement. It causes differential settlements. However, these differential
settlements also generate strains in the horizontal geosynthetic rein­
4.3. Geosynthetic-soil friction interaction forcement, and the geosynthetic is therefore clamped like a membrane
over piles. The effect of the basal reinforcement on the settlement
The behavior of GRPS embankments strongly depends on the inter­ improvement is considered in this study. The differential settlement at
action of the soil-structure elements. One of the important interaction the pile head is defined as the settlement difference between the pile
mechanisms which have not yet been understood well is the skin friction center and the mid-span of the pile spacing. A comparison of the dif­
mobilization along with the soil-geosynthetic interfaces. For large dif­ ferential settlements for the reinforced and unreinforced cases is shown
ferential settlements, the shear resistance is accumulated and mobilized in Fig. 11. According to the results, the differential settlements at the
along with the geosynthetic interfaces. Fig. 10 indicates that the shear elevation of the pile heads increase with the embankment fill height. The
stresses are mobilized in a horizontal direction along with the lower side geosynthetic considerably reduces the differential settlements. Howev­
of the soil-geosynthetic interface. However, the shear stresses decrease er, the geosynthetic role in reducing the differential settlements is more
and become rather small in the central geosynthetic area between piles. evident as the embankment height increases. It is noted that the geo­
This trend can be explained by the relationship between slope, slip, and synthetic significantly reduces the differential settlement of the GRPS
friction [20]. At a distance from the pile cap edge, the slope is suffi­ embankment, from 16.1% for the embankment of 0.75 m and 23.2% for
ciently large to increase the relative soil movements along with the the embankment of 3.0 m. The results proved that placing geosynthetic
geosynthetic interface, and friction is significantly mobilized. In layers above the pile head is a good way to reduce the differential
contrast, the relative movements between soil and geosynthetic settlements.
decrease when it approaches the geosynthetic middle and, the skin The geosynthetic deflection is shown in a three-dimensional plane
friction decreases (from 1.3 m in this analysis). for the embankment height of 1.5 m. As presented in Fig. 12, the
maximum settlement occurs at the mid-between point of the diagonal
4.4. Subsoil support line in an area covered by geosynthetic. At this position, the compression
of the subsoil under the embankment fill weight, and the surcharge is the
Pham and Dias [1] concluded that the analytical methods which
consider the subsoil support are able to produce results that are in a 20

better agreement with experimental data if compared to other methods 18


Reinforced case

which do not consider it. Several experimental studies also suggested Unreinforced case
Differential settlement (cm)

16
that the subsoil provides a substantial support, and reduces the pressure
acting on the geosynthetic [69–71]. Fig. 10 shows the upward vertical 14

stresses distribution within the soft subsoil range. Numerical results 12

proved that a portion of the load is supported by the soft subsoil. It 10


should be noted that the subsoil reaction curve increases with the dis­
tance away from the pile cap edge. It reaches a constant value starting 8

from 0.7 m in this analysis. According to physical equilibrium laws, it is 6

obvious that the embankment load is often received a portion by subsoil, 4


and the rest of the load will be carried out by geosynthetic. This finding 0.5 1 1.5 2 2.5 3

gives an important insight into the soft subsoil role. A possible upward Embankment height (m)

soft subsoil counter pressure should be considered when calculating the Fig. 11. Differential settlement in reinforced and unreinforced case.
GR strain. Among the existing design methods, EBGEO assumed that

15
Reaction of subsoil
Shear stress
Stress distribution along interface (kPa)

10

-5

-10

-15

-20
0.3 0.5 0.7 0.9 1.1 1.3 1.5

Horizontal distance away from the centerline of pile (m)

Fig. 10. Mobilization of the shear stress and subsoil reaction along with the geosynthetic interfaces.

8
T.A. Pham et al. Engineering Structures 242 (2021) 112337

Fig. 12. 3D surface of settlement in reinforced case.

larger one while the developed tensile force in the geosynthetic is the soil foundation or the geosynthetic. With the increase of the embank­
smaller one. It should be noted that the deflected geosynthetic surface is ment height, more shear resistance will be developed to enhance the soil
not completely smooth. It probably can be explained by the fact that the arching development. The efficacy approaches a constant value when
subsoil is modeled by springs with a defined spacing and by the local the embankment height increases (in this analysis, this value is equal to
action of large particles. 3 m).
The numerical results suggested that the geosynthetic improves the The efficacy component by arching in a reinforced case is lower than
overall bearing capacity of the soil-geosynthetic system in four different in the unreinforced case. The reduction of the differential settlements by
ways: (i) enhance the uniform stress distribution, (ii) increase the sub­ including a geosynthetic can be considered as a possible explanation. It
grade bearing capacity by changing the failure mode, (iii) reduction of is therefore confirmed that geosynthetic plays a role in soil arching. The
the maximum applied stress due to an applied surface load redistribu­ geosynthetic inclusion and tensioned membrane effect can be consid­
tion below the geosynthetic (i.e the load is redistributed over a wider ered as an explanation of why less soil arching is developed. However,
area), and (iv) supplementary support due to membrane effect, such as due to the tensioned geosynthetic membrane effect, the total efficacy in
the deformed geotextile provides an equivalent vertical support. the reinforced case is still greater than the unreinforced case one.
Fig. 14 shows the influence of subsoil stiffness on the efficacy of
4.6. Stress transfer mechanisms arching. The results confirmed that the soil arching is in a dependent
relationship with the subsoil stiffness. It is observed that an increase of
The performance of a geosynthetic reinforced and pile-supported the subgrade reaction modulus reduces the efficacy by arching, and a
embankment can be assessed through the efficacy parameter which is lower stiffness of the subsoil induces a higher soil arching effect. It is also
defined as the proportion of the embankment load carried by the pile interesting to note that the subsoil stiffness influence on arching be­
caps. The main difference between the unreinforced and reinforced comes more significant with the embankment height increase. An in­
cases is due to the geosynthetic inclusion. For the unreinforced case, the crease of the subgrade reaction modulus from 0.2 to 1.0 MPa/m induces
efficacy is only due to the soil arching (Ea). Meanwhile, the reinforced a reduction of 35% and 50% for respectively embankment heights of
case efficacy is not only induced by soil arching (Ea), but also by the 0.75 m and 3 m. For low embankment heights, the differential settle­
geosynthetic membrane effect (Em). The total efficacy in the reinforced ments are small, and less arching is mobilized.
case is calculated by the following expression [20]. In the existing design methods, the geosynthetic stiffness influence
was not considered when calculating arching. However, the geo­
E = Ea + Em (1)
synthetic stiffness plays a role in the load transfer mechanism as shown
As presented in Fig. 13, the efficacy increases with an increase in the in Fig. 15. The total efficacy increases with the tensile geosynthetic
embankment height. At a low embankment height, and for unreinforced
or reinforced cases, the shear resistance in the embankment fill is not Figure 13. Efficacy contribution in piled embankment
large enough to develop arching and reduce the pressure applied on the 40
H = 0.75m
H = 1.5m
50 35 H = 2.25m
H = 3.0m
45

40 30
Efficacy by arching (%)

35

25
Efficacy (%)

30

25
20
20

15 Reinforced-Membrane
Reinforced-Arching 15
10
Reinforced-Total efficacy

5 Unreinforced-Total efficacy
10
0
0.75 1.5 2.25 3
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Embankment height (m) Subgrade reaction moduluus (kPa/m)

Fig. 13. Efficacy contribution in the piled embankment. Fig. 14. Influence of subgrade reaction modulus on efficacy by arching.

9
T.A. Pham et al. Engineering Structures 242 (2021) 112337

80 40

Max. Tension in geosynthetic(kN/m)


Total - (1) + (2)
70 35 By skin friction (1)
By stretching (2)
30
60

25
Total efficacy (%)

50
20
40
15
30 10
H = 0.75m
H = 1.5m
20 5
H = 2.25m
H = 3.0m
10
0
0 1000 2000 3000 4000 5000 6000 0.75 1.5 2.25 3
Tensile stiffness of Geosynthetic (kN/m) Embankment height(m)

Fig. 15. Influence of geosynthetic stiffness on efficacy with different embank­ Fig. 17. Maximum tension in geosynthetic with embankment height.
ment height.
5. Review of the available design methods
stiffness and then approaches a constant at approximately 2,000 kN/m.
These results are in good agreement with the findings obtained by Pham 5.1. Load calculation methods
[20,21]. In this case, the load transfer mechanism to piles is maximized
by the geosynthetic layer. The geosynthetic contribution can be There are at least ten different methods currently available for the
explained by the increased stiffness of the earth platform and the design of GRPS embankments. Not all these methods were initially
additional vertical component of the geosynthetic tension. It induces developed for designing GRPS embankments, but they were later
more stress transfer from the soil to the pile. adopted for this goal. This section presents a description of ten different
methods for calculating the load distribution. The performance of the
4.7. Tension in geosynthetic load transfer mechanism by arching is assessed through the term effi­
cacy while some researchers prefer to use the stress reduction ratio.
For design purposes, the maximum tension in geosynthetic is of great The stress reduction ratio, which is defined as the ratio of the average
interest to geotechnical engineers. The distribution of geosynthetic vertical stress acting on the soft ground to the average embankment
tension as a function of the distance from the pile center is shown in weight and surcharge. SRRa = 0 represents a complete soil arching while
Fig. 16. Results indicate that the tension is not uniform along the geo­ SRR = 1 represents no soil arching.
synthetic and that the maximum tension occurs at the pile cap edge. It is
σ as
noted that the smallest load on the GR strips often occurs at the strip SRRa = (2)
γH + q
midway center between the piles while the largest load is located close
to the pile cap edges.
where SRRa is the stress reduction ratio induced by arching, σas is the
A non-linear relationship between the maximum geosynthetic ten­
vertical stress acting on the soft subsoil by the arching effect, γ is the unit
sion and the embankment height showed that the skin friction contrib­
embankment fill weight, H is the height of the embankment and q is a
uted to the geosynthetic deformation, as shown in Fig. 17. In this study,
uniform surcharge.
the strain component due to the skin friction was calculated from total
The second term is efficacy, which represents the total embankment
skin friction along the upper and lower sides of the soil-geosynthetic
load proportion by the column caps. The total efficacy is the result of two
interfaces, which like the suggestion of Pham [20]. The tensile force
components: the efficacy by arching (Ea), and the efficacy by the GR
due to skin friction is determined by considering the shear stress at the
membrane effect (Em). In contrast to the stress reduction ratio, Ea =
interfaces and the length of deformed geosynthetic curvature. The re­
0 represents no soil arching while Ea = 100% represents a complete soil
sults in Fig. 17 indicate that the geosynthetic tension by skin friction
arching
increases with the embankment height. This can be explained by the
increase of the embankment sliding resistance and the decrease of the Pac
Ea = (3)
grain rolling movements by interlocking with the embankment height (γH + q) × s2
increase. The resistance to sliding at each contact point is proportional
These two parameters are related by the following equation:
to the normal force at that contact, and hence the overall frictional
resistance increases as the vertical stress increases. As
Ea = 1 − .SRRa (4)
A
20
Pile cap Subsoil area where A is the tributary area of one column cap (A = sx⋅sy), As is the area
18 covered by the soft ground (As = sx⋅sy –a2), s is the center-to-center pile
Tension in geosynthetic (kN/m)

spacing.
16

5.1.1. Terzaghi method [9]


The arching theory developed by Terzaghi [9], is based on the trap
14

12
door experiment. It was used by many authors to describe the load
transfer mechanism in pile-supported embankments. Assuming a prism
10 of soil with shearing on the vertical sides above a trapdoor, the following
equation is derived:
8 ( )
0 0.2 0.4 0.6 0.8 1 1.2 1.4 (s2 − a2 )γ − 4aHKtanφ q − 4aHKtanφ

Distance from the center of pile, m σ as = 1 − e (s2 − a2 ) + e (s2 − a2 ) (5)


4aKtanφ γH + q
Fig. 16. Distribution of the geosynthetic tension.

10
T.A. Pham et al. Engineering Structures 242 (2021) 112337

K = 1 − sinφ (6) Eamin = min(Eacap , Eacrown ) (13)


where a is the pile cap width, K is the coefficient of earth pressure at
A
rest, ϕ is the friction angle of fill soils. σ as = (γH + q) × (1 − Ea ) × (14)
As
5.1.2. Method of Guido and co-workers [10]
This method derives from laboratory plate loading tests carried out 5.1.6. German EBGEO method [17]
by Guido et al. [10] on samples of geogrid-reinforced sand in a confined, The calculation method in EBGEO [17] is established from the lower
rigid box. The method emphasizes the significant support of subsoil and bound plasticity theory, pilot-scale tests, and numerical analyses. The
therefore reduces the mobilization of arching. However, in situations principles of the EBGEO method are adopted from the ground arching
where there is no support from the underlying soil or very small support, equilibrium model proposed by Kempfert et al. [16]. In this model, a
the method may considerably underestimate the quantity of required triangular pressure distribution is assumed for the calculations and
reinforcement. elastic subsoil support is suggested to examine the bearing mechanism of
the pile-geosynthetic-soils systems.
σ as =
(s − a)γ
√̅̅̅ (7) According to the German design guideline EBGEO [17], The vertical
3 2 stress acting on the soft subsoil by the arching effect, σ as , is determined as
follows:
5.1.3. Method of Low and co-workers [13] [ (( )− X
( q) ( )− X
Based on observations from model tests, Low et al. [13] developed an a X 2
h2g λ2
σ s = λ1 γ + H λ1 + hg λ2 + hg λ1 +
analytical solution to investigate the arching in a geosynthetic- H 4
reinforced pile-supported embankment. In this method, a semi- )]
( )− X
cylindrical arch with a thickness equal to the half-width of the cap − λ1 + h2g λ2 (15)
pile is considered. The vertical stress acting on the soft subsoil by the
arching effect,σ as , according to the method of Low and co-workers can be [ ]
expressed as follows: s2 − a2 (( ) s2
Ea = γH + q) − σ a
s + σ a
s (16)
[ ( )] (γH + q)s2 s2 − a2
γ(s − a)(Kp − 1) (s − a)Kp − 1 γs 1
σ as = + q + γH − 1+ (8) d(Kp − 1)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅
2(Kp − 1) s 2 Kp − 2 λ1 = 18(sd − d)2 ; X = λ2 sd ; sd = s2x + s2y

1 + sinφ ( φ) hg = s2d for H⩾s2d or hg = H for H < s2d where hg is the arching height, sd
KP = = tan2 450 + (9) is the diagonal distance, sx and sy is center-to-center pile spacing in di­
1 − sinφ 2
rections x and y.
where KP is the coefficient of passive earth pressure.
5.1.7. Dutch CUR 226 method [19]
5.1.4. Naughton method [72] The Dutch Design Guideline CUR 226 [19] uses the Concentric
The Naughton Method calculates the critical height of an embank­ arches model proposed by Van Eekelen et al. [18]. This model was
ment assuming that the vertical shear plane in the embankment fill due developed based on a series of laboratory tests. According to the
to arching is of log spiral shape with an initial radius equal to half the concentric arches model, the vertical load calculation consists of a set of
clear spacing between columns and final radius equal to the critical 2D and 3D concentric arches. Details of the concentric arches model can
height. According to the Naughton Method, the vertical stress acting on be found in CUR 226 [19], as well as in the paper of Van Eekelen et al.
the soft subsoil by the arching effect σ as is determined by: [19], and Pham [21]. In this paper, only the final equation of this
method is represented herein.
(γH + q)(s − a) π2tanφ
σ as = .e (10)
2H Ps = FGRsquare + FGRstrips (17)

5.1.5. British BS 8006 method (2010) [12] Pc = (γH + q).sx .sy − FGRsquare − FGRstrips (18)
An updated version of British Standard (BS 8006) was released in
2010. This updated version includes two methods for analyzing the Ea =
Pc
(19)
vertical stress on the geosynthetic reinforcement. One is the Hewlett and (γH + q)s2
Randolph method for a three-dimensional analysis while the Marston
formula has remained for two-dimensional analysis. In this study, the where Pc is the load part transferred directly to the pile by arching effect
method of Hewlett and Randolph is selected to represent BS 8006. Using (kN), Ps is the load part resting on the geosynthetic and subsoil (kN),
this method, the efficacy at the crown of the arch and the top of the FGRsquare is the total vertical load exerted by the 3D hemisphere on their
column can be respectively determined by Eqs. (11) and (12). square subsurface (kN), FGRstrip is the total vertical load exerted by the
[ 2D arches on the GR strips (kN).
(a)2 ](
Eacrown = 1 − 1 − X1 − X1 X2 + X3 ) (11)
s 5.1.8. Swedish method (2005)
[ ] [ ] The Swedish Method considers a wedge of soil under the arch formed
2K − 2 2K − 2
X1 = [1 − a/s]2(Kp − 1) ; X2 = √̅̅s 2Kp − 3 ; X3 = √s−̅̅a 2Kp − 3
2H p 2H p
between the columns. The arching angle within the soil is equal to 75
degrees from the horizontal. It adopts a critical height approach such
β that any additional overburden above the top of the wedge (above the
Eacap = (12)
1+β critical height) is transferred directly to the columns. In this method, the
[( ] critical height is 1.87 times the clear spacing between pile caps [73].
2Kp a)− Kp ( a) The vertical stress acting on the soft subsoil by the arching effect,σas ,
β=( ) 1− − 1 + Kp
Kp + 1 (1 + a/s) s s is determined as follows:
The critical pile efficacy is the lower of the two values, which can be (s − a)γ
σ as = (20)
used to determine the vertical stress on the geosynthetic reinforcement. 4.tan150

11
T.A. Pham et al. Engineering Structures 242 (2021) 112337

following expression:
5.1.9. Japanese PWRC method [15]
The total embankment volume is divided into the embankment T = JGR ε (26)
volume that acts on the improved ground and the unimproved ground or
geosynthetic. The shape of the soil volume supported by the unimproved where JGR is the tensile stiffness of geosynthetic, and ε is a strain of
ground is complex. This method was adopted by the Japanese Public geosynthetic.
Works Research Center (PWRC). Case histories and supporting infor­ In the method of Low and co-workers, some equations and charts
mation are available in Japanese. were developed to evaluate the tension and mobilized strains in the
geosynthetic reinforcement layer. The deflection and tension in geo­
( )2 [ ]
( √̅̅̅ )
π
(s − dc )2
tanθ(5dc + 4s) + (4 − π ) s s− dc
tanθ + s
2 − 1 tanθ synthetic can be obtained from the following equation:
96 2 2 6
a
σs = γ ( )
2Tsinδ yEs
(27)
2
s2 − π4dc = σ as −
s− a D
(21)
√̅̅̅ 4(y/(s − a))
dc = 2a/ π sinδ = (28)
1 + 4[y/(s − a)]2

where δ is the deflected angle, y is the maximum geosynthetic deflection,


Es is the soft subsoil modulus, D is the soft subsoil thickness.
θ = 45 + ϕ/2 for embankments without geosynthetic
In the method of Pham [20], the version of the circular deformation
θ = 45 - ϕ/2 for embankments with geosynthetic
model is adapted in this study,
where dc is the circular pile cap diameter, a is the square pile cap width, aλ3 + bλ2 + cλ + d = 0 (29)
θ = arching angle corresponding to the inclination concerning the hor­
izontal of the tangent line through each end of the arch spanning width. In which,
a = 21.33JGR + 4Ks (s − a)2 ; b = 1.7Ks αs tanφs (s − a)2 − 4(s − a).σas ;
5.1.10. Method of Pham [20] c = 1.7αp tanφp (s − a)σas + Ks (s − a)2 + 1.7αc cs .(s − a); d =
Pham [20] presented a simplified method for the analysis of − (s − a).σas where Ks is subsoil reaction modulus (subsoil stiffness), ϕp is
geosynthetic-reinforced and pile-supported embankment. This method friction angle of soil at the upper side of the interface (platform layer), ϕs
is the result of a combination of different models, including the is friction angle of soil at the lower side of the interface (subsoil), cs is
Concentric Arch model to describe arching for the soil layer, and extends total adhesion value of upper and lower sides of geosynthetic interfaces,
the tensioned membrane action for the geosynthetic with the frictional α is interaction coefficient between the reinforcement material and the
mechanism (for geosynthetic-soil interaction). The expression for the surrounding proposed soils. Value αc is relatively small (αc = 0.1).
vertical stress acting on the soft subsoil by the arching effect,σ as , is: The maximum deflection of geosynthetic is obtained from the
FGRsquare + FGRstrips following equation:
σ as = (22)
sx .sy − a2 (30)

y = λ.s = λ.(s − a)

(γH + q).sx .sy − FGRsquare − FGRstrips The maximum tension in geosynthetic is determined by the following
Ea = (23)
(γH + q)s2 expression:

FGRsquare is the total vertical load exerted by the 3D hemisphere on [ ]1 + 4λ2


T = σ as − yKs (s − a) (31)
their square subsurface (kN), FGRstrip is the total vertical load exerted by 8λ
the 2D arches on the GR strips (kN). FGRsquare and FGRstrip are determined
like the method of CUR226. 6. Comparison of the design methods with the numerical results

6.1. Efficacy
5.2. Deflection and strain of geosynthetic
The prediction results of efficacy by arching obtained from the
To find the tension developed in the geosynthetic layer, the following methods of CUR 226 and Pham are identical because the same model is
equation given in the British standard BS 8006 is used. A design strain of used. Therefore, for efficacy discussion, only the name of the CUR 226
5% is recommended for calculations. method is mentioned for representative in this section.
√̅̅̅̅̅̅̅̅̅̅̅̅̅
σ as 1
T= 1+ (24) 6.1.1. Unreinforced case
2a 6ε
Pile efficacies, obtained from the design techniques, are compared
The geosynthetic deflection is determined using the parabolic model with the numerical model results. As discussed in the previous section,
with a small strain assumption. the efficacy in the unreinforced case is contributed by only the arching
√̅̅̅̅̅ effect. The variation in efficacy with the embankment height is shown

y = (s − a) (25) for three different soil densities, namely, loose soils (ϕ = 34◦ ), medium
8
soils (ϕ = 40◦ ), and dense soils (ϕ = 46◦ ). The comparison results be­
These equations were used to calculate the deflection and tension of tween analytical and numerical models for efficacy are presented in
the geosynthetic reinforcement for the methods of Terzaghi, Guido, Figs. 18–20. According to the results, almost all methods give incon­
Naughton, Swedish, and PWRC. sistent results over the whole selected parameter range. It is interesting
The reinforcement strain for the EBGEO method was calculated by to note that the CUR 226 and Pham methods produce the best agreement
the graphs provided by EBGEO [17] assuming a support from the soil with the whole numerical results when compared to other methods. It is
foundation. Then, Eq. (24) was used to calculate the reinforcement observed that the agreement degree between the Pham, CUR 226, and
tension. Similarly, the reinforcement strain for the CUR 226 method was numerical methods increases with the fill friction angle reduction. The
calculated by the graphs provided by CUR 226 [19] or the Excel second-best choice is the EBGEO method. It is noted that the EBGEO
spreadsheet. Then, the reinforcement tension is calculated by the

12
T.A. Pham et al. Engineering Structures 242 (2021) 112337

90
Unreinforced case Terzaghi
80 Loose soil Guido
Ks = 0.2 MPa
70 Low
Naughton
60
Efficacy (%) Swedish
50 BS8006
EBGEO
40
CUR226
30 PWRC

20 Pham
Numerical
10

0
0.75m
1 1.50m
2 2.25m
3 3.0m 4

Embankment height

Fig. 18. Efficacy comparison for unreinforced case with loose soil (ϕ = 34◦ ).

90
Unreinforced case Terzaghi
80 Medium soil Guido
Ks = 0.2 MPa Low
70
Naughton
60 Swedish
Efficacy (%)

50 BS8006
EBGEO
40
CUR226
30 PWRC
Pham
20
Numerical
10

0
0.75m
1 1.50m
2 2.25m
3 3.0m
4
Embankment height

Fig. 19. Efficacy comparison for unreinforced case with medium soil (ϕ = 40◦ ).

90
Unreinforced case Terzaghi
80 Dense soil Guido
Ks = 0.2 MPa Low
70 Naughton
60 Swedish
Efficacy (%)

BS8006
50 EBGEO
CUR226
40
PWRC
30 Pham
Numerical
20

10

0
0.75m
1 1.50m
2 3
2.25m 3.0m
4
Embankment height

Fig. 20. Efficacy comparison for unreinforced case with dense soil (ϕ = 46◦ ).

13
T.A. Pham et al. Engineering Structures 242 (2021) 112337

method shows a larger difference with numerical results when compared numerical results for different embankment heights considering the
to the CUR 226 one. It should be noted that both the arching calculation porosity of the soil, as shown in Figs. 24–26. The efficacy value used for
methods in EBGEO and CUR 226 are 3D-arching models based on the the comparison in this study is the total efficacy, which is contributed
limit state equilibrium theory. However, the main difference is that the due to soil arching, and geosynthetic membrane effect. According to the
EBGEO method assumes a multi-arching model while the CUR 226 one results, the methods of EBGEO and CUR 226 agree better with the nu­
assumes a concentric arch model. merical results than the other methods. The EBGEO method provides the
None of the remaining design methods yields results that are in good best agreement with the numerical results for the loose soil case. How­
agreement with the three-dimensional model results. Regarding the ever, the EBGEO method produces inconsistent results in the cases of
Guido method, a high over-prediction is observed with a relative error of medium and dense soils. It is found that the CUR 226 method shows a
102%. The method of Guido was established with a simplifying better match than EBGEO when the fill friction angle increases. A similar
assumption that the soil wedge is like a pyramid and only depends on the overestimation can be observed for the Guido method while all the other
pile spacing. This method is therefore too simple but not practical to use design methods give overly conservative efficacy results.
in design calculation. The remaining methods (Terzaghi, Low, Naugh­ As presented in Figs. 27–29, the total efficacy also decreases with
ton, Swedish, BS8006, PWRC) significantly underpredict the arching increasing the subgrade subsoil reaction modulus. It is noted that the
efficacy. The average relative errors between the design methods and comparison between the unreinforced cases (Figs. 21–23) and the
numerical models are as follows: 77% for the Terzaghi method, 48% for reinforced ones showed that the increase in total efficacy by using a
the method of Low, 91% for the method of Naughton, 74.2% for the reinforcement layer is more effective for low subsoil reaction modulus
method of Swedish, 64% for the method of BS8006 and 75% for the values. Besides, the results indicated that none of the selected design
method of PWRC. It should be noted that except the BS8006 method, all methods yields consistent results compared with the numerical results
remaining methods are two-dimensional arching models based on plane- when the subgrade reaction modulus varies. The method of Naughton
strain equilibrium theory and are not adapted for considered embank­ under-predicts significantly while the method of Guido highly over-
ment thickness. However, the design method’s performance is increased predicts the efficacy among all considered cases. The other methods
with the embankment height increase and the fill friction angle over-predict the total efficacy when the subgrade reaction modulus in­
decrease. It can be concluded that the EBGEO and CUR 226 methods creases. It should be noted that the subsoil stiffness influence on arching
have the best match with numerical results when compared to other was not considered in the theoretical solutions. However, the numerical
methods. results indicate that the subsoil stiffness plays a role in the total efficacy.
A similar analysis can be observed in Figs. 21–23 which shows the An increase in the subsoil stiffness reduces the total efficacy for both
efficacy variation with different subsoil stiffnesses for the selected unreinforced and reinforced cases. In other words, a smaller subsoil
design techniques. The numerical results indicated that the efficacy stiffness promotes a higher soil arching effect and geosynthetic mem­
decreases with the subsoil reaction modulus increase. Furthermore, the brane effect, and the efficacy is therefore increased.
subsoil stiffness influence on the efficacy becomes more important when
the fill friction angle increases. It can be observed that the efficacy 6.2. Geosynthetic tension
reduction with increasing the subsoil stiffness for loose, medium, and
dense soils are respectively equal to 50%, 61%, and 75%. However, the Geosynthetic tension results, obtained using the selected design
subsoil modulus influence was not considered in the current design techniques, are compared with the numerical results in Fig. 30 for
methods. This is the reason why the efficacy value obtained from the different embankment heights. The methods of Naughton, Swedish, and
design methods is constant. Unfortunately, the comparison results PWRC (using Eq. (24)) produce a significant over-prediction and
indicated that none of the selected design methods yields results that are therefore are not shown in this figure. According to the results, all the
in good agreement with the numerical model results. The difference in other design methods give conservative results for the tension developed
results between numerical and theoretical methods becomes higher in the geosynthetic reinforcement, yielding uneconomical designs. The
when the subsoil reaction modulus increases. average relative error of three prominent methods, namely, CUR 226,
EBGEO, and BS 8006 are 55.42%, 219.46%, 282.69% respectively. The
6.1.2. Reinforced case CUR 226, EBGEO, and Pham methods agree well with the numerical
Efficacy obtained from the design techniques is compared with the results and better than the other design methods. The subsoil support is

100 Terzaghi
Unreinforced case
90 Loose soil Guido
H=3m Low
80
Naughton
70 Swedish
Efficacy (%)

BS8006
60
EBGEO
50 CUR226
PWRC
40
Pham
30 Numerical
20

10

0
0.1 1MPa 0.2 MPa
2 0.5 MPa
3 1 MPa
4
Subsoil reaction modulus, Ks

Fig. 21. Efficacy comparison with different subsoil stiffness for loose soil (ϕ = 34◦ ).

14
T.A. Pham et al. Engineering Structures 242 (2021) 112337

100 Terzaghi
Unreinforced case
90 Medium soil Guido
H=3m Low
80
Naughton
70
Swedish
Efficacy (%) 60 BS8006
50 EBGEO
CUR226
40
PWRC
30 Pham
20 Numerical

10

0
0.11 MPa 0.2 2MPa 0.5 MPa
3 4
1 MPa
Subsoil reaction modulus, Ks

Fig. 22. Efficacy comparison with different subsoil stiffness for medium soil (ϕ = 40◦ ).

100
Unreinforced case Terzaghi
90 Dense soil Guido
H=3m Low
80
Naughton
70 Swedish
BS8006
Efficacy (%)

60
EBGEO
50 CUR226
PWRC
40
Pham
30 Numerical
20

10

0
0.11MPa 0.2 2MPa 3 MPa
0.5 14MPa
Subsoil reaction modulus, Ks

Fig. 23. Efficacy comparison with different subsoil stiffness for dense soil (ϕ = 46◦ ).

90 Terzaghi
Reinforced case
80 Loose soil Guido
Ks = 0.2 MPa Low
70 Naughton
Swedish
60
BS8006
Efficacy (%)

50 EBGEO
CUR226
40
PWRC
30 Pham
Numerical
20

10

0
0.75m
1 1.50m
2 2.25m
3 43.0m
Embankment height

Fig. 24. Efficacy comparison for reinforced case with loose soil (ϕ = 34◦ ).

15
T.A. Pham et al. Engineering Structures 242 (2021) 112337

90
Reinforced case Terzaghi
80 Medium soil Guido
Ks = 0.2 MPa Low
70
Naughton
60 Swedish
Efficacy (%)
BS8006
50
EBGEO
40 CUR226
PWRC
30
Pham
20 Numerical

10

0
1
0.75m 1.50m
2 2.25m3 3.0m
4
Embankment height

Fig. 25. Efficacy comparison for reinforced case with medium soil (ϕ = 40◦ ).

90
Reinforced case Terzaghi
80 Dense soil Guido
Ks = 0.2 MPa Low
70
Naughton
60 Swedish
BS8006
Efficacy (%)

50
EBGEO
40 CUR226
PWRC
30
Pham
20 Numerical

10

0
0.75m
1 1.50m
2 2.25m
3 3.0m4
Embankment height

Fig. 26. Efficacy comparison for reinforced case with dense soil (ϕ = 46◦ ).

100 Terzaghi
Reinforced case
90 Loose soil Guido
H=3m Low
80
Naughton
70 Swedish
BS8006
60
EBGEO
50
Efficacy (%)

CUR226
40 PWRC
Pham
30
Numerical
20

10

0
0.11MPa 0.22MPa 0.53MPa 1 4MPa
Subsoil reaction modulus, Ks

Fig. 27. Efficacy comparison with varying subsoil stiffness for reinforced-loose soil (ϕ = 34◦ ).

16
T.A. Pham et al. Engineering Structures 242 (2021) 112337

100 Terzaghi
Reinforced case
90 Medium soil Guido
H=3m Low
80
Naughton
70 Swedish
BS8006
60
EBGEO
Efficacy (%)

50 CUR226
40 PWRC
Pham
30
Numerical
20

10

0
0.11 MPa 0.2 MPa
2 0.5 MPa
3 14MPa
Subsoil reaction modulus, Ks

Fig. 28. Efficacy comparison with varying subsoil stiffness for reinforced-medium soil (ϕ = 40◦ ).

100 Terzaghi
Reinforced case
90 Dense soil Guido
H=3m Low
80 Naughton
70 Swedish
Efficacy (%)

BS8006
60
EBGEO
50 CUR226
40 PWRC
Pham
30 Numerical
20

10

0
0.11MPa 0.2 MPa
2 0.5 MPa
3 1 4MPa
Subsoil reaction modulus, Ks

Fig. 29. Efficacy comparison with varying subsoil stiffness for reinforced-dense soil (ϕ = 46◦ ).

considered for the GR strain calculation of these three methods. A better and a comparison of the maximum geosynthetic deflection is shown in
agreement is found when compared to methods that neglected the Fig. 31. According to the results, the Pham, EBGEO, and CUR 226
subsoil support. This observation agrees with the findings of Pham and methods provide a good agreement with the numerical results. This
Dias [1]. To find a reasonable explanation for the significant difference agreement remains valid when the embankment heights change from
between the selected design methods in predicting the geosynthetic 0.75 to 3 m. The methods of CUR 226 and EBGEO under-predict 3.37%
tensile force, special attention should be paid to the design methods and 0.95% respectively while the method of BS 8006 highly over-
principle. In the EBGEO model, only the subsoil underneath the GR strip predicts the geosynthetic deflection with an average relative error of
between the adjacent piles is considered. This model also assumes that 126%. All remaining methods over-predict significantly the geo­
the vertical load distribution on the GR strips is approximately trian­ synthetic deflection. This study, therefore, suggests that the theoretical
gular (minimum load value at the pile edge). In the models of CUR 226 models should include the subsoil support in calculating the geo­
and Pham, the subsoil support is underneath all the areas covered by GR. synthetic strain and deflection. It will allow to improve and define a
Besides, the CUR 226 model also assumes that when there is no subsoil more realistic and accurate approach for the analysis of load-
support, the vertical load distribution is like an inverse triangular one deformation behavior of GRPS embankments.
(maximum load value at the pile edge) while a uniform load distribution
is used if significant subsoil support is considered. Regarding all other 7. Conclusion
models, the load distribution is assumed to be uniform as well as they do
not consider the subsoil support. This paper presents a study about the load transfer mechanism in a
geosynthetic-reinforced and pile-supported embankment by conducting
three-dimensional DEM numerical analyses. The study is then compared
6.3. Comparison of using geosynthetic deflection
with ten currently available design methods for designing GRPS em­
bankments. Several main conclusions can be drawn from this analysis as
The analysis was carried out for four different embankment heights,

17
T.A. Pham et al. Engineering Structures 242 (2021) 112337

400
Reinforced case Terzaghi

Max. Tension in geosynthetic (kN/m)


Medium soil Guido
350 Ks = 0.2 MPa Low
BS8006
300 EBGEO
CUR226
250 Pham
Numerical
200

150

100

50

0
0.75m
1 1.50m
2 2.25m
3 3.0m
4
Embankment height

Fig. 30. Comparison between analytical and numerical methods for maximum GR tensile force.

600
Reinforced case Terzaghi
Medium soil Guido
500 Ks = 0.2 MPa Low
BS8006
EBGEO
Max. GR deflection (mm)

400 CUR226
Pham
Numerical
300

200

100

0
1
0.75m 2
1.50m 3
2.25m 3.0m
4
Embankment height

Fig. 31. Comparison between analytical and numerical methods for maximum GR deflection.

follows: unreinforced one. However, the vertical stress below the geo­
synthetic increases and the total vertical stress distributed on the pile
Numerical results confirmed that the geosynthetic reinforcement cap in the reinforced case is, therefore, higher than the unreinforced
induces an efficient load transfer mechanism to the piles. The in­ one.
clusion of the geosynthetic enhances the stress transfer from the The distribution of tension in the geosynthetic layer indicates that
subsoil to piles due to the tensioned membrane action. Therefore, the tension is not uniform along the geosynthetic and that the maximum
inclusion of geosynthetic in earth platforms could reduce the dif­ tension occurs at the pile cap edge. The total geosynthetic tension is
ferential settlements as well as the possibility of soil yielding above induced due to two components, one is by the stretching of the
the pile cap. geosynthetic under the arching loads and another one is by the skin
Due to the use of geosynthetics, the vertical stress distribution above friction along with interfaces.
and below the geosynthetic is different. The results indicated that Numerical results showed that the underlying soil reaction pressure
vertical stresses below the geosynthetic are significantly concen­ between the rigid pile elements is mobilized. Therefore, the sup­
trated at the corner of the pile cap, while vertical stresses above the porting soft soil upwards counter-pressure between piles should be
geosynthetic are more uniform. When compared with the vertical introduced into the design methods to produce a more realistic
stress magnitude for an unreinforced case, the vertical stress above approach.
the geosynthetic in the reinforced case is significantly lower than the

18
T.A. Pham et al. Engineering Structures 242 (2021) 112337

Comparison results between the numerical model and analytical [18] Van Eekelen SJM, Bezuijen A, Van Tol AF. An analytical model for arching in piled
embankments. Geotext Geomembr 2013;39:78–102. https://doi.org/10.1016/j.
models showed that these available design methods differ signifi­
geotexmem.2013.07.005.
cantly when predicting the stress reduction ratio, efficacy, geo­ [19] CUR 226. Design guideline basal reinforced piled embankments. Design guideline
synthetic tension, and differential settlement. All the selected design basal reinforced piled embankments. CRC Press; 2016.
methods provide over-predictions or under predictions depending on [20] Pham TA. Load-deformation of piled embankments considering geosynthetic
membrane effect and interface friction. Geosynth Int 2020;27(3):275–300. https://
the geometric properties of the embankment. The methods of Pham, doi.org/10.1680/jgein.19.00030.
CUR 226, and EBGEO are applicable for predicting the efficacy for [21] Pham TA. Analysis of geosynthetic-reinforced pile-supported embankment with
both low and high embankments while methods of BS 8006, PWRC, soil-structure interaction models. Comput Geotech 2020;121:103438. https://doi.
org/10.1016/j.compgeo.2020.103438.
and Swedish are only applicable for high embankments. The [22] Pham TA. Behaviour of piled embankment with multi-interaction arching model.
methods of Pham, CUR 226, and EBGEO produce the best agreement Géotech Lett 2020;10(4):582–8. https://doi.org/10.1680/jgele.20.00084.
with numerical results for the prediction of differential settlements. [23] Ariyarathne P, Liyanapathirana DS. Review of existing design methods for
geosynthetic-reinforced pile-supported embankments. Soils Found 2015;55(1):
The method of Pham and CUR226 agrees well with numerical results 17–34. https://doi.org/10.1016/j.sandf.2014.12.002.
in predicting the geosynthetic tension and is generally in better [24] Lo SR, Zhang R, Mak J. Geosynthetic-encased stone columns in soft clay: a
agreement than the other presented methods. numerical study. Geotext Geomembr 2010;28(3):292–302.
[25] Zhuang Y, Ellis EA. Finite-element analysis of a piled embankment with
reinforcement and subsoil. Géotechnique 2016;66(7):596–601.
CRediT authorship contribution statement [26] Wu JT, Ye X, Li J, Li GW. Field and numerical studies on the performance of high
embankment built on soft soil reinforced with PHC piles. Comput Geotech 2019;
107:1–13.
Tuan A. Pham: Conceptualization, Methodology, Writing - original [27] Ariyarathne P, Liyanapathirana DS, Leo CJ. Comparison of different two-
draft. Quoc-Anh Tran: Conceptualization, Methodology, Writing - dimensional idealizations for a geosynthetic-reinforced pile-supported
original draft. Pascal Villard: Supervision, DEM devlopment, Writing - embankment. Int J Geomech 2013;13(6):754–68.
[28] Borges JL, Marques DO. Geosynthetic-reinforced and jet grout column-supported
review & editing. Daniel Dias: Supervision, Writing - review & editing. embankments on soft soils: Numerical analysis and parametric study. Comput
Geotech 2011;38(7):883–96.
[29] Jamsawang P, Yoobanpot N, Thanasisathit N, Voottipruex P, Jongpradist P. Three-
Declaration of Competing Interest dimensional numerical analysis of a DCM column-supported highway
embankment. Comput Geotech 2016;72:42–56.
[30] Oliveira PJV, Pinheiro JL, Correia AA. Numerical analysis of an embankment built
The authors declare that they have no known competing financial on soft soil reinforced with deep mixing columns: Parametric study. Comput
interests or personal relationships that could have appeared to influence Geotech 2011;38(4):566–76.
the work reported in this paper. [31] Rowe RK, Liu KW. Three-dimensional finite element modelling of a full-scale
geosynthetic-reinforced, pile-supported embankment. Can Geotech J 2015;52(12):
2041–54.
References [32] Yadav SK, Ye GL, Khalid U, Fukuda M. Numerical and centrifugal physical
modelling on soft clay improved with floating and fixed sand compaction piles.
Comput Geotech 2019;115:103160.
[1] Pham TA, Dias D. Comparison and evaluation of analytical models for the design of
[33] Jiang Y, Han J, Zheng G. Numerical analysis of a pile–slab-supported railway
geosynthetic-reinforced and pile-supported embankments. Geotext Geomembr
embankment. Acta Geotech 2014;9(3):499–511.
2021;49(3):528–49. https://doi.org/10.1016/j.geotexmem.2020.11.001.
[34] Meena NK, Nimbalkar S, Fatahi B, Yang G. Effects of soil arching on behavior of
[2] Chen RP, Liu QW, Wang HL, Liu Y, Ma QL. Performance of geosynthetic-reinforced
pile-supported railway embankment: 2D FEM approach. Comput Geotech 2020;
pile-supported embankment on soft marine deposit. Proc Instit Civil Engineers-
123:103601.
Geotech Eng; 2020. p. 1–18. doi: 10.1680/jgeen.19.00136.
[35] Yoo C. Performance of geosynthetic-encased stone columns in embankment
[3] Ye GB, Wang M, Zhang Z, Han J, Xu C. Geosynthetic-reinforced pile-supported
construction: numerical investigation. J Geotech Geoenviron Eng 2010;136(8):
embankments with caps in a triangular pattern over soft clay. Geotext Geomembr
1148–60.
2020;48(1):52–61. https://doi.org/10.1016/j.geotexmem.2019.103504.
[36] Potts VJ, Zdravkovic L. Finite-element study of arching behaviour in reinforced
[4] Lee T, Lee SH, Lee IW, Jung YH. Quantitative performance evaluation of GRPE: a
fills. Proc Instit Civil Engineers-Ground Improve 2010;163(4):217–29.
full-scale modeling approach. Geosynth Int 2020;27(3):342–7. https://doi.org/
[37] Ghosh B, Fatahi B, Khabbaz H, Nguyen HH, Kelly R. Field study and numerical
10.1680/jgein.19.00017.
modelling for a road embankment built on soft soil improved with concrete
[5] Briançon L, Simon B. Pile-supported embankment over soft soil for a high-speed
injected columns and geosynthetics reinforced platform. Geotext Geomembr 2020.
line. Geosynth Int 2017;24(3):293–305. https://doi.org/10.1680/jgein.17.00002.
https://doi.org/10.1016/j.geotexmem.2020.12.010.
[6] Cao WZ, Zheng JJ, Zhang J, Zhang RJ. Field test of a geogrid-reinforced and
[38] Pham HV, Dias D, Dudchenko A. 3D modeling of geosynthetic-reinforced pile-
floating pile-supported embankment. Geosynth Int 2016;23(5):348–61. https://
supported embankment under cyclic loading. Geosynth Int 2020;27(2):157–69.
doi.org/10.1680/jgein.16.00002.
[39] Messioud S, Okyay US, Sbartai B, Dias D. Dynamic response of pile reinforced soils
[7] Liu HL, Ng CW, Fei K. Performance of a geogrid-reinforced and pile-supported
and piled foundations. Geotech Geol Eng 2016;34(3):789–805.
highway embankment over soft clay: case study. J Geotech Geoenviron Eng 2007;
[40] Messioud S, Sbartai B, Dias D. Estimation of dynamic impedance of the
133(12):1483–93. https://doi.org/10.1061/(ASCE)1090-0241(2007)133:12
soil–pile–slab and soil–pile–mattress–slab systems. Int J Struct Stab Dyn 2017;17
(1483).
(06):1750057.
[8] Hoppe EJ, Hite SL. Performance of a pile-supported embankment (No. FHWA/
[41] Das AK, Deb K. Experimental and 3D numerical study on time-dependent behavior
VTRC 06-R36). Virginia Transportation Research Council; 2006.
of stone column–supported embankments. Int J Geomech 2018;18(4):04018011.
[9] Terzaghi K. Theoretical soil mechanics. New York: John Wiley & Sons; 1943.
[42] Indraratna B, Basack S, Rujikiatkamjorn C. Numerical solution of stone
p. 11–5.
column–improved soft soil considering arching, clogging, and smear effects.
[10] Guido VA, Kneuppel JD, Sweeney MA. Plate loading tests on geogrid reinforced
J Geotech Geoenviron Eng 2013;139(3):377–94. https://doi.org/10.1061/(ASCE)
earth slabs (NewOrleans). Proc Geosynth 1987;87:216–25.
GT.1943-5606.0000789.
[11] Hewlett WJ, Randolph MF. Analysis of piled embankments. In: International
[43] Jenck O, Dias D, Kastner R. Three-dimensional numerical modeling of a piled
Journal of Rock Mechanics and Mining Sciences and Geomechanics Abstracts (Vol.
embankment. Int J Geomech 2009;9(3):102–12.
25, No. 6, pp. 297–298). Elsevier Science; 1988.
[44] Huang Z, Ziotopoulou K, Filz GM. 3D Numerical Limiting Case Analyses of Lateral
[12] BS 8006-1. Code of practice for strengthened/reinforced soils and other fills.
Spreading in a Column-Supported Embankment. J Geotech Geoenviron Eng 2019;
London, UK: British Standards InstitutionBSI; 2010.
145(11):04019096.
[13] Low BK, Tang SK, Choa V. Arching in piled embankments. J Geotech Eng 1994;120
[45] Yu Y, Bathurst RJ. Modelling of geosynthetic-reinforced column-supported
(11):1917–38. https://doi.org/10.1061/(ASCE)0733-9410(1994)120:11(1917).
embankments using 2D full-width model and modified unit cell approach. Geotext
[14] Miki H, Nozu, M. Design and numerical analysis of road embankment with low
Geomembr 2017;45(2):103–20. https://doi.org/10.1016/j.
improvement ratio deep mixing method. In: Geotechnical Engineering for
geotexmem.2017.01.002.
Transportation Projects (pp. 1395-1402); 2004.
[46] Han J, Gabr MA. Numerical analysis of geosynthetic-reinforced and pile-supported
[15] PWRC, Public Work Research Center. Manual on Design and Execution of
earth platforms over soft soil. J Geotech Geoenviron Eng 2002;128(1):44–53.
Reinforced Soil Method with Use of Geotextiles, second ed. Public Work Research
[47] Badakhshan E, Noorzad A, Bouazza A, Zameni S, King L. A 3D-DEM investigation
Center, pp. 248e256; 2000 (in Japanese).
of the mechanism of arching within geosynthetic-reinforced piled embankment. Int
[16] Kempfert HG, Gobel C, Alexiew D, Heitz C. German recommendations for soil
J Solids Struct 2020;187:58–74.
reinforcement above pile-elements. In: EUROGeo3, Third Geosynthetic
[48] Chen RP, Liu QW, Wu HN, Wang HL, Meng FY. Effect of particle shape on the
Conference, Munchen (Vol. 1, pp. 279-283); 2004.
development of 2D soil arching. Comput Geotech 2020;125:103662.
[17] EBGEO. Recommendations for design and analysis of earth structures using
[49] Lai HJ, Zheng JJ, Cui MJ, Chu J. “Soil arching” for piled embankments: insights
geosynthetic reinforcements. Digital in English. German geotechnical society.
from stress redistribution behaviour of DEM modelling. Acta Geotech 2020:1–20.
ISBN: 978-3-433-60093-1; 2011.

19
T.A. Pham et al. Engineering Structures 242 (2021) 112337

[50] Rui R, van Tol F, Xia XL, van Eekelen S, Hu G, Xia YY. Evolution of soil arching; 2D [62] Almeida MSS, Fagundes DF, Thorel L, Blanc M. Geosynthetic-reinforced pile-
DEM simulations. Comput Geotech 2016;73:199–209. embankments: numerical, analytical and centrifuge modelling. Geosynth Int 2020;
[51] Han J, Bhandari A, Wang F. DEM analysis of stresses and deformations of geogrid- 27(3):301–14.
reinforced embankments over piles. Int J Geomech 2012;12(4):340–50. [63] Wijerathna M, Liyanapathirana DS. Load transfer mechanism in geosynthetic
[52] Jenck O, Dias D, Kastner R. Discrete element modelling of a granular platform reinforced column-supported embankments. Geosynth Int 2020;27(3):236–48.
supported by piles in soft soil–Validation on a small scale model test and [64] Khansari A, Vollmert L. Load transfer and deformation of geogrid-reinforced piled
comparison to a numerical analysis in a continuum. Comput Geotech 2009;36(6): embankments: field measurement. Geosynth Int 2020;27(3):332–41.
917–27. [65] McGuire M, Sloan J, Filz G. Effectiveness of geosynthetic reinforcement for load
[53] Chareyre B, Villard P. Dynamic spar elements and discrete element methods in two transfer in column-supported embankments. Geosynth Int 2020;27(2):200–18.
dimensions for the modeling of soil-inclusion problems. J Eng Mech 2005;131(7): [66] Villard P, Huckert A, Briançon L. Load transfer mechanisms in geotextile-
689–98. reinforced embankments overlying voids: Numerical approach and design. Geotext
[54] Le Hello B, Villard P. Embankments reinforced by piles and Geomembr 2016;44(3):381–95.
geosynthetics—Numerical and experimental studies dealing with the transfer of [67] Chalak C, Briançon L, Villard P. Coupled numerical and experimental analyses of
load on the soil embankment. Eng Geol 2009;106(1–2):78–91. load transfer mechanisms in granular-reinforced platform overlying cavities.
[55] Anh Tran Q, Villard P, Dias D. Discrete and continuum numerical modeling of soil Geotext Geomembr 2019;47(5):587–97.
arching between piles. Int J Geomech 2019;19(2):04018195. [68] Kozicki J, Tejchman J, Mühlhaus HB. Discrete simulations of a triaxial compression
[56] Indraratna B, Nimbalkar SS, Ngo NT, Neville T. Performance improvement of rail test for sand by DEM. Int J Numer Anal Meth Geomech 2014;38(18):1923–52.
track substructure using artificial inclusions–Experimental and numerical studies. [69] Lu W, Miao L, Wang F, Zhang J, Zhang Y, Wang H. A case study on geogrid-
Transp Geotech 2016;8:69–85. https://doi.org/10.1016/j.trgeo.2016.04.001. reinforced and pile-supported widened highway embankment. Geosynth Int 2020;
[57] Gholaminejad A, Mahboubi A, Noorzad A. Encased stone columns: coupled 27(3):261–74.
continuum–discrete modelling and observations. Geosynth Int 2020;27(6):581–92. [70] Hoppe EJ, Hite SL. Performance of a Pile-Supported Embankment (No. FHWA/
[58] Huckert A, Briançon L, Villard P, Garcin P. Load transfer mechanisms in geotextile- VTRC 06-R36). Virginia Transportation Research Council; 2006.
reinforced embankments overlying voids: experimental and analytical approaches. [71] Xu C, Song S, Han J. Scaled model tests on influence factors of full geosynthetic-
Geotext Geomembr 2016;44(3):442–56. reinforced pile-supported embankments. Geosynth Int 2016;23(2):140–53.
[59] Pham TA. Analysis of soil-foundation-structure interaction to load transfer [72] Naughton PJ. The significance of critical height in the design of piled
mechanism in reinforced piled embankments. Aust Geomech J 2019;54(1):85–100. embankments. In: Soil improvement (pp. 1–10); 2007.
[60] Fagundes DF, Almeida MS, Thorel L, Blanc M. Load transfer mechanism and [73] Rogbeck Y, Gustavsson S, Södergren I, Lindquist D. Reinforced piled embankments
deformation of reinforced piled embankments. Geotext Geomembr 2017;45(2): in Sweden-design aspects. In: Proceedings, sixth international conference on
1–10. geosynthetics (Vol. 2, pp. 755–762); 1998.
[61] Tano BFG, Stoltz G, Coulibaly SS, Bruhier J, Dias D, Olivier F, et al. Large-scale
tests to assess the efficiency of a geosynthetic reinforcement over a cavity.
Geosynth Int 2018;25(2):242–58.

20

You might also like