Geotextiles and Geomembranes 42 (2014) 302e311

Contents lists available at ScienceDirect

Geotextiles and Geomembranes

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

Instability of a geogrid reinforced soil wall on thick soft Shanghai clay

with prefabricated vertical drains: A case study
Jian-Feng Xue a, b, Jian-Feng Chen a, *, Jun-Xiu Liu a, Zhen-Ming Shi a
a
b

Department of Geotechnical Engineering, School of Civil Engineering, Tongji University, 1239 Siping Road, Shanghai 200092, China
School of Applied Sciences and Engineering, Monash University, Churchill 3842, Victoria, Australia

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 21 January 2014
Received in revised form
25 April 2014
Accepted 5 May 2014
Available online 2 June 2014

A 7.6 m high geogrid reinforced soil retaining wall (RSW) was constructed at the end of an embankment
on very thick, soft Shanghai clay with 12 m deep prefabricated vertical drains (PVDs). The settlement of
the ground, the wall movement and pore water pressure were monitored during the construction. From
day 118, halfway through the construction, unexpected pore water pressure increment was recorded
from the pore water pressure meters installed in the PVD drained zone indicating a possible malfunction
of the PVDs due to large deformation in the ground. After the last loading stage, on day 190, a sudden
horizontal movement at the toe was observed, followed by an arc shaped crack on the embankment
surface at the end of the reinforced backll zones. The wall was analyzed with a coupled mechanical and
hydraulic nite element (FE) model. The analysis considered two scenarios: one with PVDs fully functional, and the second one with PVD failure after day 118 by manually deactivating the PVDs in the FE
model. The comparison between the measured and simulated ground settlement, toe movement, and
pore water pressure supported the assumption on the malfunction of the PVDs. It is believed that the
general sliding failure in the wall was caused by the increase of pore water pressure in the foundation soil
and soils in front of the toe. It is suggested that possible failure of PVDs should be considered in the
design of such structures, and the discharge rate of the PVDs and the pore water pressure should be
closely monitored during the construction of high soil walls on soft soils to update the stability of the
structures, especially for grounds where large deformations are expected which may cause the failure of
the PVDs.
2014 Elsevier Ltd. All rights reserved.

Keywords:
Reinforced soil wall
Soft soils
Prefabricated vertical drains
Pore water pressure
General sliding failure

1. Introduction
Geosynthetic reinforced soil walls (RSWs) have been widely
used throughout the world in road embankment and retaining
structures. RSWs have many advantages including aesthetics, short
construction period, good wall stability, cost effectiveness, good
seismic response, strong adaptability on soft highly compressible
foundation soils and the ability to tolerate large differential settlement (Tatsuoka et al., 1997; Bloomeld et al., 2001; Yoo and Jung,
2004).
Research has been done on the behavior of RSWs on various soil
foundations, and various loading conditions through a large number of in-situ and laboratory tests as well as theoretical analyses
(Rowe and Skinner, 2001; Mandal and Joshi, 1996; Viswanadham
and Konig, 2009; Leshchinsky and Han, 2004; Yoo and Jung,

* Corresponding author. Tel./fax: 86 21 65983545.

E-mail address: jf_chen@tongji.edu.cn (J.-F. Chen).
http://dx.doi.org/10.1016/j.geotexmem.2014.05.003
0266-1144/ 2014 Elsevier Ltd. All rights reserved.

2008; Huang et al., 2011; Raisinghani and Viswanadham, 2011).

Despite the fact that many geosynthetic-reinforced soil walls have
been safely constructed to date, there is limited literature on the
performance and behavior of the reinforced soil wall on PVD
drained soft soils (Alfaro et al., 1997; Bloomeld et al., 2001;
Skinner and Rowe, 2005; Tanchaisawat et al., 2008; Demir et al.,
2013), and even fewer reports on failure case studies to a further
review of their original design and therefore the failure mechanism
(Leonards et al., 1994; Tatsuoka et al., 1997; Collins, 2001; Ling et al.,
2001; Borges and Cardoso, 2002; Yoo et al., 2004, 2006;
Scarborough, 2005). There are still many areas that need in-depth
studies which will help in the safe construction of RSWs on soft
clay drained with PVDs (Koerner and Koerner, 2013).
This paper presents a failure case history of a 7.6 m high geogridreinforced soil wall with vertical wrap-around facing constructed
on a typical Shanghai multi-layer soft soil ground installed with
PVDs. Unexpected increase of pore water pressure was observed in
the PVD drained soils during the construction, followed by a large
deformation in the toe, and an arc shape crack in the surface of the

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

303

embankment at the end of the reinforced zone, indicating the

failure of the wall after the last layer of backll was placed. Numerical modeling was carried out to compare with the in-situ
monitored data to identify the failure mechanism of the wall and
possible causes of the failure. Recommendations were made at the
end for the design and construction of RSWs on soft ground.
2. Description of the RSW project
2.1. Site description
The project was located in Shanghai Botanic Garden, Shanghai,
China. The RSW wall is a temporary retaining structure at the end of
a 530 m long embankment (see Fig. 1). The embankment is 37.2 m
wide on top with a 1.5H: 1V side slope, as shown in Fig. 2 with the
ground condition: ve layers of soft soils with a total thickness of
31 m from the ground level, which are (from top to bottom): 2.6 m
thick of silty clay, 4.4 m thick of mucky silty clay, 3.6 m thick of clay,
7.4 m thick of silty clay and 13.0 m thick of silty clay. Underneath is a
layer of 6.3 m thick stiff clayey silt with 31 of average SPT blow
count, underlain by stiff silty sand and clay. The ground water table
depth is shallow and it uctuates within 0.5 m below the ground
surface. Table 1 summarizes the properties of the subsoil obtained
from in situ and laboratory tests.
2.2. RSW construction
The wall was constructed in layers via staged lling. The site was
prelled with 1.6 m thick soil excavated from layer 1 silty clay to
compensate the settlement. Above the ll, a 0.6 m thick medium
sand was compacted to a bulk density of 17.0 kN/m3 as a sand
cushion for drainage as shown in Fig. 3. Prefabricated vertical drains
(PVDs) were installed in a triangle pattern with a spacing of 1.5 m to
the depth of 12 m below the surface of the sand cushion. The
embankment was constructed using the soil excavated from Layer 1
silty clay compacted at the lift of 20 cm to the bulk density of
19.0 kN/m3 using a 15 ton static double drum road roller (1.5 m
behind the wall face) with compaction width of 2.12 m, and a
lightweight walk-behind roller (within 1.5 m from the wall face).
The loading stages are shown in Fig. 4.
Fourteen layers of 10 m long high density polyethylene (HDPE)
uniaxial geogrid were placed at a vertical spacing of 0.6 m (top 6
layers) and 0.5 m (bottom layers), with 3.5 m of wrap-around
segment at the wall facing. The axial stiffness of the geogrid is
620 kN/m at 5% strain and the strength is 70 kN/m.

Fig. 1. The RSW on soft clay.

Fig. 2. Detail of the RSW and the ground condition.

2.3. In situ measurement

Below instruments were installed along the centerline of the
embankment as shown in Fig. 3:
1) : two settlement plates (C1 and C2) installed at the initial
ground surface 4 m and 20 m behind the wall toe to measure the
ground settlement;
2) : two sets of magnetic extensometers installed at 2.5 m (S1) and
15 m (S2) away from the wall toe. Five magnets were installed in
each extensometer at the depth of 4.4 m, 8.1 m, 11.8 m, 14.6 m
and 20.6 m, namely S1-1 to S1-5 and S2-1 to S2-5;
3) : two rows of pore water pressure meters installed at 2 m (P1)
and 18 m (P2) away from the wall toe at the depth of 2.6 m, 7 m,
10.6 m, 18 m and 24.5 m, namely P1-1 to P1-6, and P2-1 to P2-6;
4) : four displacement markers on the wall face at the height of 1.7,
2.7, 3.7 and 4.7 m above the toe of the wall to monitor the wall
displacement from day 139;
5) : a settlement plate (D1) was installed at 0.5 m away from the
wall toe to monitor the displacement of the toe.
3. Measurement results
3.1. Excessive pore water pressure
The measured excess pore water pressures at the monitored
depth are shown in Fig. 5. Overall the observed excess pore water
pressure in the two rows of pore water pressure meters follows the
same trend. The gure shows that at slow loading rate, excess pore
water pressure decreased steadily with time, e.g. from day 50e118.
Pore water pressure increased rapidly during the days 36e40, 138
to 140, and 187 to 189 when the loading rate was high, especially at
shallower depth, e.g. above the depth of 10.6 m. The pore water
pressure increment at the deeper locations, e.g. below 18 m, is very
low, e.g. less than 10 kPa during the whole period.
There was an unexpected rise of pore water pressure on day 118
in most of the pressure meters, while there were no construction
activities. After day 118, the pore water pressure in the ground did
not seem to dissipate at all, especially at shallower depth
(above 10.6 m). The increase of the pore water pressure observed
in the ground on day 118 and slow dissipation of pore water
pressure after that is perhaps due to the failure of the PVDs
resulting from the large deformation in the ground and wall
movement during construction. Chu et al. (2006) found that
200 mm of settlement in a vacuum pressure preloaded PVD drained
soft ground can cause the buckling of PVDs and dramatic reduction,
e.g. 84%, in the PVD discharge capacity. As shown in Fig. 6 and Fig. 7,
till day 118, there was a total settlement of more than 400 mm in C1

304

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

Table 1
Summary of geotechnical properties of the subsoil.
Layer

1
2
3
4
5

USCS classication

CL
CL
CL
CL
CL

wn

Ip

(kN/m3)

(%)

(%)

18.4
17.9
17.5
19.5
19.3

32.5
38.0
41.1
24.0
25.2

16.2
15.1
18.0
15.1
15.1

eo

0.93
1.06
1.16
0.70
0.73

c'

f'

ref
E50

kh

kv

(kPa)

( )

(MPa)

(107 cm/s)

(107 cm/s)

7.0
7.0
6.0
17.0
18.0

27.9
28.0
24.9
29.8
31.2

3.18
2.12
2.47
8.37
5.70

16.0
16.4
13.9
1.6
1.6

1.14
1.08
1.33
0.12
0.12

Note: USCS: unied soil classication system, g unit weight; wn natural water content; Ip plasticity index; e0 initial void ratio; c' effective cohesion; f' effective
ref secant modulus in standard drained triaxial test; k horizontal permeability; k vertical permeability.
friction angle; E50
h
v

Fig. 3. Prole of the RSW and the ground surface, with the location of instruments.

and C2, and more than 150 mm horizontal movement at the toe.
With such a large deformation, there is a high possibility of malfunction or partial malfunction of the PVDs. This will be further
discussed in the following sections.
3.2. Observed ground settlement and failure of the RSW
Fig. 6 shows the ground settlement measured from S1-1 to S1-4
and S2-1 to S2-4, and C1 and C2. There is little settlement observed
in S1-5 and S2-5, at the depth of 20.6 m, therefore the results from
these two magnets are not shown. It can be seen that the settlement curves follow the loading stages. There were two sharp drops
in the settlement curve. The rst one was during the days 31e40,
with a settlement of 69 mm in C1 in 10 days, then followed by a
gentle curve during a three-week resting period. This agrees with
the observed dissipation of excess pore water pressure as shown in
Fig. 5. The second sharp settlement was after day 187 when the last
layer of embankment was laid, then followed by the failure of the
wall.
The gure also shows that, below the depth of 10.6 m, at the end
of the PVDs, there is not much settlement occurred. This suggested
that there was not much consolidation below that depth during the

construction period, which agrees with the observed low excess

pore water pressure at the depth as shown in Fig. 5.
From day 187, sudden settlement was recorded in almost all the
monitored points. On day 190, an arc-shaped crack was observed at
the end of the reinforced zone, which is 10 m away from the RSW
facing as plotted in Fig. 2. The shape and location of the crack may
indicate a deep-seated failure in the wall. To prevent further sliding,
surcharge was placed in front of the wall on day 200, and 0.5 m
diameter and 28 m long concrete pipe piles were installed 5 m
away from the wall facing in a square pattern with a spacing of 3 m.
3.3. Movement of the RSW
Fig. 7 shows the ground movement at the toe of the wall. The
gure shows that, before failure, the horizontal movement at the
toe is less than the vertical movement, and the movement curves
reect the loading stages well. From day 187, there is a sudden
movement away from the embankment at the toe on horizontal
direction, while the vertical settlement is relatively steady. From
day 187 to day 189, the horizontal movement of the toe increased
about 15 mm, followed by a 67 mm movement in 24 h to day 191,
which indicates a possible block sliding of the wall.

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

305

Fig. 6. Ground settlement with time.

Fig. 4. Fill height vs. time curve.

4. Numerical modeling
PLAXIS has been used by researchers to solve large deformation
of embankment on soft soils, e.g. more than 1 m of settlement
(Brinkgreve et al., 2004; Bergadoa et al., 2002; Chaiyaput et al.,

Fig. 5. Excess pore water pressure with time.

Fig. 8 compares the settlement of the toe and measurements

from the two plates C1 and C2. It can be seen that from day 136 to
day 187, the settlement curve of the toe concave upward instead of
downward as shown in the other two curves. This may suggest that,
during that period, the toe of the wall may have been subjected to a
slightly heave due to rotation of the wall.

Fig. 7. Movement at the toe of the wall during construction.

3.4. Deformation of the wall face

Fig. 9 shows the horizontal movement of the wall face from day
139 at the level of 1.7 m, 2.7 m, 3.7 m and 4.7 m above the toe of the
wall. It can be seen that the wall moved outward dramatically after
day 187, or start of the last loading stage. Fig. 10 (a) shows the
bulging of the wall face, which indicates that the maximum bulging
is at the middle height (3.7 m) of the wall. Fig. 10 (b) shows the net
increment of displacement of the four monitored levels after day
156, which indicates a block sliding outward with slightly tilting.
More bulging was observed at the middle height of the wall during
the days 190e195, which is a common phenomenon for exible
facings walls due to compaction stresses, self-weight of the backll,
lack of section modulus or toe xity (FHWA, 2009; Ehrlich and
Mirmoradi, 2013).

Fig. 8. Settlement of the embankment at three monitored locations.

306

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

Fig. 9. Movement at different locations of the wall face during construction.

2014). The failure mechanism of the wall was studied using the
two-dimensional (2D) coupled mechanical and hydraulic FE model
on PLAXIS Version 8.2 platform. The PVDs were modeled with
drains in PLAXIS. The failure of the PVDs was modeled by turning
the drains off in the analysis. To turn off all the PVDs may be a
conservative choice in the simulation when there is limited information available. Another choice is to use equivalent permeability
of soils obtained from in-situ measured discharge rate instead of
using drains in the numerical simulation to consider the effect of
drain failure.
The stability of the RSW was examined using the Strength
Reduction Method (SRM) incorporated in the PLAXIS program. In
the model, 50 extra days were simulated after the last loading
stage. It is worthwhile to note that after the last loading stage, the
simulation is only for comparison purpose, as the main purpose of
the paper is to discuss the behavior of the wall before the failure.
4.1. Conversion of axisymmetric drainage to plane drainage
Tang et al. (2013) proposed a closed-form solution for the
consolidation of vertically drained three layer soil system. To
analyze the 3D effect of PVDs in a plane strain condition, Hird et al.
(1995) proposed to use the following equation to obtain the
equivalent permeability of the soil system:

2 B2 1
khp $ 2 $ $kh
3 De m

(1)

where kh the horizontal permeability of subsoil; khp the

equivalent horizontal permeability of the soil system in plane strain
condition; B the width of the plane-strain unit cell; De the
diameter of the effective zone of drainage; The dimensionless factor
m is to account for the smear effect, which can be dened as:

m lnn=s kh =ks ln s  0:75

(2)

where ks the horizontal permeability in the smeared zone; nDe/

dw. The factor of s is dened as the ratio of the diameter of the
smeared zone (ds) and the diameter of the vertical drain (dw), ds/dw.
The ratio of kh/ks varies with soils and ground conditions. Hansbo
(1987) found that the value can vary from 1 to 5 for clay samples
in the laboratory. According to the in-situ measurement on a
nearby site by Shen et al. (2005), the value of kh/ks is about 13.5 for

Fig. 10. Deection of the wall face during construction, (a): total displacement from
day one, (b) net displacement after day 156.

soft clays in the tested area. Therefore, a value of 13.5 was adopted
for kh/ks in the following numerical analysis.
Rixner et al. (1986) recommended the following equation to
account for the shape effect of the PVDs using the equivalent
diameter:

dw

wt
2

(3)

where w the width of a band-shaped PVD; t the thickness of

the PVD. The width and the thickness of the PVDs used in this study
were 100 mm and 4 mm respectively.
Jamiolkowski et al. (1983) found that the diameter of the smear
zone, ds, is proportional to the diameter of the cross-sectional area
of mandrel (dm):

ds 2e3dm

(4)

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

In this study, the diameter of the smear zone is dened as three

times of the diameter of the mandrel, 110 mm in this study, as
recommended by Chai and Miura (1999).
For triangle pattern of PVDs, the diameter of unit cell, De, is

De 1:05d

(5)

where d spacing of PVDs, and d 1.5m in this study.

Table 2 summarizes the parameters used in Eqs. (1) and (2).
Using the parameters, we obtained m 25.8 and khp 0.023kh. As
the discharge capacity of the PVDs is very high (up to 100 m3/year),
and staged-loading is used in the construction to avoid excess pore
water pressure buildup in the drains, the effect of well resistance
was not considered in the simulation.
4.2. Numerical model and parameters
As the embankment is very wide and long comparing to its
height, the central section of the embankment was chosen to
simulate the plane strain condition. The lower boundary was set at
31 m from the original ground surface to the bottom of layer 5.
Ninety meters long of longitudinal section is simulated in the model,
with 45 m behind the RSW and 45 m ahead of the RSW. Standard
boundary condition was used in the model, with water table at the
ground surface. An unstructured nite-element ne mesh consisting of 15-node triangular elements was used as shown in Fig. 11. The
meshes and pore water pressure were updated using update mesh
technique in the analysis to consider the large ground deformation.
The preliminary ll, the sand cushion, and the RSW backll were
modeled with linear elastic-perfectly plastic Mohr-Coulomb model,
and the hardening-soil model was used for the foundation soils.
The drain element inherent in PLAXIS was used to model PVDs, in
which excess pore water pressures are set to zero in the drain elements. As discussed in Section 3.1, the equivalent horizontal
permeability was used for soils in layers 1 to 3, in which the PVDs
were installed.
The geogrid reinforcements were modeled as elastoplastic material, using geogrid element in PLAXIS, with axial stiffness of
J 620 kN/m and the tensile strength of T 70 kN/m. The soil and
geogrid interface behavior have been studied by many researchers
(Zhou et al., 2012). For simplication and availability of the models
in the software, a rigid model was adopted for the soil and geogrid
interface, with an interface coefcient of 0.7 based on pullout tests.
Table 3 summarizes the values of the soil parameters for the FE
ref 3Eref 3E ref
analysis. The following empirical relationships: Eur
50
oed
and m 1 (Schanz et al., 1999) were used to develop the hardeningref tangent stiffness for primary oedometer
soil model, in which Eoed
ref Youngs modulus for unloading and reloading.
loading and Eur
4.3. Results of the modeling
The embankment was simulated under two scenarios: with
drains on during the whole construction period, and with drains off
after day 118 to simulate the failure of the PVDs. After deactivating

307

the drains, the horizontal permeability kh was used for the soils in
the PVD drained zone instead of khp, which may not represent the in
situ condition, as the soils have been consolidated which will result
in lower permeability.
4.3.1. Excess pore water pressure
The modeled excess pore water pressures at the monitored
points are shown in Fig. 12. The results show that with all the PVDs
on, the excess pore water pressure dissipates quickly after each
loading stage to values lower than 20 kPa. In the rst few loading
stages (before day 118), the measured and modeled values agree
well with each other. After day 118, the measured values seem
unreasonably higher than the modeled results.
As discussed in the previous section, from day 118, there might
be a failure of PVDs due to large deformation in the ground. To
validate this assumption, the drains were turned off in PLAXIS from
day 118 and a second simulation was carried out with all the other
parameters unchanged except the horizontal permeability of the
soils in the PVD zone as explained earlier. The simulated excess
pore water pressures were compared with measured values in
Fig. 12 (with PVDs off). The gure shows that the simulated value
captures the excess pore water pressure variation very well after
day 118, which supports the assumption that the PVDs have failed
after day 118.
4.3.2. Ground movement
The measured settlements at C1 are compared with the simulated values in Fig. 13. It can be seen that both simulations agree
well with the measured settlement curve of C1 before day 118. With
the PVDs on, the predicted settlements of the locations agree well
with the measured values till day 189, but underestimates the
overall settlement in the long term, e.g. till day 200. With the PVDs
off, the predicted total settlement matches the measured value
well. This again supports the assumption that the PVDs may have
failed during the construction.
4.3.3. Movement of the toe
The movement of the toe is plotted in Fig. 14. The gures show
that, for vertical settlement, the numerical model provides relatively good estimation for both situations. For lateral movement,
the model well overestimated the movement after turning off the
drains, and under estimated the total settlement after the last
loading with the drains on. The reasons for this bad performance
can be numerous (Tavenas et al., 1979; Ehrlich and Mirmoradi,
2013), such as boundary condition, soil anisotropy, the soil model
adopted in the simulation and the oversimplication of the possible
partial failure of PVDs by assuming a full failure. In this case, the
movement of the toe area is a combination of rotation due to global
failure, squeezing due to lateral pressure, ground settlement and
heave, which makes it difcult to obtain comparable results. The
using of kh for the soils in the PVD drained zone after the failure of
PVDs may also contribute to this difference.
5. Failure mechanism of the wall

Table 2
Parameters for Eqs. (1) and (2).
Item

Symbol

Value

Width of the plane-strain unit cell (m)

Equivalent drain diameter (mm)
Ratio of kh over ks in eld
Smear zone diameter (mm)
ds/dw
diameter of unit cell (m)
De/dw

B
dw
kh/ks
ds
s
De
n

1.5
52
13.5
330
6.35
1.58
30.3

The stability of the retaining structures can be analyzed using

either limit equilibrium (LEM) or strength reduction method
(Cheng et al., 2007). Since the soil strength increment, which
should be considered in LEM during consolidation, is a complex
problem and is not the purpose of this study, the factor of safety
(FS) of the structure was analyzed using the SRM built in PLAXIS in
which the soil strength increment was considered in the hardening
soil model (Brinkgreve et al., 2004):

308

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

Fig. 11. Numerical model of the RSW.

Table 3
Material properties of the backll and subsoil.
Material

Backll

Sand cushion

Layer 1

Layer 2

Layer 3

Layer 4

Layer 5

Material Model
Material Type
gunsat (kN/m3)
gsat (kN/m3)
khp (m/d)
kvp (m/d)
Eref (MPa)
ref (MPa)
E50
ref (MPa)
Eoed
ref (MPa)
Eur
m

MC
Drained
19.0
19.0
e
e
5
e
e
e
e
0.33
16
30

MC
Drained
17
20
4.32
4.32
20
e
e
e
e
0.2
0
30

HS
UU
13.6
18.4
3.18E-05
9.85E-05
e
3.18
3.18
9.54
1
0.25
7
27.9

HS
UU
12.8
17.9
3.26E-05
9.33E-05
e
2.12
2.12
6.36
1
0.25
7
28

HS
UU
12.1
17.5
2.76E-05
1.15E-04
e
2.47
2.47
7.41
1
0.25
6
24.9

HS
UU
15.4
19.5
1.38E-04
1.04E-05
e
8.37
8.37
25.1
1
0.2
17
29.8

HS
UU
15.1
19.3
1.38E-04
1.04E-05
e
5.7
5.7
17.1
1
0.2
18
31.2

n
c0 (kPa)

f0 ( )

Note: UU: unconsolidated undrained, MC: Mohr-Coulomb model, HS: Hardening soil model, gunsat unit weight above water table; gsat unit weight below water table;
ref tangent stiffness for primary oedometer loading and Eref Young's modulus for unloading and
m power for stress-level dependency of stiffness; n Poisson's ratio; Eoed
ur
reloading.

X
c
tan f
FS

Msf
cr tan fr

(6)

where c, cr and f, fr are the input and reduced cohesion and friction
P
angle respectively;
Msf is the total multiplier, which increases
until failure occurs in the numerical modeling.
The stability of the wall at the end of the last loading stage was
analyzed using SRM. The computed FS value is 1.097 with the PVDs
on, and 1.01 with the PVDs off, which is at the limit state. It is

Fig. 12. Variation of excess pore water pressure at the monitored points.

worthwhile to note that using kh for the soil permeability after

deactivating the PVDs is conservative without considering the
decrease of soil permeability due to consolidation. The predicted
critical slip surface is shown in Fig. 15. As shown in the gure, there
was no failure within the reinforced wall, and the slip surface follows the general failure mode.

Fig. 13. Comparison of settlement at C1 obtained from numerical simulation with and
without considering the failure of PVDs.

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

309

Fig. 16. Excess pore water proles under the embankment at the end of last loading
stage with, a): PVDs on, b): PVDs off after day 118.

Fig. 14. Comparison of toe movement obtained from numerical simulation with and
without considering the failure of PVDs, a): vertical displacement, b): horizontal
displacement.

The excess pore water pressure proles after the last loading
stage under the two scenarios are plotted in Fig. 16. It shows that no
excess pore water pressure was generated in the PVD drained zone
with the PVDs working, but large excess pore water pressure can be

Fig. 15. Factor of safety and the failure slip surface predicted with strength reduction
method with PVDs off after day 118.

created (up to 70 kPa) if the PVDs fail. In both scenarios, pore water
pressure in front of the toe area of the wall increased signicantly
after the construction. The gures show that in front of the wall, in
the second layer, where the slip surface passes through, the excess
pore water pressure can increase up to the range of 20e40 kPa,
even with the PVDs functional, and the value can be higher if the
PVDs fail. The increase of pore water pressure in front the toe
causes the dramatic decrease of effective stress in the zone,
therefore the reduction of shear strength and the passive resistance. This may explain the failure of the slope. To avoid the buildup
of excess pore water pressure in front of the toe area, additional
PVDs can be installed in this area for drainage purpose if space
permits and the installation of PVDs does not cause instability of
the soils.
6. Summary and conclusions
A fully monitored RSW built on very thick Shanghai soft clay
drained with PVDs was studied. Unexpected excess pore water
pressure in the PVD drained zone was observed in the middle of the
construction during resting period, when about 75% height of the
7.6 m wall had been nished. Bulging and rotation of the wall face
occurred in the last few loading stages, which is a common phenomenon in this type of wall. Large deformation in the wall and
accelerating horizontal movement at the toe of the wall was
observed after the last loading stage, with relatively steady vertical
settlement, followed by an arc shaped crack on the embankment
surface, which suggests a general shear failure mechanism.
A two-dimensional coupled mechanical and hydraulic nite
element (FE) model was used to study the behavior and the failure
mechanism of the structure. It was conrmed that a general sliding
failure occurred in the wall, with no shear failure in the geogrid
reinforced zone. The numerical simulation shows that excess pore
water pressure can be dissipated quickly via the PVDs when the
drains are fully functional. The observed increase of excess pore

310

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

water pressure was captured by manually deactivating the drains to

simulate its failure during the construction stages in the FE model.
The FE models give relatively good estimation for the settlement of
the wall. The horizontal movement of the toe compares well with
the FE results before day 118, when the PVDs are working. After day
118, the FE model underestimates the lateral movement of toe with
PVDs on, and overestimates the toe lateral movement with PVDs
off.
The stability of the RSW was examined by the SRM incorporated
in the FE program. It was found that, the last loading stage is critical
for the stability of the wall. With PVDs on, the FOS of the slope is
1.097. With deactivating the PVDs halfway through the construction, the embankment is at its limit state after the last loading stage
with FOS of 1.01. Based on the case studied, it has been found that:
1. the measured settlement at the instrumented location in the
ground showed that the settlement in the foundation soils
mainly occurred in the PVD drained zone, where the effective
stress increased the most;
2. the FE simulation showed that high excess pore water pressure
can be generated in the toe area due to lateral earth pressure,
even with the PVDs functional when excess pore water pressure
can be quickly dissipated via the drains in the drained zone.
From this point of view, if space allows, it is suggested to install
one or two rows of PVDs in front of the toe, without causing too
much ground disturbance and instability to the wall;
3. the observed high excess pore water pressure below the wall,
where the PVDs were installed, was most likely due to the
failure of the drains. In the performed FE analyses, the failure of
the drains could be well represented by manually deactivating
the drains during construction. The SRM analysis showed that
the factor of safety of the wall decreased almost to 1 by deactivating the drains. The simulation suggested that the failure of
the wall was due to the increase of pore water pressure in the
PVD zone, and soils in front of the wall, which reduced the
passive resistance against sliding;
4. in the design of embankments on PVD drained soft soils, it is
important to check the stability of the wall considering possible
failure of PVDs, especially when large deformation is anticipated
in the ground. The failure of the PVDs can be simulated by
deactivating the drains at certain stages, which is a conservative
yet passive choice. Where possible, it is recommended to use the
measured discharge rate to obtain an equivalent permeability of
the soils for numerical modeling or to carry out a full 3D
simulation.
Acknowledgments
The support from the National Natural Science Foundation of
China under grant No. 41072200 is gratefully acknowledged. The
rst author was a visiting scholar at Tongji University when the
paper was written, funding received from the Key Laboratory of
Geotechnical and Underground Engineering was much appreciated.
References
Alfaro, M.C., Hayashi, S., Miura, N., et al., 1997. Deformation of reinforced wall/
embankment system on soft clay foundation. Soils Found. 37 (4), 33e46.
Bergado, D.T., Long, P.V., Murthy, B.R.S., 2002. A case study of geotextile reinforced
embankment on soft ground. Geotext. Geomemb. 20, 343e365.
Bloomeld, R.A., Soliman, A.F., Abraham, A., 2001. Performance of mechanically stabilized earth walls over compressible soils. In: Ochiai, Otani, Yasufuku, Omine
(Eds.), Landmarks in Earth Reinforcement. A.A. Balkema, pp. 317e322.
Borges, J.L., Cardoso, A.S., 2002. Overall stability of geosynthetic-reinforced embankments on soft soils. Geotext. Geomemb. 20 (6), 395e421.
Brinkgreve, R.B.J., Broere, W., Waterman, D., 2004. Plaxis Version 8 Reference
Manual. Delft: Plaxis BV.

Chaiyaput, S., Bergado, D.T., Artidteang, S., 2014. Measured and simulated results of
a Kenaf Limited Life Geosynthetics (LLGs) reinforced test embankment on soft
clay. Geotext. Geomemb. 42, 39e47.
Chai, J.C., Miura, N., 1999. Investigation of factors affecting vertical drain behavior.
J. Geotech. Geoenviron. Eng. 125 (3), 216e226.
Cheng, Y.M., Lansivaara, T., Wei, W.B., 2007. Two-dimensional slope stability analysis by limit equilibrium and strength reduction methods. Comput. Geotech. 34
(3), 137e150.
Chu, J., Bo, M.W., Chao, V., 2006. Improvement of ultra-soft soil using prefabricated
vertical drains. Geotext. Geomemb. 24 (6), 339e348.
Collins, J.G., 2001. Lessons learned from a segmental retaining wall failure. Geotext.
Geomemb. 19 (7), 445e454.
Demir, A., Laman, M., Yildiz, A., Ornek, M., 2013. Large scale eld tests on
geogrid-reinforced granular ll underlain by clay soil. Geotext. Geomemb. 38,
1e15.
Ehrlich, M., Mirmoradi, S.H., 2013. Evaluation of the effects of facing stiffness
and toe resistance on the behavior of GRS walls. Geotext. Geomemb. 40,
28e36.
Federal Highway Administration (FHWA), 2009. Design and Construction of Mechanically Stabilized Earth Walls and Reinforced Soil Slopes. Publication
Number FHWA-NHI, pp. 10e024.
Hird, C.C., Pyrah, I.C., Russell, D., Cinicioglu, F., 1995. Modeling the effect of vertical
drains in two-dimensional nite element analyses of embankments on soft
ground. Can. Geotech. J. 32 (5), 795e807.
Hansbo, S., 1987. Design aspects of vertical drains and lime column installations. In:
Proceedings of Ninth Southeast Asian Geotechnical Conference, 2(8). Southeast
Asian Geotechnical Society, Bangkok, pp. 1e12.
Huang, C.C., Horng, J.C., Chang, W.J., Chiou, J.S., Chen, C.H., 2011. Dynamic behaviour
of reinforced walls d horizontal displacement response. Geotext. Geomemb. 29
(3), 257e267.
Jamiolkowski, M., Lancellotta, R., Wolski, W., 1983. Pre-compression and Speeding
up Consolidation, General report. Special Session 6, Proceedings of Eight European Conference on Soil Mechanics and Foundation Engineering,
pp. 1201e1226. Balkema, Rotterdam.
Koerner, R.M., Koerner, G.R., 2013. A data base, statistics and recommendations
regarding 171 failed geosynthetic reinforced mechanically stabilized earth
(MSE) walls. Geotext. Geomemb. 40, 20e27.
Leonards, G.A., Frost, J.D., Bray, J.D., 1994. Collapse of geogrid-reinforced retaining
structure. J. Perform. Constr. Facil. ASCE 8 (4), 274e292.
Leshchinsky, D., Han, J., 2004. Geosynthetic reinforced multi-tiered walls.
J. Geotech. Geoenviron. Eng. 130 (12), 1225e1235.
Ling, H.I., Leshchinsky, D., Chou, N.N.S., 2001. Post-earthquake investigation on
several geosynthetic-reinforced soil retaining walls and slopes during 1999 Ji-Ji
earthquake of Taiwan. Soil. Dyn. Earthq. Eng. 21 (4), 297e313.
Mandal, J.N., Joshi, A.A., 1996. Design of geosynthetic reinforced embankments on
soft soil. Geotext. Geomemb. 14 (2), 137e145.
Raisinghani, D.V., Viswanadham, B., 2011. Centrifuge model study on low
permeable slope reinforced by hybrid geosynthetics. Geotext. Geomemb. 29
(6), 567e580.
Rixner, J.J., Kraemer, S.R., Smith, A.D., 1986. Prefabricated Vertical Drains. Engineering Guidelines, FWHA/RD-86/168, vol. 1. Federal Highway Administration,
Washington DC.
Rowe, R.K., Skinner, G.D., 2001. Numerical analysis of geosynthetic reinforced
retaining wall constructed on a layered soil foundation. Geotext. Geomemb. 19
(7), 387e412.
Scarborough, J.A., 2005. A tale of two walls: case histories of failed MSE walls. In:
Proceedings of Geo-Frontiers 2005, pp. 2751e2762.
Schanz, T., Vermeer, P.A., Bonnier, P.G., 1999. The Hardening Soil Model: Formulation and Verication. Beyond 2000 in Computational Geotechnics-10 Years of
PLAXIS, pp. 281e290. A.A. Balkema, Rotterdam.
Shen, S.L., Chai, J.C., Hong, Z.S., Cai, F.X., 2005. Analysis of eld performance of
embankments on soft clay deposit with and without PVD-improvement. Geotext. Geomemb. 23 (6), 463e485.
Skinner, G.D., Rowe, R.K., 2005. Design and behaviour of a geosynthetic reinforced
retaining wall and bridge abutment on a yielding foundation. Geotext. Geomemb. 23 (3), 234e260.
Tanchaisawat, T., Bergado, D.T., Voottipruex, P., 2008. Numerical simulation and
sensitivity analyses of full-scale test embankment with reinforced lightweight geomaterials on soft Bangkok clay. Geotext. Geomemb. 26 (6),
498e511.
Tang, X.W., Niu, B., Cheng, G.Ch, Shen, H., 2013. Closed-form solution for consolidation of three-layer soil with a vertical drain system. Geotext. Geomemb. 36,
81e91.
Tatsuoka, F., Tateyama, M., Uchimura, T., Koseki, J., 1997. Geosynthetic reinforced soil
retaining walls as important permanent structures. Geosynth. Int. 4 (2),
81e136.
Tavenas, F., Mieussens, C., Bourges, F., 1979. Lateral displacements in clay foundations under embankments. Can. Geotech. J. 16, 532e550.
Viswanadham, B., Konig, D., 2009. Centrifuge modeling of geotextile-reinforced
slopes subjected to differential settlements. Geotext. Geomemb. 27 (2),
77e88.
Yoo, C., Jung, H.S., 2004. Measured behavior of a geosynthetic-reinforced
segmental retaining wall in a tiered conguration. Geotext. Geomemb. 22
(5), 359e376.

J.-F. Xue et al. / Geotextiles and Geomembranes 42 (2014) 302e311

Yoo, C., Jung, S.B., 2008. Performance of a two-tier geosynthetic reinforced
segmental retaining wall under a surcharge load: full-scale load test and 3D
nite element analysis. Geotext. Geomemb. 26 (6), 460e472.
Yoo, C., Jung, H.S., Jung, H.Y., 2004. Lessons learned from a failure of geosynthetic
reinforced segmental retaining wall. In: Proceedings of 3rd Asian Regional
Conference on Geosynthetics, Seoul, Korea, pp. 265e274.

311

Yoo, C., Jung, H.S., Jung, H.Y., 2006. Case history of geosynthetic-reinforced
segmental retaining wall failure. J. Geotech. Geoenviron. Eng. ASCE 132 (12),
1538e1548.
Zhou, J., Chen, J.F., Xue, J.F., Wang, J.Q., 2012. Micro-mechanism of interaction between sand and geogrid transverse rib. Geosynth. Int. 12 (6), 426e437.