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

net/publication/237370795

Finite element modeling of hexagonal wire reinforced embankment on soft clay

Article  in  Canadian Geotechnical Journal · February 2011

DOI: 10.1139/cgj-37-6-1209

CITATIONS READS
28 1,193

4 authors:

D.T Bergado Chairat Teerawattanasuk

Asian Institute of Technology King Mongkut's University of Technology North Bangkok
289 PUBLICATIONS   7,267 CITATIONS    20 PUBLICATIONS   401 CITATIONS   

SEE PROFILE SEE PROFILE

Sompote Youwai P. Voottipruex

King Mongkut's University of Technology Thonburi King Mongkut's University of Technology North Bangkok
41 PUBLICATIONS   839 CITATIONS    50 PUBLICATIONS   1,089 CITATIONS   

SEE PROFILE SEE PROFILE

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

Geofoam View project

Reinforced Cement-Deep Mixing Project View project

All content following this page was uploaded by Chairat Teerawattanasuk on 26 May 2014.

The user has requested enhancement of the downloaded file.

Color profile: Generic CMYK printer profile
Composite Default screen

1209

Finite element modeling of hexagonal wire

reinforced embankment on soft clay
D.T. Bergado, C. Teerawattanasuk, S. Youwai, and P. Voottipruex

Abstract: A full-scale test embankment was constructed on soft Bangkok clay using hexagonal wire mesh as reinforce-
ment. This paper attempts to simulate the behavior of the full-scale test embankment using the finite element program
PLAXIS. The agreement between the finite element results and the field data is quite good. The important consider-
ations for simulating the behavior of the reinforced wall embankment were the method of applying the embankment
loading during the construction process, the variation of soil permeability during the consolidation process, and the
selection of the appropriate model and properties at the interface between the soil and reinforcement. The increased
reinforcement stiffness tends to increase the reinforcement tension and increase the embankment forward rotation. The
reinforcement tensions increased with the compression of the underlying soft foundation. The appropriate interface
properties between the backfill soil and the hexagonal wire mesh reinforcement corresponding to the interaction
mechanism at working stress conditions were dominated by the direct shear mechanism. The direct shear interaction
coefficient of 0.9 was used.

Key words: soil reinforcement, hexagonal wire mesh, finite element analysis, field embankment.

Résumé : Un remblai d’essai pleine grandeur a été construit sur l’argile molle de Bangkok en utilisant des grillages
hexagonaux comme armature. Cet article tente de simuler le comportement du remblai d’essai pleine grandeur au
moyen du programme d’éléments finis PLAXIS. La concordance entre les résultats d’éléments finis et les données du
terrain est très bonne. Les considérations importantes pour simuler le comportement du remblai du mur comportaient la
méthode de mise en place du chargement du remblai durant le processus de construction, la variation de la
perméabilité au cours du processus de consolidation, et la sélection du modèle et des propriétés à l’interface entre le
sol et l’armature. L’augmentation de la rigidité de l’armature tend à accroître la traction dans l’armature et la rotation
du remblai vers l’avant. Les tractions dans l’armature ont augmenté avec la compression de la fondation molle sous-
jacente. Les propriétés pertinentes de l’interface entre le sol de remblai et le grillage hexagonal d’armature
correspondant au mécanisme d’interaction dans les conditions de contraintes de travail étaient dominées par le
mécanisme de cisaillement direct. Le coefficient d’interaction en cisaillement direct de 0,9 a été utilisé.

Mots clés : armature pour le sol, grillage hexagonal, analyse en éléments finis, remblai en nature.
[Traduit par la Rédaction] Bergado et al. 1226

Introduction the preservation of environmental conditions through

erosion protection and control. Moreover, reinforced earth
Reinforced soil or mechanically stabilized earth (MSE), constructions are flexible and can exhibit tolerance to large
an engineered composite material, has been used extensively deformations. Chai (1992) stated that for a reinforced soil
for construction of earth retaining walls and embankment wall on soft ground, under the wall loading the soft soil
slopes and in stabilization of embankments on soft ground. tends to squeeze out of the wall base and cause large relative
A savings in cost can result if locally available, fine-grained, movement between the reinforcement and the soil. There-
cohesive backfill materials can be used for reinforced earth fore, large tension forces may develop in the reinforcement.
structures (Bergado et al. 1991). The use of hexagonal link
wire reinforcements to mechanically stabilize the temporary There are many factors that affect the performance of a
retaining structure is a recent technique in reinforced earth soil reinforcement wall such as reinforcement stiffness, soil–
construction (e.g., Voottipruex 2000). An MSE structure is reinforcement interaction, and friction angle of the backfill
inexpensive, requires a simple construction operation in a soil. The effects of the parameters on a reinforced wall have
shorter period of time, and has been known to be effective in been studied by Rowe and Ho (1997) through the use of nu-
merical methods such as the finite element method (FEM).
Many researchers have used the FEM in the analysis of a re-
Received November 9, 1999. Accepted May 10, 2000. inforced soil wall (e.g., Chai 1992; Bergado et al. 1993,
Published on the NRC Research Press website on 1995; Kapurrapu and Bathurst 1995; Alfaro et al. 1997;
December 7, 2000. Rowe and Ho 1997). However, nothing was found in the lit-
D.T. Bergado, C. Teerawattanasuk, S. Youwai, and P. erature on finite element analysis of a hexagonal wire rein-
Voottipruex. School of Civil Engineering, Asian Institute of forced embankment. An important purpose of this paper is
Technology, P.O. Box 4, Klong Luang, Pathumthani 12120, to simulate the behavior of a full-scale test embankment re-
Thailand. inforced with hexagonal wire mesh constructed on soft

Can. Geotech. J. 37: 1209–1226 (2000) © 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:37:43 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1210 Can. Geotech. J. Vol. 37, 2000

Bangkok clay. The modeling parameters were obtained from Weathered clay layer
laboratory tests and back-calculated results from the behav-
ior of the full-scale test embankment. To investigate the be- In Bangkok, the uppermost 1–2 m of the soil profile con-
havior of a hexagonal reinforced embankment, a finite sists of a heavily overconsolidated weathered clay layer. The
element program (PLAXIS) was utilized to model the per- Cam Clay model is unsuitable for simulation of this heavily
formance and behavior of the field test embankment. overconsolidated soil. An elastic, perfectly plastic, Mohr-
Coulomb model with a constant value of Poisson’s ratio has
been used to simulate the behavior of the weathered clay
Full-scale test of reinforced embankment layer (Chai 1992). In this paper, the aforementioned model
with hexagonal wire mesh was also adopted for the weathered clay layer. The strength
parameters were determined from the existing test data on
A test embankment was constructed on the campus of the Bangkok clay (Balasubramaniam et al. 1978). The modulus
Asian Institute of Technology (AIT) in July 1997 of elasticity was obtained from finite element back-analysis
(Voottipruex 2000). The embankment was 6.0 m high, 6.0 m by assuming a Poisson’s ratio of 0.25 (Chai 1992; Bergado
long at the top, 6.0 m wide at the top, and 18.0 m wide at et al. 1995). The modulus of elasticity and the strength pa-
the base, as illustrated in Fig. 1. After 405 days of construc- rameters are shown in Table 2.
tion, the top of the embankment was raised by 1 m as an ad-
ditional surcharge to investigate its behavior further. The
embankment was divided into two parts along its length, and Soft and medium clay layers
each part was constructed with the same backfill, namely
The Cam Clay model has been widely used for represent-
Ayutthaya sand. Zinc-coated and PVC-coated hexagonal
ing the stress–strain relationship of the soft Bangkok clay,
wire mesh reinforcement were used in each part. The gabion
which is normally consolidated and lightly overconsolidated.
facing of the embankment was built with an inclination of
In the PLAXIS program, Cam Clay type models such as the
10° from the vertical. The side slopes and back slope were
modified Cam Clay model and the Soft Soil model are avail-
1:1 as shown in Fig. 1. The configuration of these reinforce-
able. In this paper, the Soft Soil model was used for predict-
ments is illustrated in Fig. 2.
ing the behavior of both the soft clay and medium clay
The soil profile and properties of the subsoil at the em-
layers.
bankment location are shown in Fig. 3. The soil profile is di-
For the numerical modeling, the foundation soil was di-
vided into three layers consisting: (1) the weathered clay
vided into four layers: (1) 2–4 m soft clay, (2) 4–6 m soft
layer, (2) the soft clay layer, and (3) the stiff clay layer. In-
clay, (3) 6–8 m soft clay, and (4) 8–11.5 m medium clay.
struments were installed to observe the settlements, lateral
The Soft Soil model parameters were determined from one-
movements, and excess pore-water pressures (Fig. 1).
dimensional consolidation tests. The available data were also
reproduced from Asakami (1989) as shown in Fig. 4. For
Material behavior models soft Bangkok clay, the measured deformation characteristic
of very soft clay in the laboratory is generally higher than
Generally, the relationship between stress and strain in the that in the field because of the disturbances during sampling
finite element formulation is expressed as and specimen preparation for testing. The disturbance of soil
samples tends to increase the value of the compression ratio,
[1] δ σij = Cδεij CR. Thus, the selected values of CR used in the finite ele-
ment analysis for soft clay layer were the lower bound val-
where δ σij and δ εij are stress and strain increment tensors, ues. The representative CR values are indicated by the solid
respectively; and C is a constitutive matrix that depends on line in Fig. 4. The modified compression index, λ*, has been
the model used in the analysis. found to be CR/2.303. The friction angle of the soft clay
layer was assumed to be 23°, as suggested by
Backfill soil Balasubramaniam et al. (1978) for soft Bangkok clay. The
value of the cohesion intercept was obtained by back-
The backfill soil material used in the embankment was calculation until the settlement was in agreement with the
silty sand, known locally as Ayutthaya sand. The engineer- corresponding results from the full-scale test data. The
ing properties of Ayutthaya sand are given in Table 1 strength parameters are shown in Table 2.
(Bersabe 1992). Since the Mohr-Coulomb failure criteria can For soft clay, the ratio of the compressibility parameter
represent the failure behavior of soil having an apparent co- for virgin compression and for swelling recompression,
hesion, the elastic, perfectly plastic, Mohr-Coulomb model κ*/ λ*, is commonly in the range of 0.2–0.4 (Britto and
was adopted to represent the fill embankment (Long 1996). Gunn 1987). From the results of previous studies (Chai
The apparent cohesion (c) and the friction angle (φ) were de- 1992; Alfaro et al. 1997; Long 1996), it was assumed that
termined from the large direct shear test (Long 1996; κ*/ λ* = 0.2, as suggested by Vermeer and Brinkgreve
Teerawattanasuk 1997). In this paper, the value of the modu- (1995) for all layers of soft clay.
lus of elasticity was obtained by back-calculation assuming The permeability of clays is one of most difficult parame-
a Poisson’s ratio of 0.33 until the settlement was found to be ters to determine in the analysis. The difference between the
in good agreement with the full-scale test data (Youwai laboratory permeability values and those assumed in the
1999). The modulus of elasticity and strength used in this FEM for the upper layer (2–4 m depths) is due to the pres-
analysis are presented in Table 2. ence of the uppermost weathered layer on top of the soft

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:37:44 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1211

Fig. 1. Schematic plan view layout of field instrumentation.

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:37:46 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1212 Can. Geotech. J. Vol. 37, 2000

Fig. 2. The configuration of hexagonal wire mesh reinforcement of test embankment.

Table 1. Engineering properties of Ayutthaya sand (Bersabe 1992).

Specific gravity (Gs) 2.68

Classification of the Unified Soil Classification System SP–SM
Maximum dry density (g/cm3) 1.78
Optimum moisture content (%) 12.2
Angle of friction 38.7° at 15% strain from consolidated, isotropically drained test;
26.6° at 6 mm horizontal displacement from direct shear test
Cohesion —
Relative density Dr (%) 90

clay layer. The presence of fissures in the weathered layer the in situ condition can be higher because of many hetero-
tends to increase its permeability. In addition, the presence geneities in the natural soil. Moreover, a small laboratory
of sand lenses and silt seams in the lower layers also con- specimen cannot represent the actual case in the field. There
tributes to higher in situ permeability compared with the are many uncertainties regarding the Kh/Kv due to the pres-
corresponding laboratory values. Therefore, the soil perme- ence of silt seams, sand lenses, and fissures in the in situ
ability in the upper and lower layers in the field has been as- condition. For most natural deposits, the hydraulic conduc-
sumed to be greater than the laboratory values proposed by tivity in the horizontal direction is higher than that in the
Asakami (1989). Figure 5 illustrates the available laboratory vertical direction (Hansbo 1987; Bergado et al. 1990). The
and field permeability data for soft Bangkok clay at the AIT value of Kh/Kv can vary from 1 to 15 (Jamiolkowski et al.
campus and the estimated average value. In this paper, three 1983).
sets of vertical permeability values were used, namely one, Experimental results have shown that permeability and
two, and five times the value of the vertical permeability es- compressibility vary during consolidation. Therefore, several
timated from the laboratory tests, as shown in Fig. 5. natural clays have been used to study the variation of
The horizontal permeability Kh was assumed to be ap- permeability during consolidation (Tavenas et al. 1983). In
proximately two times the vertical permeability Kv corre- this paper, the variable permeability was also used in the
sponding to the data of Asakami (1989) as plotted in Fig. 5. analysis together with the three sets of aforementioned per-
For the Bangkok clay deposit, the horizontal permeability in meability values. The initial permeability ko used for the

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:37:55 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1213

Fig. 3. General soil profile and properties of the subsoil at the Asian Institute of Technology (AIT) campus. LL, liquid limit; OCR,
overconsolidation ratio; P ′o , effective overburden pressure; Pmax
′ , effective maximum preconsolidation pressure; PL, plastic limit; Su, un-
confined shear strength; VST, vane shear test; wn, natural water content.

Table 2. Selected parameters for soil in FEM analysis.

Soil c′ E γ dry γ wet
Depth (m) modela (kPa) φ ′ (°) λ* κ* ν′ (kN/m2) Kv (×10–4 m/day) (kN/m3) (kN/m3)
0–2 M-C 30 28 — — 0.25 5000 10.0 16 18
2–4 SSM 10 23 0.13 0.026 0.33 — 4.0 13 15
4–6 SSM 10 23 0.12 0.024 0.33 — 0.5 13 15
6–8 SSM 10 23 0.12 0.024 0.33 — 0.5 13 15
8–11.5 SSM 15 23 0.11 0.022 0.25 — 2.0 13 15
11.5–13.0 Elastic — — — — 0.25 7000 20.0 13 15
Backfill M-C 10 25 — — 0.33 6000 — 18 20
Gabion M-C 20 45 — — 0.30 7000 — 18 18
Note: c ′, shear strength intercept; E, modulus of elasticity; Kv, vertical permeability (values of the horizontal permeability Kh are taken as twice those
of the corresponding values of Kv); φ′, effective angle of internal friction; γ dry and γ wet, dry and wet unit weight, respectively; κ*, modified swelling index
( κ * = κ/(1 + e), where κ is the swelling index for SSM parameters and e is the void ratio); λ*, modified compression index ( λ* = λ /(1 + e), where λ is
the compression index for SSM parameters); ν ′, Poisson’s ratio.
a
M-C, elastic-perfectly plastic Mohr-Coulomb; SSM, Soft Soil model.

case of FEM variable k was equal to two times the vertical Stiff clay layer
permeability values from the laboratory test results, as
shown in Fig. 5. The formula proposed by Taylor (1948) In the case of the stiff clay layer (11.5–13.0 m beneath the
was adopted to represent the variation of clay permeability full-scale test embankment), both the modified Cam Clay
during consolidation as follows: model and the Soft Soil model are unsuitable for heavily
overconsolidated clay because they produce extensive yield-
ing on the dry side of the critical state line where the validity
[2] k = ko × 10− ( eo − e )/ C k of models is doubtful (Long 1996). The elastic model was
used for predicting the behavior of this soil layer.
where ko is the coefficient of permeability at the initial state,
eo is the initial void ratio, e is the void ratio, k is the coeffi- Hexagonal wire mesh reinforcement
cient of permeability at the condition under consideration,
and Ck is a constant which can be estimated as half the ini- Bar elements with linear tension strain applied in the
tial void ratio or 0.5eo (Tavenas et al. 1983). PLAXIS program were used to model the hexagonal wire

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:00 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1214 Can. Geotech. J. Vol. 37, 2000

Fig. 4. Variation of compression ratio (CR) with depth of hexagonal wire mesh in air which were carried out by
Bangkok clay at the AIT campus. Cc, compression index; e, Wongsawanon (1998). The axial stiffness of hexagonal wire
void ratio. applied in this analysis is listed in Table 3.

Soil – hexagonal wire mesh interface models

The elastic, perfectly plastic model was used to simulate
the constitutive relationship of the soil – hexagonal wire
mesh interface. In the PLAXIS program, the compression
modulus was related to the shear modulus assuming a fixed
value of Poisson’s ratio of 0.45. The model parameters at the
soil–structure interface can be generated from that of the soil
using the interaction coefficient, Ri, defined as the ratio of
the shear strength of the soil–structure interface to the corre-
sponding shear strength of the soil (Vermeer and Brinkgreve
1995) as follows:
[3] tanδ = Ri tan φ
[4] ci = Ric
[5] Gi = Ri2G
where δ is the friction angle of interface element, ci is the
mesh reinforcement. The required property for this model is cohesion of interface element, Gi is the shear modulus of in-
an axial stiffness, EA, where E is the modulus of elasticity terface element, and G is the shear modulus. Therefore, if
of steel and A is the cross-sectional area of the reinforcing the values of φ, c, and G of the backfill soil have been deter-
wire. The axial stiffness was obtained from tensile tests of mined, the corresponding values of δ, ci , and Gi for the

Table 3. Selected parameters for reinforced element structure in FEM analysis.

Finite element Axial stiffness, EA, per Bending stiffness,
Identification model metre width (kN/m) EI (kN·m)
Hexagonal wire mesh reinforcement Bar element 900 —
Hexagonal wire mesh in gabion facing Beam element 900 12
Note: A, cross-sectional area of reinforcing wire; E, modulus of elasticity of steel; I, moment of inertia.

Fig. 5. Variation of measured coefficient of permeability with depth of Bangkok clay at the AIT campus.

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:03 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1215

Fig. 6. Finite element mesh used for hexagonal wire mesh embankment.

Table 4. Reduction factor from direct shear laboratory testing crushed rock inside the gabion. The Poisson’s ratio of the
(Mir 1996). crushed rock was assumed to be 0.30 (Goodman 1989). Ac-
cording to the Mohr-Coulomb failure criterion, a constraint
Shear stress, Interaction of failure is recognizing a “tension cutoff” superimposed on
Normal pressure, σn (t /m2) τ (t /m2) coefficient, R the Mohr-Coulomb criterion of failure. The actual envelope
Zinc-coated hexagonal wire mesh of critical Mohr’s circles with one principal stress negative
2.5 3.7 1.28 will lie beneath the Mohr-Coulomb criterion with the super-
5.0 5.0 1.14 imposed tension cutoff. Therefore, the c′ value of 20 kPa is
7.5 5.8 1.00 used to represent the shear strength intercept when applying
10.0 7.5 1.03 this failure criterion. The strength parameter and modulus of
PVC-coated hexagonal wire mesh elasticity were obtained by back-calculation until the com-
2.5 3.0 1.02 puted tensile force in the reinforcement agreed with the full-
5.0 4.3 0.98 scale test data. The hexagonal wire mesh in the gabion fac-
7.5 5.5 0.94 ing system was modeled by a beam element model. The pa-
10.0 7.0 0.96 rameters used in this analysis are shown in Table 3.

soil–reinforcement interface can be generated using the sin- Finite element analysis
gle parameter Ri, which can be obtained from direct shear
The hexagonal wire mesh embankment was modeled as a
tests. The mechanism governing soil–reinforcement interac-
plane strain problem. The finite element mesh and boundary
tion has been simplified into pullout of the reinforcement
condition are shown in Fig. 6. The steps in the analysis used
from the soil (the pullout mechanism) and soil sliding along
in the PLAXIS program have to deal with the plastic calcu-
the soil–reinforcement interface (the direct shear mecha-
lation referred to as undrained analysis until the maximum
nism). The direct shear interaction mode dominated the be-
excess pore-water pressure can be achieved. The consolida-
havior at the service condition with some zone having a
tion analysis is then started continuously to dissipate the ex-
pullout interaction mode (Alfaro et al. 1997). However, the
cess pore-water pressure. However, the actual consolidation
interface interaction mechanism for the hexagonal reinforced
process in the test embankment simultaneously occurred in
embankment was dominated by the direct shear interaction
the gradual loading during the embankment construction.
mode. The direct shear interaction coefficient as shown in
Thus, both undrained and consolidation analyses were carried
Table 4 was obtained from the laboratory direct shear test.
out. The effects of construction sequences were considered
In this paper, a direct shear interaction coefficient of 0.9 was
and simulated in the consolidation analysis. The embank-
adopted for all cases.
ment was constructed in three steps and completed within
2 months. After 405 days, the embankment was topped with
Gabion facing system another 1 m of fill to provide an additional surcharge of
The facing was made from large rectangular wire mesh 16.7 kPa. The construction sequence is shown in Fig. 7.
baskets wired together and filled with crushed rock. Each
basket or gabion is 1 m high and consists of a rectangular Surface settlements
unit fabricated from a triple twist of hexagonal mesh of soft-
annealed, heavily galvanized wire. The elastic, perfectly The surface settlement plates were installed 0.45 m below
plastic, Mohr-Coulomb model was used to simulate the the original ground surface. The locations of the settlement

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:06 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1216 Can. Geotech. J. Vol. 37, 2000

Fig. 7. Construction sequence of hexagonal wire mesh embankment. and the field data of the surface settlement plates is shown
in Figs. 8–10. For surface settlement plates S1 and S4 at the
front of the embankment, the back-calculated settlements
from FEM calculation agreed reasonably well with the mea-
sured field data using permeability values two times that of
the laboratory permeability (Fig. 8). The calculated values
using the high permeability yielded a settlement rate that
was too high after construction. For surface settlement plates
S2 and S5 installed at the middle of the embankment and S3
and S6 at the rear of the embankment, the back-calculated
surface settlement values agreed well with the measured
field data when the permeability was assumed to be equal to
the laboratory permeability value as shown in Figs. 9 and
10, respectively. By using a high permeability value five
times that of the laboratory permeability, the calculated val-
ues overpredicted the measured field data. In addition, the
computed surface settlement at the beginning of construction
is greater than that measured in the field due to the partially
drained behavior effect on the soft clay foundation at the
early stages of the construction which can be related to the
method of applying the embankment loading on the con-
struction process or stress level. However, the computed sur-
face settlement values after 100 days agreed well with those
measured in the field because the drained behavior is consis-
tent with the assumed permeability values.

Subsurface settlements
The comparison between the finite element results and the
field data from subsurface settlement plates is shown in
Figs. 11 and 12. Two types of subsurface settlement plates
were installed at depths of 3 and 6 m below the ground sur-
face. For the subsurface settlement plates at 3 m depth (SS1
plates are illustrated in Fig. 1. Surface settlements were ob- and SS3), the computed subsurface settlement values from
tained using permeability values one, two, and five times FEM calculation are consistent with the measured field data
that of the laboratory permeability value (k) and a variable by using the permeability equal to the laboratory value. At
permeability. The comparison between the computed value the beginning of the construction, the computed subsurface
Fig. 8. Comparison of measured and predicted surface settlements at settlement plates S1 and S4 (0.45 m depth at front).

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:10 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1217

Fig. 9. Comparison of measured and predicted surface settlements at settlement plates S2 and S5 (0.45 m depth at middle).

Fig. 10. Comparison of measured and predicted surface settlements at settlement plates S3 and S6 (0.45 m depth at rear).

settlement is overestimated when compared with the mea- quently, the agreement between the numerical results and
sured field data because of the partially drained effect on the observed data was mainly caused by the additional settle-
soft clay layer. However, the computed subsurface settle- ments due to ground subsidence. By using the permeability
ment values after 350 days agreed well with the observed based on two times that of the laboratory value for the vari-
values. For the subsurface settlement plates at 6 m depth able permeability analysis in the FEM, the measured field
(SS2 and SS4), the finite element results of the subsurface data agreed with the computed subsurface settlement values.
settlement values using the computed permeability at two Similar to the settlement characteristics at 3 m depth, the
times that of the laboratory value agreed well with the mea- computed subsurface settlements were overestimated at the
sured field data. Due to excessive groundwater pumping in beginning of construction compared with the measured field
the area, the piezometric drawdown occurred starting from data, whereas the calculated subsurface settlement values
the depth of 6 m from the original ground surface. Conse- underestimated the measured field data after 300 days. From

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:16 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1218 Can. Geotech. J. Vol. 37, 2000

Fig. 11. Comparison of measured and predicted subsurface settlements at settlement plates SS1 and SS3 (3 m depth at middle).

Fig. 12. Comparison of measured and predicted subsurface settlements at settlement plates SS2 and SS4 (6 m depth at middle).

the computed subsurface settlement results, the best settle- fore additional surcharge (405 days after construction started),
ment predictions of the hexagonal wire mesh embankment and (3) after additional surcharge (470 days after construc-
on soft Bangkok clay were obtained using a permeability tion started). The results show that the computed values are
value twice that of the laboratory value and using the vari- reasonable when compared with the measured field data.
able permeability in the finite element calculation. The 5.7 m high reinforced soil test wall – embankment
Settlement profile system using welded steel grid reinforcement was built in
1989 (Bergado et al. 1991). The steel grid reinforcement
The cross-sectional settlement profile of the embankment mats were 2.4 m wide and 5.0 m long and consisted of
at the ground surface is plotted in Fig 13. The settlement 6.0 mm and 5.4 mm diameter bars for longitudinal and
comparisons were made at three stages: (1) immediately af- transverse members, respectively, with grid size of 150 mm
ter construction (60 days after construction started), (2) be- by 225 mm. The reinforcing mats were laid continuously

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:23 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1219

Fig. 13. Comparison of measured and predicted surface settlement profiles for the zinc-coated zone.

Fig. 14. Comparison of surface settlement profiles of different types of the reinforcements.

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:27 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1220 Can. Geotech. J. Vol. 37, 2000

Fig. 15. Comparison of measured and predicted excess pore pressure at 3 m depth.

Fig. 16. Comparison of measured and predicted excess pore pressure at 6 m depth.

side by side with a uniform vertical spacing of 0.45 m. The rigid reinforced soil mass tended to rotate about the toe
other reinforced soil test wall – embankment system using compared with the less rigid reinforced soil mass (Alfaro et
polymer geogrid (Tenax TT 201 SAMP) with a height of al. 1997; Rowe and Ho 1997). Moreover, the underlying
6 m was constructed in 1993 (Alfaro 1996). The polymer weathered crust layer was cut out and then placed back but
grid reinforcements were laid continuously side by side with with less compaction around the perimeter of the steel grid
a vertical spacing of 0.3 m in the lower half and 0.6 m in the embankment to remove the effect of the weathered layer
upper half of the wall. (Shivashankar 1991). In contrast, the weathered layer around
The three full-scale test embankments were constructed the hexagonal wire mesh embankment was not removed.
on the AIT campus with the same profile. The surface settle- Therefore, the weathered layer underneath the hexagonal re-
ment profiles of the steel grid, hexagonal mesh, and geogrid inforced embankment acted like a raft foundation beneath
embankments, in order of decreasing stiffness, at the end of the embankment. Consequently, less settlements were re-
construction are presented in Fig. 14. The differential sur- corded in the hexagonal wire embankment than in the steel
face settlements between the front and the middle of the em- grid embankment, even though their load levels were similar.
bankment in the steel grid embankment are greater than
those in the hexagonal wire mesh embankment and the Excess pore pressure
geogrid embankment. The settlement patterns of the test fa-
cilities with higher stiffness or higher rigidity of reinforced In the PLAXIS program, the pore-pressure boundaries in
soil mass produced higher settlements at the toe. The more the consolidation analysis can be specified as closed flow

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:35 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1221

Fig. 17. Contours of excess pore-water pressure at 60 days (after Fig. 18. Contours of excess pore-water pressure at 405 days (af-
construction started). ter adding surcharge).

Fig. 19. Comparison of measured and predicted lateral displacement profiles.

boundaries or the closed consolidation boundaries required analyses were carried out using a different set of
to ensure flow across the boundaries does not occur (Fig. 6). permeability values, namely one, two, and five times the
The horizontal permeability was assumed to be two times value of vertical laboratory permeability, and a variable per-
the vertical permeability because there are many sand seams, meability based on two times the value of vertical laboratory
silt lenses, and fissures in the in situ conditions. However, permeability, respectively, as mentioned previously. The
the horizontal permeability has not contributed much to the vertical permeability varied from one to five times the aver-
consolidation process according to the FEM simulation. The age laboratory value.
groundwater table must be specified in finite element calcu- In the analysis, the reduction of void ratio due to soil
lations. For the test embankment site, the groundwater table compression during consolidation was considered. Thus, the
is at a depth of approximately 2 m below the ground surface. analysis of variable permeability was also conducted. The
The excess pore pressure is mainly influenced by the initial permeability value for this analysis was two times that
foundation soil permeability. In the FEM calculations, four of the laboratory value. The locations of standpipe

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:38 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1222 Can. Geotech. J. Vol. 37, 2000

Fig. 20. Predicted and measured tensions in zinc-coated hexagonal wire mesh reinforcement.

piezometers are also shown in Fig. 1. Two piezometers were decreased with time until the additional surcharge was added
installed at depths of 3 and 6 m. The comparisons between (Figs. 15, 16). Since the FEM calculation was considered an
the computed values and measured field data are illustrated undrained condition, the predictions of excess pore pressures
in Figs. 15 and 16. The following comparisons can be con- tend to be overestimated. Actually, the consolidation process
sidered: occurred simultaneously with the dissipation of excess pore
(1) By using a permeability value five times that of the pressures during each stage of construction. Therefore, the
laboratory value, the computed maximum excess pore pres- computed excess pore pressures would be more accurate if
sures (at the end of construction) are consistent with the more stages of construction were used in the calculation.
measured data. From the overall results of the FEM analysis, Figures 17 and 18 illustrate the contours of excess pore
the computed values using a permeability value twice that of pressure at the end of construction and after addition of the
the laboratory value agreed well with measured field data, surcharge, respectively. Figure 17 shows that the maximum
and the agreement is better than that obtained using values excess pore pressure at the middle of the embankment is
one or five times that of the laboratory permeability value. 52 kPa at a depth of 4.5 m below the ground surface after
(2) The variable permeability analysis, which used an ini- the end of construction (60 days after construction started).
tial value two times that of the laboratory permeability, After addition of the surcharge (405 days after construction
agreed well with the measured field values, especially during started), the maximum pore pressure is 11 kPa at a depth of
the pore-pressure dissipation after construction. 5 m below the ground surface, as shown in Fig. 18. Fig-
The predictions of excess pore pressures were overesti- ures 15 and 16 also revealed that the excess pore-water pres-
mated 60 days after construction started and then gradually sure mainly increased with the construction process,

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:41 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1223

Fig. 21. Predicted and measured tensions in PVC-coated hexagonal wire mesh reinforcement.

particularly 60 days after construction started. Then, the ex- results of vertical settlements, the partially drained consoli-
cess pore-water pressures dissipated with time until the addi- dation process at very early stages of construction is not
tional surcharge was added which caused the increase of modeled well by the undrained finite element analysis
excess pore-water pressures. Between 200 and 300 days af- (Poulos 1972; Schaefer and Duncan 1988; Chai and Bergado
ter the embankment construction, the pore-pressure data re- 1993). Moreover, the predictions of lateral displacements of
corded by the piezometers gradually increased due to the the inclinometer can be overpredicted due to the effect of
onset of the rainy season, which increased the level of the the inclinometer casing. The influence of the inclinometer
groundwater table. After 405 days of construction, the ex- casing stiffness may result in relative displacements between
cess pore-water pressure gradually reduced until 600 days the soil and casing, especially for the case where the casing
after construction started. deformed to an “S” shape (Bergado et al. 1995). Thus, the
finite element analysis can give values higher than those
measured in the field.
Lateral displacement
The comparison between the FEM results and measured
field lateral displacements is shown in Fig. 19 for 126, 343, Tension force in the reinforcement
and 490 days after the beginning of construction. The com-
puted lateral displacements of the wall face generally agreed The strains and tensile forces in the wire mesh reinforce-
with the measured field data. In the weakest zone of soft ments were measured using self-temperature compensating
clay at depths of 3–4 m below the ground surface, the com- resistance strain gauges. These measured strains were found
puted subsoil lateral displacements overpredicted the mea- to be continually changing, resulting from the movements in
sured data. Since the lateral displacement coincides with the the subsoil and the response of the test embankment to these

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:44 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1224 Can. Geotech. J. Vol. 37, 2000

Fig. 22. Distribution of maximum tensile forces in different types of reinforcements.

movements. From the strain data, the tensions in the wires inforcement, an extensible reinforcement, is lower than that
can be computed as follows: for the active (Ka) condition and is lower than that for the
hexagonal wire mesh reinforcement. The stiffness for each
[6] T = ε EA reinforcement is 38 000, 900, and 120 kN/m for the steel
grid (Chai 1992), hexagonal wire mesh (Youwai 1999), and
where T is the tension in the reinforcing wires, ε is the axial geogrid (Alfaro et al. 1997), respectively. From the FEM
strain in the reinforcing wires, E is the modulus of elasticity of analysis, the results indicate that the tensile force in the rein-
steel, and A is the cross-sectional area of the reinforcing wire. forcement increased with increasing stiffness of the rein-
The tension force distributions along zinc- and PVC- forcement, which is consistent with the previous results by
coated hexagonal wire meshes are shown in Figs. 20 and 21, Rowe and Ho (1997).
respectively. The finite element results and the measured
data for both types of hexagonal wire mesh at 240 days after
construction started agreed well. Immediately after construc- Influence of reinforcement stiffness
tion was finished, the finite element analysis overestimated
the field data due to the limitation of simulation on the con- To demonstrate the effect of the reinforcement stiffness,
solidation process. The tensions in the reinforcements in- further finite element analysis was performed by varying the
creased in response to the compressions of the underlying reinforcement system stiffness, Sr , and the stiffness of the
soft ground foundation. Due to limited data, the maximum foundation, λ*. The reinforcement system stiffness is de-
tension line can be considered to agree with either the tie- fined as follows:
back wedge or the coherent gravity failure plane as illus- EA
trated in Figs. 20 and 21. [7] Sr =
Sv
The distribution of the maximum force in the reinforce-
ment with depth is shown in Fig. 22. The three types of rein- where E is the reinforcement modulus, A is the reinforce-
forcements, hexagonal wire mesh, geogrid, and steel grid, ment cross-sectional area, and Sv is the reinforcement verti-
were compared. The maximum tensile force distribution of cal spacing. The tension force in the reinforcement is
the steel grid is greater than that for the at-rest (Ko) condi- expressed by a normalized force, F*, defined as follows:
tion, whereas the middle portion after the construction is be-
low that for the at-rest condition. The maximum tensile ∑F
[8] F* =
force distribution of the hexagonal wire mesh is below that γ H2
for the active (Ka) condition, whereas the bottom portion, 1
year after the beginning of construction, is greater than that where F is the reinforcement force, γ is the soil backfill unit
for the Ka line. The maximum tensile force of the geogrid re- weight, and H is the height of the embankment. Similar soil

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:46 AM
Color profile: Generic CMYK printer profile
Composite Default screen

Bergado et al. 1225

Fig. 23. Influence of reinforcement stiffness on normalized force of reinforced wall – embankment.

parameters were utilized as tabulated in Tables 3 and 4. The reinforced soil – wall embankment system on a soft
rigid foundation was simulated by changing the weathered foundation through good agreement between the field mea-
and soft clay layers to be rigid material similar to the under- surements and the simulated values. The important simula-
lying stiff clay layer. For a soft foundation, consolidation tion considerations in the FEM analysis consisted of the
settlement was allowed for a period of up to 600 days. method of applying the embankment loading during the con-
The effect of reinforcement stiffness, Sr , on the normal- struction process, the variation of soil permeability during
ized force, F*, is demonstrated in Fig. 23 for both soft and the consolidation process, and the selection of an appropriate
rigid foundations. The normalized force was calculated with model and properties at the interface between the backfill
summation of maximum reinforcement forces and reinforce- soil and hexagonal wire mesh reinforcement corresponding
ment forces at the facing. The results using the conventional to the interaction mechanism. The direct shear mechanism
Rankine and Coulomb methods are also plotted in Fig. 23 generally dominated the behavior of a hexagonal wire mesh
for comparison. The increase in the reinforcement system on a soft ground foundation at working stress conditions.
stiffness, Sr , yielded increased summation of maximum rein- The direct shear interaction coefficient of 0.9 was used.
forcement forces and reinforcement forces at the facing loca- Compared with other grid reinforcements with varying
tions. These results are consistent with the previous work of stiffnesses, the finite element analysis indicated that the ten-
Rowe and Ho (1997). In addition, the tensile reinforcement sile force in the reinforcement increased with an increase in
forces in the soft foundation are greater than the correspond- the reinforcement stiffness. Furthermore, the tensions in the
ing values in the rigid foundation. The increased tension in reinforcements tend to increase in response to an increase in
the reinforcement due to the compression of the underlying the compression of the underlying soft foundation. More-
soft foundation is consistent with the measured values in the over, increasing the reinforcement stiffness with a conse-
full-scale test embankment. quent increased rigidity of the reinforced mass tends to
produce greater settlements at the toe and more forward ro-
Conclusions tation of the reinforced embankment.

The numerical simulation based on finite element analysis

was carried out to study the behavior of a hexagonal wire Acknowledgement
mesh embankment on soft ground foundation. The numeri-
cal simulation techniques adopted in this paper gave a rea- Heartfelt thanks and gratitude are given to Ir. Alan Pun of
sonable representation of the overall behavior of the B.B. Trading, Kuala Lumpur, Malaysia, for donating the

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:50 AM
Color profile: Generic CMYK printer profile
Composite Default screen

1226 Can. Geotech. J. Vol. 37, 2000

Camel Brand hexagonal wire reinforcement and for sponsor- Hansbo, S. 1987. Design aspects of vertical drains and lime col-
ing the research project. umn installations. In Proceedings of the 9th Southeast Asian
Geotechnical Conference, Bangkok, Vol. 2, pp. 8-1–8-12.
Jamiolkowski, M., Lancellotta, R., and Wolski, W. 1983. Pre-
compression and speeding up consolidation. In Proceedings of
References the 8th European Conference on Soil Mechanics and Foundation
Engineering, General Report, Special Session 6. A.A. Balkema,
Ahmed, M.M. 1977. Determination of permeability profile of soft Rotterdam, The Netherlands, pp. 1201–1226.
Rangsit clay by field and laboratory tests. AIT thesis 1002, Jamshed, A.D. 1975. In-situ and laboratory permeability of Bang-
Asian Institute of Technology, Bangkok, Thailand. kok clay at Nong Ngu Hao and Rangsit. AIT thesis 765, Asian
Alfaro, M.C. 1996. Reinforced soil wall – embankment system on Institute of Technology, Bangkok, Thailand.
soft foundation using inextensible and extensible grid reinforce- Kapurrapu, R., and Bathurst, R.J. 1995. Behavior of geosynthetic
ments. Doctoral dissertation, Saga University, Saga, Japan. reinforced soil retaining walls using finite element method.
Alfaro, M.C., Hayashi, S., Miura N., and Bergado, D.T. 1997. De- Computers and Geotechnics, 17: 279–299.
formation of reinforced soil wall-embankment system on soft Long, P.V. 1996. Behavior of geotextile-reinforced embankment on
clay foundation. Soils and Foundations, 37(4): 33–46. soft ground. Doctoral dissertation GT 96-1, Asian Institute of
Asakami, H. 1989. The smear effect of vertical band drains. AIT Technology, Bangkok, Thailand.
thesis GT-88-8, Asian Institute of Technology, Bangkok, Thai- Mir, E.N. 1996. Pullout and direct shear test of hexagonal wire
land. mesh reinforcements in various fill materials including lahar
Balasubramaniam, A.S., Hwang, Z.M., Uddin, W., Chaudhry, A.R., From Mt. Pinatubo, Philippines. AIT thesis GE 95-18, Asian In-
and Li, Y.G. 1978. Critical state parameters and peak stress en- stitute of Technology, Bangkok, Thailand.
velopes for Bangkok clays. Quarterly Journal of Engineering Poulos, H.G. 1972. Difficulties in prediction of horizontal defor-
Geology, 11: 219–232. mations in foundations. Journal of the Soil Mechanics and
Foundations Division, ASCE, 98(SM8): 843–848.
Bergado, D.T., Enriques, A.S., Sampaco, C.L., and Alfaro, M.C.
Rowe, R.K., and Ho, S.K. 1997. Continuous panel reinforced soil
1990. Inverse analysis of geotechnical parameters on improved
walls on rigid foundation. Journal of Geotechnical and
soft Bangkok clay. Journal of Geotechnical Engineering, ASCE,
Geoenvironmental Engineering, ASCE, 123(10): 912–920.
118(7): 1012–1030.
Schaefer, U.R., and Duncan, J.M. 1988. Finite element analyses of
Bergado, D.T., Sampaco, C.L., Shivashankar, R., Alfaro, M.C., An-
the St. Alban test embankments. American Society of Civil En-
derson, L.R., and Balasubramaniam, A.S. 1991. Performance of
gineers, Geotechnical Special Publication 18, pp. 158–177.
welded wire wall with poor quality backfills on soft clay. In Pro-
Shivashankar, R. 1991. Behavior of mechanically stabilized earth
ceedings of the American Society of Civil Engineers
(MSE) embankment with poor quality backfills on soft clay de-
Geotechnical Engineering Congress, Boulder, Colo., pp. 909–
posits, including a study of the pullout resistances. D.Eng. dis-
922.
sertation, Asian Institute of Technology, Bangkok, Thailand.
Bergado, D.T., Chai, J.C., Abiera, H.O., Alfaro, M.C., and Tavenas, F., Jean, P., Leblond, P., and Leroueil, S. 1983. The per-
Balasubramaniam, A.S. 1993. Interaction between cohesive- meability of natural soft clays. Part II. Permeability characteris-
frictional soil and various grid reinforcements. Geotextiles and tics. Canadian Geotechnical Journal, 20: 645–660.
Geomembranes, 12(4): 327–349. Taylor, D.W. 1948. Fundamentals of soil mechanics. John Wiley &
Bergado, D.T., Chai, J.C., and Miura, N. 1995. FE analysis of grid Sons, Inc., New York.
reinforced embankment system on soft Bangkok clay. Com- Teerawattanasuk, C. 1997. Interaction and deformation behavior of
puters and Geotechnics, 17: 447–471. hexagonal wire mesh reinforcement at vicinity of shear on sand
Bersabe, N.D. 1992. Engineering properties of locally available and volcanic ash (lahar) backfill. AIT thesis GE-96-14, Asian
clayey to silty sand and their applicability as sand compaction Institute of Technology, Bangkok, Thailand.
pile material. AIT thesis GT 91-20, Asian Institute of Technol- Vermeer, P.A., and Brinkgreve, R.B.J. (Editors). 1995. Finite ele-
ogy, Bangkok, Thailand. ment code for soil and rock analysis. A.A. Balkema, Rotterdam,
Britto, A.M., and Gunn, M.J. 1987. Critical state soil mechanics The Netherlands.
via finite elements. Ellis Horwood, Chichester, U.K. Voottipruex, P. 2000. Interaction of hexagonal wire reinforcement
Chai, J.C. 1992. Interaction behavior between grid reinforcements with silty sand backfill soil and behavior of full scale embank-
and cohesive frictional soils and performance of reinforced ment reinforced with hexagonal wire. AIT doctoral dissertation,
wall/embankment on soft ground. Doctoral dissertation GT 91- Asian Institute of Technology, Bangkok, Thailand.
1, Asian Institute of Technology, Bangkok, Thailand. Wongsawanon, T. 1998. Interaction between hexagonal wire mesh
Chai, J.C., and Bergado, D.T. 1993. Performance of reinforced em- reinforcement and silty sand backfill. AIT thesis GE 97-14,
bankment or Muar clay deposit. Soils and Foundations, 33(4): Asian Institute of Technology, Bangkok, Thailand.
1–17. Youwai, S. 1999. Finite element modeling of hexagonal wire mesh
Goodman, R.E. 1989. Introduction to rock mechanics. 2nd ed. embankment on soft foundation. AIT thesis GE 98-13, Asian In-
John Wiley & Sons, Inc., New York. stitute of Technology, Bangkok, Thailand.

© 2000 NRC Canada

I:\cgj\Cgj37\Cgj06\T00-065.vp
Monday, December 04, 2000 10:38:51 AM
View publication stats