9th International Symposium on Lowland Technology
September 29-October 1, 2014 in Saga, Japan
NUMERICAL SIMULATIONS USING FEM 2D AND FEM 3D COMPARED TO OBSERVED
BEHAVIOR OF REINFORCED FULL SCALE EMBANKMENT
S. Shrestha1, P. Baral1, D.Bergado2, J.C. Chai3, T. Hino4
ABSTRACT: A full scale test embankment (6 m height) was constructed by Department of Highways, the Bureau of Road
Research and Development in Phitsanulok, Thailand. A surcharge fill of 1.2 m thick without reinforcements was added at the
top of the embankment equivalent to 2 tsm of load. One side of this embankment was reinforced with polymeric
reinforcements consisting polyester (PET), Polypropylene (PP) and high density polyethylene (HDPE) and referred as
reinforced steep slope (RSS), which is at an angle of 70 degrees from horizontal. The other side of the embankment was
reinforced with metallic reinforcements consisting of metallic strips (MS) and steel wire grids (SWG) combined with precast
concrete panel and termed as mechanically stabilized earth wall (MSEW). The comparisons of these reinforcing materials in
terms of stiffness from highest to lowest are metallic strip (MS), steel wire grids (SWG), polypropylene (PP), high density
polyethylene (HDPE) and polyester (PET). The behavior of the test embankment on hard foundation was analyzed and
compared with the simulated results using PLAXIS software. The FEM 2D and the FEM 3D simulations were compared with
the observed data. The results obtained from FEM 3D have good agreement with the field measurements in terms of vertical
and lateral deformations of the embankment. However, there were slight differences when compared with the FEM 2D
simulations due to its limitations.
Keywords: FEM 3D, FEM 2D, metallic reinforcement, polymer reinforcement, hard foundation, test embankment
BACKGROUND and direct shear mechanisms as well as the behavior of
hexagonal wire mesh reinforced embankment with silty
Soil reinforcement, has become a widely used sand backfill. Bergado and Teerawattanasuk 2008
earthwork construction method that provides technically compared the reliability of FEM 2D and FEM 3D by
attractive and cost-effective grade separations at the ground studying two full-scale embankments; steel grid
surface. The increasing use and acceptance of soil embankment having longer plan dimensions with length-to-
reinforcement has been triggered by a number of factors width ratio of 3.0 (long embankment) and hexagonal wire
including cost savings, aesthetics, simple and fast mesh reinforced embankment having shorter plan
construction techniques, good seismic performance, and the dimensions with length-to-width ratio of 1.0 (short
ability to tolerate large total and differential settlement embankment). The actual behavior of the steel grid
without structural distress. reinforced long embankment corresponded more closely to
Numerical analysis has been proved as a powerful and the results of the 2D numerical simulations. Furthermore,
convenient tool for predicting the performance of the MSE the actual behavior of the hexagonal wire mesh reinforced
wall (Bergado et al. 2003). Furthermore, many investigators short embankment corresponded more closely to the results
have demonstrated the use of the finite element method of the 3D numerical simulations. The geometric effects
(FEM) in the analysis of reinforced soil wall (Chai and were important factors that affected the results of the
Bergado 1993 a;b Bergado and Chai 1995; Tanchaisawat et numerical simulations. Suksiripattanapong et al., 2012
al. 2008). This paper deals with modeling of the full-scale performed numerical simulation on the bearing
test Mechanically Stabilized Earth wall/embankment by reinforcement earth wall constructed on the hard stratum
FEM 2D and FEM 3D as compared to the observed data. using FEM 2D. The behavior of the BRE wall was
Bergado et al. 2003 suggested FEM under plane strain simulated satisfactorily and agreed well with the predictions
condition can be successively utilized to analyze the pullout in terms of changes in foundation settlements, bearing
1
M. Eng. Graduate, Asian Institute of Technology, Bangkok, Thailand, [email protected]
2
Guest Professor, Institute of Lowland and Marine Research (ILMR), Saga University, Saga, Japan
3
Professor, Department of Civil Engineering and Architecture, Saga University, Saga, Japan
4
Professor, Institute of Lowland and Marine Research (ILMR), Saga University, Saga, Japan
Shrestha, et al.
stresses, lateral earth pressures and tensions in the MODEL PARAMETERS
reinforcements during and after construction.
A full scale test embankment (6 m height) constructed Backfill Materials
in Phitsanulok, Thailand consist of Reinforced Steep Slope
(RSS) and Mechanically Stabilized Earth Wall (MSEW) The material used as backfill in the embankment
with high strength polymeric geogrid and metallic consisted of 50% lateritic soil mixed with 50% silty sand
reinforcements. The behavior of this embankment on hard (by volume) and has moisture content and dry unit weight
foundation were re-analyzed with FEM 2D and compared as 7% and 22.05 kN/m3, respectively. The friction angle
with the simulated results from previous studies (FEM 3D). and cohesion of this backfill material obtained from direct
The soil and reinforcement parameters along with the shear test were 42 degrees and 80 kPa. From triaxial 1 (CU)
interface strengths of the MSE embankment/wall were test, the effective friction and cohesion were 37 degrees and
back-analyzed. The input parameters for metallic and 20 kPa, respectively. Moreover, from triaxial 2 (CU) test,
polymeric reinforcements were obtained from laboratory the effective friction angle and cohesion were 32.8 degrees
testing at AIT. Similarly, the input parameters for the and 0 kPa, respectively. Thus, the effective friction angle
backfill materials and surcharge were also initially obtained varied from 32.8 degrees to 37 degrees, and effective
from the laboratory test and then back calculated until the cohesion varied from 0 to 20 kPa, as obtained from two
suitable parameter was obtained. different triaxial (CU) tests.
DESCRIPTION OF THE MSE EMBANKMENT Table 1 Soil properties of the backfill soil
In the full scale embankment, reinforced steep slope Property Lateritic Soil Mixed with Sand
(RSS) of 70 degrees from the horizontal with soil bags as (50:50 by volume)
facing was utilized in one side whereas mechanically Atterberg Limit Test LL = 20.8%, PL=17.3 %,
stabilized earth wall (MSEW) with concrete panel as facing PI=3.5%.
was used in another side. Both facing (RSS and MSEW) Sieve Analysis Test Sample No.1
were designed up to a height of 6m. A surcharge fill 1.2 m Percent Finer=0.94%, Cu=40,
thick was later added without reinforcement at the top of Cc=0.34
the embankment equivalent to 2 tsm load. The length of the Sample No.2
embankment was 18 m and width was 15m. Polyester Percent Finer=0.14%,
(PET), polypropylene (PP) and high density polyethylene Cu=42.86, Cc=0.55
(HDPE) geogrids were polymeric reinforcements in the Unified Classification Poorly Graded Sand (SP)
reinforced steep slope (RSS) whereas metallic strips (MS) AASHTO A-2-4(0)
and steel wire grid (SWG) were metallic reinforcements in Classification
mechanically stabilized earth wall (MSEW) facing of the Compaction Test Maximum Dry Unit Weight
embankment. (d) = 22.05 kN/m3
The vertical spacing between each reinforcement layer Optimum Moisture Content
was 0.5 m and the length was 5 m while upper layers of (OMC) = 7.0%
metallic strip from layer 7 to layer 12 had 5.80 m length. California Bearing CBR=50.5%
Different monitoring instruments were installed to monitor Ratio (CBR) Test
vertical and lateral displacements, total stresses, excess pore Direct Shear Test Friction Angle, ’ =42 degrees
water pressure, groundwater and strains in reinforcing Cohesion,C’ = 80 kPa
material. These include inclinometers, settlement plates, Triaxial 1 (CU) Test Friction Angle, ’ =37 degrees
total pressure cells, standpipe piezometers, vibrating wire Cohesion,C’ = 20 kPa
strain gauges and fiber optic strain gauges. In addition,
Triaxial 2 (CU) Test Friction Angle, ’ =32.8
observation wells were installed to monitor the level of
degrees
groundwater at the dummy area located more than 10.0 m Cohesion,C’ = 0 kPa
from the embankment. The plan and section views of the
Plane strain from ps’=38.5 degrees
embankment are shown in Figs. 1 a, b.
Triaxial 1 (CU) Test
Plane strain from ps’=32.2 degrees
Triaxial 2 (CU) Test
pH Value 6.16
Organic Content 0.9918%
Resistivity 5088 -cm
Numerical Simulations using FEM 2D and FEM 3D Compared to observed behavior of reinforced full scale embankment
For the plane strain condition the friction angle is converted Precast concrete Panel Facing
by Lad and Lee (1976) formula as ’=38.5 degrees and 32.2
degrees, respectively. The various properties of backfill The precast concrete panel was used as wall facing. The
material are tabulated in the Table 1. dimensions of the panel are 1.5 m width, 1.5 m height and
0.15 m thick. In this study, the precast concrete panels were
Foundation Soils modeled using plate elements.. The properties of concrete
panel facing are tabulated in Table 2.
The soil profile in Phitsanulok Province consisted of
generally hard ground. One borehole was located (BH-1) in Table 2 Material properties of concrete panel facing
the middle of the embankment. Three additional boreholes
BH-2, BH-3 and BH-4 were drilled adjacent to the Parameter Name Value Unit
embankment near the RSS facing in order to obtain more Type of behavior Material type Elastic
data on the soil profile. The borehole locations and Normal stiffness EA 42000000 kN/m
instrumentations at MSE wall/embankment are shown in Flexural rigidity EI 78500 kN.m2/m
Fig.1a. Fig.1b indicates the cross section of MSE wall. The Equivalent thickness d 0.15 m
soil profiles are from dense to very dense clayey sand to Weight w 3.6 kN/m/m
hard silty clay. The level of the groundwater was found at Poisson's ratio 0.15 -
2m depth below the ground surface. Model Plate
Metallic and Polymeric Reinforcement
The metallic and polymeric reinforcement were
modeled as geogrid material in FEM 2D, specifically
designed to simulate the behavior of thin, flat and discrete
reinforcing strips. The Strip element can yield in
compression and tension. The axial stiffness was obtained
from laboratory tests. The various properties of
reinforcements are tabulated in the Table 3. The
comparison of the reinforcement stiff nesses from highest
to lowest is as follows: metallic strip (MS), steel wire grid
(SWG), polypropylene (PP), high density polyethlene
Fig. 1a Plan and instrumentations of MSE wall/ (HDPE) and polyester (PET).
embankment
Table 3 Material properties of reinforcements
Tensile Thick- Normal
Material Strength ness Stiffness,
Name Model (kN/m) (mm) EA (kN/m)
Metallic Strip
(MS) Geogrid 277.6 4.00 88000
Steel Wire
Grid (SWG) Geogrid 128.1 6.00 35000
Polyester
(PET) Geogrid 83.6 1.50 925
Polypropylen
e (PP) Geogrid 91.9 1.45 1360
Fig. 1b Cross section of MSE wall/embankment indicating High Density
the locations of monitoring instruments Polyethylene
(HDPE) Geogrid 85.8 1.91 1320
Shrestha, et al.
Soil / Reinforcement Interfaces on the foundation soil layer by layer. After the placement of
the compacted fill layer the reinforcement was placed at
Interface elements were attached on the grid elements in interval of 0.5 m vertical spacing per stage until the
order to simulate the frictional interaction between the completion of full height of the embankment. During this
geogrid and the backfill soil. The various properties of R construction stage drained Mohr Coulomb analysis is used
interface parameter from Large-Scale Direct Shear Test to simulate the layer by layer construction. After the
results are tabulated in the Table 4. completion of the full height of the embankment drained
analysis is used to simulate the consolidation process for
Table 4 Interface strengths from large-scale direct shear test 186 days. The constitutive model to simulate the behavior
result of the reinforced backfill used was a linear elastic,
perfectly-plastic model with Mohr-Coulomb’s failure
Friction criterion. This constitutive model was characterized by five
Cohesion, parameters: elastic parameters (E: Young modulus, ν:
Soil to angle, Rinter
c (kPa) Poisson’s ratio) and plastic parameters (ϕ: friction angle, c:
Ф (º)
Soil 40 23 1.00 cohesion, and ψ: dilatancy angle).
Steel
Strip 36 23 0.87
Steel
Grid 40 28 1.00
Miragrid
GX80/30
PET 33 21 0.79
Secugrid
80/80 Q1
PP 35 25 0.83
TT 090
SAMP
HDPE 33 24 0.77
NUMERICAL SIMULATIONS AND STAGED Fig. 2 FEM 2D of full scale reinforced test embankment
CONSTRUCTION OF THE EMBANKMENT
The numerical model of the MSE wall/embankment was
performed using FEM 2D (PLAXIS V 8.2, 2009). The
program allowed for plane strain idealizations including
simulation of the construction sequences. The FEM
software used required material properties (Table 5) to be
established and explicitly model the soil, facing panels,
reinforcement layers, and the interfaces. As shown in the
Fig. 2, the mesh is created and the nodal points at the
bottom boundary were fixed in both directions, and those
on the side boundaries were fixed only in the horizontal
direction. The FEM 3D (PLAXIS 3D, 2011) was done by
Baral (2012).The mesh created in FEM 3D is shown in Fig.
3 respectively. The side boundaries were placed at a
distance of two times the width of the embankment, and the
bottom boundary was fixed up to the known soil layer.
Such distances and assumed boundary conditions are Fig. 3 FEM 3D of full scale reinforced test embankment
considered to approximately simulate the semi-infinite
extent of the system. The in-situ stresses in the foundation
soil were generated by the Ko procedure. Then, the backfill
which was divided into 13 layers, as in the field was placed
Numerical Simulations using FEM 2D and FEM 3D Compared to observed behavior of reinforced full scale embankment
RESULTS AND DISCUSSIONS: larger than MS, since there is large difference in stiffness
between HDPE and MS.
Lateral Displacements
Comparison of FEM 2D and observed data
The influence parameters to lateral displacements of the
test embankment were investigated. The soil-reinforcement
interaction parameter(Rinter), modulus of elasticity (E’),
cohesion (C’) were varied to study their influences in lateral
displacements. Table 5 shows the interface parameters
(Rinter) that were used in the analysis. Based on the
simulated results, the total horizontal displacement caused
by foundation settlement was analyzed and compared with
the observed data. The lateral displacements of HDPE on
RSS side, MS on MSEW side obtained from field
measurement by inclinometers were compared with the data
from numerical simulations at 186 days after the end of the
construction. Figs. 4 to 5 show the measured and simulated
lateral deformations at PE and MS reinforcements,
respectively. Inclinometers I3 and I5 were used to measure
the lateral displacement for HDPE and MS section in the
field.
Comparison of FEM 2D (Shrestha, 2013) versus FEM 3D
(Baral, 2012) simulations
HDPE cross-section
According to the FEM 2D analyses, the lateral
displacement of PE geogrid after 186 days was found to be
7 mm. The large lateral displacement was observed at the
bottom-middle height of RSS facing. The predicted results
Fig. 4 Observed and simulated lateral displacements in
from FEM 2D slightly over predicted both the field
HDPE
measured lateral displacement and results from FEM 3D
(Baral, 2012) on the lower half of the embankment height Vertical Settlements
and under predicts for the upper half of the embankment
(Fig. 4). Surface and subsurface settlement plates were installed in
MS cross-section the embankment at different heights such as S1 to S15 at
According to the FEM 2D analyses, the lateral the foundation level - 0m and S31 to S45 at the top level-
displacement of SWG geogrid after 186 days was found to 5.5m to measure the vertical settlements. The settlement
be 4 mm. The predicted results from FEM 2D slightly over prediction on the MSE wall/embankment on hard ground
predicted both the field measured lateral displacement and foundation mostly depended on the uppermost Dense to
Very Dense Clayey Sand layer. Settlements were affected
results from FEM 3D (Baral, 2012) (Fig. 5).
when the permeability values of the subsoil layer were
varied.
The FEM 2D (Shrestha, 2013) yielded the overall behavior
of the test embankment quite closer in the MSEW facing
Comparison of FEM 2D and observed data
i.e. MS reinforced side than the RSS facing i.e. HDPE
Surface and subsurface settlement plates were installed
reinforced side when compared with the observed data.
in the embankment at different height to measure the
The results showed that observed and FEM 2D simulated
vertical settlements. Due to the construction of
displacements were under predicted above 4.50 m and over
embankment in the hard ground; the values of vertical
predicted till 4.50m height of the embankment. These may
settlements were relatively low. The settlement profile of
be due to the limitations under plane strain conditions.
the section PE-MS at different levels of the embankment
The lateral displacement of geogrids HDPE after 186 days
(0.00m at the base of embankment and 5.50m at the top
were 16mm. The lateral displacement for HDPE was found
height of embankment) are plotted together with the
simulated data in Figs. 6 to 7 ,respectively.
Shrestha, et al.
Comparison of FEM 2D (Shrestha, 2013) versus FEM 3D
(Baral, 2012) simulations
HDPE-MS cross-Section
The maximum settlement at the base of the
embankment (Level 0.00m) ranged from 50 to 70 mm at
186 days after construction. The compression of the
foundation was found to increase slightly towards the
middle, as shown in Fig. 6 for this section. Similarly, the
compression of the embankment (Level 0.00m to Level
5.50m) varied between 20 to 40 mm and is shown in Fig. 7
for this section. The FEM 2D over predicted the vertical
settlement when compared with FEM 3D (Baral, 2012) for
this section. The simulated results from FEM 2D over
predicted vertical settlements in comparison to the
simulated result of FEM 3D (Baral,2012). The over
prediction may be due to the limitation of analyses under
plane strain condition.
Fig. 5 Observed and simulated lateral displacements in MS
Table 5 Material conditions and parameters used in the analysis
sat unsat c'
Mo Depth(m Conditi Kx(m/ Ky(m/ Ф'ps
Soil Description (kN/m3 (kN/m E (kPa) (kPa
del ) on day) day) (º)
) 3) )
Draine
Backfill material MC - d 22.7 21.0 0.8 0.4 0.37 12,000 10 38.5
Dense to very 0.00- Draine
dense MC 2.00 d 19.0 17.0 0.001 0.0005 0.35 20,000 15 30
clayey sand
2.00- Draine
Loose MC 4.00 d 18.0 16.0 0.001 0.0005 0.35 15,000 7 34
clayey sand
4.00- Draine
Medium dense MC 5.50 d 18.0 16.0 0.001 0.0005 0.35 25,000 5 36
clayey sand
5.50- Undrai 0.0000
Very stiff to hard MC 10.00 ned 19.0 17.0 0.0001 5 0.35 50,000 70 28
silty clay
11.50- Draine
Dense clayey sand MC 13.00 d 19.0 17.0 0.001 0.0005 0.35 30,000 8 33
13.00- Undrai 0.0000
Hard silty clay MC 21.45 ned 22 20 0.0001 5 0.35 80,000 100 26
Rint(PET)=0.79 , Rint(PP)=0.83 , Rint(HDPE)=0.77 , Rint(SWG)=1.0 , Rint(MS)=0.87
Numerical Simulations using FEM 2D and FEM 3D Compared to observed behavior of reinforced full scale embankment
and vertical settlements in the MSEW faced side was less in
comparison with the RSS faced side from monitored data.
The RSS side has more settlement and displacement as a
result of lower stiffness than MSEW side. The stiffness
was found from highest to lowest as metallic strip (MS),
steel wire grids (SWG), polypropylene (PP), high density
polyethylene (HDPE) and polyester (PET). The results
obtained from 3D finite element method simulation (FEM
3D) have good agreement with the field measurements in
terms of vertical and lateral deformations in the
reinforcements of the embankment. However, there were
slight differences when compared with the FEM 2D
simulations due to its limitations. The discrepancy between
the measured data and the simulated data may be due to
some limitations of the boundary conditions in FEM 2D for
Fig. 6 Compression of the foundation at 186 days (Level shorter embankments with asymmetric structure and
0.00m) in HDPE-MS
varying physical properties in short distance apart. FEM 3D
is better for analyzing the behavior of the embankment with
shorter plan dimensions than FEM 2D. The boundary value
problem, the effects of boundary conditions (2D or 3D)
applied in numerical analysis should be considered as
important factors that may affect the numerical results.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the Department of
Highway, Thailand for supporting this research.
REFERENCES
Baral P., 2012. Simulation and Back-analyses of Design
Parameters of MSE wall/embankment on Hard
Fig. 7 Compression of the foundation at 186 days (Level Foundations using PLAXIS 3D Software. M. Eng.
5.5m) in HDPE-MS Thesis No. GE-11-, AIT, Bangkok, Thailand
Bergado, D.T. and Chai J. C. (1995). FE analysis of Grid
reinforced embankment system on soft bangkok clay.
CONCLUSIONS Computers and Geotechnics:17, 447-471.
Bergado, D.T., S. Youwai, C. Teerawattanasuk, P.
The full scale test embankment (6 m height) was Visudmedanukul (2003). The interaction mechanism
constructed by Department of Highway, the Bureau of and behavior of hexagonal wire mesh reinforce
Road Research and Development in Phitsanulok, Thailand embankment with silty sand backfill on soft clay.
on hard foundation. One side of this embankment was Computers and Geotechnics 30 :517-534.
reinforced with polymeric reinforcements consisting Bergado, D.T. and Teerawattanasuk, C. (2008). 2D and 3D
polyester (PET), Polypropylene (PP) and high density numerical simuations of reinforced embankments on
polyethylene (HDPE) and referred as reinforced steep slope soft ground. Geotextiles and Geomembranes, 20(6):
(RSS), which is at an angle of 70 degrees from horizontal. 343-365.
The other side of the embankment was reinforced with Chai, J.C., Bergado, D.T., 1993a. Performance of
metallic reinforcements consisting metallic strips (MS) and reinforced embankment on Muar clay deposit. Soils and
steel wire grids (SWG) combined with precast concrete Foundations 33(4), 1–17.
panel and termed as mechanically stabilized earth wall Chai, J.C., Bergado, D.T., 1993b. Some techniques for FE
(MSEW). The comparisons of the behavior of both metallic analysis of embankment on soft ground. Canadian
and polymeric reinforcements were monitored and Geotechnical Journal 30,710–719.
observed in terms of stiffness. The lateral displacements
Shrestha, et al.
PLAXIS 2D-Version (8.2), PLAXIS Manual, PLAXIS b.v., Tanchaisawat, T. , Bergado , D.T., Voottipruex, P. (2008)
The Netherlands. Numerical simulation and sensitivity analyses of full-
PLAXIS 3D-Version, 2011. PLAXIS Manual, PLAXIS scale test embankment with reinforced lightweight
b.v., The Netherlands. geomaterials on soft Bangkok clay. Geotextiles and
Suksiripattanapong, C., Chinkulkijniwat, A., Horpibulsuk, Geomembranes 26: 498-511
S., Rujikiatkamjorn, C. and Tanhsutthinon, T. (2012). Shrestha S., 2013. Reanalysis of full scale test embankment
Numerical analysis of bearing reinforcement earth wall, with polymer and metallic reinforcement on hard
Geotextiles and Geomembranes 32: l28-37. ground using FEM-PLAXIS 2D. M. Eng. Thesis No.
GE-13-, AIT, Bangkok, Thailand
