Sei sulla pagina 1di 17

ARTICLE IN PRESS

Geotextiles and Geomembranes 26 (2008) 3955 www.elsevier.com/locate/geotexmem

2D and 3D numerical simulations of reinforced embankments on soft ground


Dennes T. Bergadoa,, Chairat Teerawattanasukb
b

School of Engineering and Technology, Asian Institute of Technology, P.O. Box 4, Klong Luang, Pathumthani 10120, Thailand Department of Civil and Environmental Engineering Technology, King Mongkuts Institute of Technology North Bangkok, Bangkok 10800, Thailand Received 22 April 2006; received in revised form 14 February 2007; accepted 21 March 2007 Available online 30 May 2007

Abstract Utilizing the same constitutive models and properties of foundation soils as published by previous researchers, two full-scale test embankments, namely steel grid embankment having longer plan dimensions with length-to-width ratio of 3.0 (long embankment) and hexagonal wire mesh reinforced embankment having shorter plan dimensions with length-to-width ratio of 1.0 (short embankment), were investigated using numerical simulation in two-dimensional (2D) and three-dimensional (3D) explicit nite-difference programs, FLAC2D and FLAC3D, respectively. The 2D numerical analysis simulated the overall behavior of the steel grid reinforced long embankment with very reasonable agreement between the eld measurements and the calculated values. On the other hand, the 3D numerical analysis simulated the overall behavior of the hexagonal wire mesh reinforced short embankment. Furthermore, the simulation results from the FLAC3D used in the 2D analysis agreed with the measured settlement data in the long embankment as well as the 2D predictions from FLAC2D. The 2D and 3D numerical analyses should be considered important factors that may affect the numerical simulation results which are consistent with the current settlement predictions with SkemptonBjerrum corrections. r 2007 Elsevier Ltd. All rights reserved.
Keywords: Reinforced embankment; Numerical simulation; Soft ground; Full-scale test

1. Introduction Issues related to the design and factors affecting the performance of reinforced soil have been addressed by many researches in recent times (e.g. Bathurst et al., 2005; Kazimierowicz-Frankowska, 2005; Park and Tan, 2005; Skinner and Rowe, 2005; Varsuo et al., 2005; Al Hattamleh and Muhunthan, 2006; Hufenus et al., 2006; Nouri et al., 2006). Also, the behavior of reinforced earth structures has been comprehensively studied through eld observation of full-scale physical model, laboratory model testing, and numerical simulation. However, the cost of constructing and monitoring full-scale reinforced test embankments is quite high. An alternative method such as a numerical experiment or simulation by means of appropriate methods such as nite-element (FE) or nite-difference
Corresponding author. Tel.: +662 524 5512; fax: +662 524 6050.

E-mail addresses: bergado@ait.ac.th (D.T. Bergado), chairatsuk@yahoo.com (C. Teerawattanasuk). 0266-1144/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.geotexmem.2007.03.003

(FD) techniques (e.g. Ho and Rowe, 1994) is essentially required. In general, two-dimensional (2D) analysis can be categorized into two types: (1) 2D plane stress which is usually applied for stress analysis of thin plate structure by assuming the stress in the direction perpendicular to the plate is equal to zero and (2) 2D plane strain which is dened as the strain state in the direction perpendicular to the plane is equal to zero. Most researches assumed plane strain condition for numerical simulations of reinforced earth structures (Chai, 1992, Chai and Bergado, 1993a, b; Bergado et al., 1995, 2003; Karpurapu and Bathurst, 1995; Alfaro et al., 1997; Chai et al., 1997; Rowe and Ho, 1998; Rowe and Li, 2002; Zdravkovic et al., 2002; Hinchberger and Rowe, 2003). Many studies attempted to conduct 3D FE analyses while investigating the behavior of embankments (e.g. Smith and Su, 1997; Briaud and Lim, 1999; Auvinet and Gonzalez, 2000). Smith and Su (1997) summarized that the 3D FE analysis can be used to model the reinforced soil embankment under service loading and at collapse

ARTICLE IN PRESS
40 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955

successfully. Briaud and Lim (1999) utilized three-dimensional (3D) nonlinear FE analysis to study the inuence factors on the tieback walls. In addition, Auvinet and Gonzalez (2000) recommended that a 3D analysis must be considered under the following conditions: (a) in the case of short slopes of which boundary conditions cannot be ignored, such as earth dams built in a narrow valley or embankment at the bridge approach, (b) when soil properties vary signicantly along the longitudinal direction of the slope or embankment, (c) when the slope is subjected to concentrated loading and (d) when the potential failure is irregular. Two full-scale test embankments were constructed on Bangkok clay deposit with different types of reinforcement and backll soil: steel grid reinforced long embankment having longer plan dimensions with length-to-width ratio of 3.0 (Shivashankar, 1991) and hexagonal wire mesh reinforced short embankment having shorter plan dimensions with length-to-width ratio of 1.0 (Voottipruex, 2000). These two embankments have been fully instrumented with piezometers, settlement gauges, inclinometer casings and strain gauges (on reinforcements). Highquality eld monitoring data have been obtained. Based on the work of Teerawattanasuk (2004), the numerical simulations of these two embankment systems were realized by means of FD method using 2D and 3D explicit FD programs, FLAC2D (ITASCA and FLAC2D Version 3.4, 1998) and FLAC3D (ITASCA and FLAC3D Version 2.0, 1997), respectively. The aim of this study is to investigate the inuence of geometric congurations using 2D and 3D numerical simulations of the two full-scale tests (i.e. class C1 prediction, Lambe, 1973). Particular attention is given to the settlements, excess pore-water pressures, horizontal displacements, and tensile forces in the reinforcements. Subsequently, comparisons are made between the ndings of 2D and 3D numerical simulations and those from the actual measured eld data of the two full-scale test embankments.

2.2. Wall embankment system with hexagonal wire mesh reinforcement On the AIT campus nearby, a fully instrumented short embankment with hexagonal wire mesh as the reinforcement also was constructed. It had a length-to-width ratio of 6/6 1.0 (Voottipruex, 2000). This embankment was constructed within 60 days. This short embankment is 6.0 m in height with 6.0 m wide, and 6.0 m long at the top, and 18.0 m long at the base as shown in Fig. 2. After 405 days of construction, the top of the embankment was raised up by 1 m of additional ll to investigate its behavior (see Fig. 6). The embankment was divided into two parts along its length with zinc-coated and PVC-coated hexagonal wire mesh reinforcements backlled with Ayutthaya sand. The gabion facing of the embankment was built with 101 inclination from the vertical alignment. The side slopes and back slope have 1:1 inclination. 3. Model parameters 3.1. Foundation soils Referring to Bergado et al. (1995), the typical subsoil prole, together with the general soil properties at site, is illustrated in Fig. 3. Similar foundation soil was used for 2D and 3D numerical simulation of the two full-scale test embankments. According to the existing database of geotechnical properties of the foundation subsoils at the site (Balasubramaniam et al., 1978), the linear elasticperfectly plastic model parameters (Mohr Coulomb failure criteria) were used for the topmost heavily overconsolidated clay. The modied Cam clay model parameters were adopted for the other underlying four layers together with the estimated value of permeability (Bergado et al., 1995). For the uid properties adopted in the FLAC2D and FLAC3D analyses, the Biots modulus was applied equal to one for the incompressible grains condition. The level of groundwater was designated at 2 m depth below the ground surface. The input model parameters of foundation soils used in FLAC2D and FLAC3D analyses are tabulated in Table 1 together with the permeability coefcients and porosities for each layer of the foundation subsoils. 3.2. Backll soils 3.2.1. Lateritic backll material The lateritic backll soil was utilized in the middle portion of the steel grid reinforced embankment and was used as its representative backll soil. Theoretically, the lateritic soil is a complex engineering material which has nonlinear and nonhomogeneous properties. Many constitutive soil models were developed to represent its complicated soil behavior. However, in this study, the commonly used nonlinear elastic model with MohrCoulomb failure criteria was selected to represent the stressstrain behavior

2. Description of the two full-scale test embankments 2.1. Wall embankment system with steel grid reinforcement The reinforced long embankment with steel grid reinforcement (Fig. 1) was constructed on the campus of the Asian Institute of Technology (AIT) having a lengthto-width ratio of 15/5 3.0 (Shivashankar, 1991). The embankment was constructed over a period of 30 days (see Fig. 6). The long embankment was divided into three sections along its length with three different backll materials, namely clayey sand, lateritic soil, and weathered clay. The embankment is 5.70 m high above the existing ground surface, with 5.64 m width and 14.64 m length at the top, and about 26.0 m length at the base. This embankment has three sloping faces with 1:1 slope and one vertical wall facing.

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 41

Fig. 1. Schematic plan view and cross-section indicating instrumentation points in steel grid reinforced embankment. (a) Plan view and (b) cross-section.

of this backll material because the MohrCoulomb failure criteria can represent the failure behavior of soils having an apparent cohesion and obtaining the model parameters is also simple (Zienkiewicz et al., 1975). Bergado et al. (1993)

obtained the parameters of the lateritic backll materials from the large-scale direct shear tests as tabulated in Table 1 for the MohrCoulomb model used in FLAC2D and FLAC3D analyses.

ARTICLE IN PRESS
42 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955

Fig. 2. Schematic plan view and cross-section indicating instrumentation points in hexagonal wire mesh reinforced embankment.

3.2.2. Ayutthaya sand backll material For the hexagonal wire mesh reinforced embankment, the backll soil material, known locally as Ayutthaya sand, was modeled as a linear elasticperfectly plastic material with MohrCoulomb failure criteria. The apparent cohesion, c0 , is 10 kPa and the friction angle, f0 , is 301 (Bergado et al., 2000). Considering the average normal pressure at mid-height of the embankment (3 m high), the average normal pressure was calculated to be 50 kPa. Subsequently, in the numerical simulation of hexagonal wire mesh reinforced embankment, the input parameters of silty sand

at the average normal pressure of 50 kPa was selected to represent the backll material properties of the whole embankment. The parameters of Ayutthaya backll material for the MohrCoulomb model used in FLAC2D and FLAC3D analyses are tabulated in Table 2. 3.3. Reinforcements 3.3.1. Steel grid For the numerical simulation using FLAC3D, the steel grid reinforcement was considered to be a

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 43

Unit Weight (kN/m2) 0 1 2 3


Very soft clay

PL & LL & Water Content (%) 20 40 60 80 100 120

Preconsolidation Pressure (kPa) 0 50 100 150 200

12 14 16 18 20
Heavily overconsolidated weathered clay Lightly overconsolidated weathered clay

4 Depth (m) 5 6 7 8 9 10 11 2.0 2.5 3.0 0 10 20 30 40 0 2 4 6 OCR 8 10 Specific Gravity Undrained Shear Strength (kN/m2)
Medium stiff clay

Stiff clay

Plastic limit Liquid limit Natural water content

From Oedometer Test P'o P'max

OCR

Fig. 3. General soil prole and properties of the subsoil at Asian Institute of Technology (AIT) (Chai, 1992; Bergado et al., 1995). Table 1 Selected parameters for steel grid reinforced embankment adopted in 2D and 3D nite-difference analyses Parameters Symbol Soil layer 1 01 MCa k l M Vl P1 n kmax pco K G f0 C0 rt rd N 25.0kv 2 12 MCCb 0.04 0.18 1.1 4.256 1 0.25 12.5 14.3 3 26 4 68 5 812 Elastic 0.11 0.51 0.9 8.879 1 0.3 2.88 7.55 0.07 0.31 0.95 5.996 1 0.3 4.86 9.30 0.04 0.18 1.1 4.168 1 0.25 9.6 10.7 463,000 70,000 79.80 28.98 32 60 2000 2000 MCa Wall face Backll

Depth (m): Soil model Slope of elastic swelling line Slope of normal consolation line Frictional constant Specic volume at reference pressure (1 Pa) Reference pressure (1 Pa) Poissons ratio Maximum elastic bulk modulus ( 107 Pa) Preconsolidation pressure ( 104 Pa) Elastic bulk modulus ( 106 Pa) Elastic shear modulus ( 106 Pa) Friction angle (deg) Cohesion ( 103 Pa) Total unit weight (kg/m3) Dry unit weight (kg/m3) Porosity Permeability ( 1012 m2/(Pa s))
a b

0.25

2.67 1.6 29 29 1750 1750 0.545 17.8

1750 1750 0.545 17.8

1500 803 0.697 2.65

1650 1050 0.600 2.65

1750 1226 0.524 17.8

2400 2400

Elasticperfectly plastic MohrCoulomb model. Modied Cam clay model.

linear elastic material with Youngs modulus of 2.0 1011 Pa and Poissons ratio of 0.33 (Bergado et al., 1995). The reinforcement was represented by linear

elastic structural shell elements. The properties required for the reinforcement were density, Youngs modulus, Poissons ratio, and thickness, which were

ARTICLE IN PRESS
44 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 Table 2 Selected parameters for hexagonal wire mesh reinforced embankment adopted in 2D and 3D nite-difference analyses Parameters Symbol Soil layer 1 1 MCa k l M Vl P1 n kmax pco K G f0 c0 rt rd N 25.0kv 2 12 MCCb 0.04 0.18 1.1 4.256 1 0.25 112.5 114.3 3 26 4 68 5 812 MCa 0.11 0.51 0.9 8.879 1 0.3 2.88 7.55 0.07 0.31 0.95 5.996 1 0.3 4.86 9.30 0.04 0.18 1.1 4.168 1 0.25 9.6 10.7 5.88 2.69 45 20 1800 1800 5.00 2.31 30 10 1800 1800 MCa Wall face Backll

Depth (m): Soil model Slope of elastic swellingline Slope of normal consolation line Friction constant Specic volume at reference pressure (1 Pa) Refrence pressure (1 Pa) Possion ratio Maximum elastic bulk modulus ( 107 Pa) Preconsolidation pressure ( 104 Pa) Elastic bulk modulus ( 106 Pa) Elastic shear modulus ( 106 Pa) Friction angle (deg) Cohesion ( 103 Pa) Total unit weight (kg/m3) Dry unit weight (kg/m3) Porosity Permeability ( 1012 m2/(Pa s))
a b

0.25

2.67 1.6 29 29 1750 1750 0.545 17.8

1750 1750 0.545 17.8

1500 803 0.697 2.65

1650 10,500 0.600 2.65

1750 1226 0.524 17.8

Elasticperfectly plastic MohrCoulomb model. Modied Cam clay model.

back-calculated by matching the EI in addition to EA values as demonstrated by Teerawattanasuk et al. (2003). The input parameters of the reinforcement as structural shell element are tabulated in Table 3. In FLAC2D, the linear elastic material properties assigned to the reinforcements were the same as those applied in FLAC3D. The steel grid reinforcement was modeled using linear elastic structural cable elements with Youngs modulus of 2.0 1011 Pa, and cross-sectional area of longitudinal bar per meter width of 180 mm2 (Bergado et al., 1995). The input parameters of steel grid reinforcement as the structural cable elements are listed in Table 4.

Table 3 Selected parameters for structural shell element applied in FLAC3D Parameters Steel grid reinforcement 2 1011 0.33 0.006 2500 Hexagonal wire mesh reinforcement 5.4 108 0.33 0.003 2500

Youngs modulus, E (Pa) Poissons ratio Thickness (m) Density (kg/m3)

Table 4 Selected parameters for structural cable element applied in FLAC Parameters Steel grid reinforcement 32.5 6 104 2 1011 6 108 180 106 1.5 107

2D

Hexagonal wire mesh reinforcement 27.5 9 103 5.4 108 6 108 156 106 7.764 107

3.3.2. Hexagonal wire mesh In FLAC3D, the linear elastic structural shell elements were adopted in the numerical simulation of the hexagonal wire mesh reinforcements. The linear axial stiffnesses, EA, were determined from the in-air tensile tests conducted by Wongsawanon (1998). The axial stiffness of the reinforcement, EA, was similar to that in the numerical simulation of laboratory in-soil pullout tests (Teerawattanasuk et al. (2003)). The input parameters of hexagonal wire mesh reinforcement as structural shell element applied in FLAC3D program are tabulated in Table 3. For the numerical simulation using FLAC2D, the hexagonal wire mesh reinforcements were represented by the structural cable elements (ITASCA and FLAC3D Version 2.0, 1997). The input parameters of hexagonal reinforcement as the structural cable elements are tabulated in Table 4.

Bond friction angle of grout (deg) Bond strength of grout (N/m) Youngs modulus, E (Pa) Tensile yield strength (Pa) Cross sectional area (m2) Grout shear stiffness (N/m)

3.4. Wall face 3.4.1. Steel grid wall-facing system In both FLAC2D and FLAC3D programs, the steel grid wall-facing system was represented by the solid elements which were similar to the elements applied in the backll soil material. The wall-facing system was treated as linear

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 45

elastic material with bulk modulus of 1.56 1011 Pa and shear modulus of 7.0 1010 Pa (Bergado et al., 1995). The input parameters adopted in the numerical simulation using FLAC2D and FLAC3D analyses are listed in Table 1. 3.4.2. Hexagonal wall-facing system The hexagonal wall-facing system was made from large rectangular wire mesh baskets joined together and lled with crushed rock (Bergado et al., 2000). Similar to the case of steel grid wall-facing system, the hexagonal wall-facing system was also modeled using solid elements. However, the linear elasticperfectly plastic, MohrCoulomb model was used to simulate the hexagonal wall-facing system based on the study of Bergado et al. (2000b). The parameters required for FLAC2D and FLAC3D analyses are tabulated in Table 2. 3.5. Soilreinforcement interface In the simulation of the soilreinforcement interface, two interaction modes were considered: namely, direct shear and pullout modes. However, for the steel grid and hexagonal wire mesh, only the pullout mode is applicable. In FLAC3D, the interface element where sliding or separation occurred, is characterized by Coulomb sliding having the properties of friction, cohesion, dilation, normal stiffness and shear stiffness (ITASCA and FLAC3D Version 2.0, 1997). The interface elements are utilized to provide the sliding plane for the soilreinforcement interface. The interface resistance can be determined in terms of the interaction coefcient, R, as explained in the study of Bergado et al. (2000b). For 2D and 3D numerical analyses, the adopted interaction coefcients were 1.0 for steel grid reinforcement (Bergado et al., 1995) and 0.9 for the hexagonal wire mesh reinforcement (Teerawattanasuk et al. (2003)). The properties of soilreinforcement interface element used in the numerical simulation of steel grid and hexagonal wire mesh reinforced embankments with FLAC3D program are listed in Table 5. For FLAC2D, the applied soilreinforcement interface has been combined with the structural cable element as described previously. The input interface parameters adopted in FLAC2D program for steel grid and hexagonal wire mesh reinforcements as structural cable elements are tabulated in Table 4.

4. Numerical simulations Comparing the length-to-width ratio (L/B) of the two embankments, the L/B for the steel grid embankment is 3.0 (15/5) as indicated in Figs. 1 and 7, which is about 3.0 times greater than the hexagonal wire mesh embankment (Figs. 2 and 8) where L/B 1 (6/6). Using numerical simulations under 2D and 3D conditions, this study was carried out to investigate the inuence of geometric congurations on the calculated results such as settlements, excess pore-water pressures, and horizontal displacements, as well as tensile forces in the reinforcements. Using the same constitutive models and properties of the foundation soils utilized by Bergado et al. (1995) and Alfaro (1996), numerical simulations were conducted using 2D and 3D explicit FD programs, FLAC2D and FLAC3D, respectively. The coupled analyses were carried out in the consecutive steps. A summary of various numerical simulations that were performed on these two full-scale test embankments is tabulated in Table 6. The materials applied in the numerical simulation of the reinforced structure were classied into four types, namely (a) soil (solid brick-shaped element), (b) reinforcement (structural shell or cable element), (c) soilstructure interaction (interface element) and (d) wall-facing structure (solid brick-shaped element). The interface elements were attached to provide the sliding plane for the reinforcement and surrounding soil.

5. 2D numerical simulations of steel grid reinforced embankment (Analyses 1 and 2) 5.1. 2D FD grid discretization For 2D or plane strain condition analysis (refer to Analysis 1 in Table 6), FLAC3D was used to simulate the long steel grid reinforced embankment (refer to Fig. 1) by restricting the planes perpendicular to the side of the embankment: e.g. xed only the longitudinal directions. The materials applied in FLAC3D were presented by 3D grid elements. The FD grid discretization used in FLAC3D analysis is shown in Fig. 4. Similarly, the FD grid discretization corresponding to FLAC2D program (Analysis 2 in Table 6) is presented in Fig. 5. The x-, y-, and zdimensions of the foundation soil were 43, 1, and 12 m, respectively (see Fig. 5). Similar soil prole was utilized for the foundation soils (refer to Table 1). The uniform vertical spacing of steel grid reinforcement in the reinforced embankment was 0.45 m with a length of 5 m. In the 2D numerical simulation using FLAC3D program, the structural shell element with 5 m length, 1 m width, and 0.006 m thickness was adopted to simulate the steel grid reinforcement. As noted earlier, in FLAC2D program, the structural cable elements were used to simulate the steel grid reinforcement.

Table 5 Interface parameters adopted in nite-difference analysis with FLAC3D Parameters Types of reinforcement Cohesion of interface element, ci (kPa) Friction angle of interface element, d (deg) Interface shear stiffness, ks* (kPa/m) Interface normal stiffness, kn* (kPa/m) Values Steel grid 60 32 1.5 104 5.0 106 Hexagonal wire mesh 9 27.5 7.682 104 3.028 105

ARTICLE IN PRESS
46 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 Table 6 Numerical simulations for AIT full-scale test reinforced embankments Analysis no. 1 2 3 4 5 6 Analysis type C C C C C C Permeability 25kv 25kv 25kv 25kv 25kv 25kv Types of reinforcement Steel grid Steel grid Hexagonal wire mesh Hexagonal wire mesh Steel grid Hexagonal wire mesh Backll materials Lateritic Lateritic Ayutthaya sand Ayutthaya sand Lateritic Ayutthaya sand Boundary condition 2D 2D 2D 2D 3D 3D Numerical programs FLAC3D FLAC2D FLAC3D FLAC2D FLAC3D FLAC3D

Note: C, coupled analysis; kv, laboratory test value of vertical permeability; 2D, two-dimensional; 3D, three-dimensional; FLAC2D, two-dimensional explicit nite-difference program (FLAC2D program); FLAC3D, three-dimensional explicit nite-difference program (FLAC3D program).

Fig. 4. 2D nite-difference grid discretization for steel grid reinforced embankment with FLAC3D (Analysis 1).

Fig. 5. 2D nite-difference grid discretization for steel grid reinforced embankment with FLAC2D (Analysis 2).

order to check the differences in the results. With the FLAC3D program, the boundary condition applied in the numerical simulation of the steel grid reinforced embankment was assigned by the xed velocity boundary in x-, y-, and z-directions at the bottom of the foundation. The horizontal xed velocities of grid points in x-direction were attached to two vertical planes of the foundation soil (at x 0 and 43, see Fig. 4). In addition, the horizontal xed velocities of grid points in y-direction were attached to two vertical planes of the foundation soil (at y 0 and 1, see Fig. 4). Using FLAC2D, the boundary condition automatically xed the velocities in z-direction as the plane strain condition. Therefore, the xed velocity boundaries in xand y-directions were assigned at the bottom of the foundation (see Fig. 5). Likewise, the horizontal xed velocities of grid points in x-direction were attached to two vertical planes of the foundation soil (see Fig. 5). 5.3. Stages of construction

5.2. Boundary and initial conditions To investigate the boundary effect, two different discretization schemes have been analyzed, namely one with a nearer and the other with a farther boundary in

For the steel grid reinforced embankment, the construction sequence was divided into 13 stages with the

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 47

30 days

60 days

7 Embankment Height (m) 6 5 4 3 2 1 0 0

405 days

6.3. Stages of construction


Additional Surcharge 16.7 kPa

Steel grid reinforced embankment (5.70 m high with 13 incremental layers ) Hexagonal wire mesh reinforced embankment (6.0 m high with 12 incremental layers)

50 100 150 200 250 300 350 400 450 500 550 Time (days)

Referring to Fig. 6, the construction sequence of the hexagonal wire mesh reinforced soil embankment was divided into 12 stages with a total duration of 60 days. The procedure in numerical simulation for the hexagonal wire mesh reinforced soil embankment was done by placing the backll material and inserting the hexagonal reinforcement at an interval of 0.5 m vertical spacing per stage until the completion of full height embankment. The coupled analysis, undrained and consolidation analysis, was also taken into account in the numerical simulation. After 405 days, the traction of 16.7 kN/m2 was added on the top of the embankment. 7. 3D numerical simulations of steel grid reinforced embankment (Analysis 5) 7.1. 3D FD discretization

Fig. 6. Construction sequence of AIT full-scale test reinforced embankments.

equivalent period of 30 days (Fig. 6). The procedure in numerical simulation for the steel grid reinforced embankment was done by placing the backll material and inserting the reinforcement at an interval of 0.45 m vertical spacing per stage until the completion at the full height of the embankment. During the placement of the backll materials and inserting the reinforcement, a coupled analysis was undertaken but without pore pressure dissipation. Pore pressure dissipation was then permitted after the construction phase (about 2 days for every stage). 6. 2D numerical simulations of hexagonal wire mesh reinforced embankment (Analyses 3 and 4) 6.1. 2D FD discretization With the same numerical procedures, the hexagonal wire mesh reinforced embankment (refer to Fig. 2) was subsequently analyzed using FLAC3D and FLAC2D (refer to Analyses 3 and 4 in Table 6), respectively. The dimensions of the foundation soil were as follows 42 m length, 1 m width, and 12 m depth which were x-, y-, and zaxes, respectively. The foundation soil was divided into ve layers (see in Table 2). The reinforcement was represented by structural shell elements that were 4 m long, 1 m wide, and 0.003 m thick. The inclined wall-facing system of the hexagonal wire mesh reinforced embankment was represented by the brick-shaped elements 1.0 m long, 1.0 m wide, and 6.0 m high. 6.2. Boundary and initial conditions In the 2D numerical analyses of hexagonal wire mesh reinforced embankment using FLAC2D and FLAC3D, similar procedures were employed for assigning the boundary and initial conditions.

The numerical simulation by FLAC3D has the advantages in obtaining the exact full-scale test procedures. The 3D FD discretization is illustrated in Fig. 7. Because of the symmetry of the embankment structure, only the half section of the reinforced embankment and the foundation soil were simulated to reduce the numbers of the degrees of freedom, which is time consuming in the calculation steps. The dimensions of the foundation soil were as follows 43 m long, 28 m wide, and 12 m depth which correspond to the x-, y-, and z-axes respectively. The dimensions of wall facing with respect to x-, y-, and z-axes, were 0.2 m long, 7.5 m wide, and 5.70 m high, respectively. The reinforcement was modeled using structural shell elements in the embankment that were 5 m long, 7.5 m wide, and 0.006 m thick. 7.2. Boundary and initial conditions The boundary condition used in the numerical simulation of the embankment was assigned by the xed velocity boundary in x-, y-, and z-directions at the bottom of the foundation soil. The horizontal xed velocities of grid points in x-direction were attached to two vertical planes of the foundation soil (at x 0 and 43). The horizontal xed velocities in y-directions were attached to four vertical planes of the foundation soil that cross the x- and y-axes, respectively, as shown in Fig. 7. For 3D numerical model, the velocities in y-direction along the symmetry plane of the reinforced embankment were assigned to be zero while the velocity boundaries in x-, y-, and z-direction for the other sides of the embankment were set to be free velocity boundaries to allow the occurrence of free displacements in all directions. 7.3. Stages of construction In numerical simulation of the steel grid reinforced embankment, the stages of construction applied in 3D

ARTICLE IN PRESS
48 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955

2.5 m 5m 5.70 m 6m 14.5 m 12 m

15 m 5m 0.2 m 7m

z (+) y (+) x (+)

m 15.8

43 m

Fig. 7. 3D nite-difference grid discretization for steel grid reinforced embankment FLAC3D (Analysis 5).

6m

3m 6m

z (+) y (+)

15 m 12 m

15 m

x (+)

15 m

Fig. 8. 3D nite-difference grid discretization for hexagonal wire mesh reinforced embankment with FLAC3D (Analysis 6).

numerical analyses were parallel to the construction applied in 2D numerical analyses. 8. 3D numerical simulation of hexagonal wire mesh reinforced embankment (Analysis 6) 8.1. 3D FD discretization Fig. 8 presents the 3D FD discretization of geometry used for the hexagonal wire mesh reinforced embankment. The dimensions of the foundation soil were the following: 42 m long, 24 m wide, and 12 m depth which correspond to the x-, y-, and z-axes, respectively. The gabion facing structures were comprised of brick-shaped elements with

dimensions 3.0 m width, 1.0 m length, and 6.0 m height. The uniform vertical spacing of hexagonal wire mesh reinforcement in reinforced embankment was 0.5 m. The reinforcement was modeled using the structural shell elements in the embankment that were 4 m long, 3 m wide, and 0.003 m thick.

8.2. Boundary and initial conditions In the 3D numerical simulation of the hexagonal wire mesh reinforced embankment using FLAC3D, procedures similar to those described in Section 7.2 were employed for assigning the boundary and initial conditions.

12 m

12 m

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 49

8.3. Stages of construction In numerical simulation of the hexagonal wire mesh reinforced embankment, the stages of construction applied in 3D numerical analyses were similar to those applied in 2D numerical analyses. 9. Results and discussions 9.1. Steel grid reinforced embankment To investigate the inuence of geometric congurations, the steel grid reinforced embankments have been simulated by 2D and 3D numerical analyses (refer to Analyses 1, 2 and 5 in Table 6) by means of FD technique using FLAC2D and FLAC3D programs. The comparisons between the measured eld data and the calculated results (e.g. vertical settlements, excess pore-water pressures, lateral displacements, and tension force distribution) are discussed in the following sections. 9.1.1. Settlement The calculated and measured surface settlement (0.45 m depth below the original ground surface, refer to settlement plate S5) and subsurface settlement (3.0 m depth below the original ground surface, refer to settlement plate SS8) are compared in Figs. 9 and 10, respectively. The calculated values of surface and subsurface settlements obtained from 2D numerical analyses (refer to Analyses 1 and 2 listed in Table 6) were in agreement with the measured data as well as the FE results using CRISP program (Bergado et al., 1995; Chai, 1992). However, the calculated values for both surface and subsurface settlements attained from 3D numerical analysis (refer to Analysis 5 listed in Table 6) signicantly underestimated the measured eld data, possibly because of the three 3D loading condition and geometric effects which are signicant inuence factors on numerical simulation (Auvinet and Gonzalez, 2000). In addition, for the steel grid reinforced embankment, the measured settlement pattern is closer to 2D numerical analysis, than for 3D numerical analysis because of its longer plan dimensions. 9.1.2. Excess pore-water pressure As shown in Fig. 11, at the end of construction (i.e., after an elapsed time of 30 days), the calculated maximum excess pore-water pressures at the locations HP5 and HP6 obtained from 2D numerical analyses (refer to Analyses 1 and 2 in Table 6) overestimated the measured eld data while the calculated values from 3D analysis (Analysis 5 in Table 6) yielded satisfactory agreement. After the end of construction, the dissipation of pore-water pressures among three analysis schemes (Analyses 1, 2, and 5) have higher dissipation rate, than the measured eld data. However, referring to Analysis 1 using FLAC3D considered as 2D numerical analysis, the calculated values yielded a better agreement. Therefore, using the permeability value

equal to 25 times of kv, the results also could not predict the variation of measured pore-water pressure changes with time. 9.1.3. Lateral displacement Fig. 12 shows the comparison of lateral displacement proles of the steel grid reinforced embankment 7 months after the end of construction (or at the elapsed time of 240 days). The measured eld data only reached down to 3 m depth because the inclinometer probe could not be inserted into the deformed casing below 3 m depth (Bergado et al., 1995). In the embankment zone, the calculated results from FLAC2D (refer to Analysis 2 in Table 6) agreed well with the calculated results using CRISP program conducted by Chai (1992), but underestimated the measured eld data. For the foundation soil zone, the results calculated from

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 0 50

Measured field data FEM CRISP 25kv (Chai, 1992) Emb. S5 FDM FLAC2D 25kv FDM FLAC3D 25kv (2D analysis) FDM FLAC3D 25kv (3D analysis)

Vertical Displacement (mm)

End of Construction

100

150 200 250 300 Elapsed Time (days)

350

400

450

Fig. 9. Comparison of measured and predicted surface settlement of steel grid reinforced soil embankment, under 2D and 3D analyses at settlement plate S5 (0.45 m depth at the middle).

0 Vertical Displacement (mm) 100 200


SS8 Emb. Measured field data FEM CRISP 25kv (Chai, 1992) FDM FLAC2D 25kv FDM FLAC3D 25kv (2D analysis) FDM FLAC3D 25kv (3D analysis)

300 400 500 600 700 800 900 0 50 100


End of Construction

150 200 250 300 Elapsed Time (days)

350

400

450

Fig. 10. Comparison of measured and predicted subsurface settlement of steel grid reinforced soil embankment, under 2D and 3D analyses at settlement plate SS8 (3 m depth at the middle).

ARTICLE IN PRESS
50 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955

60
End of Construction

Excess Pore Pressure, kPa

50 40 30 20 10 0 0 50 100

Measured field data FEM CRISP 25kv (Chai, 1992) FDM FLAC2D 25kv FDM FLAC3D 25kv (2Danalysis) FDM FLAC3D 25kv (3D analysis)

Emb. HP5

150 200 250 300 Elapsed Time (days)

350

400

450

Fig. 11. Comparison of measured and predicted excess pore-water pressure of steel grid reinforced soil embankment, under 2D and 3D analyses at piezometric point HP5 (under lateritic Section 7 depth at the middle).

8 6 4 Depth/Height (m) 2 0 -2 -4 -6 -8 -10 -12 0 50 100 150 200 250 300 350 400 450 500 550 Lateral Displacement (mm)
Measured field data at 240 days FEM CRISP 25 kv at 240 days FDM FLAC2D 25 kv at 240 days FDM FLAC3D 25 kv at 240 days (2D analysis) FDM FLAC3D 25 kv at 240 days (3D analysis) Ground surface EL. = 0.0

the base of the reinforced embankment that caused large relative horizontal movement between the soil and the reinforcement. Moreover, the settlement pattern may form a concave shape at the base of the reinforced embankment. Therefore, large tension forces can be developed in the bottom reinforcements of the embankment (Bergado et al., 1995). Fig. 13 compared the calculated tension force distribution along the reinforcement length obtained from 2D and 3D numerical analyses immediately after construction (at the elapsed time of 30 days) and the measured eld data obtained from strain gauges as well as the calculated results using CRISP program (Chai, 1992). The maximum tension forces in Mat 1 occurred at 4 m from the wall face. This might be attributed to the bending effect in the reinforced embankment due to differential settlements of foundation soil between the front and rear of the reinforced embankment. These results are similar to the previous results reported by Chai (1992) and Alfaro et al. (1997). Moreover, the calculated tension force distribution from 2D and 3D numerical simulation (Analysis 1, 2, and 5) consistently shows logical results in which the tension forces of 2D numerical analysis are larger than that of 3D analysis because of higher calculated settlements of the former compared to the latter (see Figs. 9 and 10). 9.2. Hexagonal wire mesh reinforced embankment The hexagonal wire mesh reinforced embankment which has a shorter plan dimensions (length-to-width ratio, 6/6 1.0), is compared with the steel grid reinforced embankment. For the sake of comparison, 2D and 3D numerical simulations of hexagonal wire mesh reinforced embankment (refer to Analyses 3, 4 and 6 in Table 6) were also studied using the same procedure as discussed in the previous section. Similar foundation soil parameters and permeability values of foundation soils (k 25 kv) were applied in the numerical analyses. Comparisons between the ndings of 2D and 3D numerical results and the measured eld data (e.g. settlements, excess pore-water pressures, lateral displacements, and tension force distribution) are also discussed in the following sections. 9.2.1. Settlement Referring to Figs. 14 and 15, we see the calculated values of surface settlements obtained from 3D numerical analyses (Analysis 6, refer to Table 6) are closer and they slightly overestimated the measured data than the calculated results obtained from 2D numerical analyses. As shown in Fig. 16, the calculated subsurface settlements (SS2) from 3D analysis (Analysis 6 in Table 6) were slightly less than the measured eld data. The measured eld data at the settlement plate SS2 were higher than the calculated results. However, the actual settlement patterns for this embankment agreed more closely with the 3D analyses. The use of the 2D numerical simulation for the hexagonal wire mesh reinforced embankment case could not predict the actual surface and subsurface settlements. Therefore,

Fig. 12. Comparison of measured and predicted lateral displacement proles of steel grid reinforced soil embankment, under 2D and 3D analyses.

FLAC2D overestimated the measured eld data. In analysis 1 (FLAC3D under 2D numerical analysis, see Table 6), the calculated results underestimated the measured eld data. Thus, the measured eld data beneath the reinforced embankment are in between the calculated results obtained from Analyses 1 and 2 (under 2D numerical analysis). 9.1.4. Tension force in reinforcement For a reinforced soil wall constructed on rigid foundation with high stiffness reinforcement, the maximum tension force in the reinforcement is close to the value calculated by at-rest earth pressure coefcient (Adib et al., 1990; Rowe and Ho, 1997). However, for a reinforced soil wall constructed on soft ground, under embankment loading, the soft foundation soil tends to squeeze out of

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 51

0.3 H

60 40 20 0 60 40 20
Coherent gravity failure plane

Measured field data 30 days FEM CRISP 25kv (Chai, 1992) FDM FLAC2D 25 kv at 30 days FDM FLAC3D 25 kv at 30 days (2D analysis) FDM FLAC3D 25 kv at 30 days (3D analysis)

MAT 7 EL: 4.95 m

Tie-back wedge failure plane

Tension force in reinforcement at different elevation, kN/m

0 60 40 20 0 60 40 20 0 60 40 20 0 60

MAT 6 EL: 4.05 m

MAT 5 EL: 3.15 m

MAT 4 EL: 2.25 m

MAT 3 EL: 1.35 m

40 20 0 60 40 20 0 0
45+f/2 MAT 1 EL: 0 m

MAT 2 EL: 0.45 m

2 3 Distance from the wall face, m

Fig. 13. Comparison of measured and predicted tension force in the reinforcement of steel grid reinforced soil embankment, under 2D and 3D analyses.

geometric effects (2D versus 3D) can be considered to be an important factor that inuences the calculated results. 9.2.2. Excess pore-water pressure Referring to Fig. 17, at the end of construction (i.e. after an elapse of 60 days), the calculated maximum excess porewater pressure at the locations of P1 and P2 obtained from 2D numerical analyses (Analyses 3 and 4 as listed in Table 6) overestimated the measured eld data while the

calculated values from 3D analysis (Analysis 6, refer to Table 6) yielded satisfactory agreement. The dissipation of pore-water pressures after the end of construction among three analysis schemes (Analyses 3, 4, and 6 in Table 6) is higher than the dissipation rate from the measured eld data. Subsequently, once 405 days, elapses additional surcharge was added. Then, the excess pore-water pressures increased again and started to continually dissipate with time. Considering the embankment behavior during

ARTICLE IN PRESS
52 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955

0 100 Vertical Displacement (mm) 200 300 400 500 600 700 800 900 0
End of Construction

50 Excess Pore Pressure (kPa)


Measured field data (Zinc) Measured field data (PVC) FDM FLAC2D 25kv FDM FLAC3D 25kv (2D analysis) FDM FLAC3D 25kv (3D analysis)

45 40 35 30 25

End of Construction

Measured field data (Zinc) Measured field data (PVC) FDM FLAC2D 25k v FDM FLAC3D 25k v (2Danalysis) FDM FLAC3D 25k v(3D analysis)

Emb.

20 15 10 5 0 0 50 100 150 200 250 300 350 400 450 500 550 600 Elapsed Time (days)
P1

Emb. S2

50 100 150 200 250 300 350 400 450 500 550 600 Elapsed Time (days)

Fig. 14. Comparison of measured and predicted surface settlement of hexagonal wire mesh reinforced soil embankment, under 2D and 3D analyses at settlement plate S2 (0.45 m depth at the middle).

Fig. 17. Comparison of measured and predicted excess pore-water pressure of hexagonal wire mesh reinforced soil embankment, under 2D and 3D analyses at piezometer point P1 (3 m depth at the middle).

0 Vertical Displacement (mm) 100 200 300 400 500 600


End of Construction

Measured field data (Zinc) Measured field data (PVC) FDM FLAC2D 25kv FDM FLAC3D 25kv (2D analysis) FDM FLAC3D 25kv (3D analysis)

Emb. SS1

700 0 50 100 150 200 250 300 350 400 450 500 550 600 Elapsed Time (days)

construction (after elapse 030 days), the rate of excess pore-water pressure dissipation under plane strain condition is less than that of 3D analysis because the excess pore-water pressure can dissipate only in two directions (in x- and z-directions for Analysis 3 and inq x- and y-directions for Analysis 4) in the former while the excess pore-water pressures can dissipate in all directions (x-, y-, and z-directions) in the latter. However, after the 150 days elapsed, the dissipation rates of excess pore water in the 2D and 3D numerical analyses are more or less the same. 9.2.3. Lateral displacement The comparison between 2D and 3D calculated results and measured eld lateral displacement of the hexagonal wire mesh reinforced embankment are shown in Fig. 18 at 490 days after the beginning of construction. The computed lateral displacements of the wall face from 3D numerical simulation (Analysis 6 in Table 6) were generally closer to the measured eld data compared to the calculated results obtained from 2D numerical analyses (Analyses 3 and 4 in Table 6), especially for the lateral displacement in the foundation soil. For the lateral displacement in the embankment portion, the calculated results from 3D numerical analysis slightly underestimated the measured eld data, and the calculated results from 2D numerical analysis largely overestimated the eld data. However, the measured eld data are in between the 2D and 3D numerical analyses. According to the 2D numerical results for both embankments (Analyses 14), after 400 days elapsed, the calculated lateral displacements for embankment zone of the hexagonal wire mesh reinforced embankment are moderately less than that of the steel grid reinforced embankment (see Fig. 12) and the calculated settlements for both embankments have some differences (see Figs. 9 and 14).

Fig. 15. Comparison of measured and predicted subsurface settlement of hexagonal wire mesh reinforced soil embankment, under 2D and 3D analyses at settlement plate SS1 (3 m depth at the middle).

0 50 100 150 200 250


Emb. End of Construction Measured field data (Zinc) Measured field data (PVC) FDM FLAC2D 25k v FDM FLAC3D 25k v (2D analysis) SS2 FDM FLAC3D 25k v (3D analysis)

Vertical Displacement (mm)

300 350 0

50 100 150 200 250 300 350 400 450 500 550 600 Elapsed Time (days)

Fig. 16. Comparison of measured and predicted subsurface settlement of hexagonal wire mesh reinforced soil embankment, under 2D and 3D analyses at settlement plate SS2 (6 m depth at the middle).

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955
10 8 6 4 Depth/Height (m) 2
Ground surface EL. = 0.0 Legend: Additional sandbags1 m high were added Measured field data 240 days FDM FLAC2D 240 days 25k v FDM FLAC3D 240 days 25k v (2D analysis) FDM FLAC3D 240 days 25k v (3D analysis)

53

6
0.3*H =1.8 m Coherent gravity failure plane Tie-back wedge failure plane Measured field data at 490 days FDM FLAC2D 25 kv at 490 days FDM FLAC3D 25 kv at 490 days (2D analysis) FDM FLAC3D 25 kv at 490 days (3D analysis)

0 -2 -4 -6 -8 -10 -12 0 50 100 150 200 250 300 350 400 450 500 550 Lateral Displacement (mm) Wall height (m)

15 10 5

El. = 4.0 m

4 0 1 2 3

Mat. 3

0 Force (kN/m) 15 10 5

Fig. 18. Comparison of measured and predicted lateral displacement proles of hexagonal wire mesh reinforced soil embankment, under 2D and 3D analyses.

El. = 2.0 m

Mat. 2

0 20 15 10 5

9.2.4. Tension force in reinforcement Fig. 19 shows the comparisons between 2D and 3D calculated tension force distribution along the reinforcement length at 7 months after construction (elapsed time equal to 240 days), as well as the measured eld data. Similar to 2D and 3D numerical simulation of the steel grid reinforced embankment, the calculated tension force distribution (refer to Analyses 3, 4, and 6 in Table 6) also shows consistent results in which the tension forces of 2D numerical analysis are greater than that of 3D numerical analysis due to overestimated settlements of the former compared to the latter (see Figs. 1416). However, the calculated results from 3D numerical analysis yielded values closer to the measured eld data than the 2D numerical analysis because the calculated settlement and lateral movement from 2D numerical analysis predominantly overestimated the measured eld data. Considering the tension force distribution in Mat 1, the maximum tension force occurred at 3 m from the wall face similar to the steel grid reinforced embankment case which may be due to the differential settlements of the soft foundation soil at the base of the reinforced embankment. 9.3. Results summary The aforementioned simulation results of the long and short embankments corresponding to 2D and 3D conditions, respectively, are quite consistent with 2D/3D settlement predictions of embankments on soft ground. According to Skempton and Bjerrum (1957), the 3D settlement predictions are consistently low than 2D settlement predictions mainly due to lower pore pressures and lower lateral deformations in the 3D conditions. Consequently, Skempton and Bjerrum (1957) proposed a settlement correction factor (usually less than one) depending on the geometry of the problem as well as the pore pressures. Similarly, as shown in Figs. 11 and 17, the

45 + /2

El. = 0.0 m

0 0 1 2 3 Distance from embankment face (m)

Mat. 1

Fig. 19. Comparison of measured and predicted tension force in the reinforcement of hexagonal wire mesh reinforced soil embankment, under 2D and 3D analyses.

3D pore pressures were consistently low than the 2D pore pressures. Moreover, the 3D lateral deformations were consistently low than the lateral deformations as demonstrated in Figs. 12 and 18, and the trends of the results are reected in the predicted tensions in the reinforcements (see Figs. 13 and 19). Finally, the simulation results from the FLAC3D used in 2D analysis agreed with the measured settlement data in the long embankment as well as the respective 2D settlement predictions from FLAC2D and the 2D FEM CRISP of Chai (1992) as shown in Figs. 9 and 10. 10. Conclusions Utilizing the constitutive models and properties of foundation soils published by previous researchers, numerical simulations were conducted using 2D and 3D explicit FD programs, FLAC2D and FLAC3D, respectively. The 2D and 3D numerical simulations have been done to investigate the inuence of the embankment geometry. Two full-scale test embankments, namely steel grid reinforced long embankment with plan dimensions (length-to-width ratio, 15/5 3.0) and hexagonal

ARTICLE IN PRESS
54 D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 Chai, J.C., 1992. Interaction behavior between grid reinforcement and cohesive frictional soils and performance of reinforced wall/embankment on soft ground. Doctoral Dissertation No. GT-91-1, Asian Institute of Technology, Bangkok, Thailand. Chai, J.C., Bergado, D.T., 1993a. Performance of reinforced embankment on Muar clay deposit. Soils and Foundations 33 (4), 117. Chai, J.C., Bergado, D.T., 1993b. Some techniques for FE analysis of embankment on soft ground. Canadian Geotechnical Journal 30, 710719. Chai, J.C., Miura, N., Bergado, D.T., Long, P.V., 1997. Finite element analysis of geotextile reinforced embankment failure on soft subsoil. Geotechnical Engineering Journal 28 (2), 249276. Hinchberger, S.D., Rowe, R.K., 2003. Geosynthetic reinforced embankments on soft clay foundations: predicting reinforcement strains at failure. Geotextiles and Geomembranes 21, 151175. Ho, S.K., Rowe, R.K., 1994. Prediction behavior of two centrifugal model soil walls. Journal of Geotechnical Engineering, ASCE 120 (10), 18451873. Hufenus, R., Rueegger, R., Banjac, R., Mayor, P., Springman, S.M., Bro nnimann, R., 2006. Full-scale eld tests on geosynthetic reinforced unpaved roads on soft subgrade. Geotextiles and Geomembranes 24 (1), 2137. ITASCA, FLAC2D Version 3.4, 1998. Fast Lagrangian Analysis of Continua in 2 Dimensions. ITASCA Consulting Group Inc. ITASCA, FLAC3D Version 2.0, 1997. Fast Lagrangian Analysis of Continua in 3 Dimensions. ITASCA Consulting Group Inc. Karpurapu, R., Bathurst, R.J., 1995. Behavior of geosynthetic reinforced soil retaining walls using the nite element method. Computers and Geotechnics 17, 279299. Kazimierowicz-Frankowska, K., 2005. A case study of a geosynthetic reinforced wall with wrap-around facing. Geotextiles and Geomembranes 23 (1), 107115. Lambe, T.W., 1973. Predictions in soil engineering. Geotechnique 23 (2), 149202. Nouri, H., Fakher, A., Jones, C.J.F.P., 2006. Development of Horizontal Slice Method for seismic stability analysis of reinforced slopes and walls. Geotextiles and Geomembranes 24 (2), 175187. Park, T., Tan, S.A., 2005. Enhanced performance of reinforced soil walls by the inclusion of short ber. Geotextiles and Geomembranes 23 (4), 348361. Rowe, R.K., Ho, S.K., 1997. Continuous panel reinforced soil walls on rigid foundation. Journal of Geotechnical and Geoenvironmental Engineering, ASCE 123 (10), 912920. Rowe, R.K., Ho, S.K., 1998. Horizontal deformation in reinforced soil wall. Canadian Geotechnical Journal 35, 312327. Rowe, R.K., Li, A.L., 2002. Behaviour of reinforced embankments on soft rate-sensitive soils. Geotechnique 52 (1), 2940. Shivashankar, R., 1991. Behavior of mechanically stabilized earth (MSE) embankment with poor quality backlls on soft clay deposits, including a study of the pullout resistances. Doctoral Dissertation GT-90-3, Asian Institute of Technology, Bangkok, Thailand. Skempton, A.W., Bjerrum, L., 1957. A contribution to the settlement analysis of foundations in clay. Geotechnique 7 (4), 168178. Skinner, G.D., Rowe, R.K., 2005. Design and behaviour of a geosynthetic reinforced retaining wall and bridge abutment on a yielding foundation. Geotextiles and Geomembranes 23 (3), 235260. Smith, I.M., Su, N., 1997. Three-dimensional FE analysis of a nailed soil wall curved in plan. International Journal for Numerical and Analytical Methods in Geomechanics 21, 583599. Teerawattanasuk, C., 2004. Modeling of hexagonal wire reinforcement and 2D/3D simulation of full scale embankment. Doctoral Dissertation GE-03-03, Asian Institute of Technology, Bangkok, Thailand. Teerawattanasuk, C., Bergado, D.T., Kongkitkul, W., 2003. Analytical and numerical modeling of pullout capacity and interaction between

wire mesh reinforced short embankment with plan dimensions (length-to-width ratio, 6/6 1.0), were studied. The calculated results were compared to the measured eld data with particular attention to settlements, excess porewater pressures, lateral displacements, and tension force distributions in the reinforcement. The actual behavior of the steel grid reinforced long embankment corresponded more closely to the results of the 2D numerical simulations. Furthermore, the actual behavior of the hexagonal wire mesh reinforced short embankment corresponded more closely to the results of the 3D numerical simulations. Moreover, the simulation results from FLAC3D used in 2D analysis agreed with the measured settlement data in the long embankment as well as the 2D predictions from FLAC2D. Therefore, the geometric effects should be considered as important factors that can affect the results of the numerical simulations, which are consistent with the current settlement predictions with Skempton and Bjerrum corrections.

References
Adib, M., Mitchell, J.K., Christopher, B., 1990. Finite element modeling of reinforced soil walls and embankments. In: Lambe, P., Hansen, L.A. (Eds.), Design and Performance of Earth Retaining Structure. Geotechnical Special Publication, No. 25, ASCE, New York, pp. 409423. Alfaro, M.C., 1996. Reinforced soil wallembankment system on soft foundation using inextensible and extensible grid reinforcements. Ph.D. Dissertation, Saga University, Japan. Alfaro, M.C., Hayashi, S., Miura, N., Bergado, D.T., 1997. Deformation of reinforced soil wall-embankment system on soft clay foundation. Soils and Foundations 37 (4), 3346. Al Hattamleh, O., Muhunthan, B., 2006. Numerical procedures for deformation calculations in the reinforced soil walls. Geotextiles and Geomembranes 24 (1), 5257. Auvinet, G., Gonzalez, J.L., 2000. Three-dimensional reliability analysis of earth slopes. Computers and Geotechnics 26, 247261. Balasubramaniam, A.S., Hwang, Z.M., Uddin, W., Chaudhry, A.R., Li, Y.G., 1978. Critical state parameters and peak stress envelopes for Bangkok clays. Quarterly Journal of Engineering Geology and Hydrogeology 11, 219232. Bathurst, R.J., Allen, T.M., Walters, D.L., 2005. Reinforcement loads in geosynthetic walls and the case for a new working stress design method. Geotextiles and Geomembranes 23 (4), 287322. Bergado, D.T., Chai, J.C., Abiera, H.O., Alfaro, M.C., Balasubramaniam, A.S., 1993. Interaction between cohesive-frictional soil and grid reinforcements. Geotextiles and Geomembranes Journal 12 (4), 327349. Bergado, D.T., Chai, J.C., Miura, N., 1995. FE analysis of grid reinforced embankment system on soft Bangkok clay. Computers and Geotechnics 17, 447471. Bergado, D.T., Teerawattanasuk, C., Youwai, S., Voottipruex, P., 2000. FE modeling of hexagonal wire reinforced embankment on soft clay. Canadian Geotechnical Journal 37 (6), 118. Bergado, D.T., Youwai, S., Teerawattanasuk, P., Visudmedanukul, P., 2003. The interaction mechanism and behavior of hexagonal wire mesh reinforced embankment with silty sand backll on soft clay. Computers and Geotechnics 30, 517534. Briaud, J., Lim, Y., 1999. Tieback walls in sand: numerical simulation and design implications. Journal of Geotechnical and Geoenvironmental Engineering, ASCE 125 (2), 101111.

ARTICLE IN PRESS
D.T. Bergado, C. Teerawattanasuk / Geotextiles and Geomembranes 26 (2008) 3955 hexagonal wire mesh and silty sand backll under an in-soil pullout test. Canadian Geotechnical Journal 40 (5), 886899. Varsuo, R.J., Grieshaber, J.B., Nataraj, M.S., 2005. Geosynthetic reinforced levee test section on soft normally consolidated clays. Geotextiles and Geomembranes 23 (4), 362384. Voottipruex, P., 2000. Interaction of hexagonal wire reinforcement with silty sand backll soil and behavior of full scale embankment reinforced with hexagonal wire. Doctoral Dissertation GE-99-1, Asian Institute of Technology, Bangkok, Thailand. 55 Wongsawanon, T., 1998. Interaction between hexagonal wire mesh reinforcement and silty sand backll. M.Eng. Thesis GE-97-14, Asian Institute of Technology, Bangkok, Thailand. Zdravkovic, L., Potts, D.M., Hight, D.W., 2002. The effect of strength anisotropy on the behavior of embankments on soft ground. Geotechnique 52 (6), 447457. Zienkiewicz, O.C., Humpheson, C., Lewis, R.W., 1975. Associated and non-associated visco-plasticity and plasticity in soil mechanics. Geotechnique 25 (4), 671689.

Potrebbero piacerti anche