Geotextiles and Geomembranes 29 (2011) 222e232

Contents lists available at ScienceDirect

Geotextiles and Geomembranes

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

Centrifuge model tests of geotextile-reinforced soil embankments during an earthquake

Liping Wang, Ga Zhang*, Jian-Min Zhang
State Key Laboratory of Hydroscience and Engineering, Tsinghua University, Institute of Geotechnical Engineering, Department of Hydraulic Engineering, Beijing 100084, PR China

a r t i c l e i n f o
Article history: Received 9 February 2010 Received in revised form 24 October 2010 Accepted 7 November 2010 Available online 10 December 2010 Keywords: Slope Soil Geotextile Reinforcement Centrifuge model test Earthquake

a b s t r a c t
The behavior of geotextile-reinforced embankments during an earthquake was investigated using centrifuge model tests, considering a variety of factors such as gradient of slope, water content of soil, geotextile spacing, and input shaking wave. The geotextile-reinforcement mechanism was revealed on the basis of the observations with comparison of the unreinforced embankment. The geotextile signicantly decreases the deformation of the embankment and restricts sliding failure that occurs in the unreinforced embankment during an earthquake. The displacement exhibits an evidently irreversible accumulation with a uctuation during the earthquake which is signicantly dependent on the magnitude of input shaking. The peak strain of the geotextile exhibits a nearly triangular distribution in the vertical direction. The embankment can be divided into two zones, a restricting zone and restricted zone, where the soil and geotextile, respectively, play an active restriction role in the soil-geotextile interaction. The soil restricts the geotextile in the restricting zone, and this restriction is transferred to the restricted zone through the geotextile. The strain magnitude of the geotextile and the horizontal displacement of the geotextile-reinforced embankment decrease with increasing geotextile layers, with decreasing water content of the soil, with decreasing gradient of the slope, and with decreasing amplitude of the earthquake wave. 2010 Elsevier Ltd. All rights reserved.

1. Introduction Geotextiles are an effective reinforcement for various soil structures such as slopes and containing walls (e.g., Schaefer, 1997). In recent years, geotextile has been used to increase the seismic stability level of a large quantity of small-size and medium-size earth embankments, which can be regarded as a kind of typical slope. Field observation is an essential approach to obtain rst-hand data for analyzing the behavior of the geotextile-reinforced slopes/ embankments. For example, eleven reinforced soil structures were visually examined for evidence of distress resulting from an earthquake, and the results showed that these structures exhibited excellent seismic stability (Sandri, 1997). The seismic stability of an old earth-ll dam in Japan was signicantly increased by constructing a counter-balance ll using geosynthetic reinforcement (Tatsuoka et al., 2007). A large number of eld surveys were conducted on geosynthetic-reinforced embankments, and valuable understanding was obtained (e.g., Kelln et al., 2007; Indraratna et al., 2010). Nevertheless, the eld observation cannot easily change the boundary

* Corresponding author. Tel./fax: 86 10 62795679. E-mail addresses: wlp04@mails.tsinghua.edu.cn (L. Wang), zhangga@tsinghua. edu.cn (G. Zhang), zhangjm@tsinghua.edu.cn (J.-M. Zhang). 0266-1144/$ e see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.geotexmem.2010.11.002

conditions or loading styles, which means that this approach cannot be used in an investigation of the geosynthetic-reinforcement mechanism. The proper geosynthetic design of slopes/embankments is largely dependent upon systematic understanding of the behavior of such reinforced soil structures, which can be investigated by numerical simulations and model tests. The limit equilibrium methods, which have been accepted in many engineering codes, were widely employed to evaluate the stability level of geosynthetic-reinforced slopes/embankments (e.g., Srbulov, 2001). A set of equations were formulated to determine the seismic stability and permanent displacement of cover soil in a solid-waste containment system (Ling and Leshchinsky, 1997). Diverse types of analysis methods, including theoretical and numerical methods, were used to investigate the behavior and inuence parameters of the overall stability level of reinforced soil structures (e.g., Sawicki and Lesniewska, 1991; Qhaderi et al., 2005; Shukla and Kumar, 2008; Abusharar et al., 2009; Tolooiyan et al., 2009). The reliability analysis was also introduced to the safety assessment of reinforced soil structures (Genske et al., 1991). The effectiveness of numerical analysis is signicantly dependent on the accurate modeling of the soilegeosynthetic interface, which has been investigated using a number of laboratory tests (e.g., Wu et al., 2008; Zhang and Zhang, 2009; Zhang et al., 2010).

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

223

The model tests offer a powerful approach to investigate the behavior and failure mechanism of reinforced soil structures under earthquake conditions by considering various factors with efciency. For example, shaking table tests were performed on six geosynthetic-reinforced soil retaining wall models, and the results were used to analyze the effectiveness of a pseudo-static seismic stability analysis (Matsuo et al.,1998). Centrifuge model tests play an important role in such a research category as they have the advantage of reproducing the same stress level, similar deformation and failure mechanism as presented in a prototype. A series of centrifuge tests were conducted to investigate the failure mechanism, and a new distribution reinforcement force was proposed (Zornberg et al., 1998). Dynamic centrifuge tests were used to demonstrate that the earthquake loading has a signicant effect on the tension experienced by the geomembrane on a landll slope (Thusyanthan et al., 2007). A diverse range of centrifuge model tests were also conducted on the reinforced slopes with different reinforcement structures, such as geotextiles and soil nails (e.g., Porbaha and Goodings, 1996; Chen et al., 2007; Viswanadham and Knig, 2009; Hu et al., 2010; Wang et al., 2010). However, few results were reported using this type of test on the research of behavior of geotextile-reinforced embankments during an earthquake. The behavior of geotextile-reinforced embankments during an earthquake was investigated by using centrifuge model tests in this paper, considering different factors such as slope inclination, soil moisture, reinforcement layout, and input shaking wave. Apart from traditional investigations of dynamic response such as acceleration amplication, this paper focuses on full observations of earthquakeinduced deformation of the embankments and accordingly reports an examination of the response of embankment, the performance of the geotextile, and the soil-geotextile interaction. Therefore, the geotextile-reinforcement mechanism was revealed to explain how the geotextile reduces the deformation and prevents the probable failure of the embankments. This explanation can help to establish a proper design of geotextile reinforcement under earthquake conditions. In addition, the inuence characteristic of different factors on the dynamic behavior of the reinforced embankment was investigated based on the measurements of centrifuge model tests. 2. Tests 2.1. Devices The 50 g-ton geotechnical centrifuge of Tsinghua University was used for the centrifuge model tests. A specially manufactured shake table was equipped on the centrifuge to generate horizontal earthquake waves via a complex hydraulic pressure servo-system. The model container for the tests was made of aluminum alloy and was 50 cm long, 20 cm wide and 35 cm high. A transparent lucite window was installed on one container side, through which the deformation process of the soil could be observed and recorded. 2.2. Schemes Table 1 lists the centrifuge model tests conducted in this paper. The primary centrifuge tests were conducted on the geotextilereinforced (RP) and unreinforced embankments (UP), respectively, which were compared to investigate the behavior and reinforcement mechanism of the geotextile-reinforced embankments. In addition, a few comparative tests were conducted to discuss the effectiveness of the knowledge and the inuence factors by altering several aspects based on the primary tests, including the gradient of the slope (RC1), the water content of the soil (RC2), the geotextile spacing (RC3), and the amplitude of the input earthquake wave (RC4), respectively.

Table 1 List of centrifuge model tests. Case Gradient Water Geotextile Amplitude of of slope content spacing shaking wave of soil 1.5:1 1.5:1 3:1 1.5:1 1.5:1 1.5:1 17% 17% 17% 12% 17% 17% e 6 cm 6 cm 6 cm 12 cm 6 cm 9.7 g 9.7 g 9.7 g 9.7 g 9.7 g 6.5 g

Unreinforced-primary (UP) Reinforced-primary (RP) Reinforced-comparative-1 (RC1) Reinforced-comparative-2 (RC2) Reinforced-comparative-3 (RC3) Reinforced-comparative-4 (RC4)

2.3. Model Fig. 1 shows the schematic view of the geotextile-reinforced model embankment for the primary test. The unreinforced embankment is identical except for removal of the geotextile. The soil was retrieved directly from the stratum of a forest park in Beijing, China, with a plastic limit and liquid limit of 5% and 18%, respectively. The specic gravity of the soil was 2.7. The average particle size (d50), effective particle size (d10), and control particle size (d60) of the soil were 0.03 mm, 0.0012 mm, and 0.04 mm, respectively. The standard Proctor test results showed that the optimum water content of the soil was 15.2%, with a maximum dry density of 1.79 g/cm3. The model embankment was 25 cm in height for all the tests and a 6-cm-high horizontal soil layer under the embankment was set to diminish the inuence of the bottom container plate on the deformation of the embankment. In addition, silicone oil, with kinematic viscosity of 500 cSt, was painted on both sides of the container to decrease the friction between the embankment and the container. The gradient of slope was 1.5:1 (Vertical: Horizontal) for the primary tests and increased to 3:1 for a comparison. The water content of soil was 17% (corresponding maximum dry density: 1.68 g/cm3) for the primary test and decreased to 12% (corresponding maximum dry density: 1.65 g/cm3) for a comparison. The soil was compacted by 5 cm-thick layers into the container with a dry density of 1.45 g/cm3 for both water contents. It should be noted that the dry density of the soil was selected to be smaller than the actual case to obtain a more signicant deformation for a better analysis of the reinforcement rules. The consolidated undrained cyclic triaxial tests were conducted on the unsaturated soil. Fig. 2 shows the shear modulus of the soil and the Poissons ratio was taken as 0.3 by referring to the similar soils empirically. It can be seen that the peak to peak secant shear modulus decreased with increasing shear strain. The shear strength of the soil (water content: 17%) was 20 kPa in cohesion and 25 in the internal frictional angle, which was obtained under the condition that the axial strain reached 5%. The shear strength of the soil with a moisture content of 12% was somewhat larger. A type of medical gauze with a thickness of 0.14 mm was used to simulate the geotextile of the reinforced embankment. The elastic modulus of this reinforcement material was 40 kN/m according to the strain-controlled tensile loading test with a strain rate of 0.1%/min (Fig. 3). According to similarity criterion (Table 2) (Ko, 1988), it can be shown that the gauze is a reasonable substitute of the geotextile prototype, which has a typical thickness from several mm with an elastic modulus of hundreds of kN/m at a centrifugal acceleration of 50 g. Four layers of geotextile were horizontally arranged with an equal spacing of 6 cm at different elevations throughout the embankment (Fig. 1), and two layers of geotextile (layers 2 and 4 in Fig. 1b) were used for a comparison (RC3). 2.4. Measurements A series of accelerometers with a measurement accuracy of 0.3% was buried in the embankment to measure the acceleration response

224

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

40 35 Shear modulus (MPa) 30 25 20 15 10 5 0 0 .0 1 0 .1 Shear strain (%) 1 Confining pressure 100kPa 200kPa

Fig. 2. Shear modulus behavior of soil obtained from cyclic triaxial tests (water content: 17%).

using the image-based measurement system because a random gray-level distribution was yielded with a mixture of grey soil, white gauze and neighboring white terrazzo. The measurement accuracy of displacement can reach 0.02 mm based on the model dimensions for the centrifuge tests in this paper. The area within the dotted line was used for displacement measurements due to the requirement of the measurement system (Fig. 1b), and it covers the main deformation zone of the embankment. Cartesian coordinates were established with the origin as the intersection of the bottom and the axis of the embankment, specifying positive as downward in the vertical direction (y-axis) and to the right in the horizontal direction (x-axis) (Fig. 1b).

2.5. Procedures In all the tests, the centrifugal acceleration gradually increased to 50 g and was maintained during shaking. A signicant deformation of the embankment occurred in this period, for example, the dry density of the soil in the middle of the embankment increased about 8%. After the deformation became stable at 50 g-level, usually after about 30 min, an earthquake wave was input on the model container bottom. This wave lasted 1 s, with a maximum acceleration of 9.7 g in the model dimension (Fig. 4), equivalent to 0.194 g in the prototype dimension at a centrifugal acceleration of 50 g. This wave was used in all the centrifuge tests,
Fig. 1. Schematic view of geotextile-reinforced embankment in primary test (unit: cm).

8
at different elevations during the earthquake (Fig. 1b). The transducers were linked to the automatic data-acquisition system so that real-time records could be obtained during the tests. An image-record and displacement measurement system was used to record the images of the embankment during the centrifuge model tests, with which the displacement eld and its change could be obtained (Zhang et al., 2009). This system could capture 48 image frames per second. An image-correlation-analysis algorithm was used to determine the displacement vectors of the soil without disturbing the soil itself (Zhang et al., 2006). The displacement history of an arbitrary point in the soil could be measured with subpixel accuracy. The effectiveness of this measurement system could be realized by visualizing a region with a random gray-level distribution, which can be obtained by embedding white terrazzo particles laterally in the soil (Fig. 1a). It should be noted that the displacement of the gauze reinforcement layer could be measured

7 6 Tension (kN/m) 5 4 3 2 1 0 0 5 10 Strain (%) 15 20

Fig. 3. Tensile loading test result of medical gauze.

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232 Table 2 Similarity criterion for centrifuge model tests. Centrifugal acceleration (g) n Strain Stress Length Dynamic Time Et of geotextile acceleration (E: modulus, t: thickness) 1/n n n

225

10 5 a (g) 0 -5

except in the comparative test RC4 where the wave was scaled down with the amplitude of 6.5 g as a comparative shaking wave.

-10 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 t (s)
Fig. 4. Seismic wave input. t, time; a, acceleration.

3. Observations In this paper, all the measured results were presented based on the model dimension, which could be easily transformed to the prototype dimensions from the model dimension using the similarity criterion (Table 2). The displacement during spin-up of the centrifuge was subtracted from the total displacement in the earthquake-induced displacement. Fig. 5 shows the seismic response of several points at the locations of the accelerometers on the axis of the reinforced embankment in the primary test. It should be noted that the displacement refers to the one that is relative to the container during the earthquake, and its high-frequency response waves were ltered because the frame-rate of the captured images was limited (48 frames/s). Thus, a few minor displacements, especially in the horizontal direction, cannot be yielded; however, the measurement results were basically sufcient to analyze the features of displacement response. The horizontal displacement exhibited a signicant uctuation in magnitude, and the vertical displacement monotonically increased during shaking (Fig. 5). In addition, the displacement histories showed that signicant irreversible deformation, both in the horizontal and vertical directions, appeared from the beginning of the earthquake application. The irreversible deformation showed an evident dependence on the magnitude of the input earthquake wave. For example, it increased rapidly in the early earthquake period, accompanied with the remarkable input shaking acceleration, and the rate of increase became low when the wave dropped off. Thus, the post-earthquake displacement of the embankment, which reects the irreversible deformation due to shaking, can be used as a representative for the further analysis of deformation behavior of the

embankment. It can be noticed that Point 6 moved in the opposite direction of Point 1 and 3 over the entire duration of shaking. Close examination of Fig. 5 showed that the peak acceleration and displacement of the embankment both increased with increasing elevation. The amplication factors of these points can be obtained by using the acceleration histories and are summarized according to different tests (Fig. 6). These factors gradually increased with increasing elevation for both reinforced and unreinforced embankments. The amplication factors of the embankment increased if the geotextile reinforcement was used. The dynamic response was slightly increased if the water content of soil was decreased from 17% to 12% because the use of geotextile and change of soil moisture modied the dynamic properties of the embankment. Fig. 7 shows the contour lines of post-earthquake displacement over the geotextile-reinforced and unreinforced embankments in the primary tests, respectively. Here, two points of Fig. 7 should be noted: (1) the borders are designated as the dotted area in Fig. 1b and not the actual borders of the embankment; (2) the contour lines were obtained using interpolations of measured displacement of a series of points, and thus they only roughly describe the displacement distributions and point-based measurement should be used in the rened displacement analysis. It can be seen from Fig. 7 that the horizontal displacement of the embankments increased from the middle to both free surfaces, and the vertical displacement increased with increasing elevation. A distinct landslide occurred during the earthquake in the right side of the unreinforced embankment, which can be identied from the concentration of contours, whereas such a failure was avoided if the

a
a (g)

15 10 5 0 -5 -10 -15 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 0 0.2 0.4 0.6 t (s) 0.8 1

b
a (g)

15 10 5 0 -5 -10 -15 0.1

c
a (g) u (mm) v (mm)

15 10 5 0 -5 -10 -15 0 -0.1 -0.2 0 0.2 0.4 0.6 0 0.2 0.4 0.6 0.8 t (s) 1

u (mm)

u (mm) v (mm)

0.05 0 0 0.5 1 1.5 0 0.2 0.4 0.6 0.8 t (s) 1

v (mm)

1*

3*

6*

Fig. 5. Seismic response of typical points of reinforced embankment in primary test. a, acceleration; u, horizontal displacement; v, vertical displacement; t, time.

226

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

-20

-20 UP (failure) -18

(1)

-15 y (cm)
y (cm)

-16 -14 -12 -10 0

(4)

-10 UP RP RC2 1 1.1 1.2 Af 1.3 1.4 1.5

RP RC3
(2) (1) to decrease geotextile layers

-5

RC4 (3) RC2 2

RC1 (2) to increase slope inclination

(3) to decrease soil moisture (4) to decrease wave amplitude

4 6 u max (mm)

10

Fig. 6. Distribution of amplication factors of acceleration of different embankments. Af, amplication factor of acceleration.

Fig. 8. Maximum post-earthquake horizontal displacement of different embankments. umax, maximum horizontal displacement.

geotextile was used (Fig. 7). In addition, the post-earthquake horizontal displacement of the reinforced embankment was also signicantly smaller than that of the unreinforced one. It can be concluded that the geotextile signicantly decreased the deformation and increased the stability level of the embankment. Fig. 8 shows the maximum post-earthquake horizontal displacements according to different centrifuge model tests. The maximum horizontal displacement always appeared near the surface of the embankment; thus, the position is marked only using the elevation in Fig. 8. It can be seen that reduction of the water content of soil signicantly decreased the horizontal displacement of embankments due to the modulus increase associated with the decreasing moisture content. Moreover, the horizontal displacement of the embankment can be decreased by increasing the geotextile layers, by decreasing the gradient of slope, and by decreasing the amplitude of earthquake wave. The position of maximum horizontal displacement was located in the middle and upper parts of the embankment and was

signicantly affected by the geotextile layers, soil moisture, gradient of slope, and input earthquake wave. The earthquake-induced displacement distribution of the embankment was somewhat asymmetric on both sides; this asymmetry was more signicant in the horizontal displacement (Fig. 7). However, the displacement distribution of the right half of the embankment was similar with that of the left half (Fig. 7). For simplicity of description, the right half of the embankment was mainly used for further analysis of the geotextile-reinforced embankment because this half exhibited larger horizontal displacement than the left half and exhibited evident landslide in the unreinforced embankment.

4. Behavior of geotextile Fig. 9 shows the post-earthquake displacement distribution of four layers of geotextile in the right half of the primary reinforced

Fig. 7. Post-earthquake displacement contour lines of embankments in primary tests (unit: mm).

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

227

embankment. It can be seen that the displacement increased signicantly in the zone that was somewhat far from the axis, and the increase rate became fairly small near the embankment surface. This trend was similar to that of a sand slope (Zornberg and Arriaga, 2003) or a silty slope (Hu et al., 2010) under static loading conditions; however, the at region near the axis of the embankment was shorter than that of the slope. The post-earthquake displacement of a geotextile, U, can be described using the following equation with increasing horizontal coordinate, x:

A B 1 axb

(1)

where a, b, A, and B are the tting parameters, which can be determined by the least square method according to the measured displacement of the geotextile. The horizontal displacement predicted using Eq. (1) was compared with the measured results (Fig. 9), and the comparison results showed that the prediction exhibited a good t to the measurements. This demonstrated that Eq. (1) is effective to t the displacement distribution of the geotextile. The strain of the geotextile, 3, was found from the derivative of Eq. (1), as follows:

Aabxb1 2 1 axb

(2)

The strain distribution of the four layers of geotextile can be derived using Eq. (2) (Fig. 10). It should be noted that the curves do

not reach the axis, that is, they stop at a small distance from the axis because the strain of the geotextile near the central axis of the embankment, which can be concluded to be small, is not considered in the proposed equation. The strain of geotextile increased during the earthquake; however, its peak value nearly occurred at the same position for an individual layer of geotextile during shaking (Fig. 10c), similar to the observation of other studies (e.g., Fannin and Hermann, 1990). The strains of geotextile were both small near the axis and near the surface of the embankment and exhibited a peak value at a certain distance from the axis. The positions of peak strain were different on different layers of geotextile; this demonstrated that the geotextile exerted tensile force with different distribution features if it was placed at different elevations of the embankment. In other words, the reinforcement effect of the geotextile was signicantly dependent on the deformation behavior of the embankment. Fig. 11 summarizes the vertical distribution of the peak strains of geotextile of different tests. The peak strain exhibited a nearly triangular distribution, and the maximum strain appeared in the middle part of the embankment. The maximum strain appeared at the second layer of geotextile in the primary test, where the horizontal displacement was fairly large over the embankment (Fig. 7c). The peak strain of the geotextile exhibited an evident dependence on the horizontal displacement behavior of the embankment (Fig. 8). It can be inferred from Fig. 11 that the layout of geotextile, soil behavior, gradient of slope and input earthquake wave had a signicant effect on the behavior of the geotextile.

1 0.8 U (mm) 0.6 0.4 0.2 0 1 0.8 U (mm) 0.6 0.4 0.2 0 0.8 U (mm) 0.6 0.4 0.2 0 0.3 0.2 U (mm) 0.1 0 -0.1 -0.2 0 5 x (cm)
Fig. 9. Measured horizontal displacement distributions of geotextile in primary test, with comparison by model predictions. U, post-earthquake horizontal displacement of geotextile.

test fit
(%)

2 (a) Reinforcement layer 1 (top) 1.5 1 0.5 0 2.5 2 (%) 1.5 1 0.5 0 1.5 (c) Reinforcement layer 3 1 time 1s (residue) 0.5s 0.16s (%) (b) Reinforcement layer 2

(a) Reinforcement layer 1 (top)

(b) Reinforcement layer 2

(c) Reinforcement layer 3

0.5 0 1 0.8

(d) Reinforcement layer 4 (bottom)

(d) Reinforcement layer 4 (bottom) 10 15

(%)

0.6 0.4 0.2 0 0 5 x (cm) 10 15

Fig. 10. Strain distribution of geotextile in primary test. 3, tensile strain of geotextile.

228

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

-25 -20 -15 -10 -5 0 0.5 1 1.5 Peak strain (%)

Fig. 11. Distribution of peak strain of geotextile of different embankments.

f
RP RC1 RC2 RC4

ds d3 E dx dx

(4)

2.5

5. Soil-geotextile interaction The tensile force of the geotextile increased linearly with increasing strain according to the tensile loading test (Fig. 3), which indicated that the tensile modulus of the geotextile was constant before the rupture of the geotextile occurred. That is,

s E3

(3)

where E is the elastic modulus of geotextile, and s is the tension stress of geotextile. It can be noticed that the tensile force of the geotextile should be balanced with the accumulation of soil-geotextile friction stress; thus the friction stress, f, can be obtained using the derivative of the tension stress of the geotextile, as follows:

Fig. 12 shows the distribution of soil-geotextile friction stress in the primary test that was obtained using Eq. (4) according to the known distribution of the strain of the geotextile (Fig. 10). It is dened that the positive friction stress indicates that the soil pulls the geotextile, whereas the negative indicates that the geotextile pulls the soil. It can be seen that the friction stress was positive near the embankment axis and became negative near the surface, and there was denitely a zero point in friction distribution along the geotextile. This zero point of the friction stress, which is equivalent to the peak strain of the geotextile, can be used as a boundary point to distinguish different restriction states between the geotextile and soil. The positions of zero points moved to the surface with decreasing elevation of the embankment (Fig. 12). Accordingly, a surface, denoted as the P-surface, can be obtained by connecting all the zero points on four layers of the geotextile. This surface can be regarded as the boundary of different reinforcement effects of the geotextile. In other words, the zone at the inner side of the P-surface was denoted as the restricting zone where the soil restricts the geotextile, and the other zone was denoted as the restricted zone where the geotextile restricts the soil (Fig. 12). It can be inferred that the restricting zone of the embankment is smaller than that of the slope because the embankment has the restricted zone on the both sides that can be balanced by each other. Fig. 13 compares the distributions of post-earthquake horizontal displacement of the embankment along different vertical lines, located in the restricting zone and restricted zone, respectively. The horizontal displacement usually increased with increasing elevation; however, it exhibited a signicant change near the geotextile. The changes were different at different zones, which also reected different features of the soil-geotextile interaction. In the restricted zone, namely on the outer side of the P-surface, the displacement of soil signicantly decreased near the geotextile (Fig. 13). This

y (cm)

Axis
Axis

P-surface Geotextile
P-surface Geotextile

Friction stress

u (mm) -0.2 0 0.2 0.4

u (mm) 0 0.6 1.2

Measurement line

u (mm) 0 0.6 1.2

Restricting zone

Restricted zone

10

15

x (cm)

Fig. 12. Distribution of soil-geotextile friction stress in right half of embankment in primary test.

Fig. 13. Vertical distributions of post-earthquake horizontal displacement in right half of embankment in primary test. u, post-earthquake horizontal displacement.

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

229

demonstrated a remarkable trend that the geotextile restricted the soil from deforming to the embankment surface, i.e., the geotextile played an active restriction role in the soil-geotextile interaction. The displacement distributions along the vertical lines between two layers of geotextile were more notably changed near the lower geotextile than near the upper geotextile, which indicates that the soilgeotextile interaction along the lower geotextile was stronger because there was more normal stress and it was farther from the surface. On the contrary, in the restricting zone, namely on the inner side of the P-surface, the displacement of soil signicantly increased near the geotextile (Fig. 13), which demonstrated a remarkable trend where the soil restricted the geotextile from moving to the embankment surface, i.e., the soil played an active restriction role in the soil-geotextile interaction. Therefore, the two zones divided by the P-surface of the embankment exhibited signicantly different features of soil-geotextile interaction, which again conrmed the conclusions on the soil-geotextile interaction that were derived from the friction analysis (Fig. 12). Moreover, the division of the zones can be conrmed from the observations in other comparative tests. 6. Reinforcement mechanism It was found that signicant irreversible deformation of the embankment was induced by the earthquake application (Fig. 5). Furthermore, the geotextile is usually assumed to exhibit tensile stress to mainly restrict the horizontal displacement of the surrounding soil. Thus, the distribution of post-earthquake horizontal displacement of the reinforced embankment was examined to investigate the inuence of geotextile on the embankment deformation resulting from the earthquake (Fig. 14). It can be seen that the horizontal displacement increased with increasing distance from the
0.8 u (mm) 0.7 0.6 0.5 1 u (mm) 0.8
L-surface

axis. Close examinations indicated an inection at the displacement distribution curve from which the rate of increase of displacement began to decrease. Thus, a continuous surface, denoted as the Lsurface, was obtained by connecting these inections using a curve, as shown in Fig. 14 by the dashed line. The distribution of horizontal displacement of the reinforced embankment exhibited different features on different sides of the L-surface, which was also observed in other comparative tests. Fig. 15 outlines the L-surface and P-surface, which divided the right half of reinforced embankment into three different-featured zones. The L-surface was located on the outer side of the P-surface and divided the restricted zone into two zones. A close comparison of soil-geotextile friction distribution (Fig. 12) showed that the soilgeotextile interaction was more signicant on the inner side of the L-surface because there the normal stress was larger on the soilgeotextile interface. Several typical elements, as indicated in Fig. 15, were selected to analyze their strain histories to understand the features in more details. In this paper, a 1-cm-long square element was used for strain analysis. This strain can be determined directly by a common niteelement method with the displacement vector of the nodes being directly obtained using the image-measurement system. The strain at the center was thought to be the strain of this element. All the strain components, at an arbitrary plane of the element, can be easily derived. Fig. 16 compares the strain histories of the typical elements of the reinforced and unreinforced embankments in the primary tests. In the gure, the horizontal normal strain was selected because the geotextile mainly restricted horizontal deformation, and the slope-direction shear strain was selected because the slip surface of the unreinforced slope was approximately parallel to the slope surface (Fig. 7a). It can be seen that the horizontal normal strains of these elements were all negative, demonstrating

Axis

T ypical element Strong application Weak application

(a) y = -20.2cm

L-surface #4
(b) y = -16.2cm

#1

0.6 0.4 1 0.8

P-surface Slip surface in unreinforced slope

#3

#2 Geotextile

u (mm)

0.6 0.4 0.2 0 0.5 0.4 0.3 0.2 0.1 0 0 3 6 x (cm) 9 12 15 (d) y = -4.9cm (c) y = -10.6cm

Continuity Soil Soil Soil

u (mm)

Reinforment

Restricting zone Restricted zone Dominant compression effect Dominant shear effect

Fig. 14. Distribution of post-earthquake horizontal displacement at different elevations in right half of embankment in primary test. u, post-earthquake horizontal displacement.

Fig. 15. Division of embankment in primary test and reinforcement mechanism.

230

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

6 5 4 3 2 1 0 0 -1 (%) -2 -3 -4 -5 0 0.2 0.4 t (s) 0.6 0.8 1 RP UP (%)

2 1.5 (%) (%) 1 0.5 0 0 -0.5 -1 -1.5 -2 -2.5 0 0.2 0.4 0.6 t (s) 0.8 1 RP UP

a Typical element #1
0.8 0.6 (%) 0.4 0.2 0 0 -0.5 -1 (%) -1.5 -2 -2.5 0 0.2 0.4 t (s) 0.6 0.8 1
x

b Typical element #2
0.8 0.6 (%) 0.4 0.2 0

RP UP
(%)

0 -0.5 -1 -1.5 -2 -2.5 0 0.2 0.4 t (s) 0.6 0.8 1 RP UP

c Typical element #3
dilation in the horizontal direction. All the strains changed rapidly in the early period of the earthquake for both the reinforced and unreinforced embankments. They became nearly stable after about 0.4 s, when the input seismic wave began to weaken. The shear strain of Element #1 in the unreinforced embankment increased rapidly in the early period of the earthquake (Fig. 16a). This strain reached a signicant magnitude at about 0.3 s when the landslide, just across this element, occurred. However, the shear strain increased at a smaller rate and reached a lower level far from failure if the geotextile was used. In other words, the geotextile signicantly decreased the shear deformation of the embankment in the area near the slip surface. A concept, shear effect, was introduced to describe that the geotextile decreases the shear deformation of the

d Typical element #4
reinforced embankment in comparison with the unreinforced one. In addition, the horizontal dilation deformation also decreased signicantly; this change caused an evident increase in the horizontal normal strain. Accordingly, the concept of compression effect can be used to describe that the geotextile increases the horizontal compression deformation of the embankment. It can be concluded that a signicant shear effect of geotextile occurred in the area near the slip surface compared with the unreinforced embankment. In contrast to Element #1, Elements #3 and #4, located on the inner side of the L-surface, exhibited the feature that the geotextileinduced increase extent of compression strain seemed larger than the decrease extent of shear strain. The following indexes were dened to describe the extents of the shear effect and compression effect:

Fig. 16. Strain history of typical elements (location indicated in Fig. 15) during earthquake in primary tests. g, slope-direction shear strain; 3x, horizontal normal strain; t, time.

9 post earthquake shear strain of unreinforced slope = post earthquake shear strain of reinforced slope post earthquake normal strain of unreinforced slope ; : compression effect index post earthquake normal strain of reinforced slope 8 < shear effect index

(5)

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232

231

8 #1 6 shear effect index Dominant shear effect Dominant compression effect

(3) (2)

4 #2 2 #3 0 0 #4

(4)

2 4 6 compression effect index

(5)

Fig. 17. Relationship of indexes of shear effect and compression effect of different elements (location indicated in Fig. 15) in primary test.

It can be inferred from Eq. (5) that the larger index means more signicant effect. Fig. 17 shows the relationship of the indexes of shear effect and compression effect of the elements in different positions of the embankment in Fig. 15. A demarcation line with inclination of 45 can be used to divide the four elements into two groups. Elements #1 and #2, on the outer side of the L-surface, were located on the upper side of the demarcation line, whereas Elements #3 and #4, on the inner side of the L-surface, were located on the lower side of the demarcation line, which demonstrated that the L-surface can be used as a boundary to distinguish whether the shear effect or compression effect was dominant. In other words, the shear effect was dominant in the zone on the outer side of the L-surface and the compression effect was dominant in the zone on the inner side of the L-surface, though another effect was also visible (Fig. 15). The geotextile-reinforcement mechanism of the embankments can be explained as follows (Fig. 15). (1) The soil restricts the geotextile in the restricting zone of the embankment. (2) The restriction is transferred to the restricted zone of the embankment through the geotextile. (3) The geotextile exerts a strong restriction on the deformation of the soil on the inner side of the L-surface, which conrms that the friction is accordingly larger in this zone than that in the zone on the outer side of the L-surface (Fig. 12); thus the compression effect is signicant in this zone of the reinforced embankment. (4) The geotextile restriction becomes relatively small on the outer side of the L-surface; however, the compression effect is transferred to this zone due to the continuity of the embankment and prevents the occurrence of landslide. Since the shear deformation is dominant in the landslide, the shear effect becomes signicant in this zone of the reinforced embankment. 7. Remarks and conclusions A series of dynamic centrifuge model tests was conducted to analyze the behavior and reinforcement mechanism of cohesive soil embankments reinforced using geotextile during an earthquake, and different factors of inuence were considered. The main conclusions, drawn on the basis of systematic observations of these tests, as follows: (1) A distinct sliding failure occurred during the earthquake in the right side of the unreinforced embankment, and this failure was restricted by using geotextile. The geotextile also signicantly

(6)

(7)

decreased the earthquake-induced deformation of the embankment. The irreversible deformation appeared from the beginning of the earthquake and was signicantly dependent on the magnitude of input shaking. The acceleration response of the embankment increased with increasing elevation, and the amplication factors increased if the geotextile was used. The strain of a layer of geotextile increased during the earthquake, and its peak value occurred at an invariable position with a certain distance from the axis. The peak strain of geotextile exhibited a nearly triangular vertical distribution. There were zero points in the distributions of soil-geotextile friction stress, which were used to give the P-surface in the embankment. According to the P-surface, the embankment was divided into a restricting zone and restricted zone, where the soil and geotextile, respectively, played the active restriction role in the soil-geotextile interaction. According to the different features in the distribution of horizontal displacement of the embankment, the L-surface was introduced to divide the restricted zone into two sub-zones, on the inner side of which the soil-geotextile interaction is more signicant. The soil restricts the geotextile in the restricting zone, and this restriction is transferred to the restricted zone through the geotextile. The geotextile exerts a strong restriction of the deformation of soil on the inner side of the L-surface, indicating a signicant compression effect, in which the geotextile increases the horizontal compression deformation. The compression effect is transferred to the zone on the outer side of the L-surface due to the continuity of the embankment and prevents the occurrence of landslide in the unreinforced embankment, indicating a signicant shear effect whereby the geotextile decreases the shear deformation of the embankment. The strain magnitude of geotextile and horizontal displacement of the geotextile-reinforced embankment decreased with increasing geotextile layers, decreasing water content of soil, decreasing gradient of slope, and decreasing amplitude of the earthquake wave. The vertical distribution of peak strain of the geotextile was also affected by these factors.

Acknowledgements The study is supported by National Basic Research Program of China (973 Program) (No. 2007CB714108) and National Natural Science Foundation of China (No. 50979045). References
Abusharar, S.W., Zheng, J., Chen, B., Yin, J., 2009. A simplied method for analysis of a piled embankment reinforced with geosynthetics. Geotextiles and Geomembranes 27 (1), 39e52. Chen, H.T., Hung, W.Y., Chang, C.C., Chen, Y.J., Lee, C.J., 2007. Centrifuge modeling test of a geotextile-reinforced wall with a very wet clayey backll. Geotextiles and Geomembranes 25 (6), 346e359. Fannin, R.J., Hermann, S., 1990. Performance data for a sloped reinforced soil wall. Canadian Geotechnical Journal 27, 676e686. Genske, D., Klapperich, H., Adachi, T., Sugito, M., 1991. Earthquake resistant design of earth walls. A probabilistic approach. Soil Dynamics and Earthquake Engineering 5, 403e414. Hu, Y., Zhang, G., Zhang, J.M., Lee, C.F., 2010. Centrifuge modeling of geotextilereinforced cohesive slopes. Geotextiles and Geomembranes 28 (1), 12e22. Indraratna, B., Nimbalkar, S., Christie, D., Rujikiatkamjorn, R., Vinod, J., 2010. Field assessment of the performance of a ballasted railtrack with and without geosynthetics. Journal of Geotechnical and Geoenvironmental Engineering 136 (7), 907e917. Kelln, C., Sharma, J., Hughes, D., Gallagher, G., 2007. Deformation of a soft estuarine deposit under a geotextile reinforced embankment. Canadian Geotechnical Journal 44 (5), 603e617. Ko, H.-Y., 1988. Summary of the state-of-art in centrifuge model testing. Centrifuge 88. In: Craig, James, Schoeld (Eds.). Balkema, Rotterdam, pp. 11e18.

232

L. Wang et al. / Geotextiles and Geomembranes 29 (2011) 222e232 Tolooiyan, A., Abustan, I., Selamat, M.R., Ghaffari, S., 2009. A comprehensive method for analyzing the effect of geotextile layers on embankment stability. Geotextiles and Geomembranes 27 (5), 399e405. Viswanadham, B.V.S., Knig, D., 2009. Centrifuge modeling of geotextile-reinforced slopes subjected to differential settlements. Geotextiles and Geomembranes 27 (2), 77e88. Wang, L.P., Zhang, G., Zhang, J.-M., 2010. Nail reinforcement mechanism of cohesive soil slopes under earthquake conditions. Soils and Foundations 50 (4), 459e469. Wu, W., Wick, H., Ferstl, F., Aschauer, F., 2008. A tilt table device for testing geosynthetic interfaces in centrifuge. Geotextiles and Geomembranes 26 (1), 31e38. Zhang, G., Hu, Y., Zhang, J.M., 2009. New image-analysis-based displacement-measurement system for geotechnical centrifuge modeling tests. Measurement 42 (1), 87e96. Zhang, G., Liang, D.F., Zhang, J.M., 2006. Image analysis measurement of soil particle movement during a soil-structure interface test. Computers and Geotechnics 33 (4e5), 248e259. Zhang, G., Wang, L.P., Zhang, J.M., 2010. Monotonic and cyclic modeling of interface between geotextile and gravelly soil. International Journal for Numerical and Analytical Methods in Geomechanics 34, 1346e1361. Zhang, G., Zhang, J.M., 2009. Large-scale monotonic and cyclic tests of interface between geotextile and gravelly soil. Soils and Foundations 49 (1), 75e84. Zornberg, J.G., Arriaga, F., 2003. Strain distribution within geosynthetic-reinforced slopes. Journal of Geotechnical and Geoenvironmental Engineering 129 (1), 32e45. Zornberg, J.G., Sitar, N., Mitchell, J.K., 1998. Performance of geosynthetic reinforced slopes at failure. Journal of Geotechnical and Geoenvironmental Engineering 124 (8), 670e683.

Ling, H.I., Leshchinsky, D., 1997. Seismic stability and permanent displacement of landll cover systems. Journal of Geotechnical and Geoenvironmental Engineering 123 (2), 113e122. Matsuo, O., Tsutsumi, T., Yokoyama, K., Saito, Y., 1998. Shaking table tests and analyses of geosynthetic-reinforced soil retaining walls. Geosynthetics International 5 (1e2), 97e126. Porbaha, A., Goodings, D.J., 1996. Centrifuge modeling of geotextile- reinforced steep clay slopes. Canadian Geotechnical Journal 33 (5), 696e703. Qhaderi, R., Vafaeian, M., Hashemolhoseini, H., 2005. A parametric study of the behavior of geosynthetic reinforced soil slopes. International Journal of Engineering. Transactions B: Applications 18 (4), 371e389. Sandri, D.,1997. A performance summary of reinforced soil structures in the greater Los Angeles area after the Northridge earthquake. Geotextiles and Geomembranes 15 (4e6), 235e253. Sawicki, A., Lesniewska, D., 1991. Stability of fabric reinforced cohesive soil slopes. Geotextiles and Geomembranes 10 (2), 125e146. Schaefer, V.R., 1997. Ground improvement, ground reinforcement, ground treatment: developments 1987-1994. ASCE Geotechnical Special Publication 69, 603. Shukla, S.K., Kumar, R., 2008. Overall slope stability of prestressed geosynthetic-reinforced embankments on soft ground. Geosynthetics International 15 (2), 165e171. Srbulov, M., 2001. Analyses of stability of geogrid reinforced steep slopes and retaining walls. Computers and Geotechnics 28, 255e268. Tatsuoka, F., Tateyama, M., Mohri, Y., Matsushima, K., 2007. Remedial treatment of soil structures using geosynthetic-reinforcing technology. Geotextiles and Geomembranes 25 (4e5), 204e220. Thusyanthan, N.I., Madabhushi, S.P.G., Singh, S., 2007. Tension in geomembranes on landll slopes under static and earthquake loading- centrifuge study. Geotextiles and Geomembranes 25 (2), 78e95.