Seismic stability of reinforced soil walls

Koseki, J.
Institute of Industrial Science, University of Tokyo, Japan

Bathurst, R.J.
GeoEngineering Centre at Queen's-RMC, Canada

Gler, E.

Bogazici University, Turkey

Kuwano, J.

Saitama University, Japan

Maugeri, M.
University of Catania, Italy

Keywords: geosynthetic-reinforced soil retaining wall, earthquake, case history, model test, numerical analysis, seismic design, limit states design ABSTRACT: This paper provides a review of the seismic performance of reinforced soil walls in recent earthquakes using published case histories. Included are cases histories of walls used to replace conventional structures that were damaged during severe earthquakes and new construction technologies. The paper reviews recent physical model testing using shaking tables and summarizes lessons learned regarding seismic performance and potential failure mechanisms. Differences in seismic response of reinforced soil walls and conventional structures are identified. Analytical and numerical approaches for the displacement and collapse analysis of reinforced soil structures are summarized. Finally, the current status of limit states design codes for reinforced soil wall structures against earthquake is reviewed. Length of construction (km) 1 INTRODUCTION 70 60 50 40 30 20 10 0 89 Total

The good performance of geosynthetic-reinforced soil walls (GRS walls) during earthquake has been documented widely in the literature. In Japan, the greater seismic resistance of GRS walls compared to conventional retaining wall structures has led to their increasing use for new permanent structures (Tatusoka et al. 1997a) and to replace conventional structures damaged in recent earthquakes. For example, Figure 1 shows the annual and cumulative face length of GRS walls with a rigid facing constructed in Japan since 1989 (Tamura 2006). Between 1995 and 2001, the construction rate increased from less than 5 km to more than 10 km per year. This is due to the good seismic performance of GRS walls observed during the 1995 HyogokenNanbu (Kobe) earthquake. The more recent lower annual construction rates shown in the figure are due to a weaker Japanese economy. Nevertheless, over a 15-year period the total construction length of this type of reinforced soil wall is greater than 70 km. GRS walls are becoming common retaining wall structures in other countries as well, primarily because of cost savings. A study by Koerner et al. (1998) showed that GRS walls built for government projects were typically 50% cheaper than conventional retaining wall structures. Seismic issues related to GRS walls have been addressed in many keynote lectures and reports at previous geosynthetics conferences (Table 1). The

Kobe Eq. 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04

Fiscal year
Figure 1. Growth statistics for GRS walls with full-height rigid facing in Japan (Tamura 2006). Table 1. Keynote lectures and special reports in previous conferences on seismic issues for GRS walls. Lecture / report Conference Tatsuoka et al. (1995) 10th Asian Regional Conference on SMFE Bathurst and Alfaro (1997)* IS-Kyushu 96 White and Holtz (1997) IS-Kyushu 96 Tatsuoka et al. (1997) IS-Kyushu 96 Tatsuoka et al. (1998) 6ICG, Atlanta Murata (2003) IS-Kyushu 2001 * extended and updated by Bathurst et al. (2002)

Kobe case histories have been reported in detail by Tatsuoka et al. (1995, 1997b). Case history examples from the 1994 Northridge earthquake, USA, were reported by White and Holtz (1997). Bathurst

and Alfaro (1997) and Bathurst et al. (2002) provide an extensive review of the seismic design, analysis and performance of GRS walls. Tatsuoka et al. (1998) discussed the stability of GRS walls against high seismic loads. Murata (2003) reported further progress on research and development in Japan related to GRS walls. Nevertheless, it is timely to update our current knowledge of the seismic performance, analysis and design of GRS wall structures, and from this review to identify deficiencies in current design practice and areas of future research. In this paper we address several general questions: 1. How have GRS walls performed during recent large earthquakes? 2. What lessons can be learned from field performance of GRS walls during earthquake? 3. How have GRS walls been used to replace damaged conventional retaining wall structures due to earthquake? 4. What lessons can be learned from physical and numerical modeling of GRS walls under simulated seismic loading? 5. What is the current and future practice for seismic design of GRS walls including progress towards limit states design-based approaches? We attempt to answer these questions based on personal experience and a review of the literature. The paper begins with a review of field case histories from five different countries. Next we give examples of physical model testing that has been used to provide insight into seismic performance of GRS walls and other types of retaining wall systems. Current methods for the design and analysis of GRS walls are briefly reviewed in the fourth section. In the next section we review the current status of design for GRS walls for earthquake with special emphasis on developments toward limit states designbased approaches. 2 CASE HISTORIES

L = 1.8 m 12 11 10
H = 6.5 m

Layer number 9 8 7 6 5 4 3 2 1
8

L=5.5 m
a) cross-section

b) photograph taken after 1994 Northridge earthquake showing cracking at the back of shortened reinforcement lengths at top of wall. However, no displacements or distress was recorded at the front of the wall. Figure 2. Valencia wall (USA) (Bathurst and Cai 1995).

In order to answer the first three questions introduced in the previous section, we review case histories from six major earthquakes: A) 1994 Northridge Earthquake (USA) B) 1995 Hyogoken-Nanbu (Kobe) Earthquake (Japan) C) 1999 Chi-chi Earthquake (Taiwan) D) 1999 Kocaeli Earthquake (Turkey) E) 2001 El Salvador Earthquake (Central America) F) 2004 Niigataken-Chuetsu Earthquake (Japan)

2.1 1994 Northridge earthquake Sandri (1994) conducted a survey of reinforced segmental (modular block) retaining walls greater than 4.5 m in height in the Los Angeles area immediately after the Northridge earthquake of 17 January 1994 (Moment Magnitude = 6.7). The results of the survey showed no evidence of visual damage to 9 of 11 structures located within 23 to 113 km of the earthquake epicenter. Two structures (Valencia and Gould walls) showed tension cracks within and behind the reinforced soil mass that were clearly attributable to the results of seismic loading. Figure 2a shows the Valencia wall with shortened top reinforcement layers to facilitate placement of subsur2

and unreinforced segmental walls were observed to have developed cracks in the backfill during a survey by Stewart et al. (1994). They concluded that concrete crib walls may not perform as well as more flexible GRS retaining wall systems under seismic loading. 2.2 1995 Hyogoken-Nanbu (Kobe) earthquake Tatsuoka et al. (1995, 1996, 1997b) reported on the performance of a GRS wall at Tanata with a fullheight rigid facing which survived the 1995 Kobe earthquake with minor damage. A residual lateral displacement of about 20 cm was recorded at the top and 10 cm at the bottom with respect to the adjacent box culvert (Figure 3b). The standard procedure for staged construction of a GRS wall with a full-height rigid facing in Japan is illustrated in Figure 3c. On the other side of the culvert structure at the Tanata site, a cantilever-type reinforced concrete (RC) wall supported on bored piles was constructed (Figure 4). This conventional structure with a more expensive foundation treatment exhibited similar deformations due to the earthquake as the GRS wall. The conclusion drawn from this comparison is that a reinforced soil wall can be expected to perform as well as a conventional retaining wall structure constructed with a deep foundation. There were many conventional retaining wall structures without deep foundation support that showed excessive tilting and in some cases internal failure of structure. These structures had to be removed and rebuilt after the earthquake. An example is illustrated in Figure 5. The seismic stability of three different types of retaining wall structure is discussed by Tatsuoka et al. (1998) based on back-analyses. Performance differences between conventional and GRS walls are also described in Section 3.1 using the results from physical model tests. 2.3 1999 Chi-chi earthquake Several GRS walls were damaged during the 1999 Chi-chi earthquake in Taiwan (Huang 2000, Koseki and Hayano 2000, Ling and Leshchinsky 2003). An example segmental-type wall using masonry concrete blocks and a poor quality backfill is illustrated in Figure 6. The reinforcement spacing was 80 cm, which exceeded recommendations by the NCMA (1997) (i.e. maximum spacing not greater than twice the toe to heel dimension of the block). This vertical spacing may have been too large to prevent excessive deformation of the facing. In addition, there was evidence of rupture of the reinforcement at the connections and pullout of the connecting pins between block courses as the facing opened up (bulged) during failure. However, it is difficult to conclude that these post-earthquake observations are evidence that the facing triggered the wall collapse 3

a) cross-section

b) photograph taken after 1995 Kobe earthquake

c) construction sequence for GRS wall with rigid facing Figure 3. Tanata GRS wall (Tatsuoka et al. 1996).

face utilities. Despite the shortening of these layers by the contractor, there was no visual evidence of facing movement even though peak horizontal ground accelerations as great as 0.5g were estimated at the Valencia wall site (Figure 2b). Nevertheless, the need to increase the number and length of reinforcement layers close to the top of segmental retaining walls is a recurring point in this review paper. Bathurst and Cai (1995) analyzed both structures and showed that the location of cracks could be reasonably well predicted using conventional pseudostatic (Mononobe-Okabe) wedge analyses. A similar survey of three GRS walls by White and Holtz (1997) after the same earthquake revealed no visual indications of distress. Some unreinforced crib walls

Figure 4. Conventional RC cantilever wall after 1995 Kobe earthquake (Tatsuoka et al. 1996).

a) photograph of damaged structure

Precastconcrete blocks Overall instability (Possible failure plane) 80cm (Collapse) Local instability (Bulging) (Rupture of reinforcement or pull-out of connecting pin) Reinforcements (Geogrid)

b) cross-section of failed wall Figure 6. Example failed segmental GRS wall after 1999 Chichi earthquake (Koseki and Hayano 2000).

Figure 5. Example of counterfort and conventional cantilever walls after 1995 Kobe earthquake (Tatsuoka et al. 1996).

support a bridge approach fill (Figure 7). The wall has a maximum height of 10 m with a variable cross-section geometry and complicated foundation conditions. The soil reinforcement was a woven geotextile. There was no significant deformation of the wall recorded by an inclinometer or any visual indications of distress due to the earthquake. 2.5 2001 El Salvador earthquake A magnitude 7.6 earthquake occurred on 13 January 2001, 60 km off the coast of El Salvador. A large number of retaining walls were damaged. At the time of the earthquake, there were 25,000 m2 of modular block (segmental) retaining walls in place (Race and del Cid 2001). A survey of these structures by the second writer provided an opportunity to identify why some structures behaved well and others did not. All the structures were constructed on hillsides to extend property fills. The walls were up to 8 m in height. Two walls that failed were examined in detail. Contributing factors to the failures were the use of masonry privacy fences attached to the top of the walls that developed additional overturning loads at the top of the wall. A second major cause of failure of the walls was the extension of the 4

or that the facing failed as part of a larger kinematical mechanism. However, it was noted from a series of segmental GRS walls during the same earthquake that the magnitude of damage diminished with decreasing reinforcement spacing. For example, a wall with a vertical spacing of 60 cm, and thus satisfying NCMA (1997) spacing recommendations, did not show any damage. 2.4 1999 Kocaeli earthquake The Koceli earthquake occurred on 17 August 1999 with Moment Magnitude 7.4. The effects of the earthquake extended to Istanbul where many buildings suffered damage. At the time of the earthquake there was only one GRS segmental retaining wall in Turkey. The wall is located in Istanbul and is used to

a) collapse of railway embankment

Figure 7. Example cross-section of GRS wall constructed in Istanbul 1997. (Note: dimensions in meters).

b) reconstruction of embankment

Figure 8. Wall failure in El Salvador due to 1.7-m high unreinforced section at top of wall added by owner after original construction.

top unreinforced (gravity) portion of the walls that were added by the owners after construction (Figure 8), or the cutting of the reinforcement behind the walls to install subsurface utilities. All walls that were designed and built in compliance with NCMA seismic design guidelines (Bathurst 1998) were observed to have survived the earthquake without damage. 2.6 2004 Niigataken-Chuetsu earthquake During the 2004 Niigataken Chuetsu earthquake in Japan, many embankments for roads, railways and hillside widening were damaged (JSCE 2006). Figure 9 shows the collapse and reconstruction of a railway embankment (Kitamoto et al. 2006). Repair involved a combination of GRS retaining wall and earth anchors. The benefits of a similar approach to stabilize the base of these structures using soil nails is described later in Section 3.3. Figure 10 shows another case history, where both a national highway and a railway embankment were severely damaged by the same slide during the Chuetsu earthquake. The highway embankment was rec) cross-section showing reinforced soil embankment and earth anchors Figure 9. Railway embankment collapse and repair as a result of 2004 Chuetsu earthquake (Kitamoto et al. 2006).

paired using a GRS wall with segmental facing panels (Figures 11a,b). In this construction technique, the facing panels are not in contact with the reinforced soil backfill during fill placement. This allows the backfill to be well compacted without disturbing the facing. The soil is contained by an internal wrapped face and a metal mesh form. The railway embankment on the down slope side was reconstructed using a GRS wall with a full-height rigid 5

56 m

Figure 10. Failure of highway and adjacent railway embankment resulting from 2004 Chuetsu earthquake (JSCE 2006).
1500 3500 3500 2000 3000 750

a) full-height panel facing repair of railway embankment face on side opposite to highway embankment
Cast-in-situ rigid facing
13.18 m

Rail track

Panel-type facing Existing gravity-type retaining wall Before E.Q.

400

Location of existing gravity-type retaining wall after E.Q.

850 S-1 6020

Before EQ After reconstruction After 1:2 EQ .0 Silt stone

1:0.3

Collapsed gravity-type retaining wall Shinano river

1:4.0

Temporary shot crete

a) cross-section of highway embankment repair

b) concrete panel facing placement during repair of highway embankment Figure 11. Highway embankment repair as a result of 2004 Chuetsu earthquake (JSCE 2006).

facing (Figures 12a,b). The different repair decisions were decided based ground conditions, construction time and available backfill material. For example, to improve resistance against a global sliding failure, the bottom of the rigid facing was embedded in the stiff sandstone bedrock layer at the site (Figure 12b).

1500 V= 3000 m3
1000

5170

Sand stone

5%

Sand stone
5700
200

Co
100 300

b) cross-section Figure 12. Railway embankment repair as a result of 2004 Chuetsu earthquake (JSCE 2006).

PHYSICAL MODEL TESTS

Reduced-scale model testing using shaking tables is the most common approach to gain qualitative and quantitative insights into the seismic behavior of reinforced soil wall systems. In some cases, shaking tables mounted on a centrifuge have been used (e.g. Izawa et al. 2002, 2004). In very rare instances, fullscale tests have been performed. A disadvantage of reduced-scale tests is that the response of the model may be influenced by low confining pressure, farend boundary conditions of the shaking table strong box, and improperly scaled mechanical properties of the reinforcement. Nevertheless, qualitative insights are possible using this experimental approach. Furthermore, the models can be used to develop and validate numerical codes that can be used in turn to investigate wall response at prototype scale. A review of shaking table tests reported in the literature prior to 2002 is reported by Bathurst et al. (2002). An example of preparation and testing of a more recent GRS wall shaking table model with a sandy soil backfill is shown in Figures 13 and 14 (Lo Grasso et al. 2004, 2005, 2006) using sandy soil as the wall backfill. 6

Surchrge

155

Surchrge

140

53

ModelBackfill
23cm

53 ModelBackfill 20

20

(Dr=90%)

a. Cantilever type(C) 140

b. Gravity type(G) 140

20 Surchrge

a) facing construction

b) reinforcement placement
50

20

Surchrge

80 45
Extende Reinforcemen

50 ModelBackfill 20

20

ModelBackfill

c. Reinforced-soil type 1(R1)

d. Reinforced-soil type 2(R2)

Figure 15. Wall models used to investigate the influence of wall type and soil reinforcement arrangement on simulated seismic response (Watanabe et al. 2003).
-1.0

c) pluviation of sand

d) vertical columns of black sand markers

Base acceleration(G)

-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
Modified from N-S component at Kobe Marine Meteorological Observation Station during the 1995 Hyogoken-Nanbu earthquake

Figure 13. Preparation of shaking table model (model height = 0.35 m) (Lo Grasso et al. 2005).

amax

a) wall prior to shaking

b) wall after shaking

Time(sec)

Figure 14. Shaking table test (Lo Grasso et al. 2004).

In this section we investigate the effect of wall type (i.e. conventional versus GRS wall structures) on seismic response. For GRS walls, the influence of facing type, reinforcement arrangement and excitation record on wall performance is reviewed. 3.1 GRS walls with full-height rigid facing In order to investigate the reasons for the relatively good performance reported for GRS walls with fullheight rigid facings during the 1995 Kobe earthquake, a series of 1g model shaking table tests were carried out by Watanabe et al. (2003). The wall models were about 50 cm high with a level backfill Figure 15. They were seated on a horizontal soil foundation and uniformly surcharge loaded. The foundation and backfill were modeled by very dense dry sand layers. The models were subjected to several sequential base excitation records by scaling the Kobe earthquake record shown in Figure 16 in 0.1g increments (see also Koseki et al. 2003). Figure 17 shows the cumulative permanent horizontal displacements near the top of each model GRS and conventional-type wall. Up to a seismic coefficient value of about 0.5 no significant difference could be observed. However, under higher seismic loads, the residual wall displacements accu-

Figure 16. Example Kobe earthquake accelerogram (Watanabe et al. 2003).

mulated rapidly with the conventional-type retaining walls. In contrast, the GRS structures exhibited more ductile behavior, particularly the wall with extended reinforcement layers (Type 2 - Figure 15d). The final geometry of the model walls, deformation of backfill and soil failure planes are shown in Figure 18. The conventional-type walls suffered not only base sliding, but also overturning. The overturning mode was more predominant with the GRS structures. It should be noted that multiple progressive failure planes were formed in the backfill due to the effects of strain localization and strain softening behavior (Koseki et al. 1998). Note also that the reinforced backfills underwent a simple shear mode of deformation. This is different than a direct sliding mechanism, which is assumed in most limit equilibrium-based stability analyses with the reinforced backfill taken as a rigid body. The reason for the less ductile behavior of conventional retaining walls can be understood from Figure 19. The normal bearing stress at the toe of the rigid footing of the gravity wall model suddenly decreased after about 20 mm of maximum wall movement, consistent with loss of bearing capacity at the toe. A local toe failure was not observed for the reinforced soil walls because wall loads were spread 7

Wall top displacement, dtop (mm)

90 80 70 60 50 40 30 20 10 0

dtop

5cm Backfill Subsoil

Local failure due to loss of bearing capacity

35 Normal stress at bottom of base footing, (kPa) 30 25

4 3 20 2 1 5 6 1-9: Shaking step 7

Location of loadcells

LT7 6 5 4 LT7 (toe)

9 8

Gravity type Cantilever type

(with partly extended reinforcements)

15 10

LT5 LT6 LT4 (heel)

0 10

a)

Reinforced soil

5 0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3

Seismic coefficient, kh=amax/g

20 30 40 50 60 70 80 90 100 110 Wall top displacement, dtop (mm)

Figure 17. Permanent displacement at top of wall versus peak horizontal ground acceleration coefficient (Watanabe et al. 2003).

Figure 19. Footing load response during base shaking of gravity wall structure (Watanabe et al. 2003).
Tensile force (N) 15 10 5 0 15 10 5 0 0 25 Tensile force (N) 20 15 10
T y p e 3 (L = 3 5 c m ) T y p e 1 (L = 2 0 c m )

U p p e rm o s t re in fo rc e m e n t
T y p e 2 (L = 8 0 c m )

T ype 3 (L = 3 5 c m )

10
T ype 3 (L = 3 5 c m )

20

30

40

50

60

Tensile force (N)

Reinforced zone

M id d le -h e ig h t re in fo rc e m e n t

Cantilever type amax=765gal

Reinforced-soil type1 amax=1019gal

T y p e 1 (L = 2 0 c m ) T y p e 2 (L = 4 5 c m )

10

20

30

40

50

60

L o w e s t re in fo rc e m e n t

T y p e 2 (L = 2 0 c m )

Reinforced zone

T y p e 1 (L = 2 0 c m )

Gravity type amax=919gal

Reinforced-soil type2 amax=1106gal

Figure 18. Final wall geometry, soil deformations and internal soil failure planes (Watanabe et al. 2003).

5 0 0 10 20 30 40 50 60 W a ll to p d is p la c e m e n t, d to p (m m ) 70

over the wider and more flexible base of the reinforced soil zone. Figure 20 shows the accumulation of tensile loads in reinforcement layers at three different heights. The good performance of GRS walls is related to the stabilizing influence of reinforcement layers on wall facing deformations. For example, the largest tensile load was generated in the uppermost extended reinforcement layer of the Type 2 wall. This enhanced the resistance against overturning of the facing. This highlights the observation by others that the distribution of reinforcement loads may be higher at the top of the wall under earthquake loading which is opposite to the trend typically assumed for walls under static loading conditions. Figure 21 shows a global failure mechanism developed behind and below a GRS wall model constructed with a sloped toe. This highlights the influence of the foundation and toe conditions on the ultimate failure mechanism in these structures which can result in a failure pattern that is not predicted by

Figure 20. Reinforcement loads during base shaking of GRS wall structures (Watanabe et al. 2003).

Figure 21. Development of global instability mechanism during shaking of GRS model wall (Kato et al. 2002).

0.075 m

1.0 m inclinometer tube

35 30

Formation of shear bands (refer to Fig. 21.)

0.225 m

plywood back

extensometer cable sand backfill

Tensile force (N)

25 20 15 10 5 0 0 10 20 30

Reinforced(S)

1.0 m

reinforcement

glued sand layer

Reinforced(L)
(S):on slope (L):on level ground
40 50 60
slide rails

plywood base rigid steel wall

Shaking Table

Lateral displacement at wall top (mm)

a) shaking table model

7

Figure 22. Sum of tensile loads in reinforcement layers for GRS walls with a sloped foundation toe (S) and level ground (L) (Kato et al. 2002).

conventional limit equilibrium-based models that assume horizontal sliding at the base of the reinforced soil mass. The histories of the sum of reinforcement loads generated in identical walls, one constructed with a slope at the toe and the other with a level foundation, are plotted in Figure 22. The data show that reinforcement loads peaked and then softened as wall deformations accumulated for the structure with a slope while the structure with a level foundation accumulated reinforcement load monotonically with increasing wall deformation. El-Emam and Bathurst (2004, 2005) investigated the influence of toe constraint using a series of 1 mhigh shaking table tests constructed with a rigid facing panel (Figure 23a). The horizontally restrained toe in reduced-scale models attracted approximately 40% to 60% of the peak total horizontal earth load during base excitation demonstrating that a stiff facing column plays an important role in resisting dynamic forces under simulated earthquake loading. The inability of current design methodologies to account for the portion of lateral earth force resisted by the restrained toe at the base of a structural facing column leads to an under-estimation of wall load capacity. They also recorded the vertical toe force developed at the base of the model walls during base shaking and showed that down-drag forces developed at the connections resulted in vertical toe loads that were significantly higher than the self-weight of the facing panel. Current limit equilibrium-based methods of design were shown to consistently under-estimate the total load carried by the reinforcement and horizontal toe in their models, which is non-conservative for design (Figure 23b). However, the pseudo-static NCMA method (Bathurst 1998) was shown to be less conservative than the current AASHTO (2002) method. 3.2 GRS walls with segmental (modular block) facing Bathurst et al. (2002) reported the results of a series of shaking table tests that examined seismic resis-

NCMA method measured predicted facing configuration vertical thin facing (Wall 9) 6 inclined thin facing (Wall 11) vertical thick facing (Wall 8) 5 load (kN/m) 4 3 2 1 0 AASHTO/FHWA method 0 0.1 0.2 0.3 0.4 input base acceleration amplitude (g) 0.5 0.6

b) computed and measured horizontal loads in reinforcement layers versus peak base excitation acceleration level Figure 23. Reduced-scale rigid panel wall models with a hinged toe (El-Emam and Bathurst 2005).

Figure 24. Shaking table test arrangement for 1-m high GRS wall models (Bathurst et al. 2002).

tance of 1 m-high model reinforced walls with vertical and inclined facings constructed with segmental blocks, vertical incremental panel and rigid fullheight panels. The tests were focused on the influence of interface shear properties on facing column stability (Figure 24). The influence of interface shear capacity and facing batter can be seen in Figure 25. The vertical wall (Test 4) with fixed interface construction (equivalent to a rigid panel wall) required 9

800 Disp Disp Disp Disp

1435

Disp 2 1 10 200

Figure 27. Soil nails installed below base of wall facing (Kato et al. 2002). Figure 25. Wall displacement versus peak base acceleration for model walls with different facing types (Bathurst et al. 2002).

100

Tensile force(N)

80 60 40 20 0 0.0

without nails

0.4

0.8

1.2

1.6

Seismic coefficient
100

Tensile force(N)

Figure 26. Full-scale shaking table test of segmental (modular block) retaining wall (Ling et al. 2005).

80 60 40 20 0 0.0

with nails

the greatest input acceleration to generate large wall displacements during staged shaking. The vertical wall with no shear connections between segmental block layers performed worst (Test 1). However, the resistance to wall displacement was improved greatly for the weakest interface condition by simply increasing the wall batter to 8 degrees from the vertical (Test 3). The vertical wall with an incremental panel facing (i.e. segmental layers rigidly connected between geosynthetic layers) (Test 2) gave a displacement response that fell between the results of walls 1 and 4. Ling et al. (2005) reported the only experimental program to date in which full-scale model (2.8 mhigh) shaking table tests have been carried out on GRS segmental retaining walls (Figure 26). The walls were subjected to both vertical and horizontal components of the Kobe earthquake accelerogram. The test walls showed maximum deformations at the crest that were negligibly small at a scaled peak horizontal acceleration of 0.4g. The walls exhibited less than 100 mm of maximum deformation under a more severe acceleration scaled to 0.86g. The tests showed that increasing the length of reinforcement at the wall crest and reducing the reinforcement

0.4

0.8

1.2

1.6

Seismic coefficient
Figure 28. Influence of soil nails on sum of tensile loads in reinforcement layers (Kato et al. 2002).

spacing improved the seismic response of the structures. 3.3 Other new applications In order to improve the seismic performance of GRS walls constructed on slopes, Kato et al. (2002) reported the beneficial effects of installing soil nails below the base of the facing (Figure 27). This technique resulted in a potential soil failure mechanism that allowed a larger tensile load to be carried by the reinforcement layers (Figure 28). It should be noted that, in the case without nails, the tensile load was 10

reduced after formation of the global failure mechanism (see for example Figures 21 and 22 for sloped toe case). Reinforcement load-softening was prevented by using soil nails below the base of the wall facing. Another technique involves improving the mechanical properties of on-site soils by the addition of cement. Examples of this technique are described here. GRS structures have also been studied with respect to improving the seismic stability of bridge abutments supporting bridge decks. Aoki et al. (2003) carried out 1g shaking table tests on conventional and GRS bridge abutment models using cement-treated backfill and pre-loaded and prestressed GRS walls with gravel backfill and fullheight rigid facings (Figure 29). The seismic stability of cement-treated abutments was increased significantly, compared to cement-treated structures without geosynthetic reinforcement that are currently in use in Japan. An example structure using cemented-treated gravel in combination with geosynthetic reinforcement is illustrated in Figure 30 for a bridge abutment supporting the tracks for a new bullet train on Kyushu Island (Aoki et al. 2005). The construction technique resulted in a 20% saving compared to the conventional solution without the combination of cement-treated backfill and geosynthetic reinforcement. Saito et al. (2006) carried out a series of 1g shaking table tests that showed that the cement treatment of sandy backfill soils in combination with geosynthetic reinforcement can also result in improved seismic performance. An example of the model tests is shown in Figure 31. The top figure shows a structure with reinforcement, and the lower figure the nominal identical structure without reinforcement, which failed at a lower seismic loading. A variation of the combined cement-treated backfill and geosynthetic reinforcement technique has been reported by Ito et al. (2006) to construct an embankment for a national highway in Japan (Figure 32). Another potential technology is the use of expanded polystyrene (EPS) as a geofoam seismic buffer for rigid wall structures. The first use of this approach to attenuate seismic-induced dynamic forces against basement walls was reported by Inglis et al. (1996). Proof of concept has been demonstrated by Zarnani et al. (2005) who reported the results of a series of 1-m high shaking table tests with a 150-mm thick EPS seismic buffer (Figure 33a). The tests showed that the total earth force under simulated seismic load was reduced by 15% using a standard EPS material and 40% with a hollow core material (Figure 33b). Effects of the combined use of EPS and geogrid to reduce lateral earth pressures at-rest and under static active pressure conditions (i.e. without earth-

Seismic stability: High

Cement-mixed Pre-loaded & prebackfill stressed backfill
girder

Very high
Backfill soil cement-treated gravel Subsoil
Rachet system girder

Approach block by cementtreated gravel

(already in use in Japan)

Backfill soil gravel Subsoil

Figure 29. Example methods to improve the seismic stability of bridge abutments in Japan (Aoki et al. 2003).

R.L

140

Polymer geogrid Soil backfill

1:1

Sedimentary talus

0 10 20 30 40 50

1255

.5

3 4 8 14 16 26
50/30 50/23 50/16

Cement-treated gravel
100

Crystalline schist

Bedrock
150 (Unit in cm) (Unit in cm)

a) cross-section
New type abutment

b) photograph of abutment under construction Figure 30. GRS wall with geosynthetic reinforcement for bridge abutment supporting bullet train railway in Japan (Aoki et al. 2005).

quake loads) have been studied by Tsukamoto et al. (2002). 4 ANALYSIS METHODS AND NUMERICAL MODELLING

Analysis and design tools for GRS walls under dynamic loading can be classified as: a) pseudo-static methods, b) displacement methods, and c) finite element and finite difference methods. 11

a) with geosynthetic reinforcement

Figure 32. Example highway embankment project with combined cement-treated backfill and geosynthetic reinforcement (Ito et al. 2006).

a) shaking table test with geofoam EPS buffer

Horizontal wall force (kN)

b) without geosynthetic reinforcement Figure 31. Examples of shaking table models of GRS walls with cement-treated sand backfill (Saito et al. 2006).

25 20 15 10 5 0

Seismic buffer density = 16 kg/m3

Rigid wall (no buffer)

4.1 Pseudo-static methods Figure 34 shows one of the limit equilibrium-based pseudo-static methods based on a two-part wedge in the reinforced backfill (Ismeik and Gler 1998). In this method the critical wedge geometry is determined from a search of geometries giving the maximum horizontal load to be carried by the reinforcement layers (proportional to force coefficient K in Figure 35). The length of the reinforcement required to contain the critical failure wedge is also determined from the critical wedge geometry and can be presented in design chart form. Similar approaches using a single soil wedge and closed-form solutions have been reported by others (see Bathurst et al. 2002). An example is the use of additive static and dynamic pressure distributions by the NCMA to assign load to reinforcement layers based on a contributory area approach (Bathurst 1998). This method is illustrated in Figure 36, which shows that a larger dynamic portion of seismic load is distributed to the layers of reinforcement close to the top of

Cored buffer with bulk density = 1.32 kg/m3 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Acceleration (g)

b) horizontal wall force Figure 33. Shaking table tests demonstrating reduction in seismic forces against rigid walls using geofoam EPS seismic buffers (Zarnani et al. 2005).

the wall. This is a safe design practice for segmental walls, which typically have an unreinforced section at the top of the wall (Sections 2.5 and 3.2). The current AASHTO (2002) method in the USA restricts pseudo-static methods to peak horizontal ground accelerations < 0.3g. A fundamental shortcoming of pseudo-static methods is that they provide no information on wall displacements that may exceed a serviceability design criterion before an ultimate limit (collapse) state is developed.

12

Figure 34. The two-wedge failure mechanism: (a) cross section of a geosynthetic-reinforced soil wall showing the dimensions of Wedges 1 and 2; (b) forces acting on the facing and soil wedges (Ismeik and Gler 1998).

Figure 37. Example of double integration method to compute displacement of sliding mass (Cai and Bathurst 1996).

Figure 35. Results of parametric analysis illustrating relationship between force coefficient, wall geometry and soil shear strength for horizontal seismic coefficient value of kh = 0.2 (Ismeik and Gler 1998).

Figure 36. Calculation of total earth pressure distribution for GRS walls: (a) static pressure contribution; (b) dynamic pressure increment, and (c) total distribution (Bathurst 1998).

4.2 Displacement methods In order to estimate earthquake-induced displacement of GRS walls, variants of the classical Newmark sliding block method (Newmark 1965) have been proposed (e.g. Cai and Bathurst 1996, Ling et

al. 1997, Matsuo et al. 1998, Huang and Wang 2005, amongst others). These methods first require that a sliding mechanism be identified. Second the peak horizontal acceleration required to cause the mechanism to be triggered is computed (using a pseudostatic method of analysis). Finally, double integration of the horizontal ground acceleration record is carried out to compute displacements as illustrated in Figure 37. Typically the method is used to compute base sliding. However, Cai and Bathurst (1996) have used the method to compute displacements along internal sliding planes at elevations corresponding to the interface between modular block layers in segmental walls. Figure 38 shows results from 1g model shaking table tests on GRS model walls with a full-height rigid facing 50 cm high and resting on a shallow foundation layer with a thickness of 5 cm (the same as Figure 15d but with thinner foundation layer). The walls were subjected to several stepped stages of sinusoidal base excitation (Koseki et al. 2004). Permanent deformations (tilting and sliding) accumulated gradually with the increase in the peak amplitude of base acceleration. This gradual increase is difficult to reproduce numerically using a constant threshold acceleration value in Newmark-type analyses. The bottom figure in Figure 38 shows that the measured amount of base sliding increased significantly during the last excitation stage. This response stage may be simulated using the Newmark method, 13

1.0

Tilting angle (degree)

4 3 2 1 0 0 200 400 Measured Computed 600 800 Sinusoidal excitations 5cm-thick subsoil layers
Shear stress ratio, SR=/
0.5

Sinusoidal excitations 5cm-thick subsoil layers

0.0

-0.5

-1.0 0.00

0.01

0.02

0.03

0.04

0.05

Shear strain of subsoil,

Disp. of wall bottom (mm)

25 20

a) foundation soil
1.0

Shear stress ratio, SR=/

15 10 5 0 0 200 400

Measured
0.5

Computed

0.0

-0.5

600

800

-1.0 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

Base acceleration (gal)

Shear strain of reinforced backfill,

Figure 38. Comparison of computed and measured tilting and displacement of 1g GRS wall shaking table models (Koseki et al. 2004). Note 1000 gal = approximately 1g.

b) backfill soil Figure 39. Back-calculated shear-strain response of soils from 1g GRS wall shaking table models (Koseki et al. 2004).

since the excitation level corresponds to the threshold acceleration to initiate sliding failure using a pseudo-static limit equilibrium analysis together with the peak friction angle for the foundation and backfill soils. However, a sudden increase in base sliding may also be triggered by the formation of a failure plane in the unreinforced backfill immediately before the threshold acceleration is achieved, which was the case in this experiment. Similar behavior has been observed in model tests on conventional retaining wall models. Nevertheless, the magnitude of wall displacements when a distinct failure plane was formed in the backfill was much larger with the more ductile GRS wall models (Watanabe et al. 2003). In order to simulate the gradual accumulation of base sliding at relatively low earthquake loads, an attempt was made to estimate the cyclic stress-strain properties of the foundation soil supporting the reinforced backfill. Figure 39a shows the backcalculated response for the foundation soil assuming no slippage between the subsoil and the reinforced backfill. The measured response accelerations and the tensile forces in the extended reinforcement layers were used to compute the data curves in the figure (Koseki et al. 2004). A similar approach was used to generate the cyclic stress-strain response of the soil in the reinforced zone (Figure 39b). The good agreement between computed and measured

wall deformations up to the 750-gal (0.75g) excitation stage is shown in Figure 38. In Figure 40a, wall displacements are computed by considering the shear deformation of the foundation and reinforced backfill soil in combination with the Newmark method at the highest excitation level. Although there are quantitative discrepancies between computed and measured values, the agreement in qualitative trends is encouraging. Furthermore, the amount of movement during the 800-gal excitation stage is reasonably close. Similar comments apply for the nominally identical experiment carried out with an irregular excitation record (Figure 40b). Based on the results from these investigations, the Newmark sliding block method may be applied at large earthquake loads above the threshold acceleration level and after the formation of failure plane in the unreinforced backfill. However, before the formation of one or more failure planes, wall displacements due to pre-failure shear deformations of the foundation and reinforced backfill soil will control seismic response. Computed displacements generated prior to a collapse state (i.e. at low earthquake loading and a factor of safety greater than unity from a pseudo-static analysis) should not exceed a serviceability criterion.

14

athres=807gal -800

Damper

Base acceleration (gal)

-600 -400 -200 0 200 400 600 800 6.0 2 3 4 5 6 7

EP1 EP2

Disp. of wall bottom (mm)

5.5 5.0 4.5 4.0 3.5 3.0 2

Sunusoidal excitations 5 cm-thick subsoil layers

Computed
Earth pressure

Measured

a) shaking table model mounted on centrifuge

7
0.75 1.50 1.50 1.50 1.50 0.75 1.00 4.50 1.50 1.50 1.50 1.50 1.50 7.50 0.50 2.25 5.00

Time (sec)

a) sinusoidal base excitation

-800 -600 -400 -200 0 200 400 600 800 12 3 4
0.50 6.50

Embankment, 5 Embankment, 4 Embankment, 3 Embankment, 2 Embankment, 1 Foundation

15.00

Base acceleration (gal)

Irregular excitations 20 cm-thick subsoil layers

5 6 7 8

b) finite element model (dimensions in centimeters) Figure 41. Numerical modeling of centrifuge test of segmental GRS wall (Fujii et al. 2006).

Disp. of wall bottom (mm)

11 10 9 8 7 6 5 4 3 3 4 5 6 7 8 Computed

Measured

Time (sec)

b) irregular base excitation Figure 40. Back-calculated shear-strain response of soils from 1g GRS wall shaking table models (Koseki et al. 2004).

4.3 Finite element and finite difference modeling Properly conceived numerical models can be used to gather both qualitative insights and quantitative data for the performance of GRS walls under seismic loading and to guide the development of design methodologies (Cai and Bathurst 1995).

In this conference, Fujii et al. (2006) describe a numerical study aimed at simulating the results from a series of dynamic centrifuge tests on GRS segmental walls (Figure 41). They conducted 2-D finite element (FE) analyses using the program FLIP. The foundation soil and the backfill were modeled using a multi-spring approach (Iai et al. 1992). The reinforcement layers and facings were modeled by linear-beam elements, and the interfaces with the backfill were modeled by joint elements. Figure 42a shows a typical comparison between the measured and computed time histories of response acceleration in the unreinforced backfill at prototype scale. In total, thirteen test cases with different input wave forms and amplitudes were analyzed. Figure 42b summarizes the comparison between the measured and computed maximum response accelerations. In general, reasonable agreement was obtained up to an acceleration level of 600 gals (0.6g), with a tendency for the numerical model to slightly overestimate the physical test response. The largest deviations may be the result of 15

3.75

athres=807gal

6.75

600

Acceleration (gal) gal

400 200 0 -200 0 -400 -600 -800 (sec) Time (sec) 2 4 6 8 10

Centrifuge model tes t Seis mic res pons e analys is

Lateral displacement mm (mm)

800

30 20 10 0 -10 -20 -30 (sec) 0 2 4 6 8 10 12 14

12

14

Centrifuge model test Seismic response analysis

Time (sec)

a) example comparison of measured acceleration response in physical test and computed response

a) example comparison of measured lateral wall response in physical test and computed response

Maximum response acceleration of seismic response analysis (gal)

y = 1.1746x R = 0.8856
2

Maximum lateral displacement of seismic response analysis (mm)

1000 800 600 400 200 0 0

300 y = 0.9558x R = 0.9557 200

2

100

0
200 400 600 800 Maximum response acceleration of centrifuge model test (gal) 1000

100

200

300

b) peak acceleration Figure 42. Comparison of physical test results and computed acceleration response of centrifuge tests of segmental GRS walls (Fujii et al. 2006).

Maximum lateral displacement of centrifuge model test (mm)

b) lateral displacement Figure 43. Comparison of physical test results and computed lateral displacement response of centrifuge tests of segmental GRS walls (Fujii et al. 2006).

the formation of failure planes in the backfill, which was not properly modeled in the numerical code. Lateral displacements of the facing are compared between the measured and computed data in Figure 43. Accumulation of the lateral displacement was reasonably well simulated although a phase lag was sometimes observed between the measured and computed responses. Figure 44 shows measured and computed lateral earth pressures at the back of the facing. In general, the computed values largely over-estimated the measured values. This may be due to the challenge of modeling the facing interfaces. El-Emam et al. (2004) reported the results of numerical modeling of 1-m high shaking table tests that investigated full-height panel face GRS walls with different toe boundary conditions (i.e. hinged and free to rotate and slide). They used the finite difference-based program FLAC (Figure 45a). The numerical models were found to give reasonably accurate predictions of the experimental results (wall facing displacements, reinforcement loads and measured toe loads) despite the complexity of the

physical models under investigation. A constant reinforcement stiffness value was shown to be a reasonable assumption for numerical modeling of the polyester geogrid reinforcement used in the physical tests. However, reinforcement-soil slip for layers with shallow overburden depth was not considered in numerical simulations and this is thought to have led to some discrepancies in reinforcement load response close to the top of the wall. Both numerical and physical models demonstrated that the toe boundary condition has a large influence on wall performance and stability under both static and simulated seismic loading conditions. Finally, both numerical and experimental models showed that current pseudo-static seismic design methods may underestimate the size of the soil failure zone behind the wall facing, particularly at large input base acceleration amplitudes (Figure 45b,c) and these analytical methods are unable to explicitly account for the influence of the toe boundary reaction on wall response. 16

kN/m2 Horizontal earth pressure(kN/m )

25. 20. 15. 10. 5. 0. -5. 0 2 4 6 8 10 12 14 Centrifuge model tes t Seis mic res ponse analys is

stiff thin soil facing interface panel column layer 4 1.0 m

hinge

very stiff back 2.40 m reinforcement layers

3 2 1

(sec)

Time

a) example comparison of measured lateral earth pressure response in physical test and computed response

0.15 m 0.6 m thin horizontal soil layer fixed base boundary base acceleration

Maximum horizontal earth pressure 2 of seismic response analysis (kN/m )

30

very stiff foundation

a) FLAC numerical grid

shear strain < 0.2% facing panel failure surface (observed) numerically predicted shear zone reinforcement layers stiff foundation 1.0 m

20

y = 4.0315x 10 R = 0.7687
2

0 0 10 20 Maximum horizontal earth pressure of centrifuge model test (kN/m )

2

30

b) acceleration amplitude = 0.15g

shear strain < 1.0% facing panel

b) horizontal earth pressure Figure 44. Comparison of physical test results and computed horizontal earth pressure response of centrifuge tests of segmental GRS walls (Fujii et al. 2006).

failure surface (observed) numerically predicted shear zone reinforcement layers 1.0 m stiff foundation

Bathurst and Hatami (1998) reported the results of a numerical parametric study of an idealized 6-m high GRS wall with a full-height rigid facing and six layers of reinforcement. They showed that the magnitude and distribution of reinforcement loads was sensitive to the stiffness of the reinforcement materials used (Figure 46). The influence of reinforcement stiffness on seismic response of GRS walls is not considered in current design codes. Furthermore, they also showed that the relationship of the fundamental frequency of a GRS wall to the predominant frequency of the excitation record is a critical factor that may control wall response independent of the design input parameters currently considered in design codes based on pseudo-static methods (Figure 47). Numerical analyses showed no significant influence of the reinforcement stiffness, reinforcement length or toe restraint condition on the fundamental frequency of wall models (Hatami and Bathurst 2000). The review of the literature by the writers has highlighted the need for more sophisticated models

c) acceleration amplitude = 0.5g Figure 45. Predicted and observed failure surfaces from hinged toe model wall (5 reinforcement layers and L/H = 1.0) at different input base acceleration amplitudes. Note: Dark shading indicates large shear strains (El-Emam et al. 2004).

for both the soil and the reinforcement to better predict the seismic response of GRS structures. For example, hysteretic models based on Masing functions and their variants (e.g. Yogendrakumar and Bathurst 1992, Wakai et al. 2006) and strain-softening models for the backfill soil (e.g., Desai and El-Hoseiny 2005) are potentially fruitful avenues of future research.

17

Figure 46. Influence of reinforcement stiffness (J) on distribution of dynamic load increment in reinforcement layers (Bathurst and Hatami 1998). Figure 47. Influence of predominant frequency of base ground motion on wall displacements (Hatami and Bathurst 2000).

5 LIMIT STATES DESIGN FOR GRS WALLS SUBJECTED TO EARTHQUAKE Earlier in the paper we have described some general approaches for the analysis and design of GRS walls under earthquake loading. Pseudo-static methods based on a classical treatment of factor safety are the most common approach used to estimate destabilizing forces in seismic stability analyses. However, in order to bring the design of these structures in line with modern concepts for civil engineering structures, and in particular systems involving soilstructure interaction, there is a continuing movement towards a limit states design (LSD) approach. In this section we describe the history and current practice for design of GRS walls in seismic environments with special emphasis on LSD where applicable. Finally, we discuss a number of issues that should be considered as LSD codes for seismic design of GRS walls are developed in the future 5.1 Global trends for GRS wall seismic design codes Global trends toward limit states (or performancebased) geotechnical designs are summarized in Table 2. In 1994, the pre-standard version of Eurocode 7 (ENV 1997) on geotechnical design (Part 1: General Rules) was published, followed by Eurocode 8 (ENV 1998) on design of structures for earthquake resistance. They are based on the limit states design concept and intended to harmonize geotechnical design in Europe and also to harmonize geotechnical design with structural design practice (Orr 2002). An international standard (ISO 2394) identifying the general principles on reliability of structures was published in 1998. More recently, an ISO committee

Table 2. Limit states design codes for geotechnical structures. Reference Eurocode 7 (ENV 1997) Eurocode 8 (ENV 1998) Title Geotechnical design Design of structures for earthquake resistance

ISO 2394: 1998 General principles on reliability of structures ISO/CD 23469: Bases for design of structures Seismic ac2004 tions for designing geotechnical works JGS 001-2004: Principles for foundation designs grounded 2004 on a performance-based design concept AASHTO (2004)AASHTO LRFD Bridge Design Specifications (USA) CHBDC (2000) Canadian Highway Bridge Design Code GEOGUIDE 6 Guide to reinforced fill structure and slope (2002) design (Hong Kong) BS8006 (1995) Code of practice of strengthened/reinforced soils and other fills

draft (ISO/CD 23469 2004) outlining the basics for the design of geotechnical structures under seismic actions was completed. In Japan, the Japanese Geotechnical Society published its first guidance document titled Principles for foundation designs grounded on a performancebased design concept(JGS 4001 2004). The document is consistent with Principles, guidelines and terminologies for structural design code drafting founded on the performance-based design concept produced by the Japanese Society of Civil Engineering (JSCE 2003). Here we use the term performance-based design to identify codes that do not specify actual calculation methods but rather specify target performance values that must be met (i.e. allowable displacement criteria). In other words, they are used to ensure that the structure maintains an ac18

ceptable level of serviceability as well as safety during and after an earthquake. This is in contrast to specification-based design which typically means ensuring that the structure has acceptable factors of safety against prescribed failure (collapse) mechanisms based on mandated calculation methods presented in the design code. This approach is common in many of the design approaches used today for the design of GRS walls under both static and seismic load conditions. In the Canada and USA this approach is called working stress design and allowable stress design (ASD), respectively. In the discussion to follow, it is important to note that there is a fundamental difference between North American and European approaches to LSD. A factored strength approach is used in Europe, while a factored overall resistance approach is used in North America, including the National Building Code of Canada (NBCC) (NRC 1995) and the CHBDC (2000). An excellent discussion of the difference in approaches and the development of LSD methods in geotechnical engineering can be found in the paper by Becker (1996). At present there are only two Canadian guidance documents used for the design of GRS retaining walls. The Canadian Foundation Engineering Manual (CFEM 2006) gives a cursory description of the Tie-back Wedge (Simplified) method for geosynthetic reinforced soil walls but offers no guidance on the use of limit states design for static or seismic design. The Canadian Highway Bridge Design Code (CHBDC 2000) gives no guidance on GRS walls and recommends the use of analytical limit states design method that should be checked against accepted working stress design methods. However, the bridge code is currently undergoing revision to include GRS walls and guidance for LSD design for GRS walls under static and seismic load conditions. In the USA, the most current guidance documents for state agencies are the AASHTO Standard Specifications for Bridge Design (AASHTO 2002), AASHTO LRFD Bridge Design Specifications (AASHTO 2004) and the FHWA Mechanically Stabilized Earth Walls and Reinforced Soil Slopes Manual (FHWA 2001). The AASHTO (2002) document is in its last release as AASHTO moves from allowable stress design to a limit states design format (called load and resistance factor design LRFD in the USA). Nevertheless, AASHTO (2004) uses the Simplified Method for internal stability design of geosynthetic and metallic reinforced soil walls. The FHWA and USA state departments of transportation have set 2007 as the date after which LRFD is to be used for all new bridge designs. For earthquake design, a pseudo-static method is used. The FHWA (2001) document is largely consistent with the AASHTO (2002) document where they overlap for the design of reinforced soil walls. Both documents do not fully address design of modular

Table 3. Summary of current design codes for seismic design of GRS walls. Country Australia Japan Japan USA USA Reference Design by RTA (2005) Limit states design PWRC (2000) Specification-based RTRI (2006) Performance-based* FHWA (2001) Allowable stress design AASHTO Allowable stress design (2002) USA AASHTO AASHTO Allowable stress design (2004) USA NCMA Bathurst Allowable stress design (1998) Italy AGI** AGI (2005) Limit states design * with limit states design procedures, ** Italian Geotechnical Society Agency RTA PWRC RTRI FHWA AASHTO

block systems. However, a fourth related publication by the National Concrete Masonry Association (NCMA) (Bathurst 1998) is a comprehensive document for the seismic design and analysis of segmental retaining (modular block) walls. It is an allowable stress design method that calculates earth pressures from Monokobe-Okabe wedge theory, but with the orientation of any failure wedge restricted to the value for static loading only. Otherwise, the methodology recommends minimum factors of safety for internal, external and facing stability modes of failure consistent with the general approach for pseudostatic seismic analysis of other GRS types of retaining walls. In this guidance document, a Newmark sliding block analysis approach is recommended for peak ground accelerations greater than 0.3g using the method proposed by Cai and Bathurst (1996). A shortcoming of the current AASHTO (2004) LSD document is that load and resistance factors have been determined from calibration against current ASD methods with conventional factors of safety. However, this can be argued to defeat the purpose of converting to a limit states design approach. To overcome this deficiency, Allen et al. (2005) have written a Transportation Research Board guidance document that provides a detailed explanation of the methodologies to calibrate LRFD models for both metallic- and geosyntheticreinforced soil walls (i.e. to select load and resistance factors based on measured performance data). The general approach can be used for seismic design. Current seismic design codes for GRS walls are summarized in Table 3, which is updated from Zornberg and Leshchinsky (2003). One of the Australian codes (RTA 2005) adopts limit states design procedures but is written in a performance-based format. One Japanese code for railways (RTRI 2006) also adopts limit states design procedures within a performance-based framework. Another Japanese code for highways (PWRC 2000) is being updated to a limit states design format (performancebased design). 19

In the following sections, the Japanese RTRI code is briefly reviewed, together with its extension to probabilistic analyses. Next, European and British codes are described as they relate to LSD of GRS walls under seismic loading. Finally, some advantages of GRS walls over conventional retaining wall structures are identified. These performance differences should be considered when developing future LSD codes that include both classes of structures. 5.2 Design codes for railway structures in Japan 5.2.1 History The following is a brief summary of the history of limit states design code development for railway structures in Japan. a) Limit states design procedures were first introduced into design codes for concrete and steel/hybrid structures in 1992. b) LSD was introduced into design codes for foundations and retaining walls in 1997. c) A new seismic design code was published in 1999, which was written in a performance-based format for the first time. d) Design codes for concrete structures and embankments were revised into performance-based formats in 2004 and 2006, respectively. e) Other design codes for steel/hybrid structures and foundations are currently under revision to the same format. f) In order to enhance the implementation of performance-based design, a new code based on displacement limits was published in 2006. The seismic design code is also currently under revision. Both codes have been harmonized so that the same common principles are used. It should be noted that the design of retaining walls including GRS structures is described in the latest code version for embankments. The background and a brief summary of the seismic design procedures are explained in the next section. 5.2.2 Seismic design of GRS retaining walls for railway structures in Japan As illustrated in Figure 48, the design code for railway earth structures in Japan (RTRI 2006) was revised following a directive by the Ministry of Land, Infrastructure and Transport, Japan issued in 2001. This directive mandated that structures should be designed based on performance criteria. This design code is a national-level model or standard design code, accompanied by commentaries, both of which were edited by the Railway Technical Research Institute. Based on this code, individual design codes will be implemented by railway companies in Japan. Therefore, performance requirements are specified

Ministry order in 2001 (Ministry of Land, Infrastructure and Transport, Japan)

Model or standard design codes (edited by Railway Technical Research Institute)

Individual design code (by respective railway company)

Commentaries (ditto)

Figure 48. Evolution of Japanese railway design codes.

Table 4. Performance requirements for Japanese railway design codes. Action (design earthquake loads) Important structures Others Level 1 (highly expected for Tdes) Performance I: Will maintain their expected functions without repair works (no excessive displacements) Level 2 (maximum possible for Tdes) Performance II: Can restore their functions with quick repair works Performance III: Will not undergo overall instability

in this code, and analysis methods to compute performance targets are explained in the commentaries. Table 4 summarizes the required performance of railway earth structures against two levels of design earthquake loads. In this table Tdes is the design life of the structure. In general, the design life of a railway earth structure is assumed as 100 years. For a Level 1 earthquake load that is highly expected over the design life, it is required that all the earth structures will maintain their design functions without requiring repair work, i.e. will not exhibit excessive displacements (Performance I). Against a Level 2 earthquake load, which is defined as the maximum possible level over the structure design life, it is required that important earth structures can be restored to design function conditions with minimal repair (Performance II), while the other earth structures will not undergo overall instability (Performance III). In order to meet the above performance criteria, for GRS walls, the following analytical approaches are recommended: a) Performance requirements for Level 1 earthquake loads are verified through stability analyses using partial safety factors against internal instability of the reinforced backfill (using the model shown in Figure 49), external instability (using the model shown in Figure 50), and facing failure. b) Performance requirements for Level 2 earthquake loads are determined using Newmark sliding block analyses and other numerical analyses against base sliding displacement of the retaining 20

y Live load Dead load S F Block P bv P fv H P fh Pf M F O P Pb Embankment R Q

B Block

Figure 49. Internal stability of GRS walls (RTRI 2006).

rcurcular radius P 1 h reinforcement Md O slip surface TNj THj 1 Mr

a) FLAC model (Bathurst and Hatami 1998)

Figure 50. External stability of GRS walls (RTRI 2006). b) Reduced-scale shaking table test (Tatsuoka et al. 1998)
concrete facing concrete facing HF HW Vw ground surface of design RW RF VF PB HW ground surface Vw of design RW RF PB VF

Figure 52. Shear deformation of reinforced soil zone.

HF

Tgt

Tgt

(a) base sliding

(b) overturning
utop O khpL

u x dx ground surface of design

It should be also noted that, based on this design code, the tensile strength of geosynthetic reinforcement layers against earthquake load can be assigned without considering the possible effects of creep reduction. This is consistent with recent findings on the load-strain-time behavior of geosynthetic reinforcement products as reported by Greenwood et al. (2001) and Tatsuoka (2003) amongst others. 5.2.3 Probabilistic analyses In evaluating the seismic performance GRS structures, probabilistic approaches have several advantages over the deterministic approaches adopted in most existing design codes. By employing probabilistic approaches, the variability of soil and reinforcement parameters, in addition to different degrees of seismicity, can be considered rationally and quantitatively by using a reliability index, failure probability, or limit state exceedance probability. Shinoda et al. (2006a,b) conducted a series of probabilistic analyses (Monte Carlo simulations) on GRS slopes subjected to earthquake loads. Three different GRS slope configurations were considered (Table 5). The Case 1 configuration is illustrated in Figure 53a. One of the design earthquake motions specified in the RTRI (1999) seismic design code for railway structures was selected as the same input function for each analysis (Figure 53b), while the unit weight, soil strength, primary and secondary reinforcement strengths were taken as random variables described by a coefficient of variation. New21

(H-x) khtLdx

L x

(c) shear deformation of reinforced backfill Figure 51. Sliding analyses for GRS walls (RTRI 2006).

wall, overturning displacement, and shear deformation of the reinforced backfill (using the models shown in Figure 51). It should be noted that, although the reinforced backfill has been modeled as a rigid body in many existing design codes, results of numerical and physical models shows that it undergoes a simple shear-type deformation under large earthquake loads as illustrated in Figure 52. This deformation mode is evaluated in the above procedure (Figure 51c).

mark analyses assuming a circular failure plane were carried out and the computed deformations recorded for each simulation run. The allowable deformation was set at 50 cm. The computed exceedance probabilities for this limit state based on very many simulation runs are shown in Table 5. The combination of high strength primary reinforcement and high strength backfill soil (Case 1), gave a probability of exceeding the allowable deformation that was five orders of magnitude lower than the worst case investigated (Case 3). Shinoda et al. (2005) also computed the exceedance probabilities of GRS walls based on stability analyses without earthquake loads. Another series of probabilistic analyses were conducted by Miyata et al. (2001). Chalermyanont and Benson (2004, 2005) have developed reliability-based design charts for external and internal design of GRS walls under static load conditions using Monte Carlo simulations. In their analyses, only ultimate limit states defined by sliding, overturning and bearing capacity were considered for external modes of failure. The internal stability limit state was restricted to reinforcement pullout with the loading calculated using a circular slip analysis. Similar Monte Carlo simulations can also be carried out for GRS walls under seismic loadings and probabilities of exceedance computed for a range of failure mechanisms and prescribed deformation limit states. 5.3 Eurocode 8 Section 5 of Eurocode 8 (2003) offers general guidelines that are applicable to GRS walls. Some of the major points in this document are described below. Analyses may be calculated either by established methods of dynamic analysis, such as finite element models, rigid block models, or by simplified pseudostatic methods. The mechanical behavior of the soil media should include strain softening and the possible effects of pore pressure generation under cyclic loading if applicable. Simplified pseudo-static methods can be used where the surface topography and soil stratigraphy are reasonably regular. If a Mononobe-Okabe formulation is used, the point of application of the force due to the dynamic earth pressures shall be taken at the mid-height of the wall. Horizontal and vertical inertia forces should be applied to every portion of the soil mass and to any gravity loads acting at the backfill surface in pseudostatic methods of stability analysis. Vertical inertial forces shall be considered as acting upward or downward, which ever gives the most critical result. The total design force acting on the wall under seismic conditions shall be calculated at limit equilibrium of the model. For walls that are free to rotate about their toe, the dynamic force may be taken to act at the same point as the static force. The active and dynamic pressure distributions acting on the

Primary reinforcement Backfill soil Surface soil Surcharge Secondary of 10 kPa .5 1 reinforcement : 1 2m 0.3 m 1.5 m

Foundation soil
a) general arrangement for Case 1 reinforced soil slopes

Acceleration (gal)

1000 500 0 -500 -1000 5 10 15

Max. acceleraion = 924 gal

20 Time (sec)

25

30

35

b) ground motion used in Newmark analyses Figure 53. General arrangement for reinforced soil slopes used in Case 1 probabilistic analyses (Shinoda et al. 2006a,b).

Table 5. Case study parameter matrix (Shinoda et al. 2005). Backfill soil Primary reinforcement Secondary reinforcement Exceedance probability* Case 1 Case 2 Case 3 High strength High strength Low strength Yes Yes 2.7x10-6 No Yes 8.9x10-2 No Yes 4.1x10-1

wall are inclined at not greater than 2/3 with respect to the direction normal to the wall. The code requires that a limit state condition shall then be checked for the least safe potential slip surface. Serviceability limit state conditions described as a permanent displacement may be calculated using a simplified dynamic model consisting of a rigid sliding block. The calculations must be carried out using a time history representation based on the design acceleration without reductions. In the absence of a representative ground motion record, the code offers recommendations for the selection of horizontal and vertical seismic coefficient values. No reduction of the shear strength of cohesionless soils need be applied for strongly dilatant materials, such as dense sands.

22

5.4 British Standard Code of Practice 5.4.1 Design principles The British Standard Code of Practice (BS 8006 1995) is an important guidance document in Europe for the design of reinforced soil structures. The standard is written in a limit states format and guidelines are provided for selection of partial material factors and load factors for various applications and design life. One section of the standard is devoted to the design of reinforced soil walls. The Geoguide 6 (2002) code used in Hong Kong is very similar to BS 8006. However, both codes do not take into account earthquake effects on the stability of GRS structures. However, the general approach is easily adaptable to seismic design of reinforced soil walls. 5.4.2 Considered limit states and required performance In the BS 8006 standard, external and internal stability limit states are considered. For both ultimate and serviceability limit states, short- and long-term conditions must be considered in the design calculations and the effect of dead loads and other loads must be taken into account. The ultimate limit states for external stability modes of failure should be checked for bearing and tilt failure, forward sliding and slip circle failure. Potential slip surfaces passing through the structure and beyond the boundaries of the reinforced soil zone must be analysed. The possibility of large settlements in the foundation soil and/or in the reinforced soil fill should be checked as part of design calculations related to deformation serviceability limit states including differential settlements. For internal ultimate limit states, the following potential failure mechanisms should be considered in stability analysis: (i) stability of individual elements of the wall; (ii) resistance to sliding of upper portions of the structure; (iii) stability of wedges in the reinforcement fill. In the design checks, the following aspects should be taken into account: the capacity to transfer shear between the reinforcing elements; the tensile capacity of the reinforcements; and the capacity of the fill to support compression. Construction and post-construction behavior should also be considered. In the latter case, the following serviceability limits may be considered: foundation settlements; internal compression of the backfill; and internal strain of the geosynthetic reinforcement. 5.4.3 Recommendations for future seismic design development In order to extend the limits states design approach in BS 8006 to include seismic effects, the following issues should be addressed: (i) influence of inertia forces acting on the soil fill, on the surcharge applied at the fill surface, and on the wall facing mass

48 15
Figure 54. Bilinear failure surface observed during shaking table test on a GRS model wall with an inclined facing (after Lo Grasso et al. 2004).

Figure 55. Influence of seismic acceleration a and soil friction angle on the magnitude of earth pressure coefficient Ka for a GRS wall with an inclined facing i = 50 and constant static pore pressure ratio value ru = 0.25 (after Cascone et al. 1995).

(for structures with a concrete facing including modular blocks); and (ii) effect of soil cyclic behavior on active earth pressures and, consequently, the stability of the wall. The stability of a wall will be influenced by the facing angle (i.e. vertical or inclined) as demonstrated in Section 3.2. For walls with an inclined facing, experimental shaking table tests reported by Lo Grasso et al. (2004) developed a two-wedge (bilinear) failure mechanism (Figure 54). For analysis purposes, inertia forces may be considered explicitly in conventional pseudo-static wedge analyses. As an example, Figure 55 shows the effect of the seismic acceleration and soil friction angle on the magnitude of the earth pressure coefficient (Ka) for a GRS wall with an inclined facing and a constant static pore pressure ratio value (Cascone et al. 1995). No earthquake23

2,0 i =50 1,8 k H =0.25 r u=0.25 k V=0 u *=1 k a*/ k a 1,6 u *=0.6 1,4 u *=0.4 1,2 u *=0.2 1,0 20 25 30 ' 35 u *=0 40 45 u *=0.8

Figure 56. Influence of soil friction angle and static pore pressure ratio value ru on the magnitude of earth pressure coefficient Ka for a GRS wall with a vertical facing and given values of kh, kv, nq and (after Motta 1996).

Figure 57. Influence of the earthquake-induced pore pressure ratio u* on the ratio between the earth pressure coefficient Ka* evaluated taking into account the soil shear strength reduction and unmodified Ka for GRS wall with constant inclined facing angle, i = 50, static pore pressure ratio ru and horizontal kH and vertical kV seismic coefficient values (after Fazzino 2005).

induced pore pressures or vertical component of seismic acceleration were considered in these analyses. The influence of vertical component of seismic acceleration may have a large effect on GRS wall forces when pseudo-static analyses are used (e.g. Bathurst and Cai 1995). Vertical accelerations must be in analyses using Eurocode 8, as noted earlier, and in recent guidelines of the Italian Geotechnical Society (AGI 2005) focused on design procedures for GRS walls in seismic areas. For GRS walls with a vertical facing, the failure mechanisms will depend on different parameters such as the presence of the surcharge. In this case the influence of inertia forces on wall stability may be evaluated using the solutions proposed by Motta (1996). These solutions consider the influence of the surcharge magnitude q and the distance d between the surcharge and the wall facing using the nondimensional parameters nq = q/(H) and = d/H respectively, where is soil unit weight and H is wall height. As an example, Figure 56 shows the effect of soil friction angle and static pore pressure ratio on the values of the earth pressure coefficient for a horizontal seismic coefficient kh = 0.20 and no vertical acceleration (kv = 0). The influence of soil shear strength reduction due to earthquake-induced pore pressure may be evaluated using a pore-pressure ratio approach for walls with vertical or inclined facing geometry (Fazzino 2005). The analysis should be performed assuming that earthquake-induced pore pressure and wall stability are decoupled. The earthquake-induced pore pressure ratio may be evaluated using available

empirical relationships based on experimental results. The effect of soil shear strength reduction on the stability of the wall may be evaluated using the earthquake-induced pore pressure ratio u* in stability calculations. As an example, Figure 57 shows the effect of the earthquake-induced pore pressure ratio value u* on the magnitude of the earth pressure coefficient Ka from a two-part wedge analysis and a wall with an inclined face. In this plot, Ka* is the adjusted value of earth pressure coefficient and Ka is the unmodified value. The values of Ka* for other values of soil friction angle may be evaluated by combining values from curves in Figures 56 and 57. Finally as noted in Section 4.3, an estimate of the fundamental frequency of the structure should be made and this value compared to the predominant frequency of the design earthquake record (Bathurst et al. 2002). The fundamental frequency of a structure can be calculated using 1D linear elastic theory (Hatami and Bathurst 2000). Small changes in wall height may be all that is required to shift the fundamental frequency of the structure away from the predominant frequency of the design earthquake. 5.5 Advantages of GRS walls over conventional walls In Section 2, case histories showing good performance of GRS walls compared to conventional walls are reported. GRS walls develop different response mechanisms while resisting large earthquake loads. For example, the more ductile response of GRS 24

(a) view of reinforced-soil wall and bridge pier with some horizontal undulations due to differential settlement of foundation soil (photograph courtesy of H. Aydin)

Figure 59. Collapse of reinforced concrete retaining wall during 1999 Chi-chi earthquake and undulation of road surface along local road in Taiwan (connected to the south of ChangGeng Bridge) (Koseki and Hayano 2000).

that shaking table test models of these structures remained upright even when the bearing capacity of the foundation soil below the footing was exceeded. Displacement-based analyses will become more important as engineers focus on performance-based (serviceability-based) design. It is the opinion of the writers that reinforced soil walls can be expected to out-perform conventional unreinforced soil retaining wall structures with respect to displacement performance. Nevertheless, direct comparison of the displacement-time response of these two different classes of structures under nominally identical conditions using the same computational methods remains to be done. Examples of computational tools to carry out these calculations are given in Section 4. This is an area of future research investigation. In addition, reinforced-soil walls are generally more flexible than conventional walls. Thus, they may be used in areas where large uneven displacements due to surface faulting during earthquake events are expected. Figure 58a shows an approach embankment of the Arifiye overpass bridge in Turkey following the 1999 Kocaeli earthquake. It was constructed as a reinforced-soil wall using metal reinforcement strips. Although the overpass bridge collapsed completely (Figure 58b), the reinforced-soil wall survived the earthquake intact and remained in service. The reinforced-soil wall sustained large permanent deformations mainly due to large differential settlements at the foundation. The good performance of this wall contrasts with the behavior of a conventional reinforced concrete wall in Taiwan (Figure 59), which collapsed completely due to large uneven displacements (surface faulting) during the 1999 Chi-chi earthquake (Koseki and Hayano 2000). 25

(b) view of collapsed bridge superstructure and undulation of road surface (photograph courtesy of Bogazici University, Kandilli Observatory and Earthquake Research Institute) Figure 58. Damage to Arifiye overpass bridge in Turkey during 1999 Kocaeli Earthquake.

walls, was noted in model tests subjected to simulated seismic loading (Section 3). In most current design codes, factors of safety against prescribed failure (collapse) mechanisms are evaluated using pseudo-static methods for both conventional and GRS walls. Within this common framework the following benefits of using GRS walls may be realized: a) Design seismic coefficients can be reduced when designing for GRS walls. b) Minimum values for acceptable factors of safety can be varied depending on the type of wall (e.g. GRS walls or conventional walls) c) Failure mechanisms that do not occur for GRS structures can be omitted. For example, starting with the design code for railway GRS walls with rigid facings in Japan (RTRI 1992), bearing capacity failure is not considered (except for extremely weak subsoil conditions). This is based on the work of Murata et al. (1994) who showed

CONCLUSIONS

This paper has reviewed a large body of work focused on the seismic performance, analysis and design of geosynthetic-reinforced soil (GRS) walls. The results of field walls that have survived significant earthquakes and the results of model tests show that GRS walls perform well during a seismic event when properly designed. This has led to many structures remaining in service after major earthquakes while conventional structures have not. Because of their good performance, GRS structures are being considered more frequently for the replacement of conventional structures in reconstruction works. New retaining wall technologies that include geosynthetic products are identified in the paper e.g. combined soil-cement and geosynthetic reinforcement layers, and geofoam seismic buffers with and without geosynthetic reinforcement layers. Finally, the paper has highlighted the move toward limit states design for geotechnical engineering structures. Nevertheless, with the exception of Japan, the inclusion of seismic design guidelines for GRS structures within a LSD framework is lagging. ACKNOWLEDGEMENTS The authors wish to extend special thanks to Prof. C.C. Huang (National Cheng Kung University, Taiwan), Mr. H.E. Acosta-Martinez (University of Western Australia), Ms. Y. Tsutsumi (University of Tokyo, Japan), Dr. J. Izawa (Tokyo Institute of Technology, Japan), Dr. M. Tateyama, Dr. K. Kojima, Mr. K. Watanabe (Railway Technical Research Institute, Japan), Mr. Y. Tamura, Dr. M. Shinoda (Integrated Geotechnology Institute, Ltd, Japan), Dr. T. Fujii (Fukken Co., Ltd, Japan), Prof. M. El-Emam (Zagazig University, Egypt), Prof. K. Hatami (University of Oklahoma, USA) and, Mr. S. Zarnani (GeoEngineering Centre at Queens-RMC, Canada) for their help in preparing the manuscript. REFERENCES
AASHTO. 2002. Standard specifications for highway bridges, (17th Edition 2002), American Association of State Highway and Transportation Officials, Washington, DC, USA AASHTO. 2004. LRFD bridge design specifications, SI units (3rd Edition), American Association of State Highway and Transportation Officials, Washington, DC, USA AGI. 2005. Geotechnical aspects of the design in seismic area. Guidelines. Draft, March 2005. Patron Editore Bologna (in Italian). Allen, T.M., Nowak, A.S. and Bathurst, R.J. 2005. Calibration to determine load and resistance factors for geotechnical and structural design, Transportation Research Board Circular E-C079, Washington, DC, 93 p.

Aoki, H., Watanabe, K., Tateyama, M. and Yonezawa,T. 2003. Shaking table tests on earthquake resistant bridge abutment, Proc. of 12th Asian Regional Conf. on Soil Mechanics and Geotechnical Engineering, 1, 267-270. Aoki, H., Yonezawa,T., Tateyama, M. Shinoda, M. and Watanabe, K. 2005. Development of aseismic abutment with geogrid-reinforced cement-treated backfills, Proc. of 16th International Conf. on Soil Mechanics and Geotechnical Engineering, 3, 1315-1318. Bathurst, R.J. and Cai, Z. 1995. Pseudo-static seismic analysis of geosynthetic reinforced segmental retaining walls, Geosynthetics International, 2(5), 789-832. Bathurst, R.J. and Alfaro, M.C. 1997. Review of seismic design, analysis and performance of geosynthetic reinforced walls, slopes and embankments, Earth Reinforcement, Balkema, 2, 887-918. Bathurst, R.J. 1998. NCMA segmental retaining wall seismic design procedure supplement to design manual for segmental retaining walls (second edition 1997) published by the National Concrete Masonry Association, Herndon, VA, 187 p. Bathurst, R.J. and Hatami, K. 1998. Seismic response analysis of a geosynthetic reinforced soil retaining wall, Geosynthetics International, 5(1&2), 127-166. Bathurst, R.J., Hatami, K. and Alfaro, M.C. 2002. Geosynthetic reinforced soil walls and slopes: seismic aspects, Geosynthetics and Their Applications (S.K. Shukla Ed.), Thomas Telford, 327-392. Becker, D.E. 1996. Limit states design for foundations. Part I. An overview of the foundation design process, Canadian Geotechnical Journal, Vol.33, 956-983. British Standard BS8006:1995. Code of practice of strengthened/reinforced soils and other fills. 162 p. Cai, Z. and Bathurst, R.J. 1995. Seismic response analysis of geosynthetic reinforced soil segmental retaining walls by finite element method, Computers and Geotechnics, 17(4), 523-546 Cai, Z. and Bathurst, R.J. 1996. Seismic-induced permanent displacement of geosynthetic reinforced segmental retaining walls, Canadian Geotechnical J., 33, 937-955. Canadian Foundation Engineering Manual (CFEM). 2006. Canadian Geotechnical Society (in press) Canadian Highway Bridge Design Code (CHBDC) CAN/CSAS6-00. 2000. The Canadian Standards Association (CSA International), Toronto, Ontario Cascone, E., Maugeri, M. and Motta, E. 1995. Seismic design of earth-reinforced embankments, Earthquake Geotechnical Engineering, Balkema, 2, 1129-1134. Chalermyanont, T. and Benson, C.H. 2004. Reliability-based design for internal stability of mechanically stabilized earth walls, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 130(2), 163-173. Chalermyanont, T. and Benson, C.H. 2005. Reliability-based design for external stability of mechanically stabilized earth walls, International Journal of Geomechanics, 5(3), 196205. Desai C.S. and El-Hoseiny, K.E. 2005. Prediction of field behavior of reinforced soil wall using advanced constitutive model, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 131(6), 729-739. El-Emam, M. and Bathurst, R.J. 2004. Experimental design, instrumentation and interpretation of reinforced soil wall response using a shaking table, International Journal of Physical Modelling in Geotechnics, 4(4), 13-32. El-Emam, M., Bathurst, R.J. and Hatami, K. 2004, Numerical modeling of reinforced soil retaining walls subjected to base acceleration, 13th World Conference on Earthquake Engineering, Vancouver BC, 15.

26

El-Emam, M. and Bathurst, R.J. 2005. Facing contribution to seismic response of reduced-scale reinforced soil walls, Geosynthetics International, 12(5), 215- 238. Eurocode 7 (ENV 1997-1). 1994. Geotechnical design, Part 1: General rules, CEN. Eurocode 8 (ENV 1998) Design of structures for earthquake resistance, CEN. Eurocode 8. 2003. Design of structures for earthquake resistance, Part 5: Foundations, retaining structures and geotechnical Aspects. Fazzino, P. 2005. Influence of earthquake-induced pore pressure on the stability condition of earth-.reinforced walls. Diploma Thesis, University of Catania (in Italian). FHWA. 2001. Mechanically stabilized earth walls and reinforced soil slopes: design and construction guidelines, Publication FHWA NHI-00-43, Federal Highway Administration and National Highway Institute, Washington, DC, USA Fujii, T., Izawa, J., Kuwano, J., Ishihara, M. and Nakane, J. 2006. Prediction of deformation of retaining walls of geosynthetic-reinforced soil under large earthquakes, this conference. Geoguide 6. 2002. Guide to reinforced fill structure and slope design. Geotechnical Engineering Office, Civil Engineering Department, Hong Kong Greenwood, J.H., Jones, C.J.F.P. and Tatsuoka, F. 2001. Residual strength and its application to design of reinforced soil in seismic areas, Landmarks in Earth Reinforcement, Balkema, 1, 3742. Hatami, K. and Bathurst, R.J. 2000. Effect of structural design on fundamental frequency of reinforced-soil retaining walls, Soil Dynamics and Earthquake Engineering, 19, 137-157. Huang, C.C. 2000. Investigations of the damaged soil retaining structures during the Chi-Chi earthquake, Journal of the Chinese Institute of Engineers, 23(4), 417-428. Huang, C.C. and Wang, W.C. 2005. Seismic displacement of a geosynthetic-reinforced wall in the 1995 Hyogo-ken Nambu earthquake, Soils and Foundations, 45(5), 1-10. Iai, S., Matsunaga, Y., and Kameoka, T. 1992. Strain space plasticity model for cyclic mobility, Soils and Foundations, 32(2), 1-15. Inglis, D., Macleod, G., Naesgaard, E. and Zergoun, M. 1996. Basement wall with seismic earth pressures and novel expanded polystyrene foam buffer layer, Tenth Annual Symposium of the Vancouver Geotechnical Society, Vancouver, BC, Canada, 18 p. Ismeik, M. and Gler, E. 1998. Effect of wall facing on the seismic stability of geosynthetic-reinforced retaining walls, Geosynthetics International, 5(1&2), 41-53. ISO 2394:1998. General principles on reliability of structures ISO/CD 23469:2004. Bases for design of structures Seismic actions for designing geotechnical works Ito, H., Saito, T., Izawa, J. and Kuwano, J. 2006. In situ construction of the Geogrid reinforced soil wall combined with soil cement, this conference. Izawa, J., Kuwano, J. and Takahashi A. 2002. Centrifuge tilting and shaking table tests on reinforced soil wall, Proc. of 7th International Conference on Geosynthetics, Nice, 1, 229232. Izawa, J., Kuwano, J. and Ishihara Y. 2004. Centrifuge tilting and shaking table tests on the RSW with different soils, Proc. of 3rd Asian Regional Conference on Geosynthetics, Seoul, 803-810. JGS. 2004. Principles for foundation designs grounded on a performance-based design concept, JGS 4001-2004, Japanese Geotechnical Society. JSCE. 2003. Principles, guidelines and terminologies for structural design code drafting founded on the performance based design concept (code PLATFORM), Japan Society for Civil Engineers (in Japanese).

JSCE. 2006. Report on the 2004 Niigata Chuetsu earthquake, Japan Society of Civil Engineers, CD-ROM (in Japanese). Kato, N., Huang, C.C., Tateyama, M., Watanabe, K., Koseki, J. and Tatsuoka, F. 2002. Seismic stability of several types of retaining walls on sand slope, Proc. of 7th International Conference on Geosynthetics, Nice, 1, 237-240. Kitamoto, Y. Abe, H., Shimomura, H., Morishima, H. and Taniguchi, Y. 2006. Rapid and strengthened repair construction for a seriously damaged railway embankment during violent earthquakes, this conference. Koerner, R., Soong, T.Y. and Koerner, R.M. 1998. Earth retaining wall costs in USA, Geosynthetic Research Institute Report, Folsom, Pennsylvania, USA. Koseki, J., Tatsuoka, F., Munaf, Y., Tateyama, M. and Kojima, K. 1998. A modified procedure to evaluate active earth pressure at high seismic loads, Soils and Foundations, Special Issue on Geotechnical Aspects of the January 17 1995 Hyogoken-Nambu Earthquake, 2, 209-216. Koseki, J. and Hayano, K. 2000. Preliminary report on damage to retaining walls caused by the 1999 Chi-Chi earthquake, Bulletin of ERS, Institute of Industrial Science, Univ. of Tokyo, 33, 23-34. Koseki, J., Tatsuoka, F., Watanabe, K., Tateyama, M., Kojima, K. and Munaf, Y. 2003. Model tests on seismic stability of several types of soil retaining walls, Reinforced Soil Engineering, Ling, Leshchinsky and Tatsuoka (eds.), Dekker, 317-358. Koseki, J., Kato, N., Watanabe, K. and Tateyama, M. 2004. Evaluation of seismic displacement of reinforced walls, Proc. of 3rd Asian Regional Conference on Geosynthetics, Seoul, 217-224. Ling, H.I., Leshchinsky, D and Perry, E.B. 1997. Seismic design and performance of geosynthetic-reinforced soil structures, Geotechnique, 47(5), 297-316. Ling, H.I. and Leshchinsky, D. 2003. Post-earthquake investigation of several geosynthetic reinforced soil retaining walls and slopes during Ji-Ji earthquake in Taiwan, Reinforced Soil Engineering, Ling, Leshchinsky and Tatsuoka (eds.), Dekker, 297-316. Ling, H.I., Mohri, Y., Leshchinsky, D., Burke, C., Matsushima, K. and Liu, H. 2005. Large-scale shaking table tests on modular-block reinforced soil retaining walls, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 131(4), 465-476. Lo Grasso, A.S., Maugeri, M. and Recalcati, P. 2004. Response of geosynthetic reinforced soil wall under seismic condition by shaking table tests, 3rd European Geosynthetics Conference, Munich, Germany. Lo Grasso, A.S. Maugeri, M. and Recalcati, P. 2005. Seismic behaviour of geosyntethic-reinforced slopes with overload by shaking table tests, Geo-Frontiers 2005, ASCE, CDROM. Lo Grasso, A.S. Maugeri, M. and Recalcati, P. 2006. Experimental seismic analysis of geosyntethic-reinforced soil structures with three-dimensional reinforcements by shaking table tests, this conference. Matsuo, O. Tsutsumi, T., Yokoyama, K. and Saito, Y. 1998. Shaking table tests and analyses of geosythetic-reinforced soil retaining walls, Geosynthetics International, 5(1&2), 97-126. Miyata, Y., Shigehisa, S. and Kogure, K. 2001. Reliability analysis of geosynthetics reinforced soil wall, Landmarks in Earth Reinforcement, Balkema, 1, 417420. Motta, E. 1996. Earth pressure on reinforced earth-walls under general loading, Soils and Foundations, 36(4), 113-117. Murata, O. 2003. An outlook on recent research and development concerning long-term performance and extreme loading, Landmarks in Earth Reinforcement, Balkema, 2, 869893.

27

Murata, O., Tateyama, M. and Tatsuoka, F. 1994. Shaking table tests on a large geosynthetic-reinforced soil retaining wall model, Recent Case Histories of Permanent Geosynthetic-Reinforced Soil Retaining Walls, Balkema, 869-893. NCMA. 1997. Design Manual for Segmental Retaining Walls, Second Edition, Second Printing, National Concrete Masonry Association, Herndon, Virginia. Newmark, N.M. 1965. Effects of earthquakes on dams and embankments, Geotechnique, 15(2), 139-159. NRC. 1995. National building code. NRC of Canada, Ottawa, Ontario, Canada. Orr, T.L.L. 2002. Eurocode 7 - a code for harmonized geotechnical design, Foundation Design Codes and Soil Investigation in view of International Harmonization and Performance Based Design, Honjo et al. (eds), Balkema, 3-15. PWRC. 2000. Design and construction manual for geotextile reinforced soil structures, Publics Works Research Center, Japan (in Japanese). Race, R. and del Cid, H. 2001. Seismic performance of modular block retaining wall structures during the January 2001 El Salvador earthquake, International Geosynthetics Engineering Forum 2001, Taipei, Taiwan (CD-ROM), 20 p. RTA. 2005. Design of reinforced soil walls, QA specification R57, Roads and Traffic Authority, New South Wales, Australia. RTRI. 1992. Design standard for railway earth structures, Railway Technical Research Institute, Japan (in Japanese). RTRI. 1999. Seismic design standard for railway structures, Railway Technical Research Institute, Japan (in Japanese). RTRI. 2006. Design standard for railway earth structures, Railway Technical Research Institute, Japan (in Japanese). Saito, T., Ito, H., Izawa, J. and Kuwano, J. 2006. Seismic stability of the geogrid-reinforced soil wall combined with soil cement, this conference. Sandri, D. 1994. Retaining walls stand up to the Northridge earthquake. Geotechnical Fabrics Report, IFAI, St. Paul, MN, USA, 12(4), 30-31 (and personal communication). Shinoda, M., Yonezawa, T., Tateyama, M. and Koseki, J. 2005: Limit state exceedance probability of reinforced soil retaining walls, Journal of the Japan Society of Civil Engineers, No. 792/III-71, 119129 (in Japanese). Shinoda, M., Horii, K., Yonezawa, T., Tateyama, M. and Koseki, J. 2006a. Deterministic and probabilistic Newmark analyses on geosynthetic-reinforced soil slopes, this conference. Shinoda, M., Horii, K., Yonezawa, T., Tateyama, M. and Koseki, J. 2006b. Reliability-based seismic deformation analysis of reinforced soil slopes, Soils and Foundations (in press). Stewart, J.P., Bray, J.D., Seed, R.B. and Sitar, N. 1994. Preliminary report on the principal geotechnical aspects of the January 17, 1994 Northridge earthquake, UCB/EERC-94/08 Earthquake Engineering Research Center, University of California, Berkeley 1994-06, 245 p. Tamura, Y. 2006. Lessons learnt from the construction of geosynthetic-reinforced soil retaining walls with full-height rigid facing for the last 10 years, this conference. Tatsuoka, F., Tateyama, M., Koseki, J. and Uchimura, T. 1995. Geotextile-reinforced soil retaining wall and their seismic bahaviour, Proc. of 10th Asian Regional Conf. on Soil Mechanics and Foundation Engineering, Beijing, 2, 26-49. Tatsuoka, F., Tateyama M. and Koseki, J. 1996. Performance of soil retaining walls for railway embankments, Soils and Foundations, Special Issue of Soils and Foundations on Geotechnical Aspects of the January 17 1995 HyogokenNambu Earthquake, 311-324. Tatsuoka, F., Tateyama, M, Uchimura, T. and Koseki, J. 1997a. Geosynthetic-reinforced soil retaining walls as important permanent structures, the 1996-1997 Mercer Lecture, Geosynthetics International, 4(2), 81-136.

Tatsuoka, F., Koseki, J. and Tateyama, M. 1997b. Performance of reinforced soil structures during the 1995 Hyogo-ken Nanbu Earthquake, Earth Reinforcement, Balkema, 2, 9731008. Tatsuoka, F., Koseki, J., Tateyama, M., Munaf, Y. and Horii, N. 1998., Seismic stability against high seismic loads of geosynthetic-reinforced soil retaining structures, Proc. 6th Int. Conf. on Geosynthetics, Atlanta, 1, 103-142. Tatsuoka, F. 2003. Technical report -wall structures session, Landmarks in Earth Reinforcement, Balkema, Vol.2, 937948. Tsukamoto, Y., Ishihara, K., Kon, H. and Masuo, T. 2002. Use of compressible expanded polystyrene blocks and geogrids for retaining wall structures, Soils and Foundations, 42(4), 29-41. Wakai, A., Amano, M., Iizuka, Y. and Ugai, K. 2006. Numerical analysis for seismic evaluation of retaining walls consisting of concrete panels and backfill reinforced with geogrids, Journal of Geotechnical Engineering, Japan Society of Civil Engineers, 816/III-74, 157-168 (in Japanese). Watanabe, K., Munaf, Y., Koseki, J., Tateyama, M. and Kojima, K. 2003. Behaviors of several types of model retaining walls subjected to irregular excitation, Soils and Foundations, 43(5), 13-27. White, D.M. and Holtz, R.D. 1997. Performance of geosynthetic-reinforced slopes and walls during the Northridge, California Earthquake of January 17, 1994, Earth Reinforcement, Balkema, 2, 965-972. Yogendrakumar, M. and Bathurst, R.J. 1992. Numerical simulation of reinforced soil structures during blast loadings, Transportation Research Record, 1336, 1-8. Zarnani, S., Bathurst, R.J. and Gaskin, A. 2005. Experimental investigation of geofoam seismic buffers using a shaking table, 2005 North American Geosynthetics Society Conference, Las Vegas, Nevada, 14-16 December 2005, 11 p. Zornberg, J.G. and Leshchinsky, D. 2003. Comparison of international design criteria for geosynthetic-reinforced soil structures, Landmarks in Earth Reinforcement, Balkema, 2, 1095-1106.

28