Seismic stability of reinforced soil walls
Institute of Industrial Science, University of Tokyo, Japan
GeoEngineering Centre at Queen's-RMC, Canada
Bogazici University, Turkey
Saitama University, Japan
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 re-
views 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)
The good performance of geosynthetic-reinforced 60
soil walls (GRS walls) during earthquake has been 50
documented widely in the literature. In Japan, the Total
greater seismic resistance of GRS walls compared to 40
conventional retaining wall structures has led to their 30
increasing use for new permanent structures (Tatu-
soka et al. 1997a) and to replace conventional struc- 20
tures damaged in recent earthquakes. 10
For example, Figure 1 shows the annual and cu-
mulative face length of GRS walls with a rigid fac- 0 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04
ing constructed in Japan since 1989 (Tamura 2006). Fiscal year
Between 1995 and 2001, the construction rate in- Figure 1. Growth statistics for GRS walls with full-height rigid
creased from less than 5 km to more than 10 km per facing in Japan (Tamura 2006).
year. This is due to the good seismic performance of
GRS walls observed during the 1995 Hyogoken-
Nanbu (Kobe) earthquake. The more recent lower Table 1. Keynote lectures and special reports in previous
conferences on seismic issues for GRS walls.
annual construction rates shown in the figure are due Lecture / report Conference
to a weaker Japanese economy. Nevertheless, over a Tatsuoka et al. (1995) 10th Asian Regional
15-year period the total construction length of this Conference on SMFE
type of reinforced soil wall is greater than 70 km. Bathurst and Alfaro (1997)* IS-Kyushu ’96
GRS walls are becoming common retaining wall White and Holtz (1997) IS-Kyushu ’96
structures in other countries as well, primarily be- Tatsuoka et al. (1997) IS-Kyushu ’96
Tatsuoka et al. (1998) 6ICG, Atlanta
cause of cost savings. A study by Koerner et al. Murata (2003) IS-Kyushu 2001
(1998) showed that GRS walls built for government * extended and updated by Bathurst et al. (2002)
projects were typically 50% cheaper than conven-
tional retaining wall structures. Kobe case histories have been reported in detail by
Seismic issues related to GRS walls have been Tatsuoka et al. (1995, 1997b). Case history exam-
addressed in many keynote lectures and reports at ples from the 1994 Northridge earthquake, USA,
previous geosynthetics conferences (Table 1). The 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 L = 1.8 m
and performance of GRS walls. Tatsuoka et al.
(1998) discussed the stability of GRS walls against 12 Layer
high seismic loads. Murata (2003) reported further 11 number
progress on research and development in Japan re- 10
lated to GRS walls. Nevertheless, it is timely to up- 9
H = 6.5 m
date our current knowledge of the seismic perform- 8 8
ance, analysis and design of GRS wall structures, 7
and from this review to identify deficiencies in cur-
rent design practice and areas of future research. In 6
this paper we address several general questions: 5 6
1. How have GRS walls performed during recent 3 2
large earthquakes? 1
2. What lessons can be learned from field perform-
ance of GRS walls during earthquake? L=5.5 m
3. How have GRS walls been used to replace dam- a) cross-section
aged conventional retaining wall structures due
4. What lessons can be learned from physical and
numerical modeling of GRS walls under simu-
lated seismic loading?
5. What is the current and future practice for seis-
mic design of GRS walls including progress to-
wards 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 his-
tories 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 de-
sign for GRS walls for earthquake with special em- b) photograph taken after 1994 Northridge earthquake show-
phasis on developments toward limit states design- ing cracking at the back of shortened reinforcement lengths
based approaches. 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).
2 CASE HISTORIES
In order to answer the first three questions intro- 2.1 1994 Northridge earthquake
duced in the previous section, we review case histo- Sandri (1994) conducted a survey of reinforced
ries from six major earthquakes: segmental (modular block) retaining walls greater
than 4.5 m in height in the Los Angeles area imme-
A) 1994 Northridge Earthquake (USA) diately after the Northridge earthquake of 17 January
B) 1995 Hyogoken-Nanbu (Kobe) Earthquake 1994 (Moment Magnitude = 6.7). The results of the
(Japan) survey showed no evidence of visual damage to 9 of
C) 1999 Chi-chi Earthquake (Taiwan) 11 structures located within 23 to 113 km of the
D) 1999 Kocaeli Earthquake (Turkey) earthquake epicenter. Two structures (Valencia and
E) 2001 El Salvador Earthquake (Central America) Gould walls) showed tension cracks within and be-
F) 2004 Niigataken-Chuetsu Earthquake (Japan) hind the reinforced soil mass that were clearly at-
tributable to the results of seismic loading. Figure 2a
shows the Valencia wall with shortened top rein-
forcement layers to facilitate placement of subsur-
and unreinforced segmental walls were observed to
have developed cracks in the backfill during a sur-
vey 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
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 full-
height 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
a) cross-section b) photograph taken after staged construction of a GRS wall with a full-height
1995 Kobe earthquake
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 de-
formations 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 con-
structed 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 re-
moved 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 dif-
ferences between conventional and GRS walls are
also described in Section 3.1 using the results from
physical model tests.
c) construction sequence for GRS wall with rigid facing 2.3 1999 Chi-chi earthquake
Figure 3. Tanata GRS wall (Tatsuoka et al. 1996). Several GRS walls were damaged during the 1999
Chi-chi earthquake in Taiwan (Huang 2000, Koseki
face utilities. Despite the shortening of these layers and Hayano 2000, Ling and Leshchinsky 2003).
by the contractor, there was no visual evidence of An example segmental-type wall using masonry
facing movement even though peak horizontal concrete blocks and a poor quality backfill is illus-
ground accelerations as great as 0.5g were estimated trated in Figure 6. The reinforcement spacing was
at the Valencia wall site (Figure 2b). Nevertheless, 80 cm, which exceeded recommendations by the
the need to increase the number and length of rein- NCMA (1997) (i.e. maximum spacing not greater
forcement layers close to the top of segmental re- than twice the toe to heel dimension of the block).
taining walls is a recurring point in this review paper. This vertical spacing may have been too large to
Bathurst and Cai (1995) analyzed both structures prevent excessive deformation of the facing. In addi-
and showed that the location of cracks could be rea- tion, there was evidence of rupture of the reinforce-
sonably well predicted using conventional pseudo- ment at the connections and pullout of the connect-
static (Mononobe-Okabe) wedge analyses. A similar ing pins between block courses as the facing opened
survey of three GRS walls by White and Holtz up (bulged) during failure. However, it is difficult to
(1997) after the same earthquake revealed no visual conclude that these post-earthquake observations are
indications of distress. Some unreinforced crib walls evidence that the facing triggered the wall collapse
a) photograph of damaged structure
Figure 4. Conventional RC cantilever wall after 1995 Kobe
earthquake (Tatsuoka et al. 1996).
Precast- Overall instability
(Possible failure plane)
Local instability (Geogrid)
(Rupture of reinforcement or
pull-out of connecting pin)
b) cross-section of failed wall
Figure 6. Example failed segmental GRS wall after 1999 Chi-
chi earthquake (Koseki and Hayano 2000).
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 geo-
textile. There was no significant deformation of the
wall recorded by an inclinometer or any visual indi-
Figure 5. Example of counterfort and conventional cantilever cations of distress due to the earthquake.
walls after 1995 Kobe earthquake (Tatsuoka et al. 1996).
2.5 2001 El Salvador earthquake
or that the facing failed as part of a larger kinemati-
A magnitude 7.6 earthquake occurred on 13 January
cal mechanism. However, it was noted from a series
2001, 60 km off the coast of El Salvador. A large
of segmental GRS walls during the same earthquake
number of retaining walls were damaged. At the
that the magnitude of damage diminished with de-
time of the earthquake, there were 25,000 m2 of
creasing reinforcement spacing. For example, a wall
modular block (segmental) retaining walls in place
with a vertical spacing of 60 cm, and thus satisfying
(Race and del Cid 2001). A survey of these struc-
NCMA (1997) spacing recommendations, did not
tures by the second writer provided an opportunity
show any damage.
to identify why some structures behaved well and
others did not. All the structures were constructed on
2.4 1999 Kocaeli earthquake
hillsides to extend property fills. The walls were up
The Koceli earthquake occurred on 17 August 1999 to 8 m in height. Two walls that failed were exam-
with Moment Magnitude 7.4. The effects of the ined in detail. Contributing factors to the failures
earthquake extended to Istanbul where many build- were the use of masonry privacy fences attached to
ings suffered damage. At the time of the earthquake the top of the walls that developed additional over-
there was only one GRS segmental retaining wall in turning loads at the top of the wall. A second major
Turkey. The wall is located in Istanbul and is used to cause of failure of the walls was the extension of the
a) collapse of railway embankment
Figure 7. Example cross-section of GRS wall constructed in Is-
tanbul 1997. (Note: dimensions in meters).
b) reconstruction of embankment
Figure 8. Wall failure in El Salvador due to 1.7-m high unrein-
forced section at top of wall added by owner after original con-
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 ob-
served to have survived the earthquake without
2.6 2004 Niigataken-Chuetsu earthquake c) cross-section showing reinforced soil embankment and
During the 2004 Niigataken Chuetsu earthquake in
Japan, many embankments for roads, railways and Figure 9. Railway embankment collapse and repair as a result
hillside widening were damaged (JSCE 2006). Fig- of 2004 Chuetsu earthquake (Kitamoto et al. 2006).
ure 9 shows the collapse and reconstruction of a
railway embankment (Kitamoto et al. 2006). Repair paired using a GRS wall with segmental facing pan-
involved a combination of GRS retaining wall and els (Figures 11a,b). In this construction technique,
earth anchors. The benefits of a similar approach to the facing panels are not in contact with the rein-
stabilize the base of these structures using soil nails forced soil backfill during fill placement. This al-
is described later in Section 3.3. lows the backfill to be well compacted without dis-
Figure 10 shows another case history, where both turbing the facing. The soil is contained by an
a national highway and a railway embankment were internal wrapped face and a metal mesh form. The
severely damaged by the same slide during the Chu- railway embankment on the down slope side was re-
etsu earthquake. The highway embankment was re- constructed using a GRS wall with a full-height rigid
a) full-height panel facing repair of railway embankment face
on side opposite to highway embankment
Figure 10. Failure of highway and adjacent railway
embankment resulting from 2004 Chuetsu earthquake (JSCE Cast-in-situ
2006). Rail track
1500 3500 3500 2000 3000 750
Panel-type facing EQ
After re- Collapsed
Existing gravity-type Location of existing
retaining wall Before E.Q. gravity-type retaining 1:2
construction After gravity-type
400 wall after E.Q. .0 EQ retaining wall
Silt stone river
V= 3000 m3 1:4.0
Temporary Sand stone
Sand stone Co
5700 200 300 b) cross-section
a) cross-section of highway embankment repair Figure 12. Railway embankment repair as a result of 2004
Chuetsu earthquake (JSCE 2006).
3 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 re-
inforced 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, full-
scale tests have been performed. A disadvantage of
reduced-scale tests is that the response of the model
may be influenced by low confining pressure, far-
end boundary conditions of the shaking table strong
box, and improperly scaled mechanical properties of
b) concrete panel facing placement during repair of highway the reinforcement. Nevertheless, qualitative insights
embankment are possible using this experimental approach. Fur-
thermore, the models can be used to develop and
Figure 11. Highway embankment repair as a result of 2004
Chuetsu earthquake (JSCE 2006). validate numerical codes that can be used in turn to
investigate wall response at prototype scale. A re-
facing (Figures 12a,b). The different repair decisions view of shaking table tests reported in the literature
were decided based ground conditions, construction prior to 2002 is reported by Bathurst et al. (2002).
time and available backfill material. For example, to An example of preparation and testing of a more
improve resistance against a global sliding failure, recent GRS wall shaking table model with a sandy
the bottom of the rigid facing was embedded in the soil backfill is shown in Figures 13 and 14 (Lo
stiff sandstone bedrock layer at the site (Figure 12b). Grasso et al. 2004, 2005, 2006) using sandy soil as
the wall backfill.
20 23cm 20
a. Cantilever type(C) b. Gravity type(G)
20 Surchrge Surchrge
a) facing construction b) reinforcement placement 80
50 50 Extende
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).
c) pluviation of sand d) vertical columns of black -0.8
Figure 13. Preparation of shaking table model (model height = -0.4
0.35 m) (Lo Grasso et al. 2005). -0.2
Modified from N-S component at Kobe Marine Meteorological
Observation Station during the 1995 Hyogoken-Nanbu earthquake
1 2 3 4 5 6 7 8
a) wall prior to shaking b) wall after shaking
Figure 14. Shaking table test (Lo Grasso et al. 2004). Figure 16. Example Kobe earthquake accelerogram (Watanabe
et al. 2003).
In this section we investigate the effect of wall
type (i.e. conventional versus GRS wall structures) mulated rapidly with the conventional-type retaining
on seismic response. For GRS walls, the influence of walls. In contrast, the GRS structures exhibited more
facing type, reinforcement arrangement and excita- ductile behavior, particularly the wall with extended
tion record on wall performance is reviewed. reinforcement layers (Type 2 - Figure 15d).
The final geometry of the model walls, deforma-
3.1 GRS walls with full-height rigid facing tion of backfill and soil failure planes are shown in
In order to investigate the reasons for the relatively Figure 18. The conventional-type walls suffered not
good performance reported for GRS walls with full- only base sliding, but also overturning. The over-
height rigid facings during the 1995 Kobe earth- turning mode was more predominant with the GRS
quake, a series of 1g model shaking table tests were structures. It should be noted that multiple progres-
carried out by Watanabe et al. (2003). The wall sive failure planes were formed in the backfill due to
models were about 50 cm high with a level backfill the effects of strain localization and strain softening
Figure 15. They were seated on a horizontal soil behavior (Koseki et al. 1998). Note also that the re-
foundation and uniformly surcharge loaded. The inforced backfills underwent a simple shear mode of
foundation and backfill were modeled by very dense deformation. This is different than a direct sliding
dry sand layers. mechanism, which is assumed in most limit equilib-
The models were subjected to several sequential rium-based stability analyses with the reinforced
base excitation records by scaling the Kobe earth- backfill taken as a rigid body.
quake record shown in Figure 16 in 0.1g increments The reason for the less ductile behavior of con-
(see also Koseki et al. 2003). ventional retaining walls can be understood from
Figure 17 shows the cumulative permanent hori- Figure 19. The normal bearing stress at the toe of the
zontal displacements near the top of each model rigid footing of the gravity wall model suddenly de-
GRS and conventional-type wall. Up to a seismic creased after about 20 mm of maximum wall move-
coefficient value of about 0.5 no significant differ- ment, consistent with loss of bearing capacity at the
ence could be observed. However, under higher toe. A local toe failure was not observed for the rein-
seismic loads, the residual wall displacements accu- forced soil walls because wall loads were spread
Local failure due to loss of bearing capacity
dtop 5cm Location
90 35 of loadcells
Wall top displacement, dtop (mm)
80 30 1-9: Shaking step
of base footing, σ (kPa)
Normal stress at bottom
25 4 LT7 6 5 4
Gravity type 20 2 LT7 (toe)
40 Cantilever type 15
30 reinforcements) 10
Reinforced soil 5
10 LT4 (heel)
0 0 10 20 30 40 50 60 70 80 90 100 110
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 Wall top displacement, dtop (mm)
Seismic coefficient, kh=amax/g
Figure 17. Permanent displacement at top of wall versus peak Figure 19. Footing load response during base shaking of grav-
horizontal ground acceleration coefficient (Watanabe et al. ity wall structure (Watanabe et al. 2003).
15 U p p e rm o s t re in fo rc e m e n t
Tensile force (N)
T ype 3
10 T y p e 2 (L = 8 0 c m ) (L = 3 5 c m )
T y p e 1 (L = 2 0 c m )
0 10 20 30 40 50 60
Reinforced 15 T ype 3
Tensile force (N)
(L = 3 5 c m )
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
5 T y p e 1 (L = 2 0 c m )
T y p e 2 (L = 4 5 c m )
0 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 )
Tensile force (N)
15 T y p e 1 (L = 2 0 c m )
Gravity type amax=919gal Reinforced-soil type2 amax=1106gal T y p e 3 (L = 3 5 c m )
Figure 18. Final wall geometry, soil deformations and internal
soil failure planes (Watanabe et al. 2003). 0
0 10 20 30 40 50 60 70
W a ll to p d is p la c e m e n t, d to p (m m )
over the wider and more flexible base of the rein-
Figure 20. Reinforcement loads during base shaking of GRS
forced soil zone. wall structures (Watanabe et al. 2003).
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 rein-
forcement layer of the Type 2 wall. This enhanced
the resistance against overturning of the facing. This
highlights the observation by others that the distribu-
tion of reinforcement loads may be higher at the top
of the wall under earthquake loading which is oppo-
site to the trend typically assumed for walls under
static loading conditions.
Figure 21 shows a global failure mechanism de-
veloped behind and below a GRS wall model con-
structed with a sloped toe. This highlights the influ-
Figure 21. Development of global instability mechanism
ence of the foundation and toe conditions on the during shaking of GRS model wall (Kato et al. 2002).
ultimate failure mechanism in these structures which
can result in a failure pattern that is not predicted by
0.075 m 1.0 m
35 Formation of shear bands inclinometer tube
(refer to Fig. 21.)
30 0.225 m extensometer cable
25 sand backfill
Tensile force (N)
15 glued sand layer
Reinforced(L) plywood base
(S):on slope steel
5 (L):on level ground Shaking Table
0 10 20 30 40 50 60
Lateral displacement at wall top (mm) a) shaking table model
Figure 22. Sum of tensile loads in reinforcement layers for 7
measured predicted facing configuration NCMA method
GRS walls with a sloped foundation toe (S) and level ground
vertical thin facing (Wall 9)
(L) (Kato et al. 2002). 6
inclined thin facing (Wall 11)
vertical thick facing (Wall 8)
conventional limit equilibrium-based models that as- 5
sume 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 2
that reinforcement loads peaked and then softened as
wall deformations accumulated for the structure with 1
a slope while the structure with a level foundation AASHTO/FHWA method
accumulated reinforcement load monotonically with 0
0 0.1 0.2 0.3 0.4 0.5 0.6
increasing wall deformation. input base acceleration amplitude (g)
El-Emam and Bathurst (2004, 2005) investigated
the influence of toe constraint using a series of 1 m- b) computed and measured horizontal loads in reinforcement
layers versus peak base excitation acceleration level
high shaking table tests constructed with a rigid fac-
ing panel (Figure 23a). The horizontally restrained Figure 23. Reduced-scale rigid panel wall models with a
toe in reduced-scale models attracted approximately hinged toe (El-Emam and Bathurst 2005).
40% to 60% of the peak total horizontal earth load
during base excitation demonstrating that a stiff fac-
ing column plays an important role in resisting dy-
namic forces under simulated earthquake loading.
The inability of current design methodologies to ac-
count 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 ca-
pacity. They also recorded the vertical toe force de-
veloped at the base of the model walls during base
shaking and showed that down-drag forces devel-
oped 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 un- Figure 24. Shaking table test arrangement for 1-m high GRS
der-estimate the total load carried by the reinforce- wall models (Bathurst et al. 2002).
ment and horizontal toe in their models, which is
non-conservative for design (Figure 23b). However, tance of 1 m-high model reinforced walls with verti-
the pseudo-static NCMA method (Bathurst 1998) cal and inclined facings constructed with segmental
was shown to be less conservative than the current blocks, vertical incremental panel and rigid full-
AASHTO (2002) method. height panels. The tests were focused on the influ-
ence of interface shear properties on facing column
3.2 GRS walls with segmental (modular block) stability (Figure 24). The influence of interface shear
facing capacity and facing batter can be seen in Figure 25.
Bathurst et al. (2002) reported the results of a series The vertical wall (Test 4) with fixed interface con-
of shaking table tests that examined seismic resis- struction (equivalent to a rigid panel wall) required
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).
0.0 0.4 0.8 1.2 1.6
Figure 26. Full-scale shaking table test of segmental (modular 80
block) retaining wall (Ling et al. 2005). with nails
the greatest input acceleration to generate large wall
displacements during staged shaking. The vertical 40
wall with no shear connections between segmental
block layers performed worst (Test 1). However, the 20
resistance to wall displacement was improved
greatly for the weakest interface condition by simply 0
increasing the wall batter to 8 degrees from the ver- 0.0 0.4 0.8 1.2 1.6
tical (Test 3). The vertical wall with an incremental
panel facing (i.e. segmental layers rigidly connected Seismic coefficient
between geosynthetic layers) (Test 2) gave a dis-
placement response that fell between the results of Figure 28. Influence of soil nails on sum of tensile loads in
walls 1 and 4. reinforcement layers (Kato et al. 2002).
Ling et al. (2005) reported the only experimental
program to date in which full-scale model (2.8 m- spacing improved the seismic response of the struc-
high) shaking table tests have been carried out on tures.
GRS segmental retaining walls (Figure 26). The
walls were subjected to both vertical and horizontal 3.3 Other new applications
components of the Kobe earthquake accelerogram. In order to improve the seismic performance of GRS
The test walls showed maximum deformations at the walls constructed on slopes, Kato et al. (2002) re-
crest that were negligibly small at a scaled peak ported the beneficial effects of installing soil nails
horizontal acceleration of 0.4g. The walls exhibited below the base of the facing (Figure 27). This tech-
less than 100 mm of maximum deformation under a nique resulted in a potential soil failure mechanism
more severe acceleration scaled to 0.86g. The tests that allowed a larger tensile load to be carried by the
showed that increasing the length of reinforcement reinforcement layers (Figure 28). It should be noted
at the wall crest and reducing the reinforcement that, in the case without nails, the tensile load was
reduced after formation of the global failure mecha- 型式 Seismic stability: High Very high
nism (see for example Figures 21 and 22 for sloped
Cement-mixed Pre-loaded & pre-
toe case). Reinforcement load-softening was pre- girder Backfill
vented by using soil nails below the base of the wall Approach
block by cement-
facing. treated gravel cement-treated
Another technique involves improving the me- gravel
chanical properties of on-site soils by the addition of (already in use in Japan) （揺込み＋段差対策）
cement. Examples of this technique are described
GRS structures have also been studied with re- soil
spect to improving the seismic stability of bridge
abutments supporting bridge decks. Aoki et al. Subsoil
(2003) carried out 1g shaking table tests on conven-
tional and GRS bridge abutment models using ce- Figure 29. Example methods to improve the seismic stability
ment-treated backfill and pre-loaded and pre- of bridge abutments in Japan (Aoki et al. 2003).
stressed GRS walls with gravel backfill and full-
height rigid facings (Figure 29). The seismic stabil- R.L 140 Polymer geogrid
ity of cement-treated abutments was increased sig-
nificantly, compared to cement-treated structures Soil backfill
without geosynthetic reinforcement that are cur- 1:1
rently in use in Japan. An example structure using .5
cemented-treated gravel in combination with geo- Sedimentary
0 10 20 30 40 50
synthetic reinforcement is illustrated in Figure 30 for talus 3
a bridge abutment supporting the tracks for a new 8
bullet train on Kyushu Island (Aoki et al. 2005). The 14
construction technique resulted in a 20% saving Crystalline
compared to the conventional solution without the 50/30
(Unit in cm)
combination of cement-treated backfill and geosyn- 50/16
thetic reinforcement. Saito et al. (2006) carried out a
(Unit in cm)
series of 1g shaking table tests that showed that the
cement treatment of sandy backfill soils in combina-
tion with geosynthetic reinforcement can also result a) cross-section
in improved seismic performance. An example of New type abutment
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 rein-
forcement, 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 em-
bankment for a national highway in Japan (Figure
32). b) photograph of abutment under construction
Another potential technology is the use of ex-
panded polystyrene (EPS) as a geofoam seismic Figure 30. GRS wall with geosynthetic reinforcement for
buffer for rigid wall structures. The first use of this bridge abutment supporting bullet train railway in Japan (Aoki
approach to attenuate seismic-induced dynamic et al. 2005).
forces against basement walls was reported by Inglis
et al. (1996). Proof of concept has been demon- quake loads) have been studied by Tsukamoto et al.
strated by Zarnani et al. (2005) who reported the re- (2002).
sults 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 4 ANALYSIS METHODS AND NUMERICAL
simulated seismic load was reduced by 15% using a MODELLING
standard EPS material and 40% with a hollow core
material (Figure 33b). Analysis and design tools for GRS walls under dy-
Effects of the combined use of EPS and geogrid namic loading can be classified as:
to reduce lateral earth pressures at-rest and under a) pseudo-static methods,
static active pressure conditions (i.e. without earth- b) displacement methods, and
c) finite element and finite difference methods.
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
b) without geosynthetic reinforcement Seismic Rigid wall
Horizontal wall force (kN)
20 buffer density = 16 kg/m3 (no buffer)
Figure 31. Examples of shaking table models of GRS walls
with cement-treated sand backfill (Saito et al. 2006). 15
4.1 Pseudo-static methods 5 Cored buffer
with bulk density = 1.32 kg/m3
Figure 34 shows one of the limit equilibrium-based 0
pseudo-static methods based on a two-part wedge in 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
the reinforced backfill (Ismeik and Güler 1998). In Acceleration (g)
this method the critical wedge geometry is deter-
mined from a search of geometries giving the maxi- b) horizontal wall force
mum horizontal load to be carried by the reinforce-
ment layers (proportional to force coefficient K in Figure 33. Shaking table tests demonstrating reduction in seis-
mic forces against rigid walls using geofoam EPS seismic
Figure 35). The length of the reinforcement required buffers (Zarnani et al. 2005).
to contain the critical failure wedge is also deter-
mined from the critical wedge geometry and can be the wall. This is a safe design practice for segmental
presented in design chart form. Similar approaches walls, which typically have an unreinforced section
using a single soil wedge and closed-form solutions at the top of the wall (Sections 2.5 and 3.2). The cur-
have been reported by others (see Bathurst et al. rent AASHTO (2002) method in the USA restricts
2002). An example is the use of additive static and pseudo-static methods to peak horizontal ground ac-
dynamic pressure distributions by the NCMA to as- celerations < 0.3g. A fundamental shortcoming of
sign load to reinforcement layers based on a con- pseudo-static methods is that they provide no infor-
tributory area approach (Bathurst 1998). This mation on wall displacements that may exceed a
method is illustrated in Figure 36, which shows that serviceability design criterion before an ultimate
a larger dynamic portion of seismic load is distrib- limit (collapse) state is developed.
uted to the layers of reinforcement close to the top of
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 Güler 1998).
Figure 37. Example of double integration method to compute
displacement of sliding mass (Cai and Bathurst 1996).
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 mecha-
nism to be triggered is computed (using a pseudo-
static method of analysis). Finally, double integra-
Figure 35. Results of parametric analysis illustrating
tion of the horizontal ground acceleration record is
relationship between force coefficient, wall geometry and soil carried out to compute displacements as illustrated
shear strength for horizontal seismic coefficient value of kh = in Figure 37.
0.2 (Ismeik and Güler 1998). 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 in-
terface between modular block layers in segmental
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) accu-
Figure 36. Calculation of total earth pressure distribution for mulated gradually with the increase in the peak am-
GRS walls: (a) static pressure contribution; (b) dynamic
pressure increment, and (c) total distribution (Bathurst 1998). plitude of base acceleration. This gradual increase is
difficult to reproduce numerically using a constant
threshold acceleration value in Newmark-type
4.2 Displacement methods analyses.
The bottom figure in Figure 38 shows that the
In order to estimate earthquake-induced displace- measured amount of base sliding increased signifi-
ment of GRS walls, variants of the classical New- cantly during the last excitation stage. This response
mark sliding block method (Newmark 1965) have stage may be simulated using the Newmark method,
been proposed (e.g. Cai and Bathurst 1996, Ling et
Tilting angle (degree)
5cm-thick subsoil layers
Shear stress ratio, SR=τ/σ
Sinusoidal excitations 0.5
5cm-thick subsoil layers
0 Computed -1.0
0.00 0.01 0.02 0.03 0.04 0.05
Shear strain of subsoil, γ
0 200 400 600 800
25 a) foundation soil
Disp. of wall bottom (mm)
Shear stress ratio, SR=τ/σ
0 200 400 600 800 -1.0
Base acceleration (gal) 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Shear strain of reinforced backfill, θ
Figure 38. Comparison of computed and measured tilting and b) backfill soil
displacement of 1g GRS wall shaking table models (Koseki et
al. 2004). Note 1000 gal = approximately 1g. 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 thresh-
old acceleration to initiate sliding failure using a wall deformations up to the 750-gal (0.75g) excita-
pseudo-static limit equilibrium analysis together tion stage is shown in Figure 38.
with the peak friction angle for the foundation and In Figure 40a, wall displacements are computed
backfill soils. However, a sudden increase in base by considering the shear deformation of the founda-
sliding may also be triggered by the formation of a tion and reinforced backfill soil in combination with
failure plane in the unreinforced backfill immedi- the Newmark method at the highest excitation level.
ately before the threshold acceleration is achieved, Although there are quantitative discrepancies be-
which was the case in this experiment. Similar be- tween computed and measured values, the agree-
havior has been observed in model tests on conven- ment in qualitative trends is encouraging. Further-
tional retaining wall models. Nevertheless, the mag- more, the amount of movement during the 800-gal
nitude of wall displacements when a distinct failure excitation stage is reasonably close. Similar com-
plane was formed in the backfill was much larger ments apply for the nominally identical experiment
with the more ductile GRS wall models (Watanabe carried out with an irregular excitation record (Fig-
et al. 2003). ure 40b).
In order to simulate the gradual accumulation of Based on the results from these investigations, the
base sliding at relatively low earthquake loads, an at- Newmark sliding block method may be applied at
tempt was made to estimate the cyclic stress-strain large earthquake loads above the threshold accelera-
properties of the foundation soil supporting the rein- tion level and after the formation of failure plane in
forced backfill. Figure 39a shows the back- the unreinforced backfill. However, before the for-
calculated response for the foundation soil assuming mation of one or more failure planes, wall displace-
no slippage between the subsoil and the reinforced ments due to pre-failure shear deformations of the
backfill. The measured response accelerations and foundation and reinforced backfill soil will control
the tensile forces in the extended reinforcement lay- seismic response. Computed displacements gener-
ers were used to compute the data curves in the fig- ated prior to a collapse state (i.e. at low earthquake
ure (Koseki et al. 2004). A similar approach was loading and a factor of safety greater than unity
used to generate the cyclic stress-strain response of from a pseudo-static analysis) should not exceed a
the soil in the reinforced zone (Figure 39b). The serviceability criterion.
good agreement between computed and measured
Base acceleration (gal) -800
2 3 4 5 6 7
Disp. of wall bottom (mm)
5 cm-thick subsoil layers
3.5 a) shaking table model mounted on centrifuge
2 3 4 5 6 7 1.00
0.75 1.50 1.50 1.50 1.50 0.75
1.50 1.50 1.50 1.50 1.50
a) sinusoidal base excitation
athres=807gal Embankment, 2
Base acceleration (gal)
0 0.50 6.50 15.00 0.50
400 b) finite element model (dimensions in centimeters)
600 Irregular excitations
800 20 cm-thick subsoil layers
Figure 41. Numerical modeling of centrifuge test of segmental
3 4 5 6 7 8 GRS wall (Fujii et al. 2006).
Disp. of wall bottom (mm)
10 In this conference, Fujii et al. (2006) describe a
9 numerical study aimed at simulating the results from
8 Measured a series of dynamic centrifuge tests on GRS segmen-
7 tal walls (Figure 41). They conducted 2-D finite
6 Computed element (FE) analyses using the program FLIP. The
5 foundation soil and the backfill were modeled using
4 a multi-spring approach (Iai et al. 1992). The rein-
3 forcement layers and facings were modeled by lin-
3 4 5 6 7 8
ear-beam elements, and the interfaces with the back-
fill were modeled by joint elements.
b) irregular base excitation Figure 42a shows a typical comparison between
the measured and computed time histories of re-
Figure 40. Back-calculated shear-strain response of soils from
1g GRS wall shaking table models (Koseki et al. 2004).
sponse acceleration in the unreinforced backfill at
prototype scale. In total, thirteen test cases with dif-
ferent input wave forms and amplitudes were ana-
4.3 Finite element and finite difference modeling lyzed. Figure 42b summarizes the comparison be-
tween the measured and computed maximum
Properly conceived numerical models can be used to response accelerations. In general, reasonable
gather both qualitative insights and quantitative data agreement was obtained up to an acceleration level
for the performance of GRS walls under seismic of 600 gals (0.6g), with a tendency for the numerical
loading and to guide the development of design model to slightly overestimate the physical test re-
methodologies (Cai and Bathurst 1995). sponse. The largest deviations may be the result of
Lateral displacement (mm)
600 Centrifuge model tes t 実測結
Seis mic res pons e analys is
-200 0 2 4 6 8 10 12 14
0 2 4 6 8 10 12 14
-600 -20 Centrifuge model test
Seismic response analysis
a) example comparison of measured acceleration response in a) example comparison of measured lateral wall response in
physical test and computed response physical test and computed response
of seismic response analysis (mm)
y = 1.1746x
y = 0.9558x
of seismic response analysis (gal)
Maximum lateral displacement
Maximum response acceleration
800 R = 0.8856 2
R = 0.9557
0 200 400 600 800 1000 0 100 200 300
Maximum response acceleration
of centrifuge model test (gal) Maximum lateral displacement
of centrifuge model test (mm)
b) peak acceleration
b) lateral displacement
Figure 42. Comparison of physical test results and computed
acceleration response of centrifuge tests of segmental GRS
Figure 43. Comparison of physical test results and computed
walls (Fujii et al. 2006).
lateral displacement response of centrifuge tests of segmental
the formation of failure planes in the backfill, which GRS walls (Fujii et al. 2006).
was not properly modeled in the numerical code.
Lateral displacements of the facing are compared physical models under investigation. A constant re-
between the measured and computed data in Figure inforcement stiffness value was shown to be a rea-
43. Accumulation of the lateral displacement was sonable assumption for numerical modeling of the
reasonably well simulated although a phase lag was polyester geogrid reinforcement used in the physical
sometimes observed between the measured and tests. However, reinforcement-soil slip for layers
computed responses. with shallow overburden depth was not considered
Figure 44 shows measured and computed lateral in numerical simulations and this is thought to have
earth pressures at the back of the facing. In general, led to some discrepancies in reinforcement load re-
the computed values largely over-estimated the sponse close to the top of the wall. Both numerical
measured values. This may be due to the challenge and physical models demonstrated that the toe
of modeling the facing interfaces. boundary condition has a large influence on wall
El-Emam et al. (2004) reported the results of nu- performance and stability under both static and
merical modeling of 1-m high shaking table tests simulated seismic loading conditions. Finally, both
that investigated full-height panel face GRS walls numerical and experimental models showed that cur-
with different toe boundary conditions (i.e. hinged rent pseudo-static seismic design methods may un-
and free to rotate and slide). They used the finite dif- derestimate the size of the soil failure zone behind
ference-based program FLAC (Figure 45a). The the wall facing, particularly at large input base ac-
numerical models were found to give reasonably ac- celeration amplitudes (Figure 45b,c) and these ana-
curate predictions of the experimental results (wall lytical methods are unable to explicitly account for
facing displacements, reinforcement loads and the influence of the toe boundary reaction on wall
measured toe loads) despite the complexity of the response.
stiff thin soil
Horizontal earth pressure(kN/m )
very stiff back
Centrifuge model tes t facing interface
20. 2.40 m
Seis mic res ponse analys is
解析結果 panel column
15. reinforcement layers
0 2 4 6 8 10 12 14
a) example comparison of measured lateral earth pressure
response in physical test and computed response 0.15 m
fixed base boundary
30 soil layer very stiff foundation base acceleration
of seismic response analysis (kN/m )
Maximum horizontal earth pressure
a) FLAC numerical grid
shear strain < 0.2%
facing failure surface (observed)
y = 4.0315x predicted shear zone
R = 0.7687 reinforcement layers
b) acceleration amplitude = 0.15g
0 10 20 30
Maximum horizontal earth pressure
of centrifuge model test (kN/m ) shear strain < 1.0%
b) horizontal earth pressure panel
Figure 44. Comparison of physical test results and computed
horizontal earth pressure response of centrifuge tests of numerically
segmental GRS walls (Fujii et al. 2006). predicted shear zone
reinforcement layers stiff foundation
Bathurst and Hatami (1998) reported the results
of a numerical parametric study of an idealized 6-m 1.0 m
high GRS wall with a full-height rigid facing and six
layers of reinforcement. They showed that the mag- c) acceleration amplitude = 0.5g
nitude and distribution of reinforcement loads was
Figure 45. Predicted and observed failure surfaces from hinged
sensitive to the stiffness of the reinforcement materi- toe model wall (5 reinforcement layers and L/H = 1.0) at
als used (Figure 46). The influence of reinforcement different input base acceleration amplitudes. Note: Dark
stiffness on seismic response of GRS walls is not shading indicates large shear strains (El-Emam et al. 2004).
considered in current design codes. Furthermore,
they also showed that the relationship of the funda- for both the soil and the reinforcement to better pre-
mental frequency of a GRS wall to the predominant dict the seismic response of GRS structures. For ex-
frequency of the excitation record is a critical factor ample, hysteretic models based on Masing functions
that may control wall response independent of the and their variants (e.g. Yogendrakumar and Bathurst
design input parameters currently considered in de- 1992, Wakai et al. 2006) and strain-softening models
sign codes based on pseudo-static methods (Figure for the backfill soil (e.g., Desai and El-Hoseiny
47). Numerical analyses showed no significant 2005) are potentially fruitful avenues of future re-
influence of the reinforcement stiffness, search.
reinforcement length or toe restraint condition on the
fundamental frequency of wall models (Hatami and
The review of the literature by the writers has
highlighted the need for more sophisticated models
Figure 46. Influence of reinforcement stiffness (J) on distribu-
tion 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
Table 2. Limit states design codes for geotechnical structures.
SUBJECTED TO EARTHQUAKE
Earlier in the paper we have described some general
approaches for the analysis and design of GRS walls Eurocode 7 Geotechnical design
under earthquake loading. Pseudo-static methods (ENV 1997)
Eurocode 8 Design of structures for earthquake resistance
based on a classical treatment of factor safety are the (ENV 1998)
most common approach used to estimate destabiliz-
ISO 2394: 1998 General principles on reliability of structures
ing forces in seismic stability analyses. However, in
order to bring the design of these structures in line
ISO/CD 23469: Bases for design of structures – Seismic ac-
with modern concepts for civil engineering struc- 2004 tions for designing geotechnical works
tures, and in particular systems involving soil-
JGS 001-2004: Principles for foundation designs grounded
structure interaction, there is a continuing movement 2004 on a performance-based design concept
towards a limit states design (LSD) approach. In this AASHTO (2004)AASHTO LRFD Bridge Design Specifica-
section we describe the history and current practice tions (USA)
for design of GRS walls in seismic environments CHBDC (2000) Canadian Highway Bridge Design Code
with special emphasis on LSD where applicable. Fi- GEOGUIDE 6 Guide to reinforced fill structure and slope
(2002) design (Hong Kong)
nally, we discuss a number of issues that should be BS8006 (1995) Code of practice of strengthened/reinforced
considered as LSD codes for seismic design of GRS soils and other fills
walls are developed in the future
5.1 Global trends for GRS wall seismic design draft (ISO/CD 23469 2004) outlining the basics for
codes the design of geotechnical structures under seismic
actions was completed.
Global trends toward limit states (or performance- In Japan, the Japanese Geotechnical Society pub-
based) geotechnical designs are summarized in Ta- lished its first guidance document titled “Principles
ble 2. for foundation designs grounded on a performance-
In 1994, the pre-standard version of Eurocode 7 based design concept”(JGS 4001 2004). The docu-
(ENV 1997) on geotechnical design (Part 1: General ment is consistent with “Principles, guidelines and
Rules) was published, followed by Eurocode 8 terminologies for structural design code drafting
(ENV 1998) on design of structures for earthquake founded on the performance-based design concept”
resistance. They are based on the limit states design produced by the Japanese Society of Civil Engineer-
concept and intended to harmonize geotechnical de- ing (JSCE 2003). Here we use the term “perform-
sign in Europe and also to harmonize geotechnical ance-based design” to identify codes that do not
design with structural design practice (Orr 2002). specify actual calculation methods but rather specify
An international standard (ISO 2394) identifying target performance values that must be met (i.e. al-
the general principles on reliability of structures was lowable displacement criteria). In other words, they
published in 1998. More recently, an ISO committee are used to ensure that the structure maintains an ac-
ceptable level of serviceability as well as safety dur- Table 3. Summary of current design codes for seismic design
ing and after an earthquake. This is in contrast to of GRS walls.
“specification-based design” which typically means
Country Agency Reference Design by
ensuring that the structure has acceptable factors of Australia RTA RTA (2005) Limit states design
safety against prescribed failure (collapse) mecha- Japan PWRC PWRC (2000) Specification-based
nisms based on mandated calculation methods pre- Japan RTRI RTRI (2006) Performance-based*
sented in the design code. This approach is common USA FHWA FHWA (2001) Allowable stress design
in many of the design approaches used today for the USA AASHTO AASHTO Allowable stress design
design of GRS walls under both static and seismic (2002)
USA AASHTO AASHTO Allowable stress design
load conditions. In the Canada and USA this ap- (2004)
proach is called “working stress design” and “allow- USA NCMA Bathurst Allowable stress design
able stress design (ASD)”, respectively. (1998)
In the discussion to follow, it is important to note Italy AGI** AGI (2005) Limit states design
that there is a fundamental difference between North * with limit states design procedures, ** Italian Geotechnical
American and European approaches to LSD. A fac-
tored strength approach is used in Europe, while a
factored overall resistance approach is used in North
America, including the National Building Code of block systems. However, a fourth related publication
Canada (NBCC) (NRC 1995) and the CHBDC by the National Concrete Masonry Association
(2000). An excellent discussion of the difference in (NCMA) (Bathurst 1998) is a comprehensive docu-
approaches and the development of LSD methods in ment for the seismic design and analysis of segmen-
geotechnical engineering can be found in the paper tal retaining (modular block) walls. It is an allowable
by Becker (1996). stress design method that calculates earth pressures
At present there are only two Canadian guidance from Monokobe-Okabe wedge theory, but with the
documents used for the design of GRS retaining orientation of any failure wedge restricted to the
walls. The Canadian Foundation Engineering Man- value for static loading only. Otherwise, the meth-
ual (CFEM 2006) gives a cursory description of the odology recommends minimum factors of safety for
Tie-back Wedge (Simplified) method for geosyn- internal, external and facing stability modes of fail-
thetic reinforced soil walls but offers no guidance on ure consistent with the general approach for pseudo-
the use of limit states design for static or seismic de- static seismic analysis of other GRS types of retain-
sign. The Canadian Highway Bridge Design Code ing walls. In this guidance document, a Newmark
(CHBDC 2000) gives no guidance on GRS walls sliding block analysis approach is recommended for
and recommends the use of “analytical limit states peak ground accelerations greater than 0.3g using
design method” that should be checked against ac- the method proposed by Cai and Bathurst (1996).
cepted working stress design methods. However, the A shortcoming of the current AASHTO (2004)
bridge code is currently undergoing revision to in- LSD document is that load and resistance factors
clude GRS walls and guidance for LSD design for have been determined from calibration against cur-
GRS walls under static and seismic load conditions. rent ASD methods with conventional factors of
In the USA, the most current guidance documents safety. However, this can be argued to defeat the
for state agencies are the AASHTO Standard Speci- purpose of converting to a limit states design ap-
fications for Bridge Design (AASHTO 2002), proach. To overcome this deficiency, Allen et al.
AASHTO LRFD Bridge Design Specifications (2005) have written a Transportation Research
(AASHTO 2004) and the FHWA Mechanically Sta- Board guidance document that provides a detailed
bilized Earth Walls and Reinforced Soil Slopes explanation of the methodologies to calibrate LRFD
Manual (FHWA 2001). The AASHTO (2002) models for both metallic- and geosynthetic-
document is in its last release as AASHTO moves reinforced soil walls (i.e. to select load and resis-
from allowable stress design to a limit states design tance factors based on measured performance data).
format (called “load and resistance factor design - The general approach can be used for seismic design.
LRFD” in the USA). Nevertheless, AASHTO (2004) Current seismic design codes for GRS walls are
uses the Simplified Method for internal stability de- summarized in Table 3, which is updated from
sign of geosynthetic and metallic reinforced soil Zornberg and Leshchinsky (2003). One of the Aus-
walls. The FHWA and USA state departments of tralian codes (RTA 2005) adopts limit states design
transportation have set 2007 as the date after which procedures but is written in a performance-based
LRFD is to be used for all new bridge designs. format. One Japanese code for railways (RTRI
For earthquake design, a pseudo-static method is 2006) also adopts limit states design procedures
used. The FHWA (2001) document is largely consis- within a performance-based framework. Another
tent with the AASHTO (2002) document where they Japanese code for highways (PWRC 2000) is being
overlap for the design of reinforced soil walls. Both updated to a limit states design format (performance-
documents do not fully address design of modular based design).
In the following sections, the Japanese RTRI
code is briefly reviewed, together with its extension Ministry order in 2001 (Ministry of Land,
Infrastructure and Transport, Japan)
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 Model or standard design codes (edited by Individual design code
conventional retaining wall structures are identified. Railway Technical Research Institute) (by respective railway company)
These performance differences should be considered
when developing future LSD codes that include both Commentaries (ditto)
classes of structures.
5.2 Design codes for railway structures in Japan Figure 48. Evolution of Japanese railway design codes.
The following is a brief summary of the history of Table 4. Performance requirements for Japanese railway de-
limit states design code development for railway sign codes.
structures in Japan.
Action (design Level 1 (highly ex- Level 2 (maximum
earthquake pected for Tdes) possible for Tdes)
a) Limit states design procedures were first intro- loads)
duced into design codes for concrete and Important Performance I: Performance II:
steel/hybrid structures in 1992. structures Will maintain their Can restore their
b) LSD was introduced into design codes for foun- expected functions functions with quick
dations and retaining walls in 1997. without repair repair works
Others works (no exces- Performance III:
c) A new seismic design code was published in sive displacements) Will not undergo
1999, which was written in a performance-based overall instability
format for the first time.
d) Design codes for concrete structures and em-
bankments were revised into performance-based in this code, and analysis methods to compute per-
formats in 2004 and 2006, respectively. formance targets are explained in the commentaries.
e) Other design codes for steel/hybrid structures Table 4 summarizes the required performance of
and foundations are currently under revision to railway earth structures against two levels of design
the same format. earthquake loads. In this table Tdes is the design life
f) In order to enhance the implementation of per- of the structure. In general, the design life of a rail-
formance-based design, a new code based on way earth structure is assumed as 100 years. For a
displacement limits was published in 2006. The Level 1 earthquake load that is highly expected over
seismic design code is also currently under revi- the design life, it is required that all the earth struc-
sion. Both codes have been harmonized so that tures will maintain their design functions without re-
the same common principles are used. quiring repair work, i.e. will not exhibit excessive
displacements (Performance I). Against a Level 2
It should be noted that the design of retaining earthquake load, which is defined as the maximum
walls including GRS structures is described in the possible level over the structure design life, it is re-
latest code version for embankments. The back- quired that important earth structures can be restored
ground and a brief summary of the seismic design to design function conditions with minimal repair
procedures are explained in the next section. (Performance II), while the other earth structures
will not undergo overall instability (Performance III).
5.2.2 Seismic design of GRS retaining walls for In order to meet the above performance criteria,
railway structures in Japan for GRS walls, the following analytical approaches
As illustrated in Figure 48, the design code for rail- are recommended:
way earth structures in Japan (RTRI 2006) was re- a) Performance requirements for Level 1 earth-
vised following a directive by the Ministry of Land, quake loads are verified through stability analy-
Infrastructure and Transport, Japan issued in 2001. ses using partial safety factors against internal
This directive mandated that structures should be de- instability of the reinforced backfill (using the
signed based on performance criteria. This design model shown in Figure 49), external instability
code is a national-level model or standard design (using the model shown in Figure 50), and facing
code, accompanied by commentaries, both of which failure.
were edited by the Railway Technical Research In- b) Performance requirements for Level 2 earth-
stitute. Based on this code, individual design codes quake loads are determined using Newmark slid-
will be implemented by railway companies in Japan. ing block analyses and other numerical analyses
Therefore, performance requirements are specified against base sliding displacement of the retaining
S R Q
P fh Pb
Figure 49. Internal stability of GRS walls (RTRI 2006).
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 θ
Figure 52. Shear deformation of reinforced soil zone.
HW VF HW VF
ground surface Vw Vw
It should be also noted that, based on this design
ground surface of design
Tgt code, the tensile strength of geosynthetic reinforce-
RW ment layers against earthquake load can be assigned
without considering the possible effects of creep re-
(a) base sliding (b) overturning duction. This is consistent with recent findings on
the load-strain-time behavior of geosynthetic rein-
khpL forcement products as reported by Greenwood et al.
(2001) and Tatsuoka (2003) amongst others.
5.2.3 Probabilistic analyses
ground surface of design
γ In evaluating the seismic performance GRS struc-
tures, probabilistic approaches have several advan-
tages over the deterministic approaches adopted in
most existing design codes. By employing probabil-
(c) shear deformation of reinforced backfill istic approaches, the variability of soil and rein-
forcement parameters, in addition to different de-
Figure 51. Sliding analyses for GRS walls (RTRI 2006). grees of seismicity, can be considered rationally and
quantitatively by using a reliability index, failure
probability, or limit state exceedance probability.
wall, overturning displacement, and shear de- Shinoda et al. (2006a,b) conducted a series of
formation of the reinforced backfill (using the probabilistic analyses (Monte Carlo simulations) on
models shown in Figure 51). GRS slopes subjected to earthquake loads. Three
different GRS slope configurations were considered
It should be noted that, although the reinforced (Table 5). The Case 1 configuration is illustrated in
backfill has been modeled as a rigid body in many Figure 53a. One of the design earthquake motions
existing design codes, results of numerical and specified in the RTRI (1999) seismic design code for
physical models shows that it undergoes a simple railway structures was selected as the same input
shear-type deformation under large earthquake loads function for each analysis (Figure 53b), while the
as illustrated in Figure 52. This deformation mode is unit weight, soil strength, primary and secondary re-
evaluated in the above procedure (Figure 51c). inforcement strengths were taken as random vari-
ables described by a coefficient of variation. New-
mark analyses assuming a circular failure plane were Primary
carried out and the computed deformations recorded reinforcement
for each simulation run. The allowable deformation Backfill soil
was set at 50 cm. The computed exceedance prob- Surface soil
abilities for this limit state based on very many Secondary of 10 kPa
simulation runs are shown in Table 5. The combina- reinforcement 1:1.
tion of high strength primary reinforcement and high
strength backfill soil (Case 1), gave a probability of
exceeding the allowable deformation that was five 2m 0.3 m 1.5 m
orders of magnitude lower than the worst case inves-
tigated (Case 3).
Shinoda et al. (2005) also computed the ex- Foundation soil
ceedance probabilities of GRS walls based on stabil-
ity analyses without earthquake loads. Another se- a) general arrangement for Case 1 reinforced soil slopes
ries of probabilistic analyses were conducted by
Miyata et al. (2001). Chalermyanont and Benson
(2004, 2005) have developed reliability-based de-
Max. acceleraion = 924 gal
sign charts for external and internal design of GRS
walls under static load conditions using Monte Carlo 0
simulations. In their analyses, only ultimate limit -500
states defined by sliding, overturning and bearing -1000
capacity were considered for external modes of fail- 5 10 15 20 25 30 35
ure. The internal stability limit state was restricted to
reinforcement pullout with the loading calculated us- b) ground motion used in Newmark analyses
ing a circular slip analysis. Similar Monte Carlo
simulations can also be carried out for GRS walls Figure 53. General arrangement for reinforced soil slopes used
under seismic loadings and probabilities of ex- in Case 1 probabilistic analyses (Shinoda et al. 2006a,b).
ceedance computed for a range of failure mecha-
nisms and prescribed deformation limit states.
Table 5. Case study parameter matrix (Shinoda et al. 2005).
5.3 Eurocode 8
Section 5 of Eurocode 8 (2003) offers general guide- Case 1 Case 2 Case 3
lines that are applicable to GRS walls. Some of the Backfill soil High strength High strength Low strength
major points in this document are described below. Primary rein- Yes No No
Analyses may be calculated either by established forcement
methods of dynamic analysis, such as finite element Secondary re- Yes Yes Yes
models, rigid block models, or by simplified pseudo- inforcement
static methods. The mechanical behavior of the soil Exceedance 2.7x10-6 8.9x10-2 4.1x10-1
media should include strain softening and the possi- probability*
ble effects of pore pressure generation under cyclic
loading if applicable. Simplified pseudo-static meth- wall are inclined at not greater than 2φ/3 with re-
ods can be used where the surface topography and spect to the direction normal to the wall.
soil stratigraphy are reasonably regular. If a The code requires that a limit state condition shall
Mononobe-Okabe formulation is used, the point of then be checked for the least safe potential slip sur-
application of the force due to the dynamic earth face. Serviceability limit state conditions described
pressures shall be taken at the mid-height of the wall. as a permanent displacement may be calculated us-
Horizontal and vertical inertia forces should be ap- ing a simplified dynamic model consisting of a rigid
plied to every portion of the soil mass and to any sliding block. The calculations must be carried out
gravity loads acting at the backfill surface in pseudo- using a time history representation based on the de-
static methods of stability analysis. Vertical inertial sign acceleration without reductions. In the absence
forces shall be considered as acting upward or of a representative ground motion record, the code
downward, which ever gives the most critical result. offers recommendations for the selection of horizon-
The total design force acting on the wall under seis- tal and vertical seismic coefficient values. No reduc-
mic conditions shall be calculated at limit equilib- tion of the shear strength of cohesionless soils need
rium of the model. For walls that are free to rotate be applied for strongly dilatant materials, such as
about their toe, the dynamic force may be taken to dense sands.
act at the same point as the static force. The active
and dynamic pressure distributions acting on the
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 applica-
tions and design life. One section of the standard is
devoted to the design of reinforced soil walls. The 48°
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. 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).
5.4.2 Considered limit states and required
In the BS 8006 standard, external and internal stabil-
ity limit states are considered. For both ultimate and
serviceability limit states, short- and long-term con-
ditions 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. Po-
tential slip surfaces passing through the structure and
beyond the boundaries of the reinforced soil zone
must be analysed. The possibility of large settle-
ments in the foundation soil and/or in the reinforced
soil fill should be checked as part of design calcula-
tions related to deformation serviceability limit
states including differential settlements.
For internal ultimate limit states, the following Figure 55. Influence of seismic acceleration a and soil friction
potential failure mechanisms should be considered angle φ on the magnitude of earth pressure coefficient Ka for a
GRS wall with an inclined facing i = 50° and constant static
in stability analysis: (i) stability of individual ele- pore pressure ratio value ru = 0.25 (after Cascone et al. 1995).
ments of the wall; (ii) resistance to sliding of upper
portions of the structure; (iii) stability of wedges in (for structures with a concrete facing including
the reinforcement fill. In the design checks, the fol- modular blocks); and (ii) effect of soil cyclic behav-
lowing aspects should be taken into account: the ca- ior on active earth pressures and, consequently, the
pacity to transfer shear between the reinforcing ele- stability of the wall.
ments; the tensile capacity of the reinforcements; The stability of a wall will be influenced by the
and the capacity of the fill to support compression. facing angle (i.e. vertical or inclined) as demon-
Construction and post-construction behavior should strated in Section 3.2.
also be considered. In the latter case, the following For walls with an inclined facing, experimental
serviceability limits may be considered: foundation shaking table tests reported by Lo Grasso et al.
settlements; internal compression of the backfill; and (2004) developed a two-wedge (bilinear) failure
internal strain of the geosynthetic reinforcement. mechanism (Figure 54). For analysis purposes, iner-
tia forces may be considered explicitly in conven-
5.4.3 Recommendations for future seismic design tional pseudo-static wedge analyses. As an example,
development Figure 55 shows the effect of the seismic accelera-
In order to extend the limits states design approach tion and soil friction angle on the magnitude of the
in BS 8006 to include seismic effects, the following earth pressure coefficient (Ka) for a GRS wall with
issues should be addressed: (i) influence of inertia an inclined facing and a constant static pore pressure
forces acting on the soil fill, on the surcharge ap- ratio value (Cascone et al. 1995). No earthquake-
plied at the fill surface, and on the wall facing mass
i =50° r u=0.25
1,8 k H =0.25 k V=0
∆ u *=1
1,6 ∆ u *=0.8
k a*/k a
∆ u *=0.6
∆ u *=0.4
∆ u *=0.2
∆ u *=0
20 25 30 35 40 45
Figure 57. Influence of the earthquake-induced pore pressure
ratio ∆u* on the ratio between the earth pressure coefficient
Figure 56. Influence of soil friction angle φ’ and static pore Ka* evaluated taking into account the soil shear strength
pressure ratio value ru on the magnitude of earth pressure reduction and unmodified Ka for GRS wall with constant
coefficient Ka for a GRS wall with a vertical facing and given inclined facing angle, i = 50°, static pore pressure ratio ru and
values of kh, kv, nq and λ (after Motta 1996). horizontal kH and vertical kV seismic coefficient values (after
induced pore pressures or vertical component of empirical relationships based on experimental results.
seismic acceleration were considered in these analy- The effect of soil shear strength reduction on the
ses. The influence of vertical component of seismic stability of the wall may be evaluated using the
acceleration may have a large effect on GRS wall earthquake-induced pore pressure ratio ∆u* in stabil-
forces when pseudo-static analyses are used (e.g. ity calculations. As an example, Figure 57 shows the
Bathurst and Cai 1995). Vertical accelerations must effect of the earthquake-induced pore pressure ratio
be in analyses using Eurocode 8, as noted earlier, value ∆u* on the magnitude of the earth pressure co-
and in recent guidelines of the Italian Geotechnical efficient Ka from a two-part wedge analysis and a
Society (AGI 2005) focused on design procedures wall with an inclined face. In this plot, Ka* is the ad-
for GRS walls in seismic areas. justed value of earth pressure coefficient and Ka is
For GRS walls with a vertical facing, the failure the unmodified value. The values of Ka* for other
mechanisms will depend on different parameters values of soil friction angle may be evaluated by
such as the presence of the surcharge. In this case combining values from curves in Figures 56 and 57.
the influence of inertia forces on wall stability may Finally as noted in Section 4.3, an estimate of the
be evaluated using the solutions proposed by Motta fundamental frequency of the structure should be
(1996). These solutions consider the influence of the made and this value compared to the predominant
surcharge magnitude q and the distance d between frequency of the design earthquake record (Bathurst
the surcharge and the wall facing using the non- et al. 2002). The fundamental frequency of a struc-
dimensional parameters nq = q/(γH) and λ = d/H re- ture can be calculated using 1D linear elastic theory
spectively, where γ is soil unit weight and H is wall (Hatami and Bathurst 2000). Small changes in wall
height. As an example, Figure 56 shows the effect of height may be all that is required to shift the funda-
soil friction angle and static pore pressure ratio on mental frequency of the structure away from the
the values of the earth pressure coefficient for a predominant frequency of the design earthquake.
horizontal seismic coefficient kh = 0.20 and no verti-
cal acceleration (kv = 0). 5.5 Advantages of GRS walls over conventional
The influence of soil shear strength reduction walls
due to earthquake-induced pore pressure may be
evaluated using a pore-pressure ratio approach for In Section 2, case histories showing good perform-
walls with vertical or inclined facing geometry ance of GRS walls compared to conventional walls
(Fazzino 2005). The analysis should be performed are reported. GRS walls develop different response
assuming that earthquake-induced pore pressure and mechanisms while resisting large earthquake loads.
wall stability are decoupled. The earthquake-induced For example, the more ductile response of GRS
pore pressure ratio may be evaluated using available
(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 Chang-
Geng 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 ex-
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 per-
(b) view of collapsed bridge superstructure and undulation of formance. Nevertheless, direct comparison of the
road surface (photograph courtesy of Bogazici University, displacement-time response of these two different
Kandilli Observatory and Earthquake Research Institute) classes of structures under nominally identical con-
ditions using the same computational methods re-
Figure 58. Damage to Arifiye overpass bridge in Turkey during
1999 Kocaeli Earthquake.
mains 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
walls, was noted in model tests subjected to simu- more flexible than conventional walls. Thus, they
lated seismic loading (Section 3). may be used in areas where large uneven displace-
In most current design codes, factors of safety ments due to surface faulting during earthquake
against prescribed failure (collapse) mechanisms are events are expected.
evaluated using pseudo-static methods for both con- Figure 58a shows an approach embankment of
ventional and GRS walls. Within this common the Arifiye overpass bridge in Turkey following the
framework the following benefits of using GRS 1999 Kocaeli earthquake. It was constructed as a re-
walls may be realized: inforced-soil wall using metal reinforcement strips.
Although the overpass bridge collapsed completely
a) Design seismic coefficients can be reduced when (Figure 58b), the reinforced-soil wall survived the
designing for GRS walls. earthquake intact and remained in service. The rein-
b) Minimum values for acceptable factors of safety forced-soil wall sustained large permanent deforma-
can be varied depending on the type of wall (e.g. tions mainly due to large differential settlements at
GRS walls or conventional walls) the foundation.
c) Failure mechanisms that do not occur for GRS The good performance of this wall contrasts with
structures can be omitted. For example, starting the behavior of a conventional reinforced concrete
with the design code for railway GRS walls with wall in Taiwan (Figure 59), which collapsed com-
rigid facings in Japan (RTRI 1992), bearing ca- pletely due to large uneven displacements (surface
pacity failure is not considered (except for ex- faulting) during the 1999 Chi-chi earthquake
tremely weak subsoil conditions). This is based (Koseki and Hayano 2000).
on the work of Murata et al. (1994) who showed
6 CONCLUSIONS 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
This paper has reviewed a large body of work fo- Geotechnical Engineering, 1, 267-270.
cused on the seismic performance, analysis and de- Aoki, H., Yonezawa,T., Tateyama, M. Shinoda, M. and Wata-
sign of geosynthetic-reinforced soil (GRS) walls. nabe, K. 2005. Development of aseismic abutment with
The results of field walls that have survived sig- geogrid-reinforced cement-treated backfills, Proc. of 16th
nificant earthquakes and the results of model tests International Conf. on Soil Mechanics and Geotechnical
show that GRS walls perform well during a seismic Engineering, 3, 1315-1318.
Bathurst, R.J. and Cai, Z. 1995. Pseudo-static seismic analysis
event when properly designed. This has led to many of geosynthetic reinforced segmental retaining walls, Geo-
structures remaining in service after major earth- synthetics International, 2(5), 789-832.
quakes while conventional structures have not. Be- Bathurst, R.J. and Alfaro, M.C. 1997. Review of seismic de-
cause of their good performance, GRS structures are sign, analysis and performance of geosynthetic reinforced
being considered more frequently for the replace- walls, slopes and embankments, Earth Reinforcement,
Balkema, 2, 887-918.
ment of conventional structures in reconstruction Bathurst, R.J. 1998. NCMA segmental retaining wall seismic
works. design procedure – supplement to design manual for seg-
New retaining wall technologies that include mental retaining walls (second edition 1997) published by
geosynthetic products are identified in the paper – the National Concrete Masonry Association, Herndon, VA,
e.g. combined soil-cement and geosynthetic rein- 187 p.
forcement layers, and geofoam seismic buffers with Bathurst, R.J. and Hatami, K. 1998. Seismic response analysis
of a geosynthetic reinforced soil retaining wall, Geosynthet-
and without geosynthetic reinforcement layers. ics International, 5(1&2), 127-166.
Finally, the paper has highlighted the move to- Bathurst, R.J., Hatami, K. and Alfaro, M.C. 2002. Geosynthetic
ward limit states design for geotechnical engineering reinforced soil walls and slopes: seismic aspects, Geosyn-
structures. Nevertheless, with the exception of Japan, thetics and Their Applications (S.K. Shukla Ed.), Thomas
the inclusion of seismic design guidelines for GRS Telford, 327-392.
Becker, D.E. 1996. Limit states design for foundations. Part I.
structures within a LSD framework is lagging. An overview of the foundation design process, Canadian
Geotechnical Journal, Vol.33, 956-983.
British Standard BS8006:1995. Code of practice of strength-
ACKNOWLEDGEMENTS ened/reinforced soils and other fills. 162 p.
Cai, Z. and Bathurst, R.J. 1995. Seismic response analysis of
The authors wish to extend special thanks to Prof. geosynthetic reinforced soil segmental retaining walls by
finite element method, Computers and Geotechnics, 17(4),
C.C. Huang (National Cheng Kung University, Tai- 523-546
wan), Mr. H.E. Acosta-Martinez (University of Cai, Z. and Bathurst, R.J. 1996. Seismic-induced permanent
Western Australia), Ms. Y. Tsutsumi (University of displacement of geosynthetic reinforced segmental retain-
Tokyo, Japan), Dr. J. Izawa (Tokyo Institute of ing walls, Canadian Geotechnical J., 33, 937-955.
Technology, Japan), Dr. M. Tateyama, Dr. K. Ko- Canadian Foundation Engineering Manual (CFEM). 2006. Ca-
nadian Geotechnical Society (in press)
jima, Mr. K. Watanabe (Railway Technical Research Canadian Highway Bridge Design Code (CHBDC) CAN/CSA-
Institute, Japan), Mr. Y. Tamura, Dr. M. Shinoda S6-00. 2000. The Canadian Standards Association (CSA In-
(Integrated Geotechnology Institute, Ltd, Japan), Dr. ternational), Toronto, Ontario
T. Fujii (Fukken Co., Ltd, Japan), Prof. M. El-Emam Cascone, E., Maugeri, M. and Motta, E. 1995. Seismic design
(Zagazig University, Egypt), Prof. K. Hatami (Uni- of earth-reinforced embankments, Earthquake Geotechnical
versity of Oklahoma, USA) and, Mr. S. Zarnani Engineering, Balkema, 2, 1129-1134.
Chalermyanont, T. and Benson, C.H. 2004. Reliability-based
(GeoEngineering Centre at Queen’s-RMC, Canada) design for internal stability of mechanically stabilized earth
for their help in preparing the manuscript. walls, Journal of Geotechnical and Geoenvironmental En-
gineering, ASCE, 130(2), 163-173.
Chalermyanont, T. and Benson, C.H. 2005. Reliability-based
REFERENCES design for external stability of mechanically stabilized earth
walls, International Journal of Geomechanics, 5(3), 196-
AASHTO. 2002. Standard specifications for highway bridges, Desai C.S. and El-Hoseiny, K.E. 2005. Prediction of field be-
(17th Edition 2002), American Association of State High-
havior of reinforced soil wall using advanced constitutive
way and Transportation Officials, Washington, DC, USA model, Journal of Geotechnical and Geoenvironmental En-
AASHTO. 2004. LRFD bridge design specifications, SI units gineering, ASCE, 131(6), 729-739.
(3rd Edition), American Association of State Highway and
El-Emam, M. and Bathurst, R.J. 2004. Experimental design,
Transportation Officials, Washington, DC, USA instrumentation and interpretation of reinforced soil wall
AGI. 2005. Geotechnical aspects of the design in seismic area. response using a shaking table, International Journal of
Guidelines. Draft, March 2005. Patron Editore Bologna (in
Physical Modelling in Geotechnics, 4(4), 13-32.
Italian). El-Emam, M., Bathurst, R.J. and Hatami, K. 2004, Numerical
Allen, T.M., Nowak, A.S. and Bathurst, R.J. 2005. Calibration modeling of reinforced soil retaining walls subjected to
to determine load and resistance factors for geotechnical
base acceleration, 13th World Conference on Earthquake
and structural design, Transportation Research Board Cir- Engineering, Vancouver BC, 15.
cular E-C079, Washington, DC, 93 p.
El-Emam, M. and Bathurst, R.J. 2005. Facing contribution to JSCE. 2006. Report on the 2004 Niigata Chuetsu earthquake,
seismic response of reduced-scale reinforced soil walls, Japan Society of Civil Engineers, CD-ROM (in Japanese).
Geosynthetics International, 12(5), 215- 238. Kato, N., Huang, C.C., Tateyama, M., Watanabe, K., Koseki, J.
Eurocode 7 (ENV 1997-1). 1994. Geotechnical design, Part 1: and Tatsuoka, F. 2002. Seismic stability of several types of
General rules, CEN. retaining walls on sand slope, Proc. of 7th International
Eurocode 8 (ENV 1998) Design of structures for earthquake Conference on Geosynthetics, Nice, 1, 237-240.
resistance, CEN. Kitamoto, Y. Abe, H., Shimomura, H., Morishima, H. and Ta-
Eurocode 8. 2003. Design of structures for earthquake resis- niguchi, Y. 2006. Rapid and strengthened repair construc-
tance, Part 5: Foundations, retaining structures and geo- tion for a seriously damaged railway embankment during
technical Aspects. violent earthquakes, this conference.
Fazzino, P. 2005. Influence of earthquake-induced pore pres- Koerner, R., Soong, T.Y. and Koerner, R.M. 1998. Earth re-
sure on the stability condition of earth-.reinforced walls. taining wall costs in USA, Geosynthetic Research Institute
Diploma Thesis, University of Catania (in Italian). Report, Folsom, Pennsylvania, USA.
FHWA. 2001. Mechanically stabilized earth walls and rein- Koseki, J., Tatsuoka, F., Munaf, Y., Tateyama, M. and Kojima,
forced soil slopes: design and construction guidelines, Pub- K. 1998. A modified procedure to evaluate active earth
lication FHWA NHI-00-43, Federal Highway Administra- pressure at high seismic loads, Soils and Foundations, Spe-
tion and National Highway Institute, Washington, DC, cial Issue on Geotechnical Aspects of the January 17 1995
USA Hyogoken-Nambu Earthquake, 2, 209-216.
Fujii, T., Izawa, J., Kuwano, J., Ishihara, M. and Nakane, J. Koseki, J. and Hayano, K. 2000. Preliminary report on damage
2006. Prediction of deformation of retaining walls of geo- to retaining walls caused by the 1999 Chi-Chi earthquake,
synthetic-reinforced soil under large earthquakes, this con- Bulletin of ERS, Institute of Industrial Science, Univ. of
ference. Tokyo, 33, 23-34.
Geoguide 6. 2002. Guide to reinforced fill structure and slope Koseki, J., Tatsuoka, F., Watanabe, K., Tateyama, M., Kojima,
design. Geotechnical Engineering Office, Civil Engineering K. and Munaf, Y. 2003. Model tests on seismic stability of
Department, Hong Kong several types of soil retaining walls, Reinforced Soil Engi-
Greenwood, J.H., Jones, C.J.F.P. and Tatsuoka, F. 2001. Re- neering, Ling, Leshchinsky and Tatsuoka (eds.), Dekker,
sidual strength and its application to design of reinforced 317-358.
soil in seismic areas, Landmarks in Earth Reinforcement, Koseki, J., Kato, N., Watanabe, K. and Tateyama, M. 2004.
Balkema, 1, 37–42. Evaluation of seismic displacement of reinforced walls,
Hatami, K. and Bathurst, R.J. 2000. Effect of structural design Proc. of 3rd Asian Regional Conference on Geosynthetics,
on fundamental frequency of reinforced-soil retaining walls, Seoul, 217-224.
Soil Dynamics and Earthquake Engineering, 19, 137-157. Ling, H.I., Leshchinsky, D and Perry, E.B. 1997. Seismic de-
Huang, C.C. 2000. Investigations of the damaged soil retaining sign and performance of geosynthetic-reinforced soil struc-
structures during the Chi-Chi earthquake, Journal of the tures, Geotechnique, 47(5), 297-316.
Chinese Institute of Engineers, 23(4), 417-428. Ling, H.I. and Leshchinsky, D. 2003. Post-earthquake investi-
Huang, C.C. and Wang, W.C. 2005. Seismic displacement of a gation of several geosynthetic reinforced soil retaining
geosynthetic-reinforced wall in the 1995 Hyogo-ken walls and slopes during Ji-Ji earthquake in Taiwan, Rein-
Nambu earthquake, Soils and Foundations, 45(5), 1-10. forced Soil Engineering, Ling, Leshchinsky and Tatsuoka
Iai, S., Matsunaga, Y., and Kameoka, T. 1992. Strain space (eds.), Dekker, 297-316.
plasticity model for cyclic mobility, Soils and Foundations, Ling, H.I., Mohri, Y., Leshchinsky, D., Burke, C., Matsushima,
32(2), 1-15. K. and Liu, H. 2005. Large-scale shaking table tests on
Inglis, D., Macleod, G., Naesgaard, E. and Zergoun, M. 1996. modular-block reinforced soil retaining walls, Journal of
Basement wall with seismic earth pressures and novel ex- Geotechnical and Geoenvironmental Engineering, ASCE,
panded polystyrene foam buffer layer, Tenth Annual Sym- 131(4), 465-476.
posium of the Vancouver Geotechnical Society, Vancouver, Lo Grasso, A.S., Maugeri, M. and Recalcati, P. 2004. Response
BC, Canada, 18 p. of geosynthetic reinforced soil wall under seismic condition
Ismeik, M. and Güler, E. 1998. Effect of wall facing on the by shaking table tests, 3rd European Geosynthetics Confer-
seismic stability of geosynthetic-reinforced retaining walls, ence, Munich, Germany.
Geosynthetics International, 5(1&2), 41-53. Lo Grasso, A.S. Maugeri, M. and Recalcati, P. 2005. Seismic
ISO 2394:1998. General principles on reliability of structures behaviour of geosyntethic-reinforced slopes with overload
ISO/CD 23469:2004. Bases for design of structures – Seismic by shaking table tests, Geo-Frontiers 2005, ASCE, CD-
actions for designing geotechnical works ROM.
Ito, H., Saito, T., Izawa, J. and Kuwano, J. 2006. In situ con- Lo Grasso, A.S. Maugeri, M. and Recalcati, P. 2006. Experi-
struction of the Geogrid reinforced soil wall combined with mental seismic analysis of geosyntethic-reinforced soil
soil cement, this conference. structures with three-dimensional reinforcements by shak-
Izawa, J., Kuwano, J. and Takahashi A. 2002. Centrifuge tilting ing table tests, this conference.
and shaking table tests on reinforced soil wall, Proc. of 7th Matsuo, O. Tsutsumi, T., Yokoyama, K. and Saito, Y. 1998.
International Conference on Geosynthetics, Nice, 1, 229- Shaking table tests and analyses of geosythetic-reinforced
232. soil retaining walls, Geosynthetics International, 5(1&2),
Izawa, J., Kuwano, J. and Ishihara Y. 2004. Centrifuge tilting 97-126.
and shaking table tests on the RSW with different soils, Miyata, Y., Shigehisa, S. and Kogure, K. 2001. Reliability
Proc. of 3rd Asian Regional Conference on Geosynthetics, analysis of geosynthetics reinforced soil wall, Landmarks in
Seoul, 803-810. Earth Reinforcement, Balkema, 1, 417–420.
JGS. 2004. Principles for foundation designs grounded on a Motta, E. 1996. Earth pressure on reinforced earth-walls under
performance-based design concept, JGS 4001-2004, Japa- general loading, Soils and Foundations, 36(4), 113-117.
nese Geotechnical Society. Murata, O. 2003. An outlook on recent research and develop-
JSCE. 2003. Principles, guidelines and terminologies for struc- ment concerning long-term performance and extreme load-
tural design code drafting founded on the performance ing, Landmarks in Earth Reinforcement, Balkema, 2, 869-
based design concept (code PLATFORM), Japan Society 893.
for Civil Engineers (in Japanese).
Murata, O., Tateyama, M. and Tatsuoka, F. 1994. Shaking ta- Tatsuoka, F., Koseki, J. and Tateyama, M. 1997b. Performance
ble tests on a large geosynthetic-reinforced soil retaining of reinforced soil structures during the 1995 Hyogo-ken
wall model, Recent Case Histories of Permanent Geosyn- Nanbu Earthquake, Earth Reinforcement, Balkema, 2, 973-
thetic-Reinforced Soil Retaining Walls, Balkema, 869-893. 1008.
NCMA. 1997. Design Manual for Segmental Retaining Walls, Tatsuoka, F., Koseki, J., Tateyama, M., Munaf, Y. and Horii, N.
Second Edition, Second Printing, National Concrete Ma- 1998., Seismic stability against high seismic loads of geo-
sonry Association, Herndon, Virginia. synthetic-reinforced soil retaining structures”, Proc. 6th Int.
Newmark, N.M. 1965. Effects of earthquakes on dams and Conf. on Geosynthetics, Atlanta, 1, 103-142.
embankments, Geotechnique, 15(2), 139-159. Tatsuoka, F. 2003. Technical report -wall structures session,
NRC. 1995. National building code. NRC of Canada, Ottawa, Landmarks in Earth Reinforcement, Balkema, Vol.2, 937-
Ontario, Canada. 948.
Orr, T.L.L. 2002. Eurocode 7 - a code for harmonized geotech- Tsukamoto, Y., Ishihara, K., Kon, H. and Masuo, T. 2002. Use
nical design, Foundation Design Codes and Soil Investiga- of compressible expanded polystyrene blocks and geogrids
tion in view of International Harmonization and Perform- for retaining wall structures, Soils and Foundations, 42(4),
ance Based Design, Honjo et al. (eds), Balkema, 3-15. 29-41.
PWRC. 2000. Design and construction manual for geotextile Wakai, A., Amano, M., Iizuka, Y. and Ugai, K. 2006. Numeri-
reinforced soil structures, Publics Works Research Center, cal analysis for seismic evaluation of retaining walls con-
Japan (in Japanese). sisting of concrete panels and backfill reinforced with
Race, R. and del Cid, H. 2001. Seismic performance of modu- geogrids, Journal of Geotechnical Engineering, Japan So-
lar block retaining wall structures during the January 2001 ciety of Civil Engineers, 816/III-74, 157-168 (in Japanese).
El Salvador earthquake, International Geosynthetics Engi- Watanabe, K., Munaf, Y., Koseki, J., Tateyama, M. and Ko-
neering Forum 2001, Taipei, Taiwan (CD-ROM), 20 p. jima, K. 2003. Behaviors of several types of model retain-
RTA. 2005. Design of reinforced soil walls, QA specification ing walls subjected to irregular excitation, Soils and Foun-
R57, Roads and Traffic Authority, New South Wales, Aus- dations, 43(5), 13-27.
tralia. White, D.M. and Holtz, R.D. 1997. Performance of geosyn-
RTRI. 1992. Design standard for railway earth structures, thetic-reinforced slopes and walls during the Northridge,
Railway Technical Research Institute, Japan (in Japanese). California Earthquake of January 17, 1994, Earth Rein-
RTRI. 1999. Seismic design standard for railway structures, forcement, Balkema, 2, 965-972.
Railway Technical Research Institute, Japan (in Japanese). Yogendrakumar, M. and Bathurst, R.J. 1992. Numerical simu-
RTRI. 2006. Design standard for railway earth structures, lation of reinforced soil structures during blast loadings,
Railway Technical Research Institute, Japan (in Japanese). Transportation Research Record, 1336, 1-8.
Saito, T., Ito, H., Izawa, J. and Kuwano, J. 2006. Seismic sta- Zarnani, S., Bathurst, R.J. and Gaskin, A. 2005. Experimental
bility of the geogrid-reinforced soil wall combined with soil investigation of geofoam seismic buffers using a shaking
cement, this conference. table, 2005 North American Geosynthetics Society Confer-
Sandri, D. 1994. Retaining walls stand up to the Northridge ence, Las Vegas, Nevada, 14-16 December 2005, 11 p.
earthquake. Geotechnical Fabrics Report, IFAI, St. Paul, Zornberg, J.G. and Leshchinsky, D. 2003. Comparison of in-
MN, USA, 12(4), 30-31 (and personal communication). ternational design criteria for geosynthetic-reinforced soil
Shinoda, M., Yonezawa, T., Tateyama, M. and Koseki, J. structures, Landmarks in Earth Reinforcement, Balkema, 2,
2005: Limit state exceedance probability of reinforced soil 1095-1106.
retaining walls, Journal of the Japan Society of Civil Engi-
neers, No. 792/III-71, 119–129 (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 confer-
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
Stewart, J.P., Bray, J.D., Seed, R.B. and Sitar, N. 1994. Pre-
liminary 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 geo-
synthetic-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 Me-
chanics 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 Hyogoken-
Nambu 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, Geo-
synthetics International, 4(2), 81-136.