A number of conventional masonry and unreinforced concrete gravity-type retaining
walls for railway embankments were seriously damaged by the 1995 Hyogoken-Nanbu
earthquake in Japan. Many modern cantilever-type, reinforced concrete (RC) retaining
walls were also damaged, while geogrid-reinforced soil retaining walls with a full
height RC facing performed well during the earthquake (Tatsuoka et al. 1996a,b).
Pseudo-static limit equilibrium-based stability analyses that follow one of the current
aseismic design methods in Japan were conducted by the authors of the current paper
on five damaged retaining walls (Koseki et al. 1996). It was found that the critical seismic
acceleration coefficient, kh-cr , yielding a factor of safety of unity against external
instability (i.e. sliding, overturning, or bearing capacity failure) was approximately 40
to 80% of the estimated peak horizontal ground acceleration, PHGA, divided by the gravitational
acceleration, g, except for one cantilever-type RC retaining wall. This cantilever-type
wall was extremely unstable prior to the earthquake due to the existence of
an additional upper embankment on the crest behind the wall. Since the values of
kh-cr/(PHGA/g) were almost equal irrespective of the extent of damage (i.e. among the
severely damaged gravity-type retaining walls, the moderately damaged cantilevertype
RC retaining wall, and the slightly damaged geogrid-reinforced soil retaining
wall), it was inferred that the current aseismic design method for these different types
of retaining walls are inadequate. Therefore, it is required to compare the seismic performance
of different types of retaining walls to establish consistent aseismic design
methods.
In Japan, geosynthetic-reinforced soil retaining walls with a full-height RC facing
have been constructed to a total length exceeding 26 km as important permanent structures
mainly for railways (Tatsuoka et al. 1997). Many of these walls were aseismically
designed by using a pseudo-static, limit equilibrium-based stability analysis as described
by Horii et al. (1994). Use of these walls will be further promoted when their
ductile behavior against earthquake loads, as discussed by Tatsuoka et al. (1996a), is
rationally evaluated and taken into account in the aseismic design procedure.
Shaking table tests on small-scale models of geosynthetic-reinforced soil retaining
walls were conducted by several other researchers as summarized by Bathurst and Alfaro
(1996). However, comparisons of their seismic behavior to that of other types of retaining
walls is scarce; Sakaguchi (1996) compared the dynamic stability of a geogridreinforced
soil retaining model wall having wrapped-around facing with that of model
conventional-type (gravity-type, leaning-type, and cantilever-type) retaining walls.
With respect to the seismic behavior of geosynthetic-reinforced soil retaining walls
with a full-height rigid facing, Murata et al. (1992, 1994) conducted a series of shaking
table tests primarily to investigate the effect of facing rigidity on the resistance capacity
against earthquake load. However, no comparison was made with other types of retaining
walls. It should also be noted that tilt table tests have not been conducted on model
geosynthetic-reinforced soil retaining walls, even though this type of test can simulate
pseudo-static loading conditions which are assumed in most of the current aseismic design
methods based on the limit equilibrium stability analysis. Furthermore, it can be
expected that the comparison of model behavior between shaking table tests and tilt
table tests reveals the dynamic effects on wall stability and, it is hoped, any inconsistency
between the pseudo-static analysis and the actual seismic behavior. Considering the above situation, a series of shaking table tests was performed on relatively
small-scale models of a geosynthetic-reinforced soil retaining wall with a fullheight
rigid facing and three types of conventional-type retaining walls (gravity-type,
leaning-type, and cantilever-type). Tilt table tests were also conducted on models of a
geosynthetic-reinforced soil retaining wall and the leaning-type wall.
The current paper describes the critical accelerations/tilting angles and the angles of
the failure plane in the backfill layers that were observed during the experiments. Analyses
based on the recorded earth pressures, wall displacements, and response accelerations
will be reported elsewhere. These data may reveal important features with respect
to the mechanism of mobilization of seismic earth pressure, when comparing the ductile
behavior of different types of retaining walls, and when evaluating the dynamic effects
which include amplification and phase difference.