Model Tests of Seismic Stability of Several Types of Soil Retaining Walls - Reinforced Soil Engineering: Advances in Research and Practice - page 309

resist against the tilting displacement. Comparisons are also made on the resultant
force of normal earth pressures and the critical seismic coefficient at the ultimate
overall wall failure condition.
1 INTRODUCTION
The 1995 Hyogoken-Nambu earthquake caused serious damage to a number of
soil retaining walls (RWs) for railway embankments, as reported by Tatsuoka
et al. (1996). Based on field investigations and back analyses on the
performance of the damaged RWs, Koseki et al. (1996, 1999) showed that
there is a large difference between the seismic coefficients (k h ) design used in the
current design for RWs and the ratios of the highest peak horizontal ground
accelerations to the gravitational acceleration estimated at the damaged RWs.
Tatsuoka et al. (1998) argue that some factors for the above difference
include (1) the use of conservative soil strength in the design, (2) positive aspects
of dynamic effects arising from the ductility and flexibility of RWs that are not
considered in the pseudo-static approaches, and (3) the use of a global safety factor
larger than unity. Because of the above factors, it is difficult to accurately predict
the stability or performance of RWs during such a severe seismic event as the 1995
Hyogoken-Nambu earthquake when following the current seismic design pro-
cedures. It is suggested that the currently used (k h ) design -values should be increased
appropriately to avoid such collapse of RWs as observed during the earthquake. At
the same time, it is also suggested to increase the (k h ) design -value to a larger extent
in the order of (1) gravity type RWs, (2) cantilever reinforced concrete RWs, and
(3) geosynthetic reinforced soil RWs having a full-height rigid facing.
Many researchers, including Ichihara and Matsuzawa (1973), Sakaguchi
(1996), and Matsuo et al. (1998), conducted model tests to study seismic behavior
of RWs. Based on the results from these investigations, several different methods
have been proposed to predict the stability of RWs during earthquakes for both
conventional RWs and geosynthethic reinforced soil-type RWs. In engineering
practice, the limit equilibrium method by the pseudo-static approach is the most
widely used to analyze the seismic stability of RWs (e.g., Seed and Whitman,
1970; Bathurst and Alfaro, 1996; Ling et al., 1997; RTRI, 1997).
The different seismic performances of different types of RWs have not yet
been fully investigated, particularly experimentally. The possible limitations of
the pseudo-static approach have not yet been fully understood, either. In the
present study, therefore, a series of relatively small-scale model shaking tests was
conducted on the different types of RWs to observe their different performances
during irregular shaking. They were compared with the behaviors observed