1407
Analytical study of seismic slope behavior in a large-scale shaking table model test
using FEM and MPM
Étude analytique du comportement des pentes sismiques dans les essais de modèles de grandes
dimensions sur table à secousses conformément aux méthodes FEM et MPM
Abe K., Izawa J.
Railway Technical Research Institute, Tokyo, Japan
Nakamura H.
Japan Nuclear Energy Safety Organization, Tokyo, Japan
Kawai T.
Tohoku University, Sendai, Japan
Nakamura S.
Nihon University, Fukushima, Japan
ABSTRACT: Seismic safety of slopes is generally estimated through stability analysis using a simplified conservative approach or
Finite Element Method (FEM) analysis approach, focusing on the conditions of slopes before slope failure. However, it is also
important to understand the conditions of slopes after slope failure. Accordingly, the authors carried out a series of shaking table tests
with large-scale slope models. The models consist of a model with weak layer and a model which has high response acceleration at
top of the slope. Results from the tests indicated that it is important to consider large deformation along slip lines in the weak layer
and amplification and phase lag of response acceleration at the top. The one layered slope model was analyzed by FEM with a
nonlinear model as a constitutive law focusing on the amplification and phase lag of response acceleration at top of model. Both slope
models were analyzed by Material Point Method (MPM). Consequently, the same trend of the amplification and phase lag of response
acceleration and failure patterns as that seen at the shaking table test was obtained by the FEM or MPM.
RÉSUMÉ : La sécurité sismique des pentes est en général calculée par une analyse de stabilité selon une approche conservative
simplifiée du type méthode de Fellenius ou Méthode d’analyse par éléments finis (FEM), l’accent étant mis sur l’état des pentes avant
leur défaillance. Il est toutefois important, en termes de stabilité sismique des pentes, de connaître l’état des pentes après leur
défaillance. À cet effet, les auteurs ont procédé à une série d’essais de modèle de pentes de grandes dimensions sur table à secousses.
Les modèles consistaient en un modèle à couche faible (modèles à trois couches) et un modèle à accélération de réponse élevée au
sommet de la pente (modèle à une couche). Le résultat des essais montre que, dans l’estimation de la sécurité des pentes, il est
important de prendre en compte des déformations importantes le long des lignes de glissement dans la couche faible et l’amplification
et le déphasage de l’accélération de la réponse en sommet de pente. Les deux modèles de pente ont été analysés par la Méthode aux
Points Matériels (MPM). De fait, les mêmes tendances d’amplification et de déphasage de l’accélération de réponse et les mêmes
schémas de défaillance observés pour les essais sur la table à secousses ont été obtenus par la méthode FEM ou MPM.
KEYWORDS: slope model, shaking table test, FEM, MPM
1 INTRODUCTION.
In terms of seismic safety of nuclear power plants in Japan, it is
emphasized that we carefully have to consider the safety of the
nuclear power plants against slope failure as well as tsunamis
which are related incidents during earthquake. Seismic safety of
slopes is generally analyzed through stability analysis using a
simplified conservative approach such as the Fellenius method,
or the Finite Element Method (FEM) analysis approach,
focusing on the conditions of slopes before slope failure.
However, it is also important to understand the conditions of
slopes during earthquake and after slope failure. Accordingly, a
series of shaking table tests with large-scale slope models was
carried out by the authors at the world’s largest shaking table,
nicknamed “E-Defense” of NIED in Hyogo, Japan. This paper
presents results from analytical study of seismic slope behaviour
in the shaking table model test by FEM and Material Point
Method (MPM). Also, it is stated that the validity of these
methods on seismic behaviour of slopes and several issues to be
solved in future work toward developing analytical tools for
assessing the seismic safety of slopes.
2 OUTLINE OF THE SHAKING TABLE TEST
Details of the shaking table test are shown in Shinoda et al.
(2013). In this chapter, outline of the test is stated. Figure 1
shows the initial states of the test models. The test models
consisted of a slope model which had high amplitude of
response acceleration at the top of slope and a slope model with
a lower-strength cohesive sand layer. The former was called one
layered model because the model mostly consisted of one kind
of layer (general part). The model also had reinforced part at left
hand side slope consisting of geotextile, cohesive sandy soil and
sandbags in order to produce slope failure at only right hand
side slope. The latter was called three layered model because
the model consisted of three different layers: surface layer,
weak layer, which represents the lower-strength cohesive sand
layer, and base layer. The gradients and widths of the weak
layer were 45 degrees and 400 mm, respectively. The both
models were made with heights of approximately 3.0 m.
Table 1 shows the slope model soil properties as determined
from tri-axial compression tests. The base layer was made of
dense crushed stone stabilized with cement. The general part
and weak layer were made of siliceous sand mixed with
bentonite. The surface layer for three layered model was
reinforced by geotextile. To prevent sliding at the divisions
between layers of three layered model, the boundaries were
bench-cut as shown in Figure 1. Teflon-lined sheet were also
left between the slope models and the acrylic sidewalls of the
shaking table container to ensure that there were no frictional
forces between the models and the walls. The response of the
