751
3D Dynamic Numerical Modeling for Soil-Pile-Structure Interaction in Centrifuge
Tests
Modélisation numérique dynamique en 3D de l'interaction sol-pieu en centrifugeuse
Kwon S.-Y., Kim M.-M.
Seoul National University, Seoul, Republic of Korea
Kim S.g-H.
Ministry of Land, Transfer and Maritime Affairs, Seoul, Republic of Korea
Choi J.-I
University of California, Los Angeles, Los Angeles, CA, USA
ABSTRACT: 3D dynamic analysis based on the finite difference method was performed, to simulate the dynamic behavior of a soil-
pile-structure system under seismic loading, which was observed from dynamic centrifuge tests. For the centrifuge tests, model piles
were placed in a so-called equivalent shear beam box. The acceleration time histories of 12 sine waves and 10 scaled earthquake
events were used as input motions. The 3D numerical modeling was formulated in a time domain, to effectively simulate the
nonlinear behavior of soil. As a modeling methodology, the soil medium was divided into near field and far field, the latter of which
was not affected by soil-pile-structure interaction. The mesh was created only for the near field, to reduce the computing time. Far
field response was applied as a boundary condition at the boundary of the near field. Soil nonlinearity was considered by adopting a
hysteretic damping model and an interface model, which can simulate separation and slip between the soil and pile. The 3D modeling
method was calibrated, by comparing the numerical modeling result with that of the dynamic centrifuge test. Finally, the 3D modeling
method established in this research was evaluated, by comparing the numerical modeling results with those of other centrifuge tests.
RÉSUMÉ : Une analyse 3D dynamique, fondée sur la méthode des différences finies, a été réalisée afin de simuler le comportement
dynamique du système sol-pieu-structure sous chargement sismique qui a été observé d’essais dynamiques en centrifugeuse. Pour les
essais en centrifugeuse, le pieu modèle était placé dans une « boîte de cisaillement pour poutre ». Les accélérogrammes de 12 vagues
sinusoïdales et de 10 tremblements de terre ont été utilisés comme données de calcul. La modélisation numérique 3D a été formulée
dans un domaine temporel pour simuler efficacement le comportement non linéaire du sol. Comme méthode de modélisation, le sol a
été divisé en champ proche et en champ lointain, ce dernier n’étant pas affecté par l’interaction sol-pieu-structure. Le maillage a été
créé seulement pour le champ proche afin de réduire le temps de calcul. La réponse du champ lointain a été appliquée comme
condition à la limite du champ proche. La non-linéarité du sol a été considérée en adoptant un modèle d’amortissement hystérétique et
un modèle d’interface qui peut simuler la séparation et le glissement entre le sol et le pieu. La méthode de modélisation 3D a été
calibrée en comparant les résultats de la modélisation numérique avec les essais dynamiques en centrifugeuse. Enfin, la méthode de
modélisation 3D établie dans cette recherche a été évaluée en comparant les résultats de la modélisation numérique avec les résultats
d’autres essais en centrifugeuse.
KEYWORDS: Centrifuge tests, Numerical analysis, Finite difference method, Dynamic soil-pile interaction
1 INTRODUCTION
Prediction of the behavior of pile foundations under strong
earthquake loading is very important. Recently, the design
procedure for evaluating pile behavior under strong earthquake
loading has been modified, particularly after a series of mega-
earthquakes, such as the Great East Japan (3/11) Earthquake.
However, dynamic analysis of the soil-pile system is a very
complicated procedure, different from the static case, and is
affected by many factors, such as soil nonlinearity and dynamic
soil-pile interaction.
Many researchers have investigated the soil-pile-structure
interaction (SPSI) effect on pile foundations (e.g. Kaynia and
Kausel 1982, Dobry and Gazetas 1988, Markis and Gazetas
1992, Klar and Frydman 2002, Martin and Chen 2005) in the
frequency domain. However, analysis in the frequency domain
is not straightforward, and requires the Fourier transformation
for application. In addition, it is difficult to consider the
nonlinearity of soil in this analysis. Therefore, seismic analysis
in the time domain is an effective procedure to consider the
nonlinearity that occurs during strong earthquake motion. 3D
continuum modeling of the soil-pile system is one of the most
accurate techniques among dynamic analysis methods, although
this is difficult to apply, due to its complexity and lengthy
analysis time.
In this study, three-dimensional continuum modeling of the
soil-pile system using the FDM (Finite Difference Method)
program, FLAC-3D, was performed. The seismic responses of
soil-pile-structure observed in centrifuge tests were compared,
to calibrate and validate the applied 3D nonlinear FDM analysis.
The analysis mainly focused on internal responses of the pile,
such as peak bending moment of the pile foundation.
2 CENTRIFUGE TESTS
Dynamic centrifuge model tests(Yoo et al. 2012) were
performed at a condition of 40g centrifugal acceleration, using
the KOCED centrifuge at KAIST (the Korea Advanced Institute
of Science and Technology). The tests evaluate the dynamic
behavior of piles embedded in a model soil, which consists of a
dry-dense sand layer. A typical test layout is presented in Fig. 1.
The model soil consists of Jumunjin sand, with a relative
density of 80%. The model container used for the centrifuge
tests reported in this paper is the ESB (equivalent shear beam)
box. The internal dimension of the ESB box is 50cm x 50cm x
65cm. Model piles with a concentrated mass of 1.4kg were
made with aluminum pipes, and fixed at the bottom of the ESB
box, in order to simulate a rock-socketed pile. Strain gauges
were attached on both sides of the pile according to depth, in
