Abstract
Finite element simulations of two centrifuge tests on the same cantilever retaining wall model holding liquefiable backfill were conducted using the Biot formulation‐based program DIANA–SWANDYNE II. To demonstrate the effects due to different pore fluids in seismic centrifuge experiments, water was used as the pore fluid in one experiment whereas a substitute pore fluid was used in the second experiment. The cantilever wall model parameters were determined by comparing simulations with measurements from free‐vibration tests performed on the model wall without backfill. The initial stress conditions for dynamic analysis for the soil backfill were obtained by simulating static loads on the retaining wall from the soil backfill. Level‐ground centrifuge model results were used to select the parameters of the Pastor–Zienkiewicz mark III constitutive model used in the dynamic simulations of the soil. The effects due to different pore fluids were captured well by the simulations. The magnitudes of excess pore pressures in the soil, lateral thrust and its line of action on the wall, and wall bending strains, deflections, and accelerations were predicted well. Predictions of settlements and accelerations in the backfill were less satisfactory. Relatively high levels of Rayleigh damping were needed to be used in the retaining wall simulations in order to obtain numerically stable results, which is one of the shortcomings of the model. The procedure may be used for engineering purpose dealing with seismic analysis of flexible retaining walls where lateral pressures, bending strains and deflections in the wall are typically of importance. Copyright © 2008 John Wiley & Sons, Ltd.