Modelling earthquake-triggered landslide runout with the material point method

Abstract

Landslides triggered by earthquakes cause devastating consequences to downstream infrastructure. The simulation and prediction of these large-strain events remain challenging. The objectives of this work were to validate the material point method (MPM) framework for the study of coseismic landslides and to compare the capabilities of the MPM with mesh-based methods and simplified Newmark-type methods to simulate post-failure runouts. To achieve these objectives, the MPM framework was used whereby a nodal kinematic boundary condition was employed with a moving mesh. The framework was validated with a shaking table laboratory test of a saturated clay slope. A parametric analysis was then conducted using 25 real ground motions on a simple theoretical slope. The MPM results were compared with those obtained with mesh-based methods and three state-of-the-art Newmark-type approaches. The mesh-based methods were consistent with MPM predictions for small-strain instabilities associated with low-energy ground motions (i.e. Arias intensity lower than 4 m/s). When using ground motions with energy above this threshold, mesh-based methods accumulate significant errors associated with bad geometry. The MPM results consistently matched permanent displacements predicted with the Newmark-type methods employed in this analysis.