Seismic failure probability analysis of slopes via stochastic material point method

Abstract

Earthquakes, an important factor in inducing landslides, usually have great randomness. Moreover, landslides are investigated as large deformations, and traditional grid-based methods encounter mesh distortions when dealing with such problems. In this context, the spectral expression method and random functions are introduced to generate stochastic ground motions, and the material point method is adopted to perform probabilistic slope analysis. First, the dry aluminum bar collapse test and the Chiu-fen-erh-shan landslide were simulated to verify the effectiveness of this improved procedure. Then, three typical stochastic ground motions were selected to investigate the effects of material parameters (internal friction angle and cohesion) and slope geometry (slope inclination and height) on the sliding distance of the slope. Finally, the uncertainty of ground motions was taken into account when conducting the probabilistic analysis of earthquake-induced landslides. The results demonstrate that the sliding distance of slopes varies with earthquake action, exhibiting certain randomness, particularly in strong earthquakes. In addition, slope geometry influences sliding distances more than material parameters, and slope geometry should be the prior consideration for slope prevention and mitigation.