Abstract
The paper proposes a precise technique to compute the limit state exceedance probability of earth slopes using a low-discrepancy sequence Monte Carlo (LDSMC) method and an importance sampling with a low discrepancy sequence Monte Carlo (ISLDSMC) method under uncertainties of backfill and clay properties having normal distributions. To increase the computational efficiency, the low-discrepancy sequence (LDS) is adopted in this study, which is one of the quasi-random numbers having a uniform distribution. Estimation of the design points near the failure region is essential to apply the importance sampling algorithm. We found that the recently developed subset Markov chain Monte Carlo (MCMC) method is useful to find multiple design points although a first-order reliability method (FORM) is applicable for simple cases. The numerical simulation in this study revealed that the LDSMC and ISLDSMC method can effectively compute the accurate limit state exceedance probability of earth slopes with a small number of simulations by comparing with the conventionally used crude Monte Carlo simulation.
