Reliability Analysis of Sensitive Clay Slope with the Response Surface Method

Abstract

The response surface method (RSM) is increasingly useful for reliability analysis of slopes, because it can integrate numerical slope stability evaluation and reliability analysis for the class of problems where each numerical simulation of slope stability is time-consuming and numerically intensive. The RSM approach is useful for constructing global approximations of the response of, for example, a soil slope under loading. A reliability analysis of a sensitive clay slope was carried out using the commercially available program PLAXIS with the RSM. Monte Carlo simulations (MCS) were done using the response surfaces established by the RSM to estimate the probability of slope failure. The results of the reliability analyses with response surface were compared with direct Monte Carlo simulations (MCS) and First Order Second Moment (FOSM) using the full PLAXIS model. The comparison indicate that the combined RSM-MCS reliability analysis leads to more accurate predictions of the slope stability than the FOSM approach and provides practically the same results as the more time-consuming MCS with the full PLAXIS model. The approach has the advantage of being computationally efficient (fast-running) when a large number of MCSs are needed.