Effect of random field element size on reliability and risk assessment of soil slopes

Slope reliability analyses considering spatially variable soil properties generally necessitate a discretization of a slope region into disjoint zones called random field elements (RFEs). Few efforts have been devoted to investigating the effect of RFE size on the assessment of reliabilities and risks of slopes possibly due to the huge computational burdens associated with a small RFE size. This study systematically investigates the influence of RFE size on statistics of factor of safety (FS) and sliding mass (SM), distribution of critical slip surfaces, reliability, and risk of slopes. As the size of RFE is small, a stepwise covariance matrix decomposition method (CMDM) is used to effectively simulate the random field of the shear strength parameter. This stepwise CMDM is justified using a simple and straightforward method. Monte Carlo simulation (MCS) is employed to estimate the probability of failure and risk of slopes. Three cohesive slopes are adopted to investigate the effect of RFE size on the statistics of FS and SM, distribution of critical slip surfaces, system reliability, and risk of the slope. It is shown that the RFE size has a relatively larger influence on the distribution of SM than that of FS. A relatively larger size of RFE tends to overestimate the slope failure probability and risk. The critical size of RFE required for an accurate evaluation of the statistics of FS and SM is smaller than that required for the evaluation of failure probability and risk.