Multimodal Simulation Method for System Reliability Analysis

Abstract

In system reliability analysis, it is frequently necessary to resort to Monte Carlo simulation (MCS). The importance sampling method (ISM) is an advanced MCS method that may greatly improve the efficiency and accuracy of simulation approaches. Its application to structural system reliability analysis is focused here. Importance of proper determination of the sampling density is emphasized based on a critical review of other suggested sampling densities. A new sampling density, the weighted general normal sampling density (WGNSD), is proposed. This density is proportional to the ideally optimal one at the most important points of the variable space, and it is general for structural system reliability analysis. A variety of applications of this method are presented for illustration, as well as for demonstration of success with the proposed sampling density, including nondifferentiable failure surfaces and higher‐dimensional problems.