Developing an improved composite limit state method for time-dependent reliability analysis

Abstract

Time-dependent reliability analysis remains challenging in practical engineering applications, since the modeling and solution are complicated, resulting in the huge computational burden. The composite limit state (CLS) method can efficiently reduce the computational cost to a certain extent. However, the computational cost is still high. In this article, an improved composite limit state (ICLS) method is proposed for efficient time-dependent reliability analysis, which is based on the original CLS. We propose two enhancements to improve the efficiency of the ICLS approach. First, different from the original method, Kriging surrogate models at all time nodes and one CLS Kriging surrogate model are established in the ICLS method, respectively. In that case, once a new sample is selected, there is no need to evaluate the true limit state function values on the new sample at each time node, leading to a reduction in computational cost. Next, the efficiency is further improved by updating the CLS Kriging surrogate model and the surrogate model at one selected time node simultaneously. Four examples are utilized to demonstrate the efficiency and accuracy of the ICLS method for time-dependent reliability analysis.