An efficient importance sampling approach for reliability analysis of time-variant structures subject to time-dependent stochastic load

• Published in: Mechanical systems and signal processing Vol. 159; p. 107699
• Main Authors: Yuan, Xiukai, Liu, Shaolong, Faes, Matthias, Valdebenito, Marcos.A., Beer, Michael
• Format: Journal Article
• Language: English
• Published: Berlin Elsevier Ltd 01.10.2021; Elsevier BV
• Subjects: Composite limit state functions; Cumulative failure probability; Deterioration; Importance sampling; Limit states; Parameters; Random variables; Reliability analysis; Reliability engineering; Simulation-based method; Stochastic load; Stochastic processes; Structural reliability; Time dependence; Time-variant structure; Time-variant structure; Stochastic load; Cumulative failure probability; Composite limit state functions; Importance sampling; Simulation-based method
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2021.107699

•The time-variant reliability problem is transformed into a series system.•An efficient two-step importance sampling function is proposed.•The method only entails a single straightforward simulation without optimization or linearization with respect to parameters.•Examples with explicit and implicit limit state functions are presented. Structural performance is affected by deterioration processes and external loads. Both effects may change over time, posing a challenge for conducting reliability analysis. In such context, this contribution aims at assessing the reliability of structures where some of its parameters are modeled as random variables, possibly including deterioration processes, and which are subjected to stochastic load processes. The approach is developed within the framework of importance sampling and it is based on the concept of composite limit states, where the time-dependent reliability problem is transformed into a series system with multiple performance functions. Then, an efficient two-step importance sampling density function is proposed, which splits time-invariant parameters (random variables) from the time-variant ones (stochastic processes). This importance sampling scheme is geared towards a particular class of problems, where the performance of the structural system exhibits a linear dependency with respect to the stochastic load for fixed time. This allows calculating the reliability associated with the series system most efficiently. Practical examples illustrate the performance of the proposed approach.