Innovative Approaches to Sensitivity Indices for Structural Reliability

This article investigates the development of novel sensitivity indices for global sensitivity analysis with a focus on structural reliability. Through global sensitivity analysis, the reliability of structural steel components is assessed, contrasting two methodologies: one predicated on conditional mean values and the other on the conditional median of the reliability function. The research centers on the influence of input random variables on the ultimate limit state of a steel beam. Sensitivity indices are calculated via a double - nested - loop simulation utilizing the Latin Hypercube Sampling method. Analytical test cases and tailored sampling strategies are implemented to ensure rapid response times and advanced estimation of sensitivity indices. The comparison elucidates the strengths and weaknesses of various global sensitivity indices, enhancing the reliability analysis of structures. This comparative study underscores global sensitivity analysis as a valuable tool for pinpointing critical input variables that significantly affect the reliability and limit states of structures, thereby advancing the field of structural reliability analysis.