COMPUTATIONAL METHODS FOR EFFICIENT STRUCTURAL RELIABILITY AND RELIABILITY SENSITIVITY ANALYSIS

This paper proposes an efficient, adaptive importance sampling (AIS) method that can be used to compute component and system reliability and reliability sensitivities. The AIS approach uses a sampling density that is proportional to the joint probability density function of the random variables. Starting from an initial approximate failure domain, sampling proceeds adaptively and incrementally to reach a sampling domain that is slightly greater than the failure domain to minimize oversampling in the safe region. Several reliability sensitivity coefficients are proposed that can be computed directly and easily from the previous AIS-based failure points. These sensitivities can be used to identify key random variables and to adjust a design to achieve reliability-based objectives. The proposed methodology is demonstrated using a turbine blade reliability analysis problem.