Reliability analysis and reliability-based design optimization using moment methods Sang Hoon Lee, Northwestern University, Evanston, IL, USA Byung Man Kwak, Korea Advanced Institute of Science and Technology, Daejeon, Korea Jae Sung Huh, Korea Aerospace Research Institute, Daejeon, Korea

Abstract

Reliability analysis methods using the design of experiments (DOE) are introduced and integrated into reliability based design optimization (RBDO) frame work with a semi-analytic design sensitivity analysis (DSA) for the reliability measure. A procedure using the full factorial DOE with optimal levels and weighs is introduced and named as full factorial moment method (FFMM) for reliability analysis. The probability of failure is calculated using an empirical distribution system and the first four statistical moments of system performance function calculated from DOE. To enhance the efficiency of FFMM, a response surface augmented moment method (RSMM) is developed to construct a series of approximate response surface approaching to that of FFMM. A semi-analytic design sensitivity analysis for the probability of failure is proposed in combination with FFMM and RSMM. It is shown that the proposed methods are accurate and effective especially when the inputs are non-normal.