RELIABILITY-BASED DESIGN OPTIMIZATION USING CONFIDENCE-BASED MODEL VALIDATION FOR INSUFFICIENT EXPERIMENTAL DATA

The conventional reliability-based design optimization (RBDO) methods assume that a simulation model is able to represent the real physics accurately. However, the simulation model could be biased. Accordingly, when the conventional RBDO design is manufactured, the product may not satisfy the target reliability. Therefore, model validation, which corrects model bias, should be integrated in the RBDO process by incorporating experimental data. The challenge is that only a limited number of experimental data is usually available due to the cost of actual product testing. Consequently, model validation for RBDO needs to account for the uncertainty induced by insufficient experimental data as well as variability inherently existing in the products. In this paper, a confidence-based model validation process that captures the uncertainty and corrects model bias at user-specified target conservativeness level is developed. Thus, RBDO can be performed using confidence-based model validation to obtain conservative RBDO design. It is found that RBDO with model bias correction becomes a moving-target problem because the feasible domain changes as the design moves. Consequently, the RBDO optimum may not be easily found due to the convergence problem. To resolve the issue, an efficient process is proposed by carrying out deterministic design optimization (DDO) and RBDO without validation, followed by RBDO with confidence-based model validation. Finally, we demonstrate that the proposed RBDO approach can achieve a conservative and practical optimum design given a limited number of experimental data.