Uncertainty-based Multidisciplinary Design Optimization using An Approximated Second-Order Reliability Analysis Strategy

Abstract

In uncertainty-based multidisciplinary design optimization (UBMDO), all reliability limitation factors are maintained due to minimize the cost target function. There are many reliability evaluation methods for reliability limitation factors. The second-order reliability method (SORM) is a powerful most possible point (MPP)-based method. It can provide an accurate estimation of the failure probability of a highly nonlinear limit state function despite its large curvature. But the Hessian calculation is necessary in SORM, which results in a heavy computational cost. Recently, an efficient approximated second-order reliability method (ASORM) is proposed. The ASORM uses a quasi-Newton method to close to Hessian without the direct calculation of Hessian. To further improve the UBMDO efficiency, we also introduce the performance measure approach (PMA) and the sequential optimization and reliability assessment (SORA) strategy. To solve the optimization design problem of a turbine blade, the formula of MDO with ASORM under the SORA framework (MDO-ASORM-SORA) is proposed.