Concurrent subsystem uncertainty analysis in multidisciplinary design

Abstract

A modified concurrent subsystem uncertainty analysis (MCSSUA) method for uncertainty analysis in multidisciplinary design optimization (MDO) is presented in this paper a . Given the probabilistic representations of uncertain parameters and model error estimations, the method quickly evaluates the mean and the variance of a system output so as to improve the computational efficiency of probabilistic optimization such as robust design in the MDO environment. The MCSSUA method is an improved version of the original concurrent subsystem uncertainty analysis (CSSUA) method. The method utilizes the strategy of concurrent subsystem analysis to obtain the mean values of the linking variables via suboptimization. The approximation of a system output is generated at the mean values of input parameters and the mean and variance of the system output are derived. The MCSSUA method is more efficient than the original CSSUA method as the number of design variables in suboptimizations is reduced by half. It is more accurate because no assumption of independence among linking variables is required. A comparison of system uncertainty analysis (SUA) method, the CSSUA method, and the modified CSSUA method is made in the context of highly coupled analyses and multidisciplinary robust design. Two MDO examples are presented to demonstrate the proposed method.