AN EFFICIENT APPROACH TO PROBABILISTIC UNCERTAINTY ANALYSIS IN SIMULATION-BASED MULTIDISCIPLINARY DESIGN

In this paper a , computationally efficient techniques for propagating the effect of uncertainty are developed to accommodate generic probabilistic representations of uncertain parameters and error estimation models in a multidisciplinary design system. To improve the computational efficiency of probabilistic uncertainty propagation in the context of highly coupled analyses, the first order sensitivity analysis and the moment matching method are employed. This is implemented by two techniques, namely, the system uncertainty analysis method (SUAM) and the concurrent subsystem uncertainty analysis method (CSSUAM). Depending on the number of variables and the number of disciplines involved, the effectiveness of these techniques varies. A mathematical example and an electronic packaging problem are used to verify the effectiveness of these approaches.