Propagation of non-Gaussian stochastic behavior using the polynomial chaos expansion

One of the important issues in uncertainty analysis is to find an effective way for propagating uncertainty through the system. In this paper, the polynomial chaos expansion (PCE) was selected since this approach can reduce the computational effort in large-scale engineering design applications. An implementation of PCE for different probability distributions is the focus of this paper. Two existing techniques, a generalized PCE algorithm and transformation methods, are investigated and verified for their accuracy and efficiency for non-normal random variable cases. A highly nonlinear structural model of an uninhabited joined-wing aircraft and a three pin-connected rod structure are used for demonstrating the method.