Assessing probability distributions of transfer impedance from direct integration and stochastic reduced order method

This proposal aims at demonstrating the suitability of two different non-intrusive and versatile methods to determine the (cumulated) probability density function of a physical quantity of interest that depends on a couple of stochastically varying parameters. The first one is based on numerical integration using a triangulation of the parameter space, similar to matrix assembly in the finite element method. The second one, the statistical order reduction methods, utilizes orthogonal polynomials to identify an efficient “design of experiments” together with weighting factors for the data to be determined according to the identified design. Both methods are compared to each other and to the classical Monte Carlo method with respect to reliability and performance. As practically relevant benchmark problem, the determination of a cable shield's transfer impedance is studied.