Computational methods for accounting structural uncertainties - Applications in dynamic behavior of piping systems

摘要

An efficient importance sampling method is suggested to approximate the multi-dimensional integrals which arise in mechanical analysis in presence of structural uncertainties. The procedure is especially suitable when the probabilities of failure or response moments conditional on the system parameters are available. The purpose of the study is to yield sufficiently accurate results with acceptable computational cost in an industrial context. The method is an importance sampling method based on a second order asymptotic approximation combined with the HyperCube Latin simulation method. The efficiency of the method is illustrated with several examples and a test case from an industrial problem. Considerable reduction of the statistical error of the estimated failure probability is achieved. The accuracy of the method is assured provided the points of local maximum likelihood can be identified and/or are not too numerous and uniformly distributed.