PROBABILISTIC COLLOCATION FOR EFFICIENT UNCERTAINTY ANALYSIS IN GROUNDWATER FLOW

Assessment of the effects of uncertainty in model inputs on its output are now widely recognized as important parts of analyses for complex systems. In this paper, we address stochastic groundwater flow problems with random permeability fields. For this purpose, we combine low order mixed finite element method space with polynomial chaos expansions. The so-called Karhunen-Loeve expansion is employed to efficiently generate the random fields. A non-intrusive method based on probabilistic collocations is used to compute the polynomial coefficients. The computational cost is reduced by preliminarily screening the input random variables. The stochastic error analysis is investigated through numerical experiments.