Abstract. In this work, we present a numerical analysis of a method which combines a deterministic and a probabilistic approaches to quantify the migration of a contaminant, under the presence of uncertainty on the permeability of the porous medium. More precisely, we consider the flow equation in a random porous medium coupled with the advection-diffusion equation. Quantities of interest are the mean spread and the mean dispersion of the solute. The means are approximated by a quadrature rule, based on a sparse grid defined by a truncated Karhunen-Loève expansion and a stochastic collocation method. For each grid point, the flow model is solved with a mixed finite element method in the physical space and the advection-diffusion equation is solved with a probabilistic Lagrangian method. The spread and the dispersion are expressed as functions of a stochastic process. A priori error estimates are established on the mean of the spread and the dispersion. Keywords: Uncertainty quantification, elliptic PDE with random coefficients, advection-diffusion equation, collocation techniques, anisotropic sparse grids, Monte Carlo method, Euler scheme for SDE.
IntroductionMathematical modeling and numerical simulation are important tools in the prediction of pollutant transport in groundwater and in the evaluation of potential risks of contaminants. The main constraint to the development of these models is the limited knowledge of the geological characteristics and the natural heterogeneity which implies uncertainty in the parameters and data. Stochastic approaches have been developed to deal with this uncertainty, where the permeability field is modeled as a random field a = e G , where G is a correlated Gaussian field [2, 13]. Our objective is to quantify the migration of a contaminant by computing statistics of interest defined by the mean of the spread and of the dispersion of the solute. The permeability a is discretized using a Karhunen-Loève (K-L) truncation up to a suitable and moderately large order.The Monte Carlo method is the most widely used approach to deal with uncertainty. This method is used in [2,4,7,12,13] to approximate the mean spread and dispersion.Recently, stochastic collocation (see [1,9,18,19]), based on sparse tensor product approximation, has gained much attention since it is very effective and accurate for computing statistics from solutions of PDEs with random input data. In this paper, we use a truncated Karhunen-Loève expansion of the random data and an anisotropic sparse grid with Gaussian knots. Then we propose to approximate the mean values by a quadrature rule using this sparse grid.
329The cost is proportional to the number of samples in the Monte Carlo method and to the sparse grid size in the collocation method. Each sample or grid point requires solving a flow PDE and a transport PDE; this can be done in a non intrusive and parallel way. In this paper, we approximate the steady-state flow problem by a mixed finite element method in the physical space to get the velocity field. The tra...