A stability and sensitivity analysis of parametric functions in a sedimentation model

Abstract

This paper deals with the reliable and efficient numerical identification of parameters defining the flux function and the diffusion coefficient of a strongly degenerate parabolic partial differential equation (PDE), which is the basis of a mathematical model for sedimentation-consolidation processes. A zero-flux initial-boundary value problem (IBVP) posed for this PDE describes the settling of a suspension in a column. The parameters for a given material are estimated by the repeated numerical solutions of the IBVP (direct problem) under systematic variation of the model parameters, with the aim of successively minimizing a cost functional that measures the distance between a space-dependent observation and the corresponding numerical solution. Two important features of this paper are the following. In the first place, the method proposed for the efficient and accurate numerical solution of the direct problem. We implement a wellknown explicit, monotone three-point finite difference scheme enhanced by discrete mollification. The mollified scheme occupies a larger stencil but converges under a less restrictive CFL condition, which allows the use of a larger time step. The second feature is the thorough sensitivity and stability analysis of the parametric model functions that play the roles of initial guess and observation data, respectively.