Parallel multiblock multigrid algorithms for poroelastic models

Abstract

The application of parallel multigrid for two-dimensional poroelastic model is investigated. First, a special stabilized finite difference scheme is proposed, which allows one to get a monotone approximation of the differential problem. The obtained systems of linear algebraic equations are solved by a multigrid method, when a domain is partitioned into structured blocks. This geometrical structure is used to develop a parallel version of the multigrid algorithm. The convergence for different smoothers is investigated, it is shown that the box Gauss-Seidel smoother is robust and efficient. Finally, the parallel multigrid method is tested for the Poisson problem.