Generalized multiscale approximation of a multipoint flux mixed finite element method for Darcy-Forchheimer model (Q2656116)

In porous-media flow applications, when the flow velocities are relatively high, the relationship between the velocity and the pressure gradient becomes nonlinear such that Darcy's law no longer holds. The nonlinear relationship is described by Darcy-Forchheimer equation which is a corrected formula of Darcy's law by adding a quadratic nonlinear inertial term. In this paper, the Darcy-Forchheimer model in highly heterogeneous porous media is solved with a generalized multiscale finite element method (GMsFEM) combined with a multipoint flux mixed finite element (MFMFE) method. In the MFMFE method on the underlying fine grid, velocity and pressure are approximated by Brezzi-Douglas-Marini (BDM1 mixed finite elements, and the symmetric trapezoidal quadrature rule is employed for the integral of bilinear forms associated with velocity variables. In this way, local velocity is eliminated and a cell-centered system for pressure is obtained. The multiscale basis functions are constructed for the approximation of pressure and to solve the problem on the coarse grid following the GMsFEM framework. In the offline stage, local snapshot spaces and the smaller dimensional offline spaces are computed by a series of local spectral decompositions. In the online stage, the nonlinear term is handled iteratively with Newton's method and the offline solution is obtained. Based on the offline space and offline solution, the multiscale space is enriched by calculating online basis functions. The numerical results demonstrate that the number of Newton iterations is much less than Picard iterations. The multiscale method provides a good approximation on the coarse grid for large values of the Darcy-Forchheimer parameter and the online basis functions improve the accuracy of the multiscale solution.