A mass conservative generalized multiscale finite element method applied to two-phase flow in heterogeneous porous media

TL;DR

This paper introduces a mass conservative generalized multiscale finite element method for accurately modeling two-phase flow in heterogeneous porous media, ensuring local flux conservation and improved solution accuracy.

Contribution

It develops a novel flux construction technique within GMsFEM that enforces local conservation via Lagrange multipliers, enhancing flow simulation accuracy in complex media.

Findings

Flux fields are locally conservative and accurate in heterogeneous media.

Adding basis functions improves the solution's fidelity to fine-scale models.

Numerical examples validate the method's effectiveness in complex flow scenarios.

Abstract

In this paper, we propose a method for the construction of locally conservative flux fields through a variation of the Generalized Multiscale Finite Element Method (GMsFEM). The flux values are obtained through the use of a Ritz formulation in which we augment the resulting linear system of the continuous Galerkin (CG) formulation in the higher-order GMsFEM approximation space. In particular, we impose the finite volume-based restrictions through incorporating a scalar Lagrange multiplier for each mass conservation constraint. This approach can be equivalently viewed as a constraint minimization problem where we minimize the energy functional of the equation restricted to the subspace of functions that satisfy the desired conservation properties. To test the performance of the method we consider equations with heterogeneous permeability coefficients that have high-variation and discontinuities, and couple the resulting fluxes to a two-phase flow model. The increase in accuracy associated with the computation of the GMsFEM pressure solutions is inherited by the flux fields and saturation solutions, and is closely correlated to the size of the reduced-order systems. In particular, the addition of more basis functions to the enriched multiscale space produces solutions that more accurately capture the behavior of the fine scale model. A variety of numerical examples are offered to validate the performance of the method.