Upscaling the Navier-Stokes-Cahn-Hilliard model for incompressible multiphase flow in inhomogeneous porous media
Chunhua Zhang, Peiyao Liu, Cheng Peng, Lian-Ping Wang, Zhaoli Guo
TL;DR
The paper develops a macroscopic model for two immiscible incompressible fluids in inhomogeneous porous media by upscaling the pore-scale Navier–Stokes–Cahn–Hilliard equations via volume averaging. It derives closed Darcy-scale equations that embed wetting through the averaged chemical potential, and it introduces a lattice Boltzmann method to solve the resulting equations at the representative elementary volume scale. Closure is achieved for both the phase-field and momentum equations through structured ansatzes, linking microstructure to macro-scale coefficients and recovering Darcy- and Brinkman-type limits. Numerical tests on viscous fingering and bubble rise demonstrate the influence of wetting and capillary forces on interface dynamics, while highlighting the need to calibrate certain closure parameters for realistic flows.
Abstract
In this work, we present a macroscopic model for the flow of two immiscible and incompressible fluids in inhomogeneous porous medium. At the pore scale, the flow is governed by the fully Navier-Stokes equations while the evolution of the phase interface is captured by the Cahn-Hilliard equation. Using the volume averaging method, the upscaled equations describing the averaged behavior of two fluids at the Darcy scale are obtained, with unclosed terms related to spatial deviations. Then, spatial derivations are carefully modeled up to some undetermined coefficients, which could be evaluated by solving simplified closure problems in each representative volume element. In particular, the wetting behavior is incorporated into the averaged chemical potential. The differences between the proposed equations and the empirical two-phase Darcy-type models are discussed. Finally, a phase-field-based lattice Boltzmann model for the averaged equations is presented, and numerical results demonstrate the abilities of the proposed model.
