Abstract

Upscaling of flow from pore to Darcy scale is a long-standing research field within flow in porous media. It is well known that non-linearities can occur in near-well regions and high-porosity or fractured media. At the same time, the upscaled non-linear effects associated with high flow rates are hard to quantify a priori with single-scale models. Advances in pore-scale imaging combined with increased computational have made flow simulations in small pore-scale domain feasible, but computations on domains larger than at most a few centimeters are still elusive. In this work, we present a multiscale simulation framework that automatically adapts to non-linear effects as they arise. We formulate a control volume heterogeneous multiscale method (CVHMM) by coupling of a Darcy-scale control volume method with a constitutive relation that is captured based on the fine-scale physics. While the CVHMM formulation works with arbitrary upscaled laws, we emphasize its ability to be applied in fully discrete multiscale context, in particular when a finite element solver is used for solving Navier-Stokes equations on the fine-scale pore geometry. Previous versions of CVHMM are consistent only when the coarse grid is aligned with the upscaled permeability. Herein, we generalize CVHMM by introducing a new coarse solver, thus significantly improving the applicability of the method. The presented method is applied to study flow in near-well regions, as well as media with fractures and irregular grain shapes. The examples show that the method successfully copes with general grids and pore geometries and handles flows with varying degree of non-linearities even outside the domain of applicability of classical upscaled models. In terms of computational efficiency, the method seamlessly localizes computations to regions where non-linear effects are important.

Highlights

  • Flow in porous media is commonly modeled by Darcy’s law, which postulates a linear relationship between the fluid flux and pressure gradient

  • This allows for a consistent approximation; notably, it generalizes the previous formulation of control volume heterogeneous multiscale method (CVHMM) [8, 9], which used two-point flux approximation (TPFA) for the flux estimation

  • We have presented a generalized multi-point version of control volume heterogeneous multiscale method (CVHMM) for resolving non-linear flows in porous media

Read more

Summary

Introduction

Flow in porous media is commonly modeled by Darcy’s law, which postulates a linear relationship between the fluid flux and pressure gradient. Our main interest is in obtaining an effective discrete representation of non-linear upscaled Darcy-scale models, that is is flexible with respect to constitutive relationships and allows for either a priori derived nonlinear upscaled laws such as Forchheimer, or on-demand upscaling of pore-scale computations. The framework allows for flexibility of a fine-scale solver that can be an analytical model (resulting in traditional upscaling, see, e.g., [26]), or a numerical model such as a pore-network model [21, 22], or a Navier-Stokes solver [8, 9], resulting in a fully discrete multiscale method The latter approach, termed the control volume heterogeneous multiscale method (CVHMM), is the subject of this paper.

Pore-scale model
Darcy-scale model
Two-scale numerical discretization
Coarse flux approximation and quadrature rule
Two-point flux approximation
Multi-point flux approximation L-method
Non-linear estimation of coarse scale velocity
Velocity estimation by Forchheimer’s law
Velocity estimation by fine-scale finite element computation
Algorithm
Numerical study of non-linearity in applications
Near-well flow in homogeneous domain
Example setup
Fully discrete CVHMM
CVHMM with upscaled flow behavior
Computational adaptivity in low-permeable domain with a fracture
Direction-dependent flow for non-circular grains
Findings
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call