Nonlocal transport equations in multiscale media. Modeling, dememorization, and discretizations

Yalchin Efendiev, Wing Tat Leung, Wenyuan Li, Sai-Mang Pun, Petr N. Vabishchevich

TL;DR

This paper introduces a "dememorization" technique to simplify complex nonlocal transport equations with memory effects, commonly found in heterogeneous media. By transforming these memory-laden equations into a system of standard convection-diffusion-reaction equations, the authors enable more efficient numerical simulations, particularly for multiscale problems with high contrasts, using a novel partially explicit splitting scheme.

What Problem Does This Paper Address?

Many real-world phenomena, such as macro-dispersion in porous media (e.g., groundwater flow, petroleum engineering), are described by convection-diffusion equations that exhibit "memory effects." These memory terms arise from the complex interplay of processes occurring at different scales within heterogeneous materials. Mathematically, these memory terms make the equations difficult to solve, as they require storing and processing all previous time information. This computational burden becomes particularly severe in multiscale problems where features exist across vastly different length and time scales, and where high contrasts in material properties (like permeability) necessitate extremely small time steps for explicit numerical methods. The paper aims to alleviate these computational challenges by developing a method to eliminate the explicit memory terms and proposing efficient numerical schemes for the resulting system.

Methods

The core of the paper is the dememorization technique. This method transforms the original nonlocal equation with a memory integral into a coupled system of standard convection-diffusion-reaction equations. This is achieved by introducing auxiliary variables that represent the integrated memory terms.

For numerical discretization, the authors explore two approaches:

  1. Implicit Scheme: A standard fully implicit time discretization is applied to the dememorized system.
  2. Partially Explicit Splitting Scheme: This is the novel contribution. It leverages a spatial decomposition strategy, inspired by the Constraint Energy Minimizing Generalized Multiscale Finite Element Method (CEM-GMsFEM). The idea is to treat certain "fast" components of the solution implicitly and "slow" components explicitly. This allows for larger time steps compared to fully explicit methods, especially in the presence of high contrasts, without sacrificing stability. The scheme decomposes the solution into components corresponding to different spatial scales or properties, treating some implicitly and others explicitly.

Datasets

Not discussed in the paper. The paper focuses on theoretical development and numerical experiments using synthetic examples.

Key Insights and Ideas

  • Dememorization Simplifies Complexity: The key insight is that equations with memory integrals, which are computationally expensive, can be systematically converted into equivalent systems of standard PDEs. This significantly simplifies the mathematical formulation and numerical implementation.
  • Dememorized vs. Micro-Scale Equations: The dememorized equations are not the same as the original micro-scale equations (which might not have memory). Instead, they represent a macroscopic model where, for instance, convection occurs with averaged velocities.
  • Partially Explicit Schemes for High Contrast: For multiscale problems with high contrasts, explicit methods are severely limited by small time steps. The proposed partially explicit splitting scheme offers a way to overcome this by intelligently treating different parts of the solution implicitly and explicitly, leading to computational efficiency.
  • CEM-GMsFEM as a Foundation: The partially explicit scheme builds upon the CEM-GMsFEM framework, which provides a principled way to construct multiscale basis functions and decompose the solution space, enabling the identification of "fast" and "slow" components.

Key Results

The paper demonstrates the effectiveness of the dememorization technique and the partially explicit splitting scheme through numerical experiments.

  • Accuracy: The partially explicit CEM solution achieves accuracy comparable to implicit CEM solutions (both with and without additional bases) and the reference solution. For instance, in Example 4.1, the relative L2 error is around 8%, and in Example 4.2, it drops to about 1.9%.
  • Efficiency: The partially explicit scheme is shown to achieve similar accuracy to implicit schemes but with potentially less computational cost per time level due to the explicit treatment of some components.
  • Stability: Theoretical stability estimates are provided for the implicit and partially explicit schemes, particularly for the pure reaction case (absence of convection). The stability of the general convection-diffusion case is left for future work.

Limitations and Future Work

  • Stability for General Convection-Diffusion: While stability is shown for the pure reaction case, the stability analysis for the general convection-diffusion case with the partially explicit scheme is left as future work.
  • High-Contrast Limitations: The paper acknowledges that the effectiveness of the partially explicit scheme is particularly pronounced in high-contrast scenarios, which is a key motivation for its development.
  • Generalization: The authors suggest that the framework can be extended to more general nonlocal equations and multiscale problems.
  • Real-World Applications: While the paper uses examples from porous media, further validation on more complex, real-world heterogeneous media would be valuable.

Summary updated

Abstract

In this paper, we consider a class of convection-diffusion equations with memory effects. These equations arise as a result of homogenization or upscaling of linear transport equations in heterogeneous media and play an important role in many applications. First, we present a dememorization technique for these equations. We show that the convection-diffusion equations with memory effects can be written as a system of standard convection-diffusion-reaction equations. This allows removing the memory term and simplifying the computations. We consider a relation between dememorized equations and micro-scale equations, which do not contain memory terms. We note that dememorized equations differ from micro-scale equations and constitute a macroscopic model. Next, we consider both implicit and partially explicit methods. The latter is introduced for problems in multiscale media with high-contrast properties. Because of high contrast, explicit methods are restrictive and require time steps that are very small (scales as the inverse of the contrast). We show that, by appropriately decomposing the space, we can treat only a few degrees of freedom implicitly and the remaining degrees of freedom explicitly. We present a stability analysis. Numerical results are presented that confirm our theoretical findings of partially explicit schemes applied to dememorized systems of equations.
BibTeX
@article{efendiev2023nonlocal,
  author = {Yalchin Efendiev and Wing Tat Leung and Wenyuan Li and Sai-Mang Pun and Petr N. Vabishchevich},
  title = {{Nonlocal transport equations in multiscale media. Modeling, dememorization, and discretizations}},
  journal = {Journal of Computational Physics},
  year = {2023},
  doi = {10.1016/j.jcp.2022.111555},
}
Figure