A High Order Cartesian Grid, Finite Volume Method for Elliptic Interface Problems

Will Thacher, Hans Johansen, Daniel F. MartinJournal of Computational Physics 2023 · DOI · OpenReview

TL;DR

This paper introduces a novel high-order finite volume method for solving elliptic partial differential equations (PDEs) with interfaces. The method achieves arbitrary high-order accuracy on a Cartesian grid by precisely handling complex interface geometries and jump conditions, offering a robust and conservative approach for challenging problems in computational science.

What Problem Does This Paper Address?

Many physical phenomena, such as multi-material systems and multi-phase flows, are modeled by elliptic PDEs that exhibit discontinuities in their coefficients, source terms, or solutions due to interfaces. Numerically solving these equations is difficult because standard methods assume smooth solutions, which is not the case at interfaces. Existing methods often struggle with accuracy, conservation, and the ability to handle complex interface shapes and high-contrast properties. This paper aims to bridge this gap by developing a high-order, conservative finite volume method that can accurately capture these interface phenomena on a Cartesian grid.

Methods

The core of this work is a high-order finite volume method built upon a Cartesian grid with embedded boundaries (EB). Key methodological contributions include:

  • Generalized Truncation Error Analysis: A rigorous analysis based on cell-centered Taylor series expansions is developed to derive stencils of arbitrary high order. This analysis precisely defines the stencil weights based on local discrete solution data and geometric information.
  • Exact Geometric Moment Computation: A novel approach using Green's theorem is presented to compute exact geometric moments of the embedded interface directly from its implicit function definition. This is crucial for accurately representing the geometry of "cut cells" (cells intersected by the interface).
  • High-Order Stencil Construction: The method constructs stencils with a simple bilinear representation, allowing for the straightforward incorporation of spatially varying coefficients and jump conditions. This is achieved by carefully defining moment matrices and using a weighted least-squares approach for stencil construction.
  • Finite Volume Conservation: The method ensures conservation by using a single, averaged flux across each non-interface face, derived from neighboring stencils.

Datasets

Not discussed in the paper. The paper focuses on developing and validating a numerical method using manufactured solutions and well-defined test cases, rather than training on large datasets.

Key Insights and Ideas

  • Geometric Exactness is Key: Accurately computing geometric moments of the interface, even from an implicit representation, is fundamental to achieving high-order accuracy on a Cartesian grid.
  • Taylor Series as the Foundation: The generalized truncation error analysis, rooted in Taylor series expansions, provides a systematic way to design stencils for arbitrary orders of accuracy.
  • Bridging Finite Difference and Finite Volume: The method effectively combines the geometric flexibility of embedded boundaries with the conservation properties of finite volume schemes, overcoming limitations of both approaches.
  • Interface Dominates Error: The truncation error is primarily concentrated at the interface, which, being a lower-dimensional feature, can lead to higher-order convergence in certain norms (e.g., L¹ norm).
  • Robustness to High Contrast: The method demonstrates remarkable robustness in handling large jumps in coefficients and solutions, a critical feature for many physical applications.

Key Results

The paper demonstrates the method's capability to achieve second, fourth, and sixth-order accuracy on various test problems. Specifically:

  • High-Order Accuracy: The method achieves the expected order of convergence for both truncation and solution errors, with clean convergence rates observed in L¹ and L∞ norms.
  • Robustness: The scheme handles problems with high-contrast coefficients (ratios up to 10⁴) and large solution jumps (up to 4 orders of magnitude) without significant degradation in accuracy.
  • Complex Geometries: The method successfully handles complex interface geometries.
  • Conservation: The finite volume formulation ensures conservation of flux across cell faces.

Limitations and Future Work

  • Limitations:
    • The current implementation is for 2D problems.
  • The paper acknowledges that the efficiency of the solver is not the primary focus, suggesting potential for optimization.
  • The analysis of the combined effects of homogeneous jump conditions and operator regularity on solution error is left for future work.
  • Future Work:
    • Developing an efficient geometric multigrid solver.
  • Extending the method to three dimensions.
  • Incorporating adaptive mesh refinement.
  • Further exploring the theory of undetermined stencil systems to enhance sparsity, conditioning, or stability.

Summary updated

Abstract

We present a higher-order finite volume method for solving elliptic PDEs with jump conditions on interfaces embedded in a 2D Cartesian grid. Second, fourth, and sixth order accuracy is demonstrated on a variety of tests including problems with high-contrast and spatially varying coefficients, large discontinuities in the source term, and complex interface geometries. We include a generalized truncation error analysis based on cell-centered Taylor series expansions, which then define stencils in terms of local discrete solution data and geometric information. In the process, we develop a simple method based on Green's theorem for computing exact geometric moments directly from an implicit function definition of the embedded interface. This approach produces stencils with a simple bilinear representation, where spatially-varying coefficients and jump conditions can be easily included and finite volume conservation can be enforced.
BibTeX
@article{thacher2023high,
  author = {Will Thacher and Hans Johansen and Daniel F. Martin},
  title = {{A High Order Cartesian Grid, Finite Volume Method for Elliptic Interface Problems}},
  journal = {Journal of Computational Physics},
  year = {2023},
  doi = {10.1016/j.jcp.2023.112351},
}
Figure