A general positivity-preserving algorithm for implicit high-order finite volume schemes solving the Euler and Navier-Stokes equations

Qian-Min Huang, Hanyu Zhou, Yu-Xin Ren, Qian WangJournal of Computational Physics 2024 · DOI

TL;DR

This paper introduces a novel, general positivity-preserving algorithm for implicit high-order finite volume schemes used in computational fluid dynamics. It addresses the critical issue of maintaining non-negative density and pressure in simulations, particularly for complex flows, by leveraging asymptotic analysis near vacuum states, offering a more robust and less restrictive solution than existing methods.

What Problem Does This Paper Address?

Simulating fluid dynamics, especially with compressible Euler and Navier-Stokes equations, often encounters a critical robustness issue: the numerical solutions can produce non-physical negative values for density and pressure. This problem is particularly challenging for high-order numerical methods and implicit time-stepping schemes, which are essential for efficiently simulating complex phenomena like turbulence or flows with stiff source terms. Existing positivity-preserving methods often rely on low-order schemes or specific mathematical properties that are difficult to extend to implicit, high-order solvers. This limitation restricts the practical application of these advanced numerical techniques in engineering.

Methods

The core of this paper is a new positivity-preserving algorithm based on an asymptotic analysis of the governing equations near "local vacuum minimum points" (PLVMs). The analysis reveals that solutions near these points exhibit exponential decay, which can be leveraged to ensure positivity. The proposed method involves two key steps:

  1. Linear Residual Correction (LRC): This step directly corrects the linear residual of the numerical scheme to ensure positivity. This is achieved by introducing a correction factor, $\alpha$, which modifies the evolution of the solution.
  2. Conservative Linear Residual Correction (CLRC): The LRC, while ensuring positivity, can destroy the conservation property of the numerical scheme. The CLRC procedure is introduced to recover this conservation. It involves determining interfacial correction factors that ensure the overall scheme remains conservative while maintaining positivity.

The algorithm is designed to be implemented iteratively within implicit dual time-stepping schemes, preserving positivity at intermediate and converged states without requiring time step size reduction. It is also noted that the algorithm is independent of the specific spatial discretization scheme and does not rely on the existence of low-order positivity-preserving baseline schemes or convex decomposition.

Datasets

Not discussed in the paper. The paper focuses on developing and testing a numerical algorithm, not on training or evaluating on specific datasets.

Key Insights and Ideas

  • Asymptotic Analysis as a Foundation: The paper demonstrates the power of using asymptotic analysis of solutions near critical points (PLVMs) to derive fundamental properties that can guide the design of robust numerical algorithms.
  • Exponential Decay for Positivity: The insight that solutions near vacuum states exhibit exponential decay provides a physical basis for developing positivity-preserving strategies.
  • Two-Stage Correction for Positivity and Conservation: The separation of concerns into an LRC for positivity and a CLRC for conservation is a key algorithmic innovation, addressing two critical aspects of numerical scheme robustness.
  • Generality and Reduced Restrictions: The algorithm's independence from low-order schemes, convex decomposition, and time step restrictions makes it a more general and practical solution compared to prior methods.
  • Implicit Scheme Compatibility: The ability to integrate this positivity-preserving technique into implicit time-stepping schemes is crucial for efficient simulation of challenging flow problems.

Key Results

The paper presents numerical results for several benchmark cases, including the Sedov blast-wave, double rarefaction, Le Blanc shock-tube, Mach 2000 jet, and shock reflection/diffraction problems.

  • Positivity Preservation: The proposed algorithm successfully preserves the positivity of density and pressure in all tested cases, including those with strong discontinuities and near-vacuum states.
  • Accuracy Preservation: The algorithm is shown to be accuracy-preserving, with the correction factors not adversely affecting the order of accuracy of the numerical scheme.
  • Robustness Improvement: Numerical tests demonstrate a significant improvement in the robustness of the high-order finite volume schemes, preventing abnormal termination due to negative density or pressure.
  • Efficiency: The algorithm does not require time step size reduction and can be implemented iteratively within implicit schemes, contributing to efficient simulations. The CLRC procedure, in particular, is computationally less demanding than LRC.
  • Viscous Flow Application: The paper highlights the successful application of the positivity-preserving algorithm to implicit high-order Navier-Stokes solvers, a novel contribution.

Limitations and Future Work

  • Computational Cost of CLRC: While CLRC is less computationally intensive than LRC, it still involves additional computations, particularly for cells in set I. The paper notes that this cost is manageable as it applies to a small number of cells.
  • Conjecture on Set I: The validity of the CLRC procedure relies on a conjecture regarding the behavior of solutions in set I cells. While stated to be always valid in practical simulations, a formal proof might be a future direction.
  • Extension to Other Schemes: The paper suggests that the algorithm's principles can be extended to other numerical schemes beyond finite volume methods.
  • Explicit Schemes: While the algorithm is designed for implicit schemes, it is mentioned that it can also be applied to explicit time-stepping schemes, though this was not studied in the paper.

Summary updated

BibTeX
@article{huang2024general,
  author = {Qian-Min Huang and Hanyu Zhou and Yu-Xin Ren and Qian Wang},
  title = {{A general positivity-preserving algorithm for implicit high-order finite volume schemes solving the Euler and Navier-Stokes equations}},
  journal = {Journal of Computational Physics},
  year = {2024},
  doi = {10.1016/j.jcp.2024.112999},
}
Figure