TL;DR
This paper introduces a novel matrix stability analysis method to quantitatively assess the shock instability of second-order MUSCL finite-volume schemes. The authors reveal that shock instability is influenced by various factors, including the Riemann solver, Mach number, limiter function, numerical shock structure, and grid, and they provide a deeper understanding of the underlying mechanisms.
What Problem Does This Paper Address?
Modern shock-capturing schemes, widely used in computational fluid dynamics for simulating flows with shocks (especially in hypersonic regimes), often suffer from numerical anomalies like the "carbuncle phenomenon" when encountering strong shocks. These anomalies, such as spurious shock distortions and oscillations, significantly degrade the reliability of simulation results, particularly for predicting phenomena like hypersonic heating. While first-order schemes have been analyzed for stability, a quantitative method for higher-order schemes, like the commonly used second-order MUSCL (Monotonic Upstream-centered Scheme for Conservation Laws) schemes, has been lacking. This paper aims to fill that gap by developing and applying a matrix stability analysis method to understand and predict shock instability in these higher-order schemes.
Methods
The core contribution of this paper is the development of a matrix stability analysis method specifically tailored for second-order finite-volume MUSCL schemes. This method extends existing techniques for first-order schemes by linearizing the discretized Euler equations around a steady mean flow and analyzing the eigenvalues of the resulting stability matrix. The sign of the maximal real part of these eigenvalues determines the stability of the scheme. The authors also explore the use of primitive variables in their analysis, which can offer computational advantages. They systematically investigate the influence of various factors on shock stability, including:
- Riemann Solvers: Comparing different solvers (Roe, HLL, HLLC, AUSM+, van Leer) with varying dissipative characteristics.
- Spatial Accuracy: Comparing first-order and second-order schemes.
- Limiters: Analyzing the impact of different flux limiters (superbee, van Leer, van Albada, minmod) with varying levels of dissipation.
- Mach Number: Examining its effect on stability across a range of values.
- Numerical Shock Structure: Investigating how the internal structure of the captured shock influences stability.
- Computational Grid: Analyzing the role of aspect ratio and distortion angle.
Datasets
Not discussed in the paper. The study relies on numerical experiments and theoretical analysis rather than training on datasets.
Key Insights and Ideas
- Second-order schemes are more prone to instability than first-order schemes with low-dissipative solvers: While higher-order schemes generally offer better accuracy, they can be more susceptible to shock instabilities, especially when using solvers like Roe that have minimal dissipation.
- Dissipative solvers and limiters are crucial for stability: Schemes with higher dissipation (e.g., HLL solver, minmod limiter) tend to be more stable, effectively damping out perturbations that can lead to shock anomalies.
- The numerical shock structure is the origin of instability: The analysis strongly suggests that the instability originates within the numerical representation of the shock itself, rather than solely from upstream or downstream regions.
- Grid design matters: The aspect ratio and distortion angle of computational cells can significantly influence shock stability, with elongated cells (larger aspect ratio) and increased distortion angles generally leading to more stable solutions.
- The matrix stability analysis is a powerful predictive tool: The developed method provides a quantitative way to predict shock stability and understand the underlying mechanisms, correlating well with numerical experiment results.
Key Results
The study demonstrates that:
- Riemann solvers play a critical role: Roe and HLLC solvers, known for low dissipation, lead to unstable behaviors (positive real parts of eigenvalues), while HLL and van Leer solvers, being more dissipative, show stable characteristics (negative real parts). AUSM+ is found to be marginally stable.
- Spatial accuracy's impact is solver-dependent: For stable or marginally stable solvers (HLL, AUSM+), enhancing spatial accuracy to second-order improves stability. However, for unstable solvers (Roe), second-order accuracy worsens stability.
- Limiters significantly affect stability: The superbee limiter (low dissipation) leads to more instability, while the minmod limiter (high dissipation) enhances stability for second-order schemes.
- Mach number's influence is significant but limited: While increasing Mach number generally exacerbates instability, its effect plateaus beyond a certain threshold.
- Numerical shock structure is key: The stability is highly dependent on the internal shock conditions, with conditions closer to downstream states promoting stability.
- Grid aspect ratio and distortion angle can alleviate instability: Increasing the aspect ratio (elongating cells in the y-direction) and distortion angle of the grid can improve stability.
Limitations and Future Work
- Focus on second-order schemes: The current work specifically addresses second-order MUSCL schemes. The authors mention that a follow-up study will present the matrix stability analysis method for high-order finite-volume schemes.
- Quantitative analysis of high-order schemes: The paper highlights the need for quantitative tools to investigate the shock stability of high-order schemes, which are known to be at a higher risk of instability and are more challenging to analyze. This is identified as a key area for future research.
- Understanding the "why" of instability: While the paper identifies the numerical shock structure as the origin, further in-depth investigation into the underlying mechanisms of high-order shock-capturing methods is suggested for future work.
Summary updated