On a linear stability issue of split form schemes for compressible flows

Vikram Singh, Praveen Chandrashekar

Abstract

Split form schemes for Euler and Navier-Stokes equations are useful for computation of turbulent flows due to their better robustness. This is because they satisfy additional conservation properties of the governing equations like kinetic energy preservation leading to a reduction in aliasing errors at high orders. Recently, linear stability issues have been pointed out for these schemes for a density wave problem and we investigate this behaviour for some standard split forms. By deriving linearized equations of split form schemes, we show that most existing schemes do not satisfy a perturbation energy equation that holds at the continuous level. A simple modification to the energy flux of some existing schemes is shown to yield a scheme that is consistent with the energy perturbation equation. Numerical tests are given using a discontinuous Galerkin method to demonstrate these results.

Introduction

High-order numerical methods, such as Discontinuous Galerkin (DG), are essential for simulating complex turbulent flows in aerospace or climate modeling. However, these methods are notoriously fragile and often suffer from aliasing instabilities—mathematical noise that grows until the simulation crashes. To mitigate this, researchers use "split form" schemes designed to preserve physical properties like kinetic energy or entropy.

While these schemes are generally more robust, recent studies found they fail a deceptively simple test: the density wave problem. In this scenario, a wave of varying density moves through a fluid at constant velocity and pressure. Many state-of-the-art schemes that should be stable actually exhibit linear instability here, leading to unphysical results or solver failure. This paper investigates why this happens and how to fix it.

Methodology: The "Reduced Energy" Law

The authors employ a rigorous mathematical approach called linear stability analysis, linearizing the compressible Euler equations around a base density wave solution. They discovered that while the full energy of a density wave perturbation isn't strictly bounded, there is a "reduced energy" that is conserved at the continuous level.

Most existing schemes, including the popular Kennedy and Gruber (KG) and Ducros forms, fail to conserve this reduced energy discretely. This failure leads to transient growth that triggers non-linear instabilities, even if the scheme's eigenvalues appear stable.

The mKEP Scheme

To solve this, the authors propose a modified Kinetic Energy Preserving (mKEP) flux. Unlike previous versions, this scheme is specifically engineered to satisfy the reduced energy conservation law for perturbations. By modifying the energy flux to use primitive variables in a specific quadratic/cubic form, the mKEP scheme ensures that small errors don't grow uncontrollably in the presence of density gradients.

Numerical Results and Validation

The authors validated their approach using several standard benchmark problems, including 2D density waves and 3D Taylor-Green vortices.

Stability in Density Waves

In the 2D density wave test, the KG scheme demonstrated significant instability, blowing up almost instantly. The mKEP and KEEP-PE schemes remained stable significantly longer than the standard Central and Ducros schemes when perturbations were added.

Table 1: Comparison of stability limits (time to blow-up) for various split form schemes on the density wave problem. The KG scheme shows significantly lower stability thresholds compared to the proposed mKEP and other schemes.

As shown in the data, the KG scheme's stability limit is notably lower (e.g., $T \approx 0.13$ for $N=3$) compared to the mKEP scheme ($T \approx 0.51$).

Turbulence Accuracy

In 3D Taylor-Green Vortex simulations, the mKEP scheme showed very low kinetic energy dissipation, matching DNS results closely. In contrast, the KEEP-PE scheme—while stable for density waves—showed unphysically high dissipation once the flow became turbulent. The mKEP scheme also successfully simulated turbulent channel flow ($Re_\tau = 180$), with mean velocity and Reynolds stress statistics showing excellent agreement with high-fidelity DNS data.

Conclusion

This research highlights that discrete consistency with the perturbation energy equation is a critical requirement for the linear stability of split form schemes. The proposed mKEP scheme provides a more robust framework for high-order simulations, avoiding the non-physical blow-ups common in existing methods while maintaining high accuracy in turbulent regimes. Future work will focus on extending these stability proofs to more complex physics, such as multi-species flows or magnetohydrodynamics.

Summary updated

Abstract

Split form schemes for Euler and Navier-Stokes equations are useful for computation of turbulent flows due to their better robustness. This is because they satisfy additional conservation properties of the governing equations like kinetic energy preservation leading to a reduction in aliasing errors at high orders. Recently, linear stability issues have been pointed out for these schemes for a density wave problem and we investigate this behaviour for some standard split forms. By deriving linearized equations of split form schemes, we show that most existing schemes do not satisfy a perturbation energy equation that holds at the continuous level. A simple modification to the energy flux of some existing schemes is shown to yield a scheme that is consistent with the energy perturbation equation. Numerical tests are given using a discontinuous Galerkin method to demonstrate these results.
BibTeX
@misc{singh2021linear,
  author = {Vikram Singh and Praveen Chandrashekar},
  title = {{On a linear stability issue of split form schemes for compressible flows}},
  year = {2021},
  eprint = {2104.14941},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2104.14941},
}

Loading related content...

Figure