On physical and numerical instabilities arising in simulations of non-stationary radiatively cooling shocks

TL;DR

This paper investigates physical and numerical instabilities in simulations of radiatively cooling shocks, highlighting how grid choice affects observed instabilities and emphasizing the importance of accurate modeling to avoid artifacts.

Contribution

It identifies the conditions under which numerical and physical instabilities arise in shock simulations and compares effects across different numerical grid geometries.

Findings

Cartesian grids induce bending and fragmentation of shells due to numerical noise.

Polar meshes better preserve shell stability, reducing numerical artifacts.

Rayleigh--Taylor like instabilities are triggered by physical and numerical factors.

Abstract

We describe our modelling of the radiatively cooling shocks and their thin shells with various numerical tools in different physical and calculational setups. We inspect structure of the dense shell, its formation and evolution, pointing out physical and numerical factors that sustain its shape and also may lead to instabilities. We have found that under certain physical conditions, the circular shaped shells show a strong bending instability and successive fragmentation on Cartesian grids soon after their formation, while remain almost unperturbed when simulated on polar meshes. We explain this by physical Rayleigh--Taylor like instabilities triggered by corrugation of the dense shell surfaces by numerical noise. Conditions for these instabilities follow from both the shell structure itself and from episodes of transient acceleration during re-establishing of dynamical pressure balance after sudden radiative cooling onset. They are also easily excited by physical perturbations of the ambient medium. The widely mentioned Non-linear Thin Shell Instability, in contrast, in tests with physical perturbations is shown to have only limited chances to develop in real radiative shocks, as it seems to require a special spatial arrangement of fluctuations to be excited efficiently. The described phenomena also set new requirements on further simulations of the radiatively cooling shocks in order to be physically correct and free of numerical artefacts.