The role of physical and numerical viscosity in hydrodynamical instabilities

TL;DR

This paper evaluates how physical and numerical viscosities influence the Kelvin-Helmholtz Instability in hydrodynamical simulations, comparing SPH and MFM methods, and assesses SPH's suitability for galaxy cluster studies.

Contribution

It provides a detailed comparison of SPH and MFM in modeling KHI, quantifies numerical viscosity effects, and demonstrates SPH's capability to accurately simulate physical viscosity in galaxy clusters.

Findings

SPH accurately reproduces the reduction of KHI growth due to physical viscosity.

Numerical viscosity in SPH is at least an order of magnitude smaller than physical viscosity in galaxy clusters.

SPH methods are suitable for studying physical viscosity effects in galaxy cluster environments.

Abstract

The evolution of the Kelvin-Helmholtz Instability (KHI) is widely used to assess the performance of numerical methods. We employ this instability to test both the smoothed particle hydrodynamics (SPH) and the meshless finite mass (MFM) implementation in OpenGadget3. We quantify the accuracy of SPH and MFM in reproducing the linear growth of the KHI with different numerical and physical set-ups. Among them, we consider: $i)$ numerical induced viscosity, and $ii)$ physically motivated, Braginskii viscosity, and compare their effect on the growth of the KHI. We find that the changes of the inferred numerical viscosity when varying nuisance parameters such as the set-up or the number of neighbours in our SPH code are comparable to the differences obtained when using different hydrodynamical solvers, i.e. MFM. SPH reproduces the expected reduction of the growth rate in the presence of physical viscosity and recovers well the threshold level of physical viscosity needed to fully suppress the instability. In the case of galaxy clusters with a virial temperature of $3\times10^7$ K, this level corresponds to a suppression factor of $\approx10^{-3}$ of the classical Braginskii value. The intrinsic, numerical viscosity of our SPH implementation in such an environment is inferred to be at least an order of magnitude smaller (i.e. $\approx10^ {-4}$), re-ensuring that modern SPH methods are suitable to study the effect of physical viscosity in galaxy clusters.