CRASH Simulation of Rayleigh-Taylor, Richtmyer-Meshkov, and Magnetic Rayleigh-Taylor Instability.

Abstract

The research discussed in this thesis was motivated by the supernova Rayleigh-Taylor (SNRT) experiments conducted in 2009. Originally designed as laboratory astrophysics experiments relevant to the Rayleigh-Taylor Instability (RTI) at the He-H interface during supernova explosion, these experiments exhibited unusual late-time morphology development, characterized by the lack of mushroom caps and uniform width of the spikes. In response, a "Magnetic Straitjacket" hypothesis was proposed to explain the discrepancy, based on the Biermann Battery mechanism. In order to test this hypothesis, we used the Center for Radiative Shock Hydrodynamics (CRASH) code developed for a sufficiently similar problem and with the necessary capabilities. We validated this alternative usage of the CRASH code with simulations of pure hydrodynamic RTI and Richtmyer-Meshkov Instability (RMI) and identified the suitable combinations of numerical schemes and parameters. For the RTI, we compared the results of simulations to the analytical solution for the early time behavior, examined the late-time morphology, and tested the low-resolution limit for the RTI simulations using CRASH. For the RMI, we modeled Collins and Jacobs' experiment and compared the results of CRASH simulations to the experimental observations as well as to the simulation results of several other code packages. Finally, we modeled the original SNRT experiments with magnetohydrodynamics (MHD) and Biermann battery effect. Unfortunately, the results were inconclusive due to insufficiently resolved simulations, limited by the explicit time integration of the magnetic diffusion. Furthermore, pilot runs with higher resolution indicated that simulations that fully resolve the gradients necessary to calculate the Biermann battery effect may be susceptible to the development of extraneous small-wavelength instabilities. Developments of implicit time integration of the magnetic diffusion and possibly new numerical schemes are therefore necessary for further progress, either with the CRASH code or other Eulerian code packages.