Extension of a Modular Particle-Continuum Method for Nonequilibrium, Hypersonic Flows.

Abstract

As a hypersonic vehicle travels through the atmosphere, it crosses many flow regimes, from rarefied to continuum. Even in mostly continuum flow, there may be local flow features, such as shocks, boundary layers, or wakes, that display collisional nonequilibrium effects. In these regions, the molecular nature of the gas must be taken into account to accurately simulate the flow. While numerically efficient, application of continuum methods to simulate the entire flow field introduces physical inaccuracies throughout the simulation domain. Conversely, while physically accurate, simulation of the entire flow field with a kinetic method, such as the direct simulation Monte Carlo (DSMC), is computationally expensive. Instead, a hybrid method may be used that uses the DSMC method only in regions that exhibit rarefied effects, while employing the continuum description to simulate the rest of the flow field. This dissertation extends a modular particle-continuum (MPC), hybrid method to include consistent physical models for internal energy relaxation in both flow modules. The MPC method uses a breakdown parameter to determine the interface location between flow modules, while state-based coupling procedures are used to transfer information between each module. The capabilities of the MPC method are expanded by parallelizing the method for distributed memory systems in order to decrease processor memory and wall clock time requirements. Comparison with full DSMC simulations are performed to verify each extension of the MPC method and compare computational requirements over full DSMC. The MPC method is tested for hypersonic flow of nitrogen over various blunt body configurations and Knudsen numbers and is shown to reproduce full DSMC results with a high degree of accuracy for macroscopic flow quantities, surface properties, velocity, and energy probability density functions. Careful consideration of the changes to the evaluation of the breakdown parameter and coupling procedures due to the inclusion of internal energy relaxation models is described. Dynamic domain decomposition routines that take into account the inhomogeneous nature of the MPC method are developed and tested. The computational speedup achieved by the MPC method over full DSMC ranges from 1:67 to over 28 and varies with the reduction in number of simulation particles.