A modular particle -continuum numerical algorithm for hypersonic non-equilibrium flows.

Hypersonic vehicles traveling at high altitudes experience conditions ranging from rarefied to continuum flow. Even within a mostly continuum flow, there may be local regions of rarefied (or non-equilibrium) flow generated by sharp leading edges, by rapid expansion in the wake of a vehicle, as well as by strong gradients in shock waves and boundary layers. Although a particle method is necessary for accurate simulation of non-equilibrium regions, particle simulation of the entire flowfield is computationally expensive. A modular particle-continuum (MPC) numerical algorithm for steady-state hypersonic flows is presented that solves the Navier-Stokes equations in regions of near-equilibrium and uses the direct simulation Monte Carlo (DSMC) method to simulate regions of non-equilibrium gas flow. Existing state-of-the-art DSMC and Navier-Stokes codes are loosely-coupled using state-based information transfer and a novel modular implementation. The MPC algorithm allows for complete spatial and temporal scale decoupling whereby particle and continuum regions are simulated using different mesh densities and different sized timesteps. Particle and continuum regions are identified using a continuum breakdown parameter and automatically adapt during the hybrid simulation. The MPC method is tested for hypersonic flow of nitrogen over a 2D cylinder at various Mach numbers where the global Knudsen number is 0.01. The MPC method is shown to reproduce full DSMC simulation results with a high degree of accuracy for flowfield quantities, surface properties, and local velocity distribution functions. The computational speedup achieved by the MPC method over full DSMC simulation ranges from 2.7 to 3.3 for the cylinder flows and is found to scale directly with the number of particles replaced by a continuum description. In addition, MPC simulations of axi-symmetric planetary probe and hollow cylinder-flare configurations with global Knudsen numbers of 0.001, are compared with full DSMC solutions and experimental data. MPC simulations of both configurations are shown to accurately reproduce DSMC results for velocity slip, temperature jump, thermal non-equilibrium, surface properties, and the extent of separated flow. For these mainly continuum flows, the MPC method is shown capable of reproducing full DSMC solutions with an order-of-magnitude decrease in both computational time and memory.PhDAerospace engineeringApplied SciencesUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/126965/2/3287627.pd