Enhancing DSMC simulations of rarefied gas mixtures using a fast-converging and asymptotic-preserving scheme

TL;DR

This paper introduces a synthetic iterative scheme that accelerates DSMC simulations of rarefied gas mixtures by combining particle simulations with macroscopic equations, achieving faster convergence and preserving continuum limits.

Contribution

The authors develop a general synthetic iterative scheme that enhances DSMC with macroscopic equations, significantly speeding up convergence for rarefied gas mixture simulations.

Findings

The method accelerates DSMC convergence by roughly 30 times at Kn=0.01.

It preserves the Navier-Stokes limit asymptotically in continuum regimes.

Validated on Poiseuille flow and hypersonic flow past cylinders with different gas models.

Abstract

The numerical simulation of rarefied gas mixture dynamics with disparate masses using the direct simulation Monte Carlo (DSMC) method is slow, primarily because the time step is constrained by that of the lighter species, necessitating an enormous number of evolution steps to reach a steady state. Here, we address this issue by developing a general synthetic iterative scheme, in which the traditional DSMC simulation is intermittently enhanced using a macroscopic synthetic equation. Specifically, after running the DSMC for a certain number of time steps, the high-order constitutive relations for stress and heat flux, as well as the momentum and energy exchange terms from inter-species collisions, are extracted from the DSMC and incorporated into the macroscopic synthetic equations. These equations are solved to obtain the steady state, and the solution is then used to update the particle distribution in DSMC, thereby skipping unnecessary intermediate evolutions. This two-way coupling not only accelerates convergence to the steady state but also asymptotically preserves the Navier-Stokes limit in the continuum flow regime, allowing the spatial cell size to be much larger than the molecular mean free path. The accuracy of our method is validated for one-dimensional force-driven Poiseuille flow and two-dimensional hypersonic flow past cylinders, considering Maxwell and hard-sphere gases with mass ratios of 10 and 100. Although our in-house DSMC method is approximately an order of magnitude slower than the open-source DSMC code SPARTA, intermittently augmenting it with the synthetic equation makes it roughly 30 times faster at a Knudsen number of 0.01, with even greater computational gains anticipated at smaller Knudsen numbers. This work represents a critical step toward developing fast-converging and asymptotic-preserving schemes for hypersonic chemical reactions.