Probing double-distribution-function models in discrete-velocity Boltzmann methods for highly compressible flows: Particles-on-demand realization.

The double distribution function approach is an efficient route toward an extension of kinetic solvers to compressible flows. With a number of realizations available, an overview and comparative study in the context of high-speed compressible flows is presented. We discuss the different variants of the energy partition, analyses of hydrodynamic limits, and a numerical study of accuracy and performance with the particles on demand realization. Out of three considered energy partition strategies, it is shown that the nontranslational energy split requires a higher-order quadrature for proper recovery of the Navier-Stokes-Fourier equations. The internal energy split, on the other hand, while recovering the correct hydrodynamic limit with fourth-order quadrature, comes with a nonlocal-both in space and time-source term that contributes to higher computational cost and memory overhead. Based on our analysis, the total energy split demonstrates the optimal overall performance.