Two dimensional local adaptive discrete velocity grids for rarefied flow simulations

We propose a deterministic method designed for unsteady flows, based on a discretization of the Boltzmann (BGK) equation with local adaptive velocity grids. These grids dynamically adapt in time and space to the variations of the width of the distribution functions. This allows a significant reduction of the memory storage and CPU time, as compared to standard discrete velocity methods, and avoid the delicate problem to construct a priori a suffcient global velocity grid.