On the combined performance of nonlocal artificial boundary conditions with the new generation of advanced multigrid flow solvers

Abstract

We develop theoretically and implement numerically a unified flow solution methodology that combines the advantages relevant to two independent groups of methods in computational fluid dynamics that have recently proven successful: The new factorizable schemes for the equations of hydrodynamics that facilitate the construction of optimally convergent multigrid algorithms, and highly accurate global far-field artificial boundary conditions (ABCs). The primary result that we have obtained is the following. Global ABCs do not hamper the optimal (i.e., unimprovable) multigrid convergence rate pertinent to the solver. At the same time, contrary to the standard local ABCs, the solution accuracy provided by the global ABCs deteriorates very slightly or does not deteriorate at all when the computational domain shrinks, which clearly translates into substantial savings of computer resources.