Criteria and Methods for Evaluating Solutions to Transient, Turbulent Flow, with Different Mesh Topologies

We discuss the context of mesh generation for discretizing transient flow analyses, and evaluation of the attending solutions, in terms of accuracy and computational expense. We arrive at methodologies for developing workflows with a predictable level of accuracy and computational cost. We propose criteria to compare solution accuracy for transient flows from inviscid to laminar and turbulent, and expended computational effort, in addition to grid spacing, for meshes of different topologies. We propose that measures of accuracy per computational cost can be more useful to the analyst than accuracy per grid spacing, at least for transient flows. One class of single phase, single species, incompressible, free shear vortical flows is studied, using one flow solver. The merits of hexahedral, tetrahedral, and polyhedral meshes of various refinements are compared. Future work is planned to expand the array of simple test cases, to include more complex turbulent flows and additional physics.