Grid Independence Via Automated Unstructured Adaptation

Abstract

Grid independence is frequently an overlooked item in computational fluid dynamics (CFD) analyses. Results obtained from grid dependent solutions may prove to be costly, in that engineering design decisions can be made using potentially faulty information. An automated method for grid independence is developed for two-dimensional unstructured wall function grids. Grid independence is achieved via successive levels of adaptive refinement. Adaptive refinement is performed in an automated manner and is based on multiple field variables. Sensors are placed at strategic locations within the flow field, which are determined by examining the CFD solution of a uniform grid. Three cases are examined, the backward-facing step, flow over an asymmetric transonic airfoil, and hydrogen combustion in a channel. Grid independent solutions are obtained for all three cases. Results for each case compare well with experimental data and/or other numerical predictions.