Insight derived from high order structured finite difference CFD simulations of flow past generic simplified car models

This thesis focuses upon using a high-order finite-difference method on a structured overset grid to study flow features around a generic simplified car model adjusting mesh refinements, rear slant angles, solvers, and turbulence models. These processes are explored to develop a procedure for simulating more complex and realistic car models. Three different mesh refinements from 17 to 108 million vertices were tested with three different solvers (RANS, URANS, and DES) on an Ahmed body with a subcritical slant angle ascertain the optimum mesh parameters for subsequent simulations. Using a 57 x 106 vertex mesh, multiple rear slant angles near the critical angle (30 degrees) were investigated with RANS and URANS approaches to compare drag, lift, and flow fields with experimental and CFD data found in literature. Similar trends were observed in CFD predictions and experimental data, including flow separation at the critical angle (30 degrees), but all predicted results were within 15% of experimental measurements for both time-averaged and unsteady simulations. At the sub-critical angle (25 degrees), CFD predictions using multiple hybrid RANS/LES approaches were compared against time-averaged and unsteady experimental measurements. These did not disagree with previous results and drag values were over predicted by a maximum of 4%, while lift values were under predicted by a maximum of 15% when compared to experimental results. Subsequent studies investigating elongated mesh refinement areas were inconclusive. The procedures outlined compare reasonably well to experimental data and can be used as a starting point for simulating more realistic models including complex dynamic pitch, heave, and roll simulations involving road vehicles.