Aerodynamic performance assessment of automotive shapes is typically performed in wind tunnels. However, with the rapid progress in computer hardware technology and the maturity and accuracy of Computational Fluid Dynamics (CFD) software packages, evaluation of the production-level automotive shapes using a digital process has become a reality. As the time to market shrinks, automakers are adopting a digital design process for vehicle development. This has elevated the accuracy requirements on the flow simulation software, so that it can be used effectively in the production environment. Evaluation of aerodynamic performance covers prediction of the aerodynamic coefficients such as drag, lift, side force and also lift balance between the front and rear axle. Drag prediction accuracy is important for meeting fuel efficiency targets, prediction of front and rear lifts as well as side force and yawing moment are crucial for high speed handling.
In this paper, we have focused on the evaluation of aerodynamic coefficients on a wide range of BMW validation shapes. This model was chosen as the fidelity of the model was well controlled between experimental measurement and the digital model, which is quite important when correlation studies are carried out. The model that was tested had modular front and rear sections. The rear top quadrant was swapped to represent typical automotive shapes on the road, ranging from a shape with sharper rear end to a more rounded rear end shape, leading to significant variations in the wake flow and flow separation on the backlite and decklid. A section at the front of the hood and grille was replaced to change the radius of the hood corner and inclination angle of the front grille.
In this study we have used PowerFLOW 4.3 flow simulation software based on the Lattice-Boltzmann Method, to evaluate the aerodynamics of the shapes under consideration. We have assessed the accuracy of aerodynamic drag as well as front and rear lift. To obtain the best results, the simulations were performed using an accurate representation of rotating wheel geometry using the sliding mesh approach. For all the shapes that have been tested, the drag coefficient was predicted within 2% of the experimental measurements, with the majority of drag values within 1%. All of the front and rear lift coefficients were within 0.015 (15 counts) of the experimental measurements, with the majority of values within 0.010 (10 counts). These accuracy levels allow the use of digital simulation tools in vehicle design. The paper also considers the improved accuracy of the sliding mesh approach compared to other traditional methods of modeling wheel rotation.