Improving Models and Simulations for Automotive Drag Predictions

CFD simulations for automotive bluff bodies often rely on simplified models for both the vehicle and the computational domain. Current models maintains simplified shape but fail in capturing the full flow behaviour over a real vehicle. A model including vehicle body and tunnel geometry is presented in the literature and has been modified in order to improve its fidelity and reduce the di”erences with real life facilities. This model features an easy approach for the generation of the vehicle and wind tunnel models through 21 parameters as well as an automatic process for the setup of CFD cases, from meshing to simulation. Changes to the numerical schemes used by the CFD software are made to account for the variable quality of the grid. Moreover a moving ground system has been included to replicate the real system present in some automotive wind tunnels. Another key aspect for the prediction of drag coe!cient for automotive blu” bodies is that the same vehicle in di”erent facilities often leads to di”erent drag coe!cients, with an average discrepancy of 6 to 10%. A parametric setup based on the improved models is carried out and a correction method is proposed by combining new sampling points with the results obtained previously in the literature. This new combined correction shows promising results while featuring a significanlty lower amount of samples, reducing the standard deviation by 43% compared to the current state of art when applied to a small dataset of 10 experimental points (5 for each vehicle in 2 di”erent wind tunnels).