Aerodynamic simulation results are most of the time compared to wind tunnel results. It is too often simplistically believed that it suffice to take the CAD geometry of a car, prepare and run a CFD simulation to obtain results that should be comparable. With the industry requesting accuracies of a few drag counts when comparing CFD to wind tunnel results, a careful analysis of the element susceptible of creating a difference in the results is in order.
In this project a detailed 1:4 scale model of the Hyundai Genesis was tested in the model wind tunnel of the FKFS. Five different underbody panel configurations of the car were tested going from a fully paneled car to a car without panels. The impact of the moving versus static ground was also tested, providing over all ten different experimental results for this car model. These ten different configurations were simulated having in mind to reproduce the testing condition in every possible aspect to be able to truly evaluate the accuracy of the CFD results. In the setup process, three aspects were considered: geometry precision, wind-tunnel environment and conditions and finally evaluation of the proper reference values.
In the geometry preparation process an additional step was added to morph the geometry prepared from the CAD model. The target used was scanned data obtained from the physical model after the test session. The wind tunnel environment was reproduced in the setup to include all side effects (pressure gradient, jet expansion, etc.) and boundary conditions found in the FKFS scale model wind tunnel during the testing session. When reproducing wind tunnel environment in a simulation it is necessary to evaluate the proper values of dynamic pressure and static pressure needed to non-dimensionalize the results. A new method was used to evaluate the proper reference values. The empty wind-tunnel method was also used and results are compared to the new method.
This paper shows that it is possible to achieve very accurate CFD results using PowerFLOW if all aspects of the testing condition are systematically reproduced in the simulation setup.