Abstract
Keywords: Vehicle Aerodynamics, Computational Fluid Dynamics (CFD), Large Eddy Simulation (LES), High Performance Computing (HPC), Unsteady Aerodynamic Force
In the previous study, we intensively modified and optimized the unstructured-grid LES code "FrontFlow/red (FFR)", which had been originally developed by us under the national project called "Revolutionary Simulation Software (RSS21)", for the execution on the Earth Simulator (ES). We then successfully applied our High Performance Computing (HPC) LES to the ASMO shape and a formula car using 25 to 125 million (world-largest class engineering LES) numerical elements, and showed the validity of the method compared with Reynolds-Averaged Navier-Stokes Simulation (RANS), with regard to the prediction of steady aerodynamic forces.
The objective of this study is to validate the HPC-LES for the prediction of unsteady aerodynamic forces on road vehicles. The unsteady forces will be remarkable in such conditions as sudden steering action, overtaking, or cross wind, and are difficult to estimate by a conventional wind tunnel test. Large Eddy Simulation (LES) will be an encouraging solution for this matter, because it can reproduce unsteady turbulence characteristics with high accuracy.
It is assumed that the unsteady aerodynamic forces acting on a vehicle can be explained as a solid-fluid interaction through the organized eddy structures appearing around the body. Thus, we first investigate how the HPC-LES properly reproduce eddy structures around a vehicle by comparing those with an experimental visualization. For this purpose the method is applied to the production vehicle, Mazda Atenza, on which reliable wind-tunnel data had been obtained at Mazda Motor Corp. Total of about 40 million numerical elements were required to properly reproduce the complicated shapes such as the engine room and the underbody geometry. About 500 CPU with 300GB memory and 40 hours of real computational time were consumed on the ES. It was demonstrated that LES properly captures characteristic flow structures and excellent agreement between LES and experimental visualization was achieved.
We then studied the unsteady aerodynamic forces acting on the vehicle during the dynamic yaw-angle change, and their relationships with transient flow structures were focused. The unsteady aerodynamic forces such as the yawing-moment were found to be relatively large during the vehicle dynamic motion compared with the case when the motion is static.
As a result, we have demonstrated that HPC-LES can be a dominant aerodynamic assessment tool especially for the unsteady vehicle aerodynamics, which would contribute to the innovative aerodynamic vehicle design in the near future