High Performance Computing: Analytical Aerodynamics for Automotive Vehicles

Abstract

High Performance Computing (HPC) represents a significant resource whereby automotive manufacturers can utilize analytical methodologies to reduce experimental testing and design time, resulting in lower costs and higher quality. Optimization of styling and aerodynamics requires multiple CFD simulations which have been enabled by the commercial availability of parallel algorithms, as well as enhancements in computer architectures. We have developed a Virtual Aerodynamic Wind Tunnel (VAWT) which uses PowerFLOW® and can simulate conditions similar to experimental wind tunnels. One key element of this methodology is the use of PowerFLOW. Two of the major attributes of PowerFLOW are its inherent parallelization and automeshing capabilities. In this paper, we will focus on the scalability and feasibility of PoweFLOW, which is essential for the optimization of styling and aerodynamics. Timing and scalability results on an Origin 2000 server are presented for a number of different configurations.