In the past few decades, a number of environmentally friendly technologies in the automotive industry have been developed. Automotive styling is also regarded as an important factor in resolving environmental issues by reducing drag force, which results in high fuel efficiency. This paper demonstrates an efficient automotive aerodynamic optimization process and automatic system based on CFD.
The optimization process consists of three stages, DOE, RSM and Optimization. A Latin Hypercube (LHC) sampling was adopted in the optimization process, performed by modeFRONTIER, and three types of RSM were compared – Radial Basis Function (RBF), Gaussian process (GP) and Kriging (KR). This process was refined by employing AMS-RSM methodology in order to cope with a more complicated parameter study. The AMS-RSM algorithm contributed to the following: 1. searching for lower optimal CD, 2. supplying response surface accuracy, and 3. supplying sampling number and sampling pattern robustness. So as to realize the optimization flow, the CFD solver, mesh morphing and optimization software were systematically integrated and, consequently, the optimal CD values are obtained automatically. As a result, aerodynamic design can be efficiently and precisely forwarded in aerodynamic development. Man hours are also effectively reduced as compared to conventional parametric studies. In our 12-parameter study, the optimum aerodynamic design was obtained within a week. Consequently, valuable guidance in how to advance aerodynamic design works was offered in the early development phase.