Abstract
Keywords: Surrogate model, response surface, optimization, Pareto-optimal solution
The present paper investigates use of the surrogate models for the multi-objective shape optimization of vehicles. The objectives of interest are related to the aerodynamic performance of vehicles such as air resistance (drag), lift force, moments, aeroacoustic properties or soiling and accumulation of water. The suggested technique is tested here on a two-dimensional vehicle geometry with four design variables and two objective functions. Faced centered composite design is used to define the data points in the design space that will be used for the computer experiments (CFD) with a steady solver and standard k - turbulence model. The second order polynomial response surface model is then constructed from the computer calculated responses from CFD. Search for the optimal design is done using the -constraint method. In order to investigate the correlation between the two objectives (the drag and the lift) the Pareto-optimal solutions were computed. Hierarchical cluster algorithm is finally used to analyze the Pareto optimal solutions and to draw conclusions about the design. Finally some improvements of the technique that are required in order to make the suggested technique to become an engineering tool are discussed.