Abstract
This research aims at performing aerodynamic design optimization of automotive vehicle configuration represented by vehicle modelling function for drag reduction by controlling COANDA flow developed around the rear part of the trunk. For this sake, a vehicle modeling function is exactly defined in the form of an exponential function to smoothly express the complex 2D and 3D curved shapes of an automobile. Thirty two 3D aerodynamic virtual car models are developed and CFD analyses are performed for the each case. Neural network models are constructed for efficient aerodynamic optimization especially considering the design variables related to the rear trunk shape of the car. Using deterministic optimization method, drag minimized trunk configuration is deduced. For robust design optimization, approximate moment approach is employed to evaluate the robustness of performance. Subsequently, a relation between the robustness of performance indices and the parameters representing automobile geometry is investigated.
Keywords: Vehicle Modelling Function, 3D Virtual Car Model, COANDA Flow, Drag Coefficient, Robust Design Optimization