A Gradient Based Optimization Workflow Based on CFD Adjoint Solver and Advanced RBF Mesh Morphing
Over the last decades the maturity of the CFD solvers and their companion adjoint solvers made possible the computation of the sensitivity of an observed performance (a pressure drop, a drag coefficient, a downforce) with respect to the shape of a component. It is nowadays a very well consolidated practice to explore shape sensitivity maps helping the designer to decide where moving outward/inward surfaces to improve the performance.
In the present study, we show how advanced mesh morphing allows creating a set of shape variations which intensities can be combined to get the resulting performance together with its derivatives, thanks to the adjoint sensitivity. The computational cost (a CFD run and an adjoint one) stays the same whatever is the number of shape parameters introduced and an effective gradient based optimization can be enabled. The approach is demonstrated for the shape optimization of high performance automotive runners adopting modeFRONTIER, Ansys Fluent and RBF Morph.