Proper Orthogonal Decomposition, surrogate modelling and evolutionary optimization in aerodynamic design

• We propose a reduced order model of aerodynamic flow for shape optimization problems.
• The metamodel is based on Proper Orthogonal Decomposition and Radial Basis Functions.
• Zonal approach improves the quality of shock-wave prediction in transonic flow.
• Adaptive sampling improves the surrogate prediction quality all over the design space.
• Evolutionary optimization assisted with zonal-adaptive model shows superior results.

A computational methodology is proposed for CFD-based aerodynamic design to exploit a reduced order model as surrogate evaluator. The model is based on the Proper Orthogonal Decomposition of an ensemble of CFD solutions. A zonal approach is presented to better solve the shock wave region and improve the surrogate prediction in transonic flow. Model validation and in-fill criteria are shown as valid tools to examine the accuracy of the surrogate and, therefore, to feed the model back with “intelligent” information. The reduced order model is integrated in an evolutionary optimization framework and used as fitness evaluator to improve the aerodynamic performances of a two-dimensional airfoil. Finally, the performances of the surrogate-based shape optimization are compared to the efficiency of a meta-model assisted optimization and to the accuracy of a plain optimization, where, instead, each aerodynamic evaluation is performed with the high-fidelity model.