A hybrid approach for surrogate-based aerodynamic optimization with constraints

Kurzfassung

An optimization framework combining TAU code with different options regarding the solver’s type, different design of experiment methods, optimization algorithms and various surrogate modelling methodologies has been developed. For the sake of finding the global optimum accurately, the surrogate model is adaptively refined based on the Kriging error, the expected improvement function and the cost function on the surrogate model. Two test cases, optimizing an artificial objective function with two optima for a NACA 4-digit airfoil with two design variables and minimizing the drag of the transonic RAE2822 airfoil with 10 design parameters, indicate that the developed framework combined with a hybrid sample refinement strategy can efficiently capture the global optimum. Furthermore, a high-lift configuration with 10 design parameters is optimized under constraints on the lift and moment coefficient, demonstrating a promising industrial application of the optimization framework.