Optimization of suction control on separation flow around an airfoil at low Reynolds number

In order to obtain the optimal set of suction parameters for different objectives, an optimization method is developed to solve single-objective and multi-objective problems, by combining RBF neural network and genetic algorithm. Considering the separation flow over a NACA0012 airfoil surface, some local porous suction regions are mounted on the upper surface. RBF neural network is used as the surrogate model to substitute CFD computation, so as to reduce the amount of computation. The corresponding single-objective and Pareto multi-objective optimizations are performed using genetic algorithm. The optimization results show that the present optimization method has satisfactory convergence and accuracy. The maximum increase in lift-to-drag ratios up to 2.4 times is achieved after performing single-objective optimization of maximum lift-to-drag ratio. The uniform distributing and satisfied Pareto front is gained by Pareto multi-objective optimization, which provides a selective database with effective solutions. As long as the suction angle, hole diameter and hole spacing are reasonable, only smaller suction coefficient is required to get significant lift enhancement, and keep a high FOM value, so that the entire control system maintains a high level of energy efficiency ratio.