Two schemes of multi-objective aerodynamic optimization for centrifugal impeller using response surface model and genetic algorithm

This paper presents two schemes of multi-objective aerodynamic optimization design for centrifugal impeller blade. One is genetic algorithm(GA) combined with a commercial computational fluid dynamics(CFD) software, and the other is GA combined with the surrogate model. The two schemes are respectively applied to multi-objective optimization for the same centrifugal impeller blade. For multi-objective genetic algorithm(MOGA), non-uniform mutation and Pareto ranking and fitness-sharing technique are used to obtain fast convergence speed and good capability to search the Pareto front of GA. For the surrogate model based on radial basis function(RBF), design of experiments(DOE) technology is adopted to select samples. The parameters and weight coefficients in the surrogate model are solved by GA instead of traditional least square method. According to the geometrical feature of centrifugal impeller, a three-dimensional reconstruction method for the blade shape based on nonuniform rational B-spline(NURBS) is introduced. The numerical simulation is used to evaluate the aerodynamic performance of the optimal and initial impeller. The computational results show that the aerodynamic performance of impellers designed by both optimization schemes is improved to some extent. At the same time, the main reasons for the improvement in aerodynamic performance of the optimal impeller are revealed. For the optimal impellers, the isentropic efficiency and total pressure ratio are increased by about 1.0% and 3.0% respectively. Through comparison of two schemes applied to the centrifugal impeller optimization design, it is found that the computational performance of the second optimization scheme is superior to that of the first optimization scheme.