Abstract: In order to improve hydraulic efficiency and head of a low specific-speed centrifugal pump, a low specific-speed centrifugal pump with 30 specific-speed was selected as a research object in this study. By taking the maximum head and hydraulic efficiency of centrifugal pump as two optimization objectives, the basic equation of centrifugal pump and the Plackeet-Burman test are combined for screening parameters to be optimized. By doing so. the vane outlet placement angle of centrifugal pump impeller was selected, and the wrap angle and outlet width of the blades were selected too as variables for optimization in our study. In the optimization process, 30 groups of experiments were arranged using the optimal Latin super-cubic design method, and then an approximate model between the optimization target and the variables was fitted numerically by using the RBF neural network model. The multi-objective optimization was finally carried out in terms of NSGA-Ⅱ genetic algorithm. The optimization results show that the head of the optimized impeller basically remains unchanged, the hydraulic efficiency of the impeller is 5.82% higher than that before optimization. The optimization eliminates the hump on the flow-head curve of the impeller and makes the operation of the centrifugal pump more stable. The pressure gradient in the impeller flow channel is reduced after optimization, and the area as well as size of eddies are reduced apparently to some extent. The turbulence area in the impeller flow channel becomes uniformly distributed, and blades to work are enhanced as a result. Thus the hydraulic efficiency has been improved.

Key words: centrifugal pumps, low specific speed, head, hydraulic efficiency