Multiobjective hydraulic optimization of the diffuser vane in an axial flow pump

Abstract

Separation flows tend to induce a chaotic flow field that eventually leads to energy losses and reduced efficiency. The present study performed a multiobjective optimization to improve the hydraulic performance of an axial flow pump at the best efficiency point (BEP) and critical stall point based on the diffuser vane (DV) geometry. Computational fluid dynamics were applied to predict the hydraulic performance of a series of DV models with design points generated through design of experiment. Six different surrogate models were evaluated based on the R-squared criteria. The nondominated sorting genetic algorithm II was also employed to search for optimum solutions for design variables. Hydraulic performance balance between low and high flow rate conditions was analyzed based on the velocity triangle. After optimization, the efficiency and total head at the BEP of the optimum model were increased by 2.341% and 2.779%, respectively, compared to the reference model. Despite the minimal changes to the hydraulic performance at the critical stall point, the optimal operating range was notably expanded in the high flow rate region. Thorough evaluation of losses attributed to horseshoe, corner, and trailing-edge vortices was conducted in meridional planes, multiple spans, and various cross sections in the DV domain. Additionally, the formation and development of turbulent flow were analyzed in detail by transient simulation. Vibration and noise caused by instabilities in the flow characteristics of the reference model were substantially reduced by 36.76% and 67.342% at the first higher-harmonic frequencies at the BEP and the critical stall point, respectively.