Optimization Design of Energy-Saving Mixed Flow Pump Based on MIGA-RBF Algorithm

Machines

Yong Zhang¹  Guangjuan Wei¹  Rong Lu²  Jianping Yuan²  Qiaorui Si²  Xiaohui Lei³ 

[1] Jiangsu Xugong Construction Machinery Institute Co., Ltd., Xuzhou 221004, China;Research Center of Fluid Machinery Engineering and Technology, Jiangsu University, Zhenjiang 212013, China;State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research, Beijing 100038, China;

关键词: mixed flow pump;    optimization platform;    surrogate model;    entropy generation;    MIGA;   

DOI  :  10.3390/machines9120365

来源: DOAJ

【 摘 要 】

Mixed flow pumps driven by hydraulic motors have been widely used in drainage in recent years, especially in emergency pump trucks. Limited by the power of the truck engine, its operating efficiency is one of the key factors affecting the rescue task. In this study, an automated optimization platform was developed to improve the operating efficiency of the mixed flow pump. A three-dimensional hydraulic design, meshing, and computational fluid dynamics (CFD) were executed repeatedly by the main program. The objective function is to maximize hydraulic efficiency under design conditions. Both meridional shape and blade profiles of the impeller and diffuser were optimized at the same time. Based on the CFD results obtained by Optimal Latin Hypercube (OLH) sampling, surrogate models of the head and hydraulic efficiency were built using the Radial Basis Function (RBF) neural network. Finally, the optimal solution was obtained by the Multi- Island Genetic Algorithm (MIGA). The local energy loss was further compared with the baseline scheme using the entropy generation method. Through the regression analysis, it was found that the blade angles have the most significant influence on pump efficiency. The CFD results show that the hydraulic efficiency under design conditions increased by 5.1%. After optimization, the incidence loss and flow separation inside the pump are obviously improved. Additionally, the overall turbulent eddy dissipation and entropy generation were significantly reduced. The experimental results validate that the maximum pump efficiency increased by 4.3%. The optimization platform proposed in this study will facilitate the development of intelligent optimization of pumps.

【 授权许可】

Unknown