Multi-objective optimization on diffuser of multistage centrifugal pump base on ANN-GA

Tianxin Wu,Denghao Wu,Yun Ren,Yu Song,Yunqing Gu,Jiegang Mou
DOI: https://doi.org/10.1007/s00158-022-03278-x
IF: 4.279
2022-06-16
Structural and Multidisciplinary Optimization
Abstract:As the core equipment of the water supply system, multistage centrifugal pumps produce high energy consumption every year. Thus, in the context of global warming, it is urgent to improve the efficiency of multistage centrifugal pumps. The diffuser is a key component of the pump, and it directly affects the efficiency of the pump. This research proposes a multi-objective optimization method for the diffuser based on the artificial neural network (ANN) and genetic algorithm (GA). First, diffusers are modeled parametrically by adjusting the five geometrical variables. The multi-objective optimization design of diffusers is carried out with head and Minimum Efficiency Index (MEI) as optimization objectives. MEI is an official and comprehensive index to evaluate pump efficiency under three different flow rates of 0.75 Q d , 1.0 Q d , and 1.1 Q d . Meanwhile, an entropy production method is introduced as an energy loss identification tool in the proposed ANN-GA method to help evaluate the internal flow losses quantitatively and understand the root causes directly. The result shows that the multi-objective optimization method is suitable for the optimization design of diffusers under different flow rates. Compared with the original model, the head of the optimized model is improved by 1.47 m at the designed point and the C MEI is reduced by 1.89. And more stable flow and lower energy loss were realized by introducing the optimized diffuser. These results provide a new solution for the optimal design of multistage centrifugal pump diffusers.
mechanics,computer science, interdisciplinary applications,engineering, multidisciplinary
What problem does this paper attempt to address?