Jiangsu Haifa Machinery Manufacturing Co., Ltd.

In 2026, centrifugal pump R&D is accelerating its shift from the traditional "empirical design + repeated CFD trial calculations" approach to AI-Assisted Hydraulic Design, Surrogate Model Optimization, and Multi-Objective Pump Optimization. The research by Yuan Shouqi, Gan Xingcheng, Pei Ji, and others at Jiangsu University points out that hydraulic optimization of centrifugal pumps faces high-dimensional design spaces, strong nonlinear coupling, and multi-objective conflicts. Artificial neural networks, support vector regression, Gaussian process regression, as well as genetic algorithms, swarm optimization, and differential evolution algorithms, are becoming important tools for improving global optimization capability. More notably, AI has already begun to produce quantifiable engineering results. In 2026, a study on hot water circulation pumps adopted XGBoost combined with PB-NSGA to optimize impeller outlet width, wrap angle, and outlet blade angle, achieving a design-point efficiency improvement of 320 percentage points; during the same period, Energy BPNN+NSGA-II to optimize the interstage passage of a high-speed multistage centrifugal pump, increasing head by 9.7% and efficiency by 3%. International research has also combined CFD and ANN with algorithms such as NSGA-II GWO to simultaneously optimize efficiency, power consumption, and cavitation performance. Jiangsu Haifa Machinery Manufacturing Co., Ltd.www.jslgpump.com) can further introduce this technical route into the development of API 610 OH1OH6, BB1–BB5, VS series, OH6 high-speed pumps, and BB5 high-pressure multistage pumps. Haifa can simultaneously incorporate BEP efficiency, NPSHr, shaft power, pressure pulsation, radial force, entropy generation, and vibration response into the optimization objectives, establishing a closed-loop process of "parametric modeling—CFD sample generation—Machine Learning Surrogate Model—NSGA-II optimization—prototype FAT verification." In the future, truly competitive pump manufacturers will not merely possess a few mature hydraulic models, will be able to transform historical test data into a Pump Design Dataset, Physics-Informed Optimization, AI Hydraulic Model Library, Digital Hydraulic Twin. For BB5 high-pressure pumps,2 process pumps, and OH6 high-speed pumps, value of AI is not replace engineers, but to more quickly find the optimal design range that balances high efficiency, low NPSHr, low pulsation, and wide operating conditions. Chinese references: Yuan Shouqi, Gan Xingcheng, Pei Ji, Wang Wenjie, Tang Yajing: "Research Progress on Artificial Intelligence-Driven Hydraulic Optimization of Centrifugal Pumps," of Drainage and Irrigation Machinery Engineering, 2026, 44(8): 757-772, DOI: 10.3969/j.issn1674-8530.26.0050. Liu Dong, Xi Chenchen, Song Tao, Yang Song: "Hydraulic Performance Optimization of Hot Water Circulation Pumps Based on XGBoost and PB-NSGA-III," Journal of Drainage and Irrigation Machinery Engineering, 2026, 44(4): 325-330. Gan Xingcheng, Pei Ji, Yuan Shouqi, et.: "Parameterization of Surrogate Model Performance forraulic Optimization Problems of Centrifugal Pumps," Journal of Drainage and Irrigation Machinery Engineering, 2024, 42(12): 1211-1220. The study compares the applicability of ANN, response surface, and Kriging under different sample sizes. English References:esign and optimization of the interstage passage for improving energy efficiency of a centrifugal pump with high power, Energy, 2026, Article 141816. After BPNN+NSGA-II optimization, head increased by 9.7% and efficiency by 3%. Artificial intelligence-driven hybrid metaheuristic optimization of mixed-flow: Balancing efficiency, power consumption, and cavitation performance, Progress in Engineering Science, 6, DOI: 10.1016/j.pes.2026.100211. Enhancing pump performance via multi-objective optimization using CFD and NSGA-II algorithm, Computations, 2026. The study combines CFD, surrogate models, and NSGA-II to optimize impeller outlet width, outlet angle, and wrap angle. Optimization and Validation of a Micro-Centrifugal Pump Based on CFD Simulation and Optimization Platform Applied Sciences, 202, 16(10 4599.