Enhancing pump performance via multi-objective optimization using CFD and NSGA-II algorithm

Purpose This article aims to improve the centrifugal pump’s head performance and hydraulic efficiency through the application of an integrated multi-objective optimization (MOO) approach. Optimization of impeller design parameters seeks to enhance operating efficiency and increase the pump’s operating range in terms of changing flow rates. Design/methodology/approach The suggested methodology combined computational fluid dynamics (CFD) simulations, surrogate modeling, and Non-dominated Sorting Genetic Algorithm II (NSGA-II). Three geometric parameters (outlet width, blade outlet angle and wrap angle) were systematically varied. In order to accurately forecast pump performance, radial basis function (RBF) networks and linear regression surrogate models have been developed. Subsequently, NSGA-II was utilized for MOO targeting improvements in head and hydraulic efficiency. Two optimization strategies have been implemented: one focus on best efficiency point and the second on different flow ranges. Findings The first optimization approach resulted in improvements of 4.39% in head and 5.94% in hydraulic efficiency at BEP. The second optimization process enhanced head performance by 6.42% and efficiency by 7.63% and produced a steeper H-Q curve for uses where high control sensitivity of the flow is needed. The width of the impeller outlet was the controlling factor, while blade outlet angle had little effect in range of the test. Originality/value Using simulation coupled with efficient surrogate modeling techniques, it builds computationally efficient framework to reduce optimization time while preserving accuracy. It highlights the strong impact of impeller outlet width on pump performance and introduces a multi-optimization strategy to promote operational flexibility, providing significant insights and guidelines for designing superior turbomachinery.