Hydraulic optimization of a high-temperature molten salt canned motor pump using an ANN-NSGA-II approach

High-temperature molten salt canned motor pumps are critical components in Generation IV molten salt reactors, and their hydraulic performance directly affects the safety and efficiency of reactor operation. To improve pump performance, this paper presents a multi-objective optimization framework combining artificial neural network (ANN) surrogate models and the NSGA-II algorithm. Computational fluid dynamics (CFD) simulations were conducted to evaluate pump head and efficiency at the design operating point, which were selected as the optimization objectives. The relationships between impeller and guide vane geometric parameters and hydraulic performance were established using ANN surrogate models, and multi-objective optimization was then performed using the NSGA-II algorithm. The optimized pump exhibits improved head over the entire investigated flow rate range, with enhanced efficiency at the design and high flow rates. At the design operating point, the head and efficiency are increased by 1.42% and 0.94%, respectively, with larger improvements achieved at higher flow rates. The performance improvements are attributed to optimized geometric design and improved hydraulic matching of the impeller and guide vane, which reduce fluid friction and impact losses. The proposed optimization framework provides a practical reference for the hydraulic design of pumps used in molten salt reactor systems.