Double-suction centrifugal pumps form an integral part of power plant systems in maintaining operational stability. However, there has been a common problem of achieving a better cavitation performance over a wider operating range because the traditional approach for impeller design often leads to the design effect not meeting the operational needs at off-design conditions. In addressing the problem, an optimization scheme was designed with the hub and shroud inlet angles of the double-suction impeller to minimize the suction performance at non-design flow conditions. A practical approach that speeds up the cavitation simulation process was applied to solve the experimental design, and a multi-layer feed forward artificial neural network (ANN) was combined with the non-dominated sorting genetic algorithm II to solve the multi-objective problem into three-dimensional (3D) Pareto optimal solutions that meet the optimization objective. At the design point, the suction performance was improved by 6.9%. At non-design flow conditions, the cavitation performance was improved by 3.5% at 1.2Q_d overload condition, 4% at 0.8Q_d, and 5% at 0.6Q_d. Additionally, there was significant reduction in the attached cavity distribution in the impeller and suction domains when the optimized model was compared to the original model at off-design points. Finally, the optimization established a faster method for a three-objective optimization of cavitation performance using ANN and 3D Pareto solutions.

In order to improve the operation performance of the multi-stage double-suction centrifugal pump and reduce the internal energy loss of the pump, this paper proposes a single-objective optimization design method based on the non-hierarchical response surface methodology (RSM) and the multi-island genetic algorithm (MIGA). Nine parameters, such as the blade outlet width and blade wrap angle, were used as design variables, and the optimization objective was efficiency under design conditions. In total, 149 sets of valid data were obtained under the Latin hypercube sampling method (LHS), the corresponding thresholds were set for efficiency and head, and 99 sets of valid data were obtained. A cross-validation analysis of the sieved data was carried out based on non-hierarchical RSM, global optimization of the efficiency was carried out using MIGA, and numerical verification was carried out via CFD. The research results show that compared with hierarchical RSM, non-hierarchical RSM can approximate the nonlinear relationship between the objective function and the design variables with higher accuracy, and the model fitting R² value was 0.919. The efficiency was improved by 3.717% after optimization. The overall prewhirl of the impeller inlet after optimization decreased, the internal speed of the volute significantly improved, the large-area vortex at the volute and the outlet pipe was eliminated, the impact loss at the volute separating tongue disappeared, and the overall hydraulic performance of the pump was improved. The total entropy output value of the optimized pump was reduced by 4.79 (W/K), mainly concentrated on the reduction in the entropy output value of the double volute, and the overall energy dissipation of the pump was reduced.

In order to improve the operation performance of the multi-stage double-suction centrifugal pump and reduce the internal energy loss of the pump, this paper proposes a single-objective optimization design method based on non-hierarchical response surface model (RSM) and the multi-island genetic algorithm (MIGA). Nine parameters, such as the blade outlet width and blade wrap angle, were used as design variables, and the optimization objective was the efficiency under design conditions. In total, 149 sets of valid data were obtained under the latin hypercube sampling method (LHS), the corresponding thresholds were set for efficiency and head, and 99 sets of valid data were obtained. A cross-validation analysis of the sieved data was carried out based on non-hierarchical RSM, global optimization of the efficiency was carried out using MIGA, and numerical verification was carried out via CFD. The research results show that compared with hierarchical RSM, non-hierarchical RSM can approximate the nonlinear relationship between the objective function and the design variables with higher accuracy, and the model fitting R2 value was 0.919. The efficiency was improved by 3.717% after optimization. The overall prewhirl of the impeller inlet after optimization decreased, the internal speed of the volute significantly improved, the large-area vortex at the volute and the outlet pipe was eliminated, the impact loss at the volute separating tongue disappeared, and the overall hydraulic performance of the pump was improved. The total entropy output value of the optimized pump was reduced by 4.79 (W/K), mainly concentrated in the reduction in the entropy output value of the double volute, and the overall energy dissipation of the pump was reduced.

Double-suction centrifugal pumps are extensively employed in industrial applications owing to their high efficiency, low vibration, superior cavitation resistance, and operational durability. This study analyzes how impeller oblique cutting angles (0°, 6°, 9°, 12°) affect a double-suction pump at a fixed 4% trimming ratio and constant average post-trim diameter. Numerical simulations and tests reveal that under low-flow (0.7Q_d) and design-flow conditions, the flat-cut (0°) minimizes reflux ratio and maximizes efficiency by aligning blade outlet flow with the mainstream. Increasing oblique cutting angles disrupts this alignment, elevating reflux and reducing efficiency. Conversely, at high flow (1.3Q_d), the 12° bevel optimizes outlet flow, achieving peak efficiency. Pressure pulsation at the volute tongue (P11) peaks at the blade-passing frequency, with amplitudes significantly higher for 9°/12° bevels than for 0°/6°. The flat-cut suppresses wake vortices and static–rotor interaction, but oblique cutting angle choice critically influences shaft-frequency pulsation. Entropy analysis identifies the volute as the primary loss source. Larger oblique cutting angles intensify wall effects, increasing total entropy; pump chamber losses rise most sharply due to worsened outlet velocity non-uniformity and turbulent dissipation. The flat-cut yields minimal entropy at Q_d. These findings provide a basis for tailoring impeller trimming to specific operational requirements. Furthermore, the systematic analysis provides critical guidance for impeller trimming strategies in other double-suction pumps and pumps as turbines in micro hydropower plants.