A multi-objective aerodynamic performance optimization of a centrifugal vacuum pump and the flow physics analysis

This work performs a multi-objective aerodynamic optimization for a high-vacuum centrifugal vacuum pump. The hub and shroud profiles of the impeller were chosen as the geometrical factors for the optimization and were parameterized using the Bezier curves. The total pressure ratio (πtt) and polytropic efficiency (η) were set as the optimization objectives. The sample models were generated using the Reynolds-averaged Navier–Stokes simulations based on the Latin hypercubic sampling method. The backpropagation neural network was employed as the surrogate model, and the multi-objective optimization was performed using the non-dominated sorting genetic algorithm. The Pareto front of the optimization was obtained, and two optimized models, named OPT1 and OPT2, were selected for the simulation and analysis of the internal flow. The numerical results reveal that compared with the baseline model, the two optimized models improve πtt by 1.009% and 1.863%, respectively, and the absolute magnitude of η by 1.206% and 1.019%, respectively. The analysis of the internal flow of the baseline and optimized centrifugal vacuum pumps demonstrates that the optimized hub and shroud produce a comparably better uniform flow in the impeller, with remarkably reduced separation and reduced Mach number. The significant static entropy and entropy generation formed near the shroud are substantially weakened in magnitude and size. The flow in the radial vaneless region of the optimized models presents uniform distributions of flow angle and Mach number over the whole circumference. The reduced energy loss and uniform flow field contribute to the improved performance of the centrifugal vacuum pump.