A mixed-flow waterjet pump with a vaneless diffuser is treated to improve its hydraulic efficiency as well as cavitation performance. In order to conduct the design optimization, the authors apply a multiobjective strategy combined with design of experiments (DOE), computational fluid dynamics (CFD), inverse design method, surface response method (RSM) and non-dominated sorting genetic algorithm-II (NSGA-II). The hydraulic efficiency and the total vapor volume are selected as the optimization targets, and nine parameters are used to describe the blade shape with the same meridional section.
For numerical simulation, RANS method is applied with SST k-ω turbulence model and a mass transfer cavitation model based on the Rayleigh-Plesset equation. Optimal Latin hypercube design method is used in the design of experiments to uniformly sample in variation ranges and global optimization is then conducted by using non-dominated sorting genetic algorithm-II (NSGA-II) based on the input-target approximation functions built by the response surface model (RSM).
The optimization results demonstrate that both hydraulic efficiency and cavitation performance are improved at the design point through this multiobjective strategy. Based on analysis of the internal flows, secondary flows would be important contributor to the hydraulic loss as well as the nonuniform flow at impeller exit, and can be suppressed by adjusting the blade load along the hub or shroud by using the inverse design method.