Optimization of cavitation characteristics of aviation fuel centrifugal pump inducer based on surrogate model

This paper presents a method for optimizing the cavitation performance of aviation fuel centrifugal pump inducer based on surrogate model with highly accurate simulations. To mitigate the interface blurring effect caused by the dissipation of convective terms, this study first introduces an artificial convective term into the phase control equation to enhance the numerical discretization accuracy and stability. The numerical simulation considering cavitation phase transition effects is implemented based on the OpenFOAM platform; next, a GPR surrogate model for the inducer cavitation performance is constructed using a full-factorial experimental design method. The global sensitivity of the inducer's structural parameters and the mechanism of their interaction effects are revealed, and the numerical errors of the surrogate model are quantified; finally, particle swarm optimization is employed to optimize the design of the inducer component, and the global convergence of results is verified based on the expected improvement function used in Bayesian optimization. Results show that the cavitation number obtained through simulation is 0.052, with an error of only 3.6% compared to experimental results, and the GPR model for fitting the cavitation performance simulator exhibits very small prediction uncertainty. After optimization, the blade inlet angle and edge position are adjusted, and as a result, the cavitation number is reduced from 0.0456 to 0.0407, indicating an obvious improvement of cavitation performance.