Development of a standard series for pump-jet propulsors and data-driven regression for performance prediction

Pump-jet propulsors are widely used in marine applications because of their advantages in propulsive efficiency, cavitation performance, and low noise performance. However, the lack of a standard series database still limits rapid preliminary design and systematic performance comparison of pump-Jet propulsors. To address this issue, this study develops the Standard Pump-Jet Propulsor Series A (SPJP-A) and corresponding data driven regression formulae for open water performance prediction. The proposed series comprises 240 systematically parameterized configurations covering variations in pitch ratio, blade area ratio, stator exit angle, rotor blade number, and stator blade number. A stator exit angle based characterization is introduced to improve the geometric description of rotor–stator matching. Open water performance is evaluated using validated RANS simulations, from which 3801 grid independent operating points are obtained. For the representative validation case, the average error in open water characteristics is below 5%. On this basis, stepwise polynomial regression models with forward selection are established for the main performance coefficients within the SPJP-A design envelope. The models show high predictive accuracy, with cross validation R2 values above 0.999 for all modeled coefficients, while independent test cases yield MAE below 0.021 and RMSE below 0.024. The results also show that rotor–stator torque equilibrium does not necessarily correspond to maximum propulsive efficiency. For some configurations, the highest efficiency occurs under non balanced torque sharing. The proposed SPJP-A series and regression formulae provide a transparent and reproducible tool for preliminary Pump-Jet design, enabling rapid performance evaluation within the prescribed parameter range without repeated CFD simulations.