Prediction of the maneuvering performance of autonomous underwater vehicles equipped with pump–jet propulsion remains computationally intensive when relying solely on high-fidelity computational fluid dynamics. To overcome this limitation, a surrogate maneuvering model is developed to achieve comparable accuracy with drastically reduced computational cost. The model is constructed from numerical results obtained using unsteady Reynolds-averaged Navier–Stokes equations with the k–ω shear stress transport turbulence model, and formulated through a Taylor-expansion-based framework. The propulsion and rudder modules are refined to enhance physical representation and efficiency: a conventional open-water-based formulation is adopted to embed the pump–jet propulsive model, incorporating axial flow velocities near the duct inlet for improved thrust prediction; meanwhile, the rudder force model minimizes the number of captive simulations by employing a kinematic approach that compensates for limited datasets. The surrogate model is applied to free-running simulations and validated against high-fidelity computational results. The findings confirm that the proposed framework reproduces the dominant trends of kinematic responses, forces, and moments with high consistency, providing a practical and time-efficient alternative for maneuvering prediction of underwater vehicles equipped with pump–jet propulsion systems.
Kwon et al. (Tue,) studied this question.