Comparison of Propeller and Self-Propulsion Performance of a Submarine Using RANS and LES Turbulence Models

Underwater radiated noise (URN) from submarine propulsors requires reliable flow prediction as a prerequisite for cavitation and acoustic analyses. This study validates a computational fluid dynamics (CFD) workflow to predict the propeller open-water (POW) and self-propulsion performance of a SUBOFF 5470 hull equipped with an E1619 propeller, while focusing on the effect of turbulence modeling and a numerical approach. Reynolds-averaged Navier–Stokes (RANS) simulations were performed using the shear stress transport k– and Reynolds stress model (RSM), while a large eddy simulation (LES) was conducted using wall-modeled and wall-resolved treatments for the hull and propeller, respectively. The POW calculations were conducted using both the moving reference frame (MRF) and sliding-mesh unsteady approaches, and the mesh convergence was assessed using the grid convergence index (GCI). The GCI analysis showed monotonic convergence, with numerical uncertainties for the thrust and torque coefficients within 2%. In POW predictions, differences between MRF and sliding-mesh results were small, indicating consistent performance trends across both numerical approaches. Among the tested advance ratios, the RSM showed the closest agreement with the experimental thrust and torque characteristics, while LES more clearly resolved blade-induced wake nonuniformity and coherent vortical structures. In the self-propulsion analysis, the model-to-model comparison confirmed that RANS-based approaches were effective for global performance prediction, while LES provided additional advantages in resolving detailed wake structures relevant to subsequent cavitation and acoustic analyses. These findings help establish a validated CFD procedure for submarine propulsor analysis and provide a practical baseline for future cavitation and URN predictions under cavitation-tunnel conditions.