Computational Fluid Dynamics (CFD) simulations using Reynolds Averaged Navier-Stokes (RANS) equations are increasingly adopted as an analysis tool to predict the hydrodynamic coefficients of underwater vehicles. These simulations have shown to offer both a high degree of accuracy comparable to experimental methods and a greatly reduced computational cost compared to Large Eddy Simulation (LES) and Direct Numerical Simulation (DNS). However, one of the major challenges faced with CFD simulations is that the results can vary greatly depending on the numerical model settings. This paper uses the DARPA SUBOFF hull form undergoing straight-line and rotating arm manoeuvres at different drift angles to analyse the hydrodynamic forces and moments on the vehicle against experimental data, showing that the selection of the boundary conditions and turbulence models, and the quality of the mesh model can have a considerable and independent effect on the computational results. Comparison between the Baseline Reynolds Stress Model (BSLRSM) and Shear Stress Transport with Curvature Correction (SSTCC) were carried out for both manoeuvres, showing that with a sufficiently fine mesh, appropriate mesh treatment, and simulation conditions matching the experiments; the BSLRSM predictions offer good agreement with experimental measurements, while the SSTCC predictions are agreeable with the longitudinal force but fall outside the experimental uncertainty for both the lateral force and yawing moment.