An evaluation of the RANS method for the prediction of steady ship rudder performance compared to wind tunnel measurements

Description/Abstract

Using Reynolds-Averaged Navier-Stokes (RANS) methods detailed investigations are carried out, focussing on different cases of steady rudder flow. During the introducing two-dimensional cases the hybrid meshing scheme is optimized for rudder use, and grid parameters are studied. Also the effect of solver settings and the capabilities of implemented turbulence models are investigated. Two different verification strategies are applied and compared - eventually concentrating on parametric grid independence studies, rather than using global systematic grid refinement.

The three-dimensional investigation concentrates on the all-movable square tipped NACA0020 rudder tested within the large wind tunnel of the University of Southampton (3.5m x 2.5m). This case is modeled in free stream as well as within the wind tunnel, using the Spalart-Allmaras and the κ-ε RNG turbulence models. The numerical prediction of the tunnel blocking effect is investigated. In particular, the tip vortex flow with its effect on the rudder performance is studied and how this is influenced by the numerical discretization. Here the span-wise distribution of the normal force coefficient is an informative tool. The three-dimensional grids consist of between 2 - 10 to the power of 6 to 4.5 - 10 to the power of 6 cells.

Verification studies are carried out and results are validated against experimental data as far as available. Generally, the lift prediction turned out to show closer agreement with experimental benchmark (within 10%) than the drag, which used to deviate more (within 30%). The prediction of the tip vortex shows the expected characteristics. The tip peak in the span-wise normal force distribution is found to be captured well.