Evaluation of drag estimation methods for ship hulls

Abstract: This study aims to evaluate to which degree CFD can complement traditional towing-tank experiments, and to develop a computationally cheap and robust methodology for these type of simulations. Two radically different surface ship hulls were chosen for the trials: a capesize bulk carrier and a fast displacement hull. A bare submarine hull was also used to benchmark the software in the early stages of the study. All simulations were Reynolds–Averaged Navier–Stokes (RANS) simulations using the k-w-SST turbulence model. The chosen software was OpenFOAM 5.x and foam-extend 4.1 coupled with the commercial expansion Naval Hydro Pack, which is supposed to handle high Courant numbers well. Producing a high-quality mesh which is able to capture both the free surface and the boundary layers proved to be of utmost importance and the meshing procedure is thus discussed in detail. A majority of the surface ship simulations suffered from a phenomenon known as numerical ventilation. The effect seemed to be much worse for the fast-displacement hull and the drag estimates for this hull deviated around 16.1% from experimental data. The bulk carrier was less affected and the drag estimates for this hull only deviated around 6.3% from experimental data. In order to reduce the computational cost, all surface ship simulations were run with a maximum Courant number of 25 and some consequences of this are also discussed.