Mesh and domain optimization strategies for a RANS solver to accurately estimate the open water propulsive characteristics of fixed pitch propellers are proposed based on examining the effect of different mesh and computation domain parameters. The optimized mesh and domain size parameters were selected using Design of Experiments (DoE) methods enabling simulations to be carried out in a limited memory environment, and in a timely manner; without compromising the accuracy of results. A Reynolds-Averaged Navier Stokes solver is used to predict the propulsive performance of a fixed pitch propeller. The predicted thrust and torque for the propeller were compared to the corresponding measurements.
A total of six meshing parameters were selected that could affect the computational results of propeller open water performance. A two-level fractional factorial design was used to screen out parameters that do not significantly contribute to explaining the dependent parameters: namely simulation time, propeller thrust and propeller torque. A total of 32 simulations were carried out only to find out that the selected six meshing parameters were significant in defining the response parameters. Optimum values of each of the input parameters were obtained for the DOE technique and additional simulations were run with those parameters. The simulation results were validated using open water experimental results of the same propeller.
It was found that with the optimized meshing arrangement, the propeller opens simulation time was reduced by at least a factor of 6 as compared to the generally popular meshing arrangement. Also, the accuracy of propulsive characteristics was improved by up to 50% as compared to published simulation results. The methodologies presented in this paper can be similarly applied to other simulations such as calm water ship resistance, ship propulsion to systematically derive the optimized meshing arrangement for simulations with minimal simulation time and maximum accuracy. This investigation was carried out using STAR-CCM+, a commercial CFD package; however the findings can be applied to any RANS solver.