Part 2: Simulation methodology and numerical uncertainty for RANS-CFD for the hydrodynamics of wind-assisted ships operating at leeway angles

Abstract

A Reynolds-averaged Navier Stokes computational fluid dynamics (RANS-CFD) package will be one of the primary tools used during the development of a performance prediction program for wind-assisted commercial ships. This paper describes the simulation verification exercise, performed in support of the experimental validation presented in Part 1 of this two-part series describing the RANS-CFD method employed in this research. The predominance of large-scale separated flow structures in the wake of the sailing ship, an artefact of sideforce production necessary for sailing, points to a careful verification exercise and estimate for the numerical uncertainty to support the systematic investigation of wind-assisted ship hydromechanics and meshing guidelines within the available computer resources. Methods for CFD uncertainty quantification are defined and implemented for verification cases at leeway angles equal to 0^ᵒ, 6^ᵒ, and 9^ᵒ. Analysis for four sets of grids with different meshing strategies and for varying time steps results in a grid definition and time step for simulation validation. Numerical uncertainty as adopted in Part 1 for validation is defined. Finally, the meshing strategy for full-scale simulation is described, as used for the production runs of the Delft Wind Assist Series.