Quantifying uncertainty in turbulence resolving ship airwake simulations.

High fidelity computational fluid dynamics simulations of ship airwakes are often performed with time accurate, turbulence resolving methods, such as large eddy simulation or detached eddy simulation. In our review of the literature we find that the finite sample error in the results of these simulations is rarely expressed in terms of uncertainty in the statistics. This makes it difficult to determine if any disagreement with experiment or other simulations can be attributed to finite sample time, grid resolution, or modelling approximations. In this paper, turbulence resolving simulations of the airwake over the SFS1 geometry are used to demonstrate a method for quantifying the uncertainty due to finite sample error with an estimated confidence interval. Based on this quantification, a stopping criterion is then formulated so that the simulation time can be set by a predetermined requirement for the confidence in the statistics. [Display omitted] • A comprehensive review of the literature on CFD simulations of ship airwakes is presented. • The dependence of velocity spectra on grid resolution in ship airwake simulations is studied. • A method for estimating the confidence intervals for flow statistics from ship airwake simulations is formulated. • A stopping criterion based on estimated confidence interval widths is proposed. • A concise set of recommendations for grid resolutions, time steps, and simulation lengths for ship airwake CFD is made. [ABSTRACT FROM AUTHOR]