Study on Estimation of Flow Around Marine Propeller by Large-Scale LES and Affection of Grid Resolution and Reynolds Number

Abstract

Abstract To develop a ship with higher propulsion performance, accurate estimation method is required. Tthe performance of ships is estimated by both experimental methods using scale models and numerical methods based on computational fluid dynamics (CFD) using Reynolds average Navier-Stokes equation (RANS) models. Experiments with scale models have limitations regarding fast execution and cost reduction due to physical constraints, and we believe it is important to improve numerical methods. Adjustment of turbulent models can improve the estimation accuracy of numerical methods, but it is necessary to continuously compare and verify the numerical and experimental results. There is a limit to the improvement of the estimation accuracy of the numerical method with RANS models. Direct numerical simulation (DNS), which can be solved without modeling the turbulence field, can acquire an accurate flow field, but it is difficult to implement because it requires a huge number of grids. In a large eddy simulation (LES) model, if the grid resolution (filter size) is fine, the limitations imposed by the model itself are reduced. LES performed with a sufficiently fine grid resolution may be able to achieve the comparable estimation accuracy as DNS without a high computational load. We considered this method to be promising for accurately estimating ship performance. This is an initial study of a ship performance estimation calculation method using large-scale LES. We performed LES calculations with up to 6.4 billion cells for propeller open water test and obtained basic knowledge about large-scale LES calculations. We carried out calculations with different grid resolutions under the same operating conditions, and with different Reynolds numbers at the same resolution to show how the grid resolution and Reynolds number affect the estimation of propeller performance and flow around the propeller.