Large Eddy Simulation of an asymmetric spinnaker

Abstract

This work presents a numerical study into the aerodynamics of downwind sailing using different methods for modelling turbulence, comparing Large Eddy Simulation (LES) and Reynolds-Averaged Navier-Stokes (RANS) methods. Previously published experimental data of downwind sails have shown areas of flow separation and vortex structures at the leading and trailing edge of the gennaker, making these sails challenging to model accurately. The ability of LES to model the transition to turbulence within separated shear layers can lead to an accurate prediction of the leading edge separation bubble, which significantly influences the flow field around the top half of the sail, and downstream of the layer's reattachment. The transient nature of the LES solution allows the computation of the creation and shedding of unsteady vortices at the leading edge and downstream of the sail draft. The effect of the vortex rolls being convected towards the trailing edge is to generate a boundary layer which is more resistant to separation. Comparison with the pressure distribution measured in experiment shows that LES correctly predicts the location of separation, while RANS erroneously predicts a larger region of stalled flow which limits the suction on the sail. As a result, the overall drive and side forces computed by the LES are in better agreement with the experimental data, with less than 3.5% error, that the RANS which underestimates their magnitude by more than 14%. However, it should be noted that the accuracy of the LES calculations has not been demonstrated, and so the results should be treated with caution and not considered conclusive.