CFD Prediction of Turbulent Flow on an Experimental Tidal Stream Turbine using RANS modelling

A detailed computational fluid dynamics (CFD) study of a laboratory scale tidal stream turbine (TST) is presented. Three separate Reynolds Averaged Navier Stokes (RANS) models: the k-epsilon and k-omega-SST eddy-viscosity models, and the Launder-Reece-Rodi (LRR) Reynolds stress model, are used to simulate the turbulent flow-field using a new sliding-mesh method implemented in EDF's open-source Computational Fluid Dynamics solver, Code_Saturne. Validation of the method is provided through a comparison of power and thrust measurements for varying tip-speed ratios (TSR). The SST and LRR models yield results within several per cent of experimental values, whilst the k-epsilon model significantly under-predicts the force coefficients. The blade and turbine performance for each model is examined to identify the quality of the predictions. Finally, detailed modelling of the turbulence and velocity in the near and far wake is presented. The SST and LRR models are able to identify tip vortex structures and effects of the mast as opposed to the standard k-epsilon model.