Abstract

Accurate prediction of wind turbine (WT) wake is essential for optimising wind farm layouts and maximising energy production. Traditional wake models, such as the Jensen and Park models, are commonly used in WT simulations but often struggle to capture wake characteristics accurately. Furthermore, high-fidelity Computational Fluid Dynamics simulations are computationally intensive, limiting their applicability for large-scale simulations. This research introduces an innovative approach to WT wake modelling using the Spalart-Allmaras (SA) turbulence model within a steady Reynolds-Averaged Navier-Stokes framework, specifically applied to horizontal-axis WT. The turbulent length scale, which is essential for wake predictions in SA turbulence model, has been derived from the standard k-ε turbulence model based on the neutral atmospheric boundary layer assumption. WT is modelled as actuator disk (AD), with thrust as a momentum source term distributed across the AD using a radial distribution function. Wake velocities are measured from 2.5 to 10 times the WT diameter downstream. The model’s accuracy is validated using four WTs of varying sizes and operational conditions. The average mean absolute percentage error (MAPE) of 5.5% confirming that the SA model effectively captures wake profiles at multiple downstream locations. Additionally, the SA model achieves these results with significantly reduced computational costs compared to traditional two-equation turbulence models. These findings offer valuable insights for optimising turbine placement and improving wind farm performance, positioning this research as highly relevant for both academic and industrial applications.