Abstract
Prediction of the loading and energy yield of a turbine within an array requires knowledge of the onset flow profile due to the wakes of turbines located upstream. This depends on many factors and processes including the geometry of the turbine and supporting structure, turbine operating point, array layout and the profiles of velocity and turbulence of the ambient flow. Various studies have employed CFD to simulate the flow-field downstream of single and multiple turbines using actuator disk and RANS-BEM methods. However limited evaluation has been presented for wake interaction. In this study two numerical approaches for modelling the wakes of arrays of turbines are evaluated by comparison to experimental data from a study of a single wake and the merging wakes of a single row of five turbines. With RANS-BEM the transverse profile of velocity is simulated to reasonable accuracy over the far-wake region provided that the ambient flow turbulent kinetic energy and dissipation rate are representative. Improved agreement of the velocity-profile over the near-wake region is obtained by representing the elevated turbulent kinetic energy over the tip vortex region. The magnitude of TKE in this region is based on prior blade modelled simulations and experiments. The second approach assumes that the self similar far wake profile of a single turbine may be superimposed for multiple turbines based on local velocity conditions. This is highly efficient but does not account for the device-scale bypass flow. Mean loading of a rotor is a function of the mean of velocity squared integrated over the rotor plane. For downstream rows located at 8D downstream this is predicted by both methods to within 4% and 12% for aligned and staggered turbines respectively