Comparison of wind turbine wake numerical simulation with scanning lidar measurements

Wind turbines wake can have an impact of 10 to 20 % on the production of an offshore wind farm [1]. Thus an accurate modelling of these wakes remains an important but still difficult issue for the estimation of the Annual Energy Production (AEP). Although the final objective of this work is to compare scanning lidar and numerical simulation at an offshore site, it is first performed on a flat onshore site in order to focus on the wake of a single row of turbines. A four month campaign has been carried out with an instrumental set-up including a scanning lidar WINDCUBE 200S, a vertical profiler (lidar WINDCUBE v2) and an instrumented mast. A comparison between scanning lidar and cup anemometers measurements is presented in [2] for the same campaign. Reynolds Averaged Navier Stokes numerical simulations (RANS) of wind turbine wake have been performed with the open source CFD (Computational Fluid Dynamic) code Code-Saturne (fully non-linear 3D flow model) [3]. LES (Large Eddy Simulation) allows to take into account the flow unsteadiness but are much computationally expensive and still difficult to use in an operational context and therefore is not used in this study. The scanning lidar allows to obtain wind speed data in a large cone inside the wind farm. However a methodology has to be defined for the comparison with the model. One of the challenges is to define a relevant method to compare wind speed measured with the lidar with a low rate sampling (four scans per hour for a given elevation) to an ensemble average wind speed value (obtained from averaged Navier Stokes equations). For RANS simulations, the comparison is more relevant with an averaging of several lidar scans for a given elevation. The results are shown for a steady meteorological situation allowing an averaging over 1 hour (corresponding to 4 scans). Despite a larger variability in the measurements, these results show a good agreement with the model for distances behind the turbine larger than about 5 turbine diameters. It confirms the ability of a CFD code associated with an actuator disc representation of the turbine to simulate the far wake correctly.