Electricity production by wind turbines as a means for the verification of wind simulations

Wind measurements are rare in heights of interest in terms of wind turbines. Consequently, well-verified model simulations of the wind conditions are fundamental during the planning stage of wind farms. In this paper, a new approach is presented where data from operational wind turbines is used to verify these wind simulations. Several issues are discussed: model wind speed has to be combined with the power curve from the wind turbine in order to be comparable with the turbine data. Furthermore, in operational mode a turbine does not perfectly perform at any time step. Thus, those time steps have to be identified and replaced by missing values since a model cannot reproduce an unsual turbine operation. In addition, wind turbines are influenced by surrounding turbines resulting in a reduction of power production at some time steps. Consequently, a wake model is applied on the model simulation. In a next step, the edited turbine power data is compared to the power time series resulting from a mesoscale simulation for Germany with the mesoscale model WRF. The simulation is improved by a remodelling-method during the post-processing. This method alters the model wind speed and particularly results in a reduced wind speed bias. Comparing these model simulation's power production time series to the turbine production data reveals that they correlate well with most values in the range of 0.82 to 0.87 for an hourly temporal resolution. The relative bias for the simulations is between 10 to 25% in terms of power. Finally, a comparison with a model verification with wind measurement data is carried out. It is shown that the verification with data from operational wind turbines and from wind measurements are in the same range in terms of correlations.