Publications on wake modelling
Brand et al. Accurate wind farm development and operation – Advanced wake modeling. EWEA Offshore 2013
Summary The ability is demonstrated to calculate wind farm wakes on the basis of ambient conditions that were calculated with an atmospheric model. Specifically, comparisons are described between predicted and observed ambient conditions, and between power predictions from three wind farm wake models and power measurements, for a single and a double wake situation. The comparisons are based on performance indicators and test criteria, with the objective to determine the percentage of predictions that fall within a given range about the observed value. The Alpha Ventus site is considered, which consists of a wind farm with the same name and the met mast FINO1. Data from the 6 REpower wind turbines and the FINO1 met mast were employed. The atmospheric model WRF predicted the ambient conditions at the location and the measurement heights of the FINO1 mast. May the predictability of the wind speed and the wind direction be reasonable if sufficiently sized tolerances are employed, it is fairly impossible to predict the ambient turbulence intensity and vertical shear. Three wind farm wake models predicted the individual turbine powers: FLaP-Jensen and FLaP-Ainslie from ForWind Oldenburg, and FarmFlow from ECN. The reliabilities of the FLaP-Ainslie and the FarmFlow wind farm wake models are of equal order, and higher than FLaP-Jensen. Any difference between the predictions from these models is most clear in the double wake situation. Here FarmFlow slightly outperforms FLaP-Ainslie.
Download Brand et al. (2013).
Özdemir, Versteeg and Brand. Improvements in ECN wake model. ICOWES 2013
Summary In this study improvements to the ECN wake model WakeFarm, - the aerodynamic heart of the ECN wind farm wake model FarmFlow - are presented in two folds. First of all, an improved near wake model is shown where the wake is modeled by a thin vortex sheet represented by discrete vortex rings of variable strength. The solution is obtained analytically with the Biot-Savart law, where the elliptic integrals are evaluated numerically. It is shown that the induced velocity is lower than for the original near wake model, which is in accordance with the theory. Furthermore the diabatic wind profile model is improved by implementing the Gryning model which is valid for the entire boundary layer. The results are compared with a diabatic wind model valid for the surface layer and with the data obtained from measurements in the ECN Wind turbine Test site Wieringermeer (EWTW). Although the results seem closer to the data, the EWTW data is a single data set representing all stability regions. An initial attempt to categorize the EWTW data in to two stability regions depending on the time of the day of the measured data did not lead to better conclusions. A need for a better data mining to distinguish the different stability regions and data from different sites is obvious. In addition, accurate temperature measurements should be standard for wake measurement campaigns.
Download Ozdemir, Versteeg and Brand (2013).
Bot, Eecen and Brand. The influence of wind farms on the wind speed above the wind farms. Torque 2012
Summary This paper addresses the effect of clusters of wind farms on the flow above the wind farms and their power productions. In the first part past and present modelling approaches are reviewed. In the second part computational results are presented and discussed. These results comprise the wind speeds within, in between and above two offshore wind farms, plus the power of these wind farms, as calculated with the wind farm design tool FarmFlow. It is shown that the wind speed deficit due to a cluster of wind turbines may extend to large heights, and that increasing the space between wind farms may not always assist in the recovery of the external wind speed. It is concluded that the wakes of large offshore wind farms may cause production losses at other wind farms even when they are separated at huge distances.
Download Bot, Eecen & Brand (2012).