Power performance verification in complex terrain using nacelle lidars: the Hill of Towie (HoT) campaign

Nacelle lidars are an attractive alternative to meteorological masts for power performance testing in complex terrain, because of the ease of deployment. This report presents the comparison of wind speed and power curve measurements using two commercial nacelle lidar system – one Avent 4-beam Wind Iris and one ZephIR Dual Mode – and a ground-based profiling lidar (ZP300), in a complex site.

The model-fitting wind field reconstruction technique using measurements between 0.5D and 1D upstream, which has previously been demonstrated in flat terrain (Nørrekær Enge campaign), is here tested in complex terrain. The two nacelle lidars have been deployed on a Siemens 2.3MW turbine at the Hill of Towie wind farm in Scotland between July and October 2017.

The data analysis has been performed with three different wind speed estimates applied to the exact same dataset: 1) the ZP300 wind speed measurements at hub height, located 2.7D (220m) from the turbine. 2) the nacelle lidar wind speed estimate using the wind model fitting to the measurement at 2.5D upstream. 3) the nacelle lidar wind speed estimate using the wind-induction model fittingto the measurements between 0.5D and 1D upstream.

With the wind model, the wind speed estimate is within 2% from the ZP300 measurements, corresponding to an error in AEP in the order of 4%. With the wind-induction model, the free stream wind speed estimate is within 1% from the ZP300 corresponding to an AEP error of approximately 2%. In the second case, the reference wind speed is the ZP300 wind speed measurements corrected using the site calibration.

The power curve measured using the three measurement systems were compared to the turbine manufacturer warranted power curve as reference. The reduction in the statistical power uncertainty (type A) usually clearly observed in flat terrain when using nacelle lidars was demonstrated to be of a lesser extent.

Here, the wind-induction model performed slightly better than the wind model. In this study, measurements from nacelle lidars close to the turbine rotor were used to estimate the free stream wind speed. The resulting measured power curve was at least as accurate as the one obtained using the ground-based profiler measurements corrected with the site calibration.

Thus, it was demonstrated that it is possible to measure a turbine’s power curve at a (moderately) complex site without the need for a site calibration.