Presentations

Abstract

Accurate measurements of wind turbines' nacelle direction are necessary to correctly assess the directional performance of onshore wind farms and understand how turbine wakes propagate across the farm. This knowledge, in turn, is essential to implement sector-based wind-farm control strategies. Although the nacelle direction is important for both on- and off-shore wind farm, it is especially crucial for onshore wind farms where wind turbine wakes can interact with terrain features and systematically change directions. Nacelle direction measurements are well-known to have low accuracy since they are not required for most of the control functionalities of individual turbines, which primarily react to their relative direction to the flow. This study evaluates two data-driven methods to correct the nacelle direction signal and estimate the calibration offset from 10-min measurements: the median-based method and the lidar-GPS method. The median-based method considers the median nacelle direction of the wind farm’s up-stream turbines (i.e. those unaffected by wakes) as a reference. The up-stream turbines are identified based on the median nacelle direction of all the wind farm for each 10-min. The calibration offsets of each turbine (Δθmedian-based,i) can then be estimated as the average difference between the turbines yaw encoder (θi) and the median nacelle direction of the wind farm’s up-stream turbines (Θ). Δθmedian-based,i=‹Θ-θi›    for i=1,...,Nwtg This method disregards the natural wind direction variability among the up-stream turbines of the wind farm, which can impact its accuracy especially in complex terrain sites. Therefore, a second methodology based only on geometrical considerations can be used for verification: the lidar-GPS method uses the GPS system incorporated in nacelle-mounted lidars to track the heading of the turbine. If the lidars are mounted some lateral distance from the turbine yawing axis, when the turbine yaws the GPS-measured easting and northing coordinates trace a circle. Since the centre of rotation corresponds to the tower centre coordinates, the relative distance between the tower centre and the GPS location is calculated, converted to Cartesian coordinates (x, y) and used to compute the absolute direction of the nacelle: θlidarGPS,i=-(90+arctan⁡(y/x)) Where corrections are applied to compensate for the GPS measuring location at the nacelle rear and to convert from mathematical to geographic direction convention. The calibration offset is then computed as: ΔθlidarGPS,i=‹θlidarGPS,i-θi›   for i=1,…,Nwtg The lidar-GPS method’s precision depends on the lateral distance of the lidar GPS from the turbine yaw axis, but is unaffected by terrain. In an example application of these two approaches, yaw calibration offsets were computed for 21 turbines of a wind farm located in mid-complex terrain in which every turbine has a nacelle-mounted lidar. Both techniques identified yaw encoder inaccuracies (above 60° in one case), and had good agreement, with the calculated offsets differing by less than 5° for 14 turbines, while the remaining 7 turbines showed yaw offsets discrepancies of up to 17°, with the largest differences observed in turbines positioned centrally in the park. These discrepancies will be further analysed and discussed.