Inflow Measurements by Nacelle Mounted Lidars for Wind Turbine and Farm Control

A major challenge in the operation and control of wind turbines is the adjustment to changes in the wind condition. Currently, a variety of sensors is being utilized to determine characteristics in the inflow to maximize the power production of the turbine while keeping its structural loads within the design limits. Among those, light detection and ranging (lidar) is a novel technology that allows to remotely sense the wind and has the possibility to overcome drawbacks of more traditional nacelle anemometry. Lidar systems mounted on the nacelle of wind turbines can be used in a forward-looking setup to probe the incoming wind field at several positions. By making suitable assumptions important inflow characteristics can be derived, such as wind speed, turbine yaw misalignment, vertical shear and gusts.

Several approaches to determine the dominant frequency in a Doppler spectrum exist. Here we investigate the performance of the maximum, centroid and median methods on the spatial averaging effect of the lidar and its accuracy compared to a sonic anemometer. Surprisingly, the maximum method, which uses the least information from the Doppler spectrum, performed best at reducing the spatial averaging. However, this benefit is gained at the expense of an increased root mean squared error (RMSE) compared to the sonic measurement. The overall best performance was achieved by the median method, which showed the lowest RMSE and had a slight mitigation of the spatial averaging effect compared to the centroid method.

By providing a preview of the rotor-effective wind speed (REWS), lidar systems can be used to assist pitch controllers to reduce structural loading on turbines. In this project, the coherence of the REWS estimated from turbine and lidar measurements is evaluated experimentally and compared to a model based on the Mann turbulence and a model for the spatial averaging of continuous-wave lidars. The comparison shows improved agreement with the field data compared to previously used models. It can be applied as a computationally efficient tool to optimize the focus point positions of a lidar system. In that way the coherence of the REWS estimated from turbine and lidar can be maximized.

Wakes can severely violate the flow assumptions applied when deriving inflow characteristics from lidar measurements. The effect on the yaw misalignment measurement is investigated in this work. Large biases occur in half-wake situations, where one of the beams of a lidar system is affected by a wake. It is thus necessary to detect and correct situations where the lidar is influenced by wakes. Here a wake detection algorithm is proposed that uses the broadening of the Doppler spectrum due to small-scale turbulence that is generated inside wake flows. As a detection parameter the line-of-sight equivalent turbulence intensity is used to quantify the amount of turbulence within the probe volume and the standard deviation is defined as the spectral width of the Doppler spectrum. The algorithm is tested on a test turbine and it was possible to detect all half- and full-wake situations. Also, an empirical correction method using with the undisturbed wind direction information from a meteorological mast is proposed and it was shown that the bias in the yaw misalignment measurement can be removed for the case investigated.