Turbulence characterization using a multiple-, a four- and a two-beam nacelle lidar system

Description

Turbulence characteristics are essential for the load estimation and power performance assessment of wind turbines. Lidars of different types and configurations have been mounted on the nacelle and used in a forward-looking mode to scan the inflow in front of wind turbines. They can potentially estimate the turbulence characteristics of the inflow that actually influence the wind turbine performance better than other types of anemometry, because they are able to yaw with the turbine and scan over an area of the inflow.

Turbulence measurements with lidars can suffer from two problems. One is the averaging effect due to the large sample volume in which lidars measure the radial velocity, and the other is the contribution of all components of the Reynolds stress tensor to the estimated velocity variances, which depends on the scanning geometry. Peña et al. (2017) demonstrated two methods to characterize turbulence in the inflow using radial velocities from nacelle lidars. The first uses the Mann turbulence model (Mann 1994) combined with a model of the spatial radial velocity averaging of the lidars, which describes the degree of spatial-average and cross-contamination effects of turbulence measures. The second uses the ensemble-averaged Doppler radial velocity spectrum, which results in turbulence measures that are not affected by the averaging effect. Moreover, Peña et al. (2019) performed an analysis of measurements of a SpinnerLidar (a multiple-beam scanning lidar system) that was mounted on the nacelle of a V52 wind turbine. They showed that mean wind speeds and components of the Reynolds stress tensor can be accurately estimated when compared to estimates from sonic anemometer measurements.

Here, we are further investigating the averaging effect and the cross-contamination for turbulence characterization. We use measurements from three lidar systems, which are mounted on the V52 wind turbine at DTU Risø campus (Fig.1): the SpinnerLidar (on the top), a 4-beam WindVision (in the middle) and a 2-beam WindEye (at the bottom). The latter two lidars are from Windar Photonics A/S. All lidars are based on a continuous-wave system. The SpinnerLidar scans in a rose-curve pattern and generates 400 radial velocities with one full scan (Fig.2); it performs 30 full scans per minute at a focus distance of 62 m. The WindVision uses the same focus distance and has half-cone opening angles of 18 deg. The WindEye measures at 37 m with half-cone opening angles of 30 deg. The focus distances were decided so the probe length volume of all units are similar. Both lidars from Windar Photonics perform a full scan in 1 s.

In this work, we show the estimates of variances and covariances of the fluctuations of wind velocity components. Furthermore, we compare the estimates from the measurements with those from numerical simulations for three lidar systems. The numerical simulations provide means to estimate the variances of the velocity components for given idealized homogeneous turbulent fields. We also want to study the impact that the lidar scanning geometry has on characterizing turbulence by comparing the estimates from the SpinnerLidar to those from the WindEye and WindVision.

Period	26 May 2021
Event title	Wind Energy Science Conference 2021
Event type	Conference
Conference number	3
Location	Hannover, Germany, Lower SaxonyShow on map
Degree of Recognition	International