Improved modeling of lidar wind preview for wind turbine control

Abstract

Lidar has been used in the wind energy field for decades. Most of the time, it was mainly used for wind resource assessments of potential wind farms. In recent years, the cost of lidar systems has decreased significantly, and the applications of lidar have become increasingly diverse. In some cases where it is more expensive to erect meteorological mast towers, such as offshore wind farm development, lidar has become the preferred resource assessment choice. In addition to resource assessment, lidar has attracted extensive research interest in wind turbine control, power performance characterization, wind power prediction, and wind farm control. In terms of lidar-assisted control, this technology has been proven to reduce wind turbine loads, rotor speed variations, and power fluctuations. In recent years, lidar-assisted wind turbine control has been applied to commercial projects. Wind turbines operate mainly in atmospheric turbulence, where wind speeds are highly random and fluctuating. The wind rotates the turbine to generate electricity while also imposing fatigue loads. Traditional wind turbine control only relies on feedback control, which adjusts the blade pitch angles and the generator torque by measuring changes in the rotational speed of the turbine. This control effect is achieved after the turbulent wind disturbance has already acted on the wind turbine and it is lagging. Differently, lidar-assisted control uses the wind preview provided by nacelle- or spinner-lidars to achieve feedforward control. For example, when the wind speed changes, the lidar can sense that and inform the pitch control system to achieve an earlier pitch adjustment. Some key features of nacelle lidar measurements make it necessary to process the lidar measurement before it can be used for turbine control. For instance, the lidar measurement is the projection of the three-dimensional wind velocity vector onto the lidar beam direction, but the turbine rotor is mainly interacting with the axial component in the velocity vector. In addition, the lidar system provides only limited measurements of the upstream position while the turbine rotor interacts with the downstream turbulence through its three blades. Usually, differences in turbulence at upstream and rotor positions are described as wind evolution. Further, the lidar measurement can be unavailable due to blade blockage or special weather patterns. Therefore, a lidar preview quality study needs to be performed to determine the usable part of the wind preview provided by the lidar. A filter design is necessary to filter out uncorrelated information in the lidar wind preview. In actual operation, wind turbines suffer from different atmospheric stability conditions. In different atmospheric stability conditions, the spectrum and coherence of the turbulent wind can vary. Because of the limited measurement at upstream positions, lidar preview is highly linked to the spatial coherence of turbulence. Thus far, it is still unclear whether the variation of the turbulence characteristics will have an impact on the lidar preview quality and the benefits of lidar-assisted control. Although there are already commercial use cases for lidar-assisted control, the existing lidar-assisted control simulation environment does not fully cover the characteristics of lidar measurements discussed above. Therefore, the main objective of this thesis is to improve the lidar wind preview modeling and assess the benefits of lidar-assisted control using the improved modeling. First, the improvement is achieved by incorporating the wind evolution phenomenon into conventional three-dimensional turbulence models: the Mann and Kaimal models, resulting in four-dimensional turbulence models. The performances of these two extended models in predicting turbulence spectra and spatial coherences are evaluated using measurements from a pulsed lidar and a meteorology mast. Furthermore, the analytical models which represent the correlation between the lidar-previewed rotor effective wind speed and the rotor effective wind speed at the rotor position, are derived based on the extended turbulence models. Spectra and coherences of turbulence in different atmospheric stabilities are summarized, and their impacts on lidar preview qualities are investigated. A preliminary study on whether the upstream wind turbine wake affects the lidar preview of a downwind wind turbine is also conducted. Moreover, a realistic lidar module is updated in the aeroelastic simulation tool, which allows for studying the impact of unavailable lidar data on lidar wind preview quality. Through simulations using realistic data availability, this study indicates that the blade blockage and low data availability event have negligible impacts on the lidar preview quality. Finally, this thesis evaluates the benefits of lidar-assisted control under various turbulence characteristics representative of different atmospheric stabilities using the updated aeroelastic simulation tool. This study exhibits that the benefits of lidar-assisted control are related to the mean wind speed, turbulence spectrum, turbulence spatial coherence, and used turbulence models. For the NREL 5.0 MW turbine and a four-beam lidar, the benefits of lidar-assisted control are primarily reductions in rotor speed variation, power fluctuation, and tower-base bending load. The benefits are observed in all three investigated stabilities: unstable, neutral, and stable.