Wind speed feedforward control for a wind turbine using LIDAR

Wind energy has relatively minor environmental impact compared to traditional energy sources. It doesn’t consume fuel or emits air pollution, while the land used for a wind farm can still be used mostly for other purposes such as agriculture. Some people state that wind turbines consume subsidies as a fuel, since they aren’t profitable without it. They use a blurred view on the real costs of energy, since they do not consider the costs associated with the pollution of the fossil energy sources. Despite this, wind energy has to deal with the public opinion that wind energy is expensive and should look for opportunities to reduce the prices. This can be done by reducing cost, increasing the yield or by extending the lifetime of the turbine. The latter can be achieved by reducing the loads on the structure. The rotor speed of a wind turbine is the result of the balance between the aerodynamic and the generator torque. Most of the utility scale variable speed wind turbines currently use a rotor speed feedback controller to control this torque balance. They can adjust the captured aerodynamic torque by changing the angle of the blades relative to the wind. But, the effect of changes in the wind on the rotor speed is delayed due to the inertia of the rotor. Meanwhile the construction suffers from the variations in the loads. For this reason it would be advantageous to measure the wind speed and use it for control, in order to anticipate on changes in the wind earlier and reduce the loads on the structure resulting in an extended lifetime of the turbine. The wind speed is currently measured on most turbines, however this measurement is useless for control, since it is contaminated by the wake of the rotor of the turbine. LIDAR is a measurement technique, which is able to measure the upcoming wind speed in front of the turbine using a laser. This measurement technique is relatively new to the wind energy market and an optimal configuration for the measurement is yet to be defined. So are the possibilities for using these measurements in a control system. A realistic simulation environment is required to investigate this and come to a reliable conclusion. This environment is created using a coupling between the simulation software packages Bladed and Simulink. It provides the possibility to co-simulate the nonlinear model of the turbine in Bladed together with a new controller in Simulink. Bladed provides simulated ‘line-of-sight’(LOS) LIDAR wind speed measurements from the wind field used. Multiple LOS measurements are combined to determine a mean wind speed representative for the whole rotor plane. How well the reality is approximated depends on the number of LOS measurements considered, the measurement distance(s) and number of measurement points per distance. The performance of the wind turbine using various LIDAR configurations is compared. The knowledge about the upcoming wind is used in the controller by adding a wind speed feedforward loop to the existing rotor speed feedback controller. The performance of the wind turbine using two different feedforward controllers is investigated. Controller A is based on the optimal pitch action for the wind at the turbine. While controller B is based on the difference between the actual pitch angle and the optimal pitch angle related to the wind speed measured at some distance. The feedback controller parameters are relaxed while using these feedforward controllers in order to achieve load reductions without deteriorating the rotor speed control. Feedforward controller B performs best for the load reduction, with reduced pitch rate demands, while having still sufficient disturbance rejection in the rotor speed control. This is because it takes more time to compensate for the virtual pitch error, resulting in lower pitch rate demands and thrust related loads. Filtering the measured wind speed adequately is a crucial step in the load reduction for controller A while controller B has this property by design. The filter effect of controller B changes with the wind speed as higher wind speeds provide less time to compensate for the upcoming changes in the wind. The optimal configuration of the LIDAR for the XD115 appears to be the pulsed system with four points on the azimuth at each of the three measurement distances. This configuration provides early knowledge about upcoming changes in the wind speed, while having sufficient coverage of the complete rotor plane.