Keeping power reserves by wind farms (WF) is necessary for frequency and voltage stability with provision of voltage source behavior to the grid. This results in additional challenges for WF compared to an individual wind turbine (WT), as the WT can influence each other aerodynamically via the wind wake. A holistic approach for the task of power reserve is required, in which the design of the WF management is coupled with the selection of the reserve method on WT control level. The overall approach also has to take into account selectable power limitations, stored kinetic energy reserve by using the rotating mass of the drive train as well as an optimum result in the partial and full load range. In this paper, such a holistic method for the provision of reserve power in a WF is presented. It is analyzed on the basis of stationary and dynamic simulation scenarios. For the dynamic simulation a scenario with wind rotation is selected, in which time delays are modeled according to the change in WT wind direction and wind speed due to the wind wakes.

The change in the electrical grid from conventional power plants to decentralized renewable energy sources like wind turbines (WT) as well as the provision of grid side control with voltage source behavior lead to a demand of power reserves by WT. To keep the reserve, WT have to reduce power output. In the partial load range, the overspeeding method without pitching the blades is usually preferred, as it is possible to keep additional reserve power in the form of the kinetic energy of the rotating mass of drive train and wind rotor. However, thrust forces on the WT increase. This paper presents a novel reserve method, in which the thrust coefficients are unchanged by slowly adjusting the pitch angle despite keeping reserve power. This constant thrust force method has advantages in terms of limiting the thrust forces together with increasing the stored energy in the rotating mass. The reserve method is analyzed in ideal wind scenarios as well as with real wind field conditions with wind height profile, wind speed reduction by the tower as well as wind wakes by wind turbines located upwind. The method is extended to an observer-based overall approach without wind speed measurement.

With the increasing penetration of wind energy into the electrical grid, the requirements for grid supporting wind turbines are becoming more and more important. As the rotational frequency of variable-speed wind turbines is decoupled from the grid’s frequency, such systems can theoretically operate at any operation point within their limits, holding a certain power reserve. With such an operational behavior, wind turbines can actively participate in the primary frequency control. Nevertheless, in combination with active-pitch control, this can lead to a counter intuitive system response, endangering system stability. This paper therefore presents resulting system behavior, analyses the cause theoretically and proposes an effective solution with low implementation effort. The overall effectiveness and functionality of the implemented method is verified with simulation models, including aerodynamic and mechanical characteristics of the wind turbine.

The implementation of observers is necessary in a variety of different drive train applications and enables the capturing of state and disturbance variables that are not available or could only be measured with expensive measurement effort. As an application comes foremost the controller design as state-feedback, for which all state variables must be known. But also for condition monitoring and superimposed operational management techniques, an observer can be beneficial. When operating wind turbines, both the electrical and mechanical state variables can be tracked by an observer. In this paper an observer is designed for the drive train of a wind turbine. It enables the estimation of both the mechanical state space variables as well as the disturbance in form of the torque on the rotor. After a description of the modeling of the wind turbines drive train in form of a three mass model, the observer design by Kalman filtering is presented. For validation, the observer is analyzed in a simulation and is also implemented on a FPGA of a wind turbine test bench. Both the simulation models and the test bench replication emulate aerodynamic characteristics of the rotor and typical operational management strategies. Finally, the performance of the observer is evaluated based on measurement results regarding possible application purposes.

When increasing the number of renewable energy sources coupled to the grid, the number of conventional power plants is decreased. To prevent higher frequency gradients and nadirs in the future grid, wind turbines (WT) have to provide inertial and primary power. For the last one a WT has to keep power reserves by overspeeding the drive train or pitching the blades. To avoid unnecessary losses in this operation point, the maximal available power must be determined accurately. An opportunity is to use the current wind speed value, whereby the measurement at the roof of the nacelle is usually not precise enough. Therefore in this paper a novel method to reconstruct the wind speed value by using a disturbance observer and an iterative algorithm is shown. Its robustness to measurement delays and model uncertainties is investigated. This wind reconstruction is included in a total control structure with primary and inertia controller for the grid support.