The development of operational strategies for wind farms as an integrated plant system to achieve a variety of goals from elevating power production to reducing maintenance needs has generated a great deal of interest in recent years. Achieving these operational goals requires an estimate of the energy available and the wind conditions affecting each turbine. The importance of the aerodynamic interaction of wind turbines with the dynamic atmospheric resource means that wakes (the momentum deficit due to power extraction) and their interactions through the farm have the largest influence on the available energy. Predicting the influence of wakes and their interactions, therefore, form the basis of wind farm control strategies to reduce power production losses, track a power signal, mitigate structural loading, or balance the wear and tear on wind turbines to decrease operation and maintenance costs. The articles in the “Advances in Wind Plant Controls: Strategies, Implementation, and Validation” Special Topic in the Journal of Renewable and Sustainable Energy describe the further development and evaluation of wake models and new approaches to wake steering that exploit advances in sensing or estimation to improve control performance.
The development and validation of most wind plant control strategies rely on a wake model to estimate the benefit of taking control actions, e.g., to either assess the additional power production potential of constituent wind turbines or to evaluate the energy available to the wind plant in the atmospheric flow. A number of the papers in the collection employ the FLow Redirection and Induction in Steady-state (FLORIS) tool to evaluate their approaches.1,2 FLORIS includes a number of the different wake models proposed in the scientific literature into a central platform. FLORIS enables a modular representation that combines momentum deficit, wake-added turbulence, and superposition models. The work of Hamilton et al.3 exploits this modularity to examine seven combinations of different formulations for the velocity deficit, added turbulence, and wake superposition methods to represent the wake physics. They validate these models using data from Lilligrund and demonstrate the relative benefits of different modeling choices in different regions of the farm.
Given the time-varying conditions under which wind farms operate, it is not surprising that sensing, estimation algorithms, and operational data are playing an increasingly important role to improve both the model prediction and control performance. The dynamic nature of the conditions under which wind farms operate has been included in wake models, and the control through probabilistic forecasting methods, such as Markov chains and a Weibull wind model, has been integrated into model-predictive control strategies.4 Data-driven methods, including modal decomposition methods and machine learning algorithms, are also pushing wind plant control science forward. Ali et al.5 use the proper orthogonal decomposition in conjunction with a long-/short-term memory approach to optimize sensor locations for wake measurements, critical to validating wake modeling strategies and implementing wind plant controls in the field. Stanfel et al.2 use a reinforcement learning algorithm to predict wind plant wake losses in dynamic conditions and compare to a static lookup table provided by FLORIS.
Wake steering has garnered a great deal of attention as a means of increasing power output and a number of these articles focusses on that as a control goal.1,4,6,7 These benefits were quantified in Ref. 8, which estimated the benefits to annual energy production (AEP) of 60 wind plants in the U.S. They found that wake steering can reduce the levelized cost of energy (LCOE) through an AEP gain of 0.8% and allow a 30% reduction in turbine spacing.7 develop a model for the power output for a yawing turbine under various wind speed and direction that takes into account changes in atmospheric conditions over a diurnal cycle. Their model is validated in a full-scale field experiment.
The control of power output through yaw combined with sensing and estimation techniques can lead to better control performance. The work of Ref. 1 uses local sensor measurements and information sharing between turbines to describe and respond to heterogeneity operating conditions and offer a preview to changes in atmospheric conditions to wind turbines located downwind. Sensing and estimation of clusters of turbines that interact through wakes using temporal correlations of power production are shown to real-time improve the control performance.6 Time-dependent wind estimation models are also shown to improve the control performance in Ref. 4.
The body of work featured in the “Advances in Wind Plant Controls: Strategies, Implementation, and Validation” Special Topic demonstrates the advances in validation, sensing, and real-time algorithms that leverage combined techniques that are changing the nature of wind farm control research. Each of the contributed articles serves to mature wind plant controls as a practice that can be implemented in existing wind plants to optimize power production or reduce loads or be integrated into the design and optimization operations of new wind plants. The wind plant control problem will continue to be one of the most pressing technical challenge areas and among the sources of the greatest progress in global wind energy development.
Conflict of Interest
The authors have no conflicts to disclose.
Dennice F Gayme: Conceptualization (equal). Nicholas Hamilton: Conceptualization (equal). Raul Bayoan Cal: Conceptualization (equal).