Abstract
Wind turbine curtailment is oftentimes required as a means to mitigate environmental impacts of a wind energy installation. For example, curtailment may be required to satisfy constraints on the occurrence of shadow flicker on nearby homes or to reduce wildlife fatalities below a certain limit. This paper introduces an optimal curtailment strategy, which seeks to maximize wind plant revenue while meeting imposed environmental constraints by intelligently selecting curtailment times. To formulate the problem, a discrete set of curtailment decision times is defined, and long‐term forecast data are used to predict demand‐weighted power production at each time. Dynamic programming is used in conjunction with in situ meteorological sensors to compute the probabilistically optimal curtailment decision in real time. By leveraging both forecast data and real‐time measurements, the algorithm ensures that the constraint budget is expended strategically, curtailing when revenue is low and saving operating hours for times when revenue is likely to be higher. Through a series of simulation studies involving shadow flicker constraints, algorithm performance is characterized and compared with two simpler approaches: a greedy scheme and a threshold‐based scheme. These studies highlight the benefit of the optimal algorithm as well as the performance tradeoffs inherent in the three different solution approaches. Overall, the dynamic programming algorithm is shown to exhibit significant benefits in cases where sufficient meteorological and demand data are available.