Identification of wind turbine clusters for effective real time yaw control optimization

Abstract

Control algorithms seeking to maximize wind plant power production may not require that all turbines communicate with each other for the purpose of coordinating an optimal control solution. In practice, an efficient and robust control solution may result by coordinating only turbines that are aerodynamically coupled through wake effects. The implementation of such control strategy would require information of which clusters of turbines are coupled in this way. As the wind changes direction, the clusters of coupled turbines may vary continuously within the array. Hence, in practical applications, the identification of these clusters has to be performed in real time in order to efficiently apply a coordinated control approach. Results from large eddy simulations of the flow over a wind farm array of 4 × 4 turbines are used to mimic Supervisory Control And Data Acquisition (SCADA) data needed for the cluster identification method and to evaluate the effectiveness of the yaw control applied to the identified clusters. Results show that our proposed method is effective in identifying turbine clusters, and that their optimization leads to a significant gain over the baseline. When the proposed method does not find clusters, the yaw optimization is ineffective in increasing the power of the array of turbines. This study provides a model-free method to select the turbines that should communicate with another to increase power production in real time. In addition, the analysis of the flow field provides general insights on the effect of the local induction, as well as of the wind farm blockage, on yaw optimization strategies.