Data-driven yaw optimization of a full-scale 60 MW wind farm

Due to the prevalence of wake interactions in large, utility-scale horizontal axis wind farms, wake steering based power optimization has received significant recent attention. Upstream wind turbines may intentionally misalign the axis of rotation with respect to the incoming wind to create a yaw angle. The misalignment of thrust force and wind velocity vectors results in a lateral deflection of the momentum deficit region behind the yawed wind turbine. While the yawing of a turbine decreases its power production, it enhances the performance of the downstream turbine due to the deflection of the wake. Recent literature has proposed a supplement to wake models which accounts for the wake deflection in yawed conditions. Model parameters for low-order wake models are calibrated using five years of operational power SCADA data from a 60 MW wind farm. The low-order wake models are used to improve the power production of the operational wind farm through the use of a novel optimization technique. Various optimization algorithms and wake models are also compared to study sensitivity and robustness.