Surrogate-based optimization under uncertainty of wind farm control using combined control strategies

Abstract

Due to the limited availability of sites on both land and sea, there is a need to maximize the power density of wind farms whilst limiting the adverse wake effect between the wind turbines that cause power losses and/or increased load cases. The study focuses on analyzing the potential of combining yaw-based wake-steering and constant blade pitch control down-regulation to mitigate the detrimental wind turbine wake interactions. Data-driven surrogate models based upon polynomial chaos theory have been used to model the statistical distributions of the wind farm power production and short-term Damage Equivalent Loads (DEL) in function of the control inputs. The model calibration data has been generated for a range of control settings for two V27 wind turbines aligned with the wind direction through large eddy simulations using the flowsolver EllipSys3D and the aeroelastic code Flex5. A power-based optimization with DEL constraints has been performed and median power gains ranging from +1% to +3% have been observed at close spacings depending on the severity of the imposed constraints. It was identified that larger power gains corresponded to an increase in DEL. At larger spacings, the wind farm control strategies shows limited performance increases and revert to baseline operations.