Data-driven wind plant control

Abstract

Each wind turbine in a cluster of wind turbines (a wind power plant) can influence the performance of other turbines through the wake that forms downstream of its rotor. The wake has a reduced wind velocity, since the turbine extracts energy from the flow, and the obstruction by the wind turbine rotor causes increased turbulence. If another turbine is standing in the path of a wake, the reduced wind speed in the wake results in a lower electrical energy production of that turbine. Also, the increased turbulence and velocity gradients in the wake may induce an increase in fatigue loads on the downstream turbine. The topology and amount of wake interaction depends on time-varying atmospheric conditions, and on the operating point of each turbine that can be adjusted by changing their control settings. In high-fidelity computational fluid dynamics simulations, we have evaluated the potential of the different control degrees of freedom of the turbine to influence the wake interaction effects. Further, in the thesis we propose coordinated plant-wide control algorithms to optimize the performance of the wind plant, by adjusting control settings in a way that takes into account the wake interaction effects. These algorithms use data-driven techniques in order to adapt to the time-varying atmospheric conditions in a time-efficient manner.