Closing the loop in model-based wind farm control

Abstract

Open-loop wake steering inconsistently improves the energy yield of wind farms in field experiments in the literature. This thesis matures wind farm control technologies for power maximization in a model-based closed-loop framework towards real-world practical applicability. Firstly, open-loop wake steering is evaluated on a commercial onshore wind farm in this thesis. While a net increase in energy yield is measured, the experiment highlights the inaccuracy of the simplified wind farm model, leading to situational losses in energy yield. Secondly, this thesis demonstrates that wind turbine power and wind vane measurements can be used to calibrate the simplified model online to maintain a high model accuracy. This closed-loop concept is tested in high-fidelity simulation, showing a consistent energy yield increase of 1.4%. Finally, this thesis explores the usage of dynamic wind farm models for control, which would further push the accuracy of simplified models yet is held back by the large computational costs. The contributions in this dissertation greatly advance the status quo of wind farm control solutions and their applicability in commercial wind farms.