Hardware-in-the-loop wind turbine system test benches and their usage for controller validation
Jassmann, Uwe; Abel, Dirk (Thesis advisor); Monti, Antonello (Thesis advisor)
Dissertation / PhD Thesis
The subject of this thesis is the research on hardware in-the-loop (HiL) systems for system test benches, which allow the operation of modern wind turbines with activated, unmodified main control at such test benches. Another focus of this thesis is the development of a Model-based Predictive Control for wind turbines. The aim of this thesis is to develop and implement various HiL-methods and investigate to what extent these methods enable the testing of wind turbines, equipped with a model-based control, at system test benches. Within this thesis, a Model-based Predictive Control, based on a reduced wind turbine model is developed. It reduces the mechanical loads of the wind turbine and ensures steady electrical output power, without using additional sensors. By implementing a Move-Blocking strategy, it is shown that the computational power, required by the Model-based Predictive Control, can be significantly lowered. This allows implementation of this control and a Kalman filter based wind speed estimator on a state-of-the-art Programmable Logic Controller and executing both functionalities it in real time. The HiL system developed in this dissertation consists of a signal-level HiL, for the emulation of missing sensors and actuators, and mechanical-level HiL, for the emulation of the missing rotor inertia and the associated eigenfrequencies. Four mechanical-level HiL concepts are designed and either emulate the inertia, the inertia and the pre-calculated eigenfrequencies or the inertia and the eigenfrequencies that evolve at runtime. The varying level of accuracy of the different methods comes along with different levels of robustness. As a result, individual, very robust methods are particularly suitable for initial commissioning. This is shown by experimental results, presented in this thesis. Other methods, however, exhibit an excellent trade-off between robustness and accuracy, and are therefore suitable for testing wind turbines equipped with Model-based Predictive Control, which is shown in extensive simulation studies in this work.