Design, Implementation and Optimization of a LQRI control for a floating offshore wind turbine in MOST

In the last seventy years global demand for energy has led to several problems, most notably air pollution, climate change and natural resources depletion, due to the intense dependence on fossil fuels. Consequently, a large amount of renewable energy technologies are being developed in order to replace fossil fuels and to contrast their huge negative effects. Among them, wind energy appears to be a promising and consolidated renewable alternative. Accordingly, in the last years, many simulators for floating off shore wind turbine (FOWT) have been developed. For the aim of this work, a controller for the simulator named MOST, developed by MOREnergy Lab, has been designed. One of the controller equipped to MOST is a baseline PI controller, developed by NREL, which is a simple use controller. However, for this thesis’ project, its low capacity to stabilize the power production around its nominal value has been identified as a limit to overcome. Moving from such limitation, the research focus is to develop a likewise simple use controller to solve such an issue. The new controller is an LQRI developed on a State Space linear representation of a FOWT. The methodology of research started by the analysis of LQR in order to understand its function. Then after, the analysis proceeded with a simplification of the mathematical model which describes the physics and the mechanism of FOWTs. From such a simplified model, the LQR controller has been first developed, and then optimized trough an integral action which has generated a new controller named LQRI. This new controller, resulted in both the stabilization of the power production, which was a limitation of the controller, and further, through the addition of different modules it has been showed its capacity to work in different operative conditions. The first chapter will introduce some basic aspects of wind turbines and it will trace back the evolution of the energy market. Additionally, an estimation of the scenario in the next years will be provided to investigate the future development of such a technology. The second chapter will focus on state-of-art FOWTs mathematical models taking into account their aim, pros and cons. Furthermore, it will introduce some state-of-art mathematical controllers like PI and NREL/ROSCO, clarifying the choice to develop a LQR controller in this work. The third chapter will introduce some basic information about MOST. It will proceed showing the rationale of the linear representation outlined and the design process of the LQR controller. Additionally it will illustrate the results through a comparison of the two controllers applied to MOST. In the fourth chapter different optimizations of the model will be explored and presented. Firstly, the LQRI controller will be investigated and implemented. Secondly, results about linearization with different wind speeds and a gain scheduling controller will be analysed. The results of the addition of the control torque input will be as well discussed. In conclusion, the last chapter will compare the results of the baseline PI controller and the new controller developed. An explanation about the simplifications of the model will be provided to show how, deleting them, a more accurate controller may be realized.