par Oudich, Younes 
Président du jury Gyselinck, Johan
Promoteur Kinnaert, Michel
Co-Promoteur De Belie, Frederik
Publication Non publié, 2025-01-24

Président du jury Gyselinck, Johan

Promoteur Kinnaert, Michel

Co-Promoteur De Belie, Frederik
Publication Non publié, 2025-01-24
Thèse de doctorat
Résumé : | The global warming crisis leads countries to invest in renewable energies, mostly in wind energy. Indeed, the latter has the highest growth rate in terms of installed capacity in modern power systems. However, due to the variability of the wind and the limitations of forecasting, it is challenging for wind farm operators to compete with the traditional power plants in the electricity market. The goal of the present work is to investigate different power reserve strategies for wind energy, enabling wind farm operators to participate in ancillary services. The core of the work is divided in three main chapters, each focusing on a power reserve strategy. The first power reserve strategy focuses on primary frequency regulation, and is the most common one in wind energy literature. It is achieved by derating the wind turbines, in other words, the power supplied to the grid is less than the maximum power the wind farm can extract from the wind, and the difference is the reserve. The latter can be realized either by acting on the pitch, or the torque, or a combination of both. The objective of this chapter is to suggest a new derating method, and compare it with the classical derating method in terms of lifetime damage equivalent load. The classical method is based on a modification of the generator torque set point to achieve derating, while keeping the standard pitch controller. The proposed method realizes derating by acting on the pitch, while preserving the same tip speed ratio set point as for power maximization, in the second operating region of the wind turbine. Both strategies were simulated under various wind conditions covering all the wind turbine operating regions. For a constant power reserve, simulation results on a single turbine show that the proposed strategy achieves similar derating as the classical one, while inducing lower lifetime fatigue loads.The second power reserve strategy is based on the power difference between the classical configuration of a wind farm, where all the wind turbines are facing the wind direction, and the configuration where the yaw angles of the turbines are optimized to maximize the overall wind farm power. This power difference can be seen as a power reserve for secondary frequency regulation. The aim of this work is to assess the feasibility of wind farm operators to participate in the secondary frequency regulation thanks to this reserve emerging from yaw control. To do so, a distributed yaw optimization method is developed to estimate the total power gained through wake redirection. This method is based on a static wake model that was built and validated through synthetic data generated with the FAST.Farm simulator. Considering a scaled Belgian wind farm as study case, results show that the requirements of the Belgian transmission system operator are fulfilled both in terms of response time and level of power reserve for most wind directions. The study is limited to wind speeds below the rated speed of the considered wind turbines, where the wake model is valid.The third approach to power reserve is based on green hydrogen production, which can be used in fuel cells for both primary and secondary frequency regulations. In that framework, an offshore islanded microgrid is considered, composed of a wind turbine, a battery, and an electrolyzer. The aim is to suggest a wise choice of the battery rated power to ensure the microgrid stability, taking into account the wind variations. To achieve the latter aim, a detailed electrical model of the microgrid, including its controllers, is built and validated through PSCAD simulations. By linearizing the model around a variable operating point corresponding to wind speed variations, a linear parameter varying descriptor system is obtained. Robust stability of the microgrid is assessed via two different methods. The first one exploits Barmish's extension of the Kharitonov theorem. It considers the descriptor system as an uncertain system, neglecting the slow time varying character of the wind variability compared to the electric time-constants. The second method is more rigorous as it considers the time-varying nature of the uncertain parameters. This method is based on linear matrix inequalities resulting from Lyapunov's stability theory. The results of both methods show that the rated power of the battery should be at least $6.25%$ of the wind turbine's one to ensure microgrid stability. These results are also validated by simulations. Yet the proposed approach guarantees stability without requiring extensive simulations. |