Model Predictive Control and Moving Horizon Estimation for Solar Tower Power Plants: Analysis and Adaptations to the Experimental Facility Solar Tower Jülich

Kurzfassung

This thesis presents the extension of a model predictive control design by a moving horizon estimator, applied at the experimental facility Solar Tower Jülich. The control addresses challenges posed by cloud passages, aiming to maximize the receiver outlet power and maintain a stable outlet temperature despite solar radiation fluctuations. A safe operation has to be ensured by preventing overheating and thermal stresses in the receiver caused by fast temperature changes. These occur when the solar radiation increases rapidly after a cloud passage. The control adjusts the heat transfer fluid's mass flow and the heliostat aim points, incorporating cloud predictions to proactively handle solar radiation changes. The moving horizon estimator adapts the receiver model's parameters, attempting to improve the model accuracy and thus, enhance the efficiency and robustness of the control. The control performance was tested in simulation for different cloud coverages. By applying inaccurate cloud prediction, the robustness of the control was evaluated, yielding improvements of up to 60 % compared to the control without the parameter estimation. Tests conducted at the Solar Tower Jülich revealed shortcomings in the control design. The applied receiver model showed insufficient accuracy even when applying the parameter estimation. The real-time capability of the control was not ensured due to long delay times of the heliostats and increased process times. The thesis suggests to adapt and validate the receiver model and to enhance the real-time capability through improvements in the computational performance and the heliostat field operating system.