Advanced modeling, control and observation concepts for thermal energy systems

Abstract

The present thesis provides novel methods in the area of model-based optimization of thermal energy systems with special focus on a hybrid energy storage. The contributions of the thesis can be structured into: 1.) modeling, 2.) model reduction and 3.) state estimation for a hybrid energy storage and finally 4.) operation optimization of an entire thermal energy system. First, the hybrid energy storage system is investigated in more detail. Therefore, a cosimulation methodology is introduced, which efficiently couples sub-models of different complexity with guaranteed energy conservation. The required number of sub-models to accurately represent the hybrid storage system is determined by an optimization criterion. One of the sub-models describing a phase change material (PCM) is computational expensive and not real-time capable. A data-based model reduction approach is developed to replace the laborious solution of the Navier-Stokes equations in the phase change problem without compromising model accuracy.The obtained real-time capable PCM cell model serves as basis in a state observer, but is still of high order. In the observer, two models of different order are employed to estimate the distributed system state in the PCM. The real-time capable but high order PCM cell model predicts future system outputs. The observer uses the predicted outputs, measurements and a dominant-states only model to correct the high-order PCM cell model. As a result the states inside the PCM can be estimated accurately while reducing the computational load and ensuring observability. Finally, a broader perspective is taken and the control of an entire thermal energy system is examined. The investigated industrial use case is comprised of components with different complexities and stochastic disturbances. The proposed hierarchical operation optimization framework incorporates a novel cooperation concept between the optimization layers and leads to optimal expected plant operating costs. All developed tools can be easily adapted and transferred to other components or plant setups. They enable advanced modeling, control and observation of complex systems and thus result in improved operation efficiency.