Implementation and Validation of a Multi Layer Model Predictive Controller for Energy Supply Systems

In this thesis, a newly developed multi-layer model predictive control concept, the Energy Demand Control System (EDCS), is implemented and validated on a simulation model. The simulation model is developed, verified, and validated with industrial measurement data. The model consists of three main components: a heat supply controlled by a hysteresis controller, sensible heat storage, and heat demand. The EDCS replaces the hysteresis controller and calculates the optimal trajectories for the heat pump operation. The trajectory is calculated by solving a linear mixed-integer optimization problem considering the heat demand prediction, the power price prediction, model constraints, and objectives. The EDCS enables a reduction of energy costs by utilizing the flexible power price market while ensuring process quality with optimal heat supply. In this thesis, the online heat demand predictor of the EDCS is developed further with focus on robustness against unpredicted time shifts in the heat demand. A simulation study is conducted to examine the effect of unpredicted heat demand starts and the robustness of the EDCS on the process quality. This thesis shows that the EDCS is applicable for the examined process and reached a software status, which ensures an error-free and comprehensible execution. The study results illustrate that the EDCS can reduce insufficient heat supply while reducing the cost of power consumption. A Pareto front is presented visualizing the trade-off between process safety and consumed power cost.