Distributed Heating Networks

Abstract: There is a shortage of models and analysis methods of fourth generation district heating networks, which are capable of both extracting and depositing heat energy to some thermal network grid. This thesis fills that gap by combining mathematical models of components into a network that is capable of sending heat energy between its nodes. Questions regarding good heating strategies for controlling the nodes were posed, and based on these, some simplifications were made to produce simpler systems to work with. Near-optimal distributed control strategies were produced and tested on simulations of the full mathematical models. For comparison, tuned P- and PI-controllers were also simulated on the full system. The results showed that the optimal controllers induce less oscillations and had less stationary error, however this caused larger control signals and more grid-wise interaction, causing neighboring users to feel more of the impact when a single node changed its operating point. This effect can be suppressed if a heating battery is connected to the system.