• Morten Kajhøj Riis-Vestergaard
3. term, Control and Automation, Master (Master Programme)
This report analyses how district heating for a single-family home can be controlled, leading to two control schemes. Model predictive controllers are designed based on identified models. Models are identified with a grey-box method. For which a model structure is derived based on one-phase energy balance, where after model parameters are estimated. The collaboration partner Neogrid provides the data used for parameter estimation. As part of the flexibility Heat Grid Bornholm, a research and demonstration project\cite{Flexibility_Heat_Grid_Bornholm}, their gateways collect data from 35 single-family home's district heating installations. The resulting models were affected by a wrong time delay inherent from the model structure and disturbances in the collected data from fireplaces. Resulting in only a few models being approved. A cost function for an MPC is designed with slack variables to give a desired behaviour. The MPC is simulated successfully and proves the concept of estimating a model and constructing a controller without manual inputs. It additionally showcases how the MPC can be used to move the peak energy usage. Due to a lack of consistency, the design structure and subsequent controller are not implemented.
Publication date5 Jan 2023
Number of pages80
External collaboratorNeogrid Technologies ApS
R&D Scientist Anders Larsen ala@neogrid.dk
Place of Internship
ID: 508856603