Model predictive control of hydronic heating systems in buildings: Evaluation of energy performance, peak shaving and thermal comfort

Abstract

With the increasing need for energy efficiency in the building sector, advanced control strategies might prove to become essential for heating systems. This thesis investigates the implementation and performance of utilizing model predictive control (MPC) to optimize the hydronic heating system’s supply temperature in a high complexity, mod ern office building. The evaluation focuses on energy performance, peak shaving and thermal comfort. To evaluate the control strategy, a reference building within the advanced building sim ulation tool IDA ICE was utilized. Since the software does not support model predictive control, a co-simulation was established between IDA ICE and Python. As a predictor for the MPC, simplified resistance capacitance models representing the building zones were constructed. The simulations were conducted over a two-week period to optimize computational efficiency and in March to evaluate the impact of solar gains. Initially, the results indi cated energy savings of 20–30% when implementing the MPC comparedto atraditional control system. However, after eliminating temperature discrepancies to ensure a cor rect comparison, the actual savings related to the control logic were determined to be 3–5%. Regarding peak shaving, the analysis indicated a potential average reduction in peak heating demand of 26% over a one-week period, while indoor thermal comfort was successfully maintained within the defined boundaries of 21–25°C. Furthermore, the robustness of the MPC was tested against forecast uncertainties in outdoor tempera ture and occupancy. The results demonstrated that the closed-loop feedback effectively compensated for prediction errors and maintained thermal comfort. Multiple sources of error and limitations were identified. For instance, the results of the shortened two-week simulation period may not apply for other periods of the heating season. Additionally, focusing solely on March presents an idealized scenario since the conditions in March are typically favorable for an MPC due to moderate heating demands and significant solar gains. Nonetheless, the results clearly demonstrate the potential of implementing MPC. While further testing on a physical building is required to validate these findings, MPC offers a viable path forward for enhancing energy effi ciency in modern buildings.