Optimising energy flexibility in Finnish residential buildings: A comparative study of PI, rule-based and model predictive control strategies

As integral components of urban infrastructure, buildings play a crucial role in maintaining occupant well-being, especially during extreme weather conditions. This research presents a model predictive control (MPC) approach to harnessing the energy flexibility of buildings by utilising their thermal mass to cost-effectively manage the energy use. The study compares two apartment buildings located in the Nordic climate of Helsinki, Finland: one built in the 1970s and a modern positive energy building (PEB) with a high-performance envelope exceeding the minimum requirements of national building regulations. Three control strategies are evaluated for building thermal mass activation: Proportional-Integral (PI) control as a standard strategy for thermal comfort, Rule-Based Control (RBC) as a cost-based benchmark strategy and an advanced MPC as an innovative energy-flexible strategy for cost-savings. The three investigated control strategies are implemented by interfacing IDA ICE building energy performance simulation software with the programming environment, Python as a master controller. The study aims to optimise the operation of the building's energy systems in real-time, minimising energy costs while maintaining comfort constraints by adjusting temperature setpoints based on dynamic weather conditions and occupant behaviour by applying the adaptive thermal comfort model. The results, obtained from simulations, demonstrate that the MPC provided the highest cost savings, particularly under high and fluctuating price conditions. In the 1970s building, MPC achieved up to 29.9 % cost savings compared to PI control, while RBC achieved up to 17.2 % savings. In the modern PEB, MPC resulted in up to 14.8 % cost savings, with RBC achieving up to 7.9 % savings. These findings highlight MPC's potential to improve energy efficiency and resilience in buildings, especially in cold climates.