Experimental evaluation of model predictive control and fuzzy logic control for demand response in buildings

Energy flexibility is essential for aligning the energy demand with the intermittent electricity generation from renewable energy sources. In the European Union, buildings account for 40 % of the final energy consumption, thus offering significant potential for energy flexibility through load shifting, peak clipping, valley filling, and flexible load shaping. While experimental studies are crucial for providing realistic estimates of cost savings, comparing various control algorithms in the real world is inherently difficult. The present paper addresses this challenge by simultaneously controlling three architecturally identical buildings with different controllers to shift space heating in response to dynamic pricing. Model predictive control (MPC) and fuzzy logic control (FLC) are compared to a baseline control across various experiments, encompassing different objectives, price signals, and comfort levels. Over the course of a one-month experimental study, both MPC and FLC improved thermal comfort while achieving cost savings ranging from 7.8 % to 33.4 % and from 4.4 % to 8.6 %, respectively. The additional savings provided by MPC compared to FLC increase with greater price variability, indicating that MPC is particularly advantageous in markets with high price spreads. Conversely, when prices fluctuate less, the computationally more efficient FLC is sufficient. When minimizing costs, the MPC reduces the heating costs by 33.4 % but merely reduces the CO 2 emissions by 2.9 %. Consequently, focusing solely on cost minimization is insufficient to achieve substantial emission reductions. • Real-world comparison of advanced heating controls for building demand response. • Three identical buildings enable a reliable control performance evaluation. • Broad comparison includes various objectives, temperature bounds, and disturbances. • CO2 and cost savings range from 2.1 % to 33.4 % while improving comfort.