MPC Using a Digital Twin Versus Simpler Traditional Heating Control in Swedish Multi-Family Buildings—A Measurement-Based Study

Our buildings must become more energy efficient. Multi-family buildings account for a significant part of the total energy use. However, many energy-saving measures are very costly and require extensive effort. In this context, improved heating control is a cost-effective supplement. Over the past decade, sensors, machine learning, and related technologies have enabled more advanced heating control. Much of the development relates to model predictive control (MPC), but feedback control (FBC) remains useful on its own and as a stabilising component in various MPC strategies. This field study’s novelty lies in comparing two neighbouring, similar Swedish multi-family buildings with traditional feedforward heating control (FFC). One building was equipped with an established pure FBC strategy and the other with a pure MPC strategy. The latter is based on a tailor-made digital twin with continuous input of real-time weather and household electricity data, etc. In the MPC building, the savings amounted to 11 kWh/m2, and in the FBC building to 6 kWh/m2. The FBC results are consistent with a recent similar field study in the same city. However, the MPC results are unique as they are based on a highly detailed digital twin. The results indicate that, under similar conditions, MPC strategies should involve FBC sparingly if energy efficiency is a high priority.