Scalable, reconfigurable Model Predictive Control for building heating systems

Inefficient design and operation of building heating systems can have a large impact on global energy consumption. Traditional heuristic approaches often supply excess heat and cannot adapt to faults and changes in the building and heating system. Model Predictive Control (MPC) based strategies can incorporate future building usage and weather conditions to achieve more efficient heating. While MPC can produce an improved performance over standard strategies, many approaches taken in the literature are not easily scalable and do not allow for intuitive reconfiguration. Two possible MPC strategies for control of a building heating system are designed and compared here. In the first strategy, the thermal comfort of the occupants of the building is balanced with the energy use in a single objective function. In the second strategy, a lexicographic, multi-objective formulation is used to split the competing goals of energy reduction and thermal comfort. The strategies are assessed in a validated simulation platform in terms of energy efficiency, comfort performance, scalability and reconfigurability in times of system changes or faults.