An MPC framework for all-air systems in non-residential buildings

By the end of 2020 all newly constructed buildings have to be nearly zero energy buildings (nZEB). The transmission- and ventilationlosses have already been significantly decreased by the increasing insulation thickness and the improved airtightness of the building envelope. Further energy reductions can be achieved by a better design and optimization of energy efficient building installations. In school- and office buildings the ventilation system has a large contribution to the total energy consumption. A control strategy that adjusts the operation to the actual demand can significantly reduce the energy consumption. This is important in rooms with a highly fluctuating occupancy profile, such as classrooms and offices. However, standard rule-based 'classic” control strategy, is reactive, making the installation 'lag behind' in relation to the question. As a result, (1) a good indoor climate is not always guaranteed and (2) the actual energy saving potential is lower than predicted. A model-based predictive control (MPC) addresses this criticism. An MPC takes into account the current situation (= feedback) and the future demand, weather conditions and occupation (= feedforward). Application of MPC for the control of ventilation systems is not new. However, CO2 concentration, as an indicator of indoor air quality, is not included as a physical parameter in the model identification so far. In addition, multi-zone models are not common. Furthermore, MPC is a time-consuming and complex process that requires a lot of computational power. As a result, the application in office and school buildings is still limited, especially in ventilation systems. The objective of this PhD project is to realize energy savings by optimizing the control of ventilation systems in (nZEB) office- and school buildings. First, we want to develop a model-based predictive control (MPC) for a ventilation system that takes into account both the indoor temperature and the CO2 concentration. Secondly, on basis of these insights, we want to draft guidelines to improve the rule-based control strategy.