In summer, a ceiling fan has a tremendous potential to cool rooms, resulting in energy saving and a peak load reduction. During the critical peak pricing (CPP) period of the electricity, one method to attain substantial energy cost saving is to turn off the heating, ventilating and air conditioning (HVAC) system, but this may result in overheating. Fortunately, precooling and ceiling fans can be used to reduce the feeling of hot sensation. Therefore, the best control strategy for an HVAC system and ceiling fans is to maximize energy cost saving without much thermal comfort sacrifice. This paper presents a multi-objective optimization model predictive control (MO-MPC) of precooling temperature setpoint and ceiling fan speed. The proposed MO-MPC integrates an optimization engine, Dakota, and a detailed energy simulation tool, EnergyPlus, to determine the optimal control strategies considering the energy cost and thermal comfort simultaneously. A sub-program of substituting EnergyPlus with decision tree regression models for building climate prediction is deployed. If the calculation time is too long to achieve real-time control, this sub-program can be adopted to accelerate the calculation. In this study, the proposed MO-MPC framework is applied to a small and large building in four different cities with different CPPs. For a summer day, compared with the baseline cases of no precooling and no ceiling fan, the optimal strategies for the small building achieve energy cost savings between 10.3% and 33.3% and a peak power load decrease from 44.0% to 51.9%. Without precooling but with a ceiling fan control, on a particular summer day, for the large building case in Shanghai, 7.4% energy saving is achieved. The optimal control of the precooling temperature and ceiling fans result in a 14.4% energy cost saving. By using the decision tree regression models instead of EnergyPlus, the calculation time of the big office reduces from about 10 hours 40 minutes to 24.4 minutes.
Read full abstract