Real-time building heat gains prediction and optimization of HVAC setpoint: An integrated framework. (15th May 2022)

• Record Type: Journal Article
• Authors: Wang, Zu; Calautit, John; Wei, Shuangyu; Tien, Paige Wenbin; Xia, Liang
• Is Part Of: Journal of building engineering. Volume 49(2022)
• Publication Date: 2022-05-15
• Subjects: Artificial intelligence -- HVAC temperature Setpoint optimization -- Building internal heat gains -- Building energy reduction -- Occupants' thermal satisfaction; Building -- Periodicals; 690.05
• DOI: 10.1016/j.jobe.2022.104103
• Languages: English
• ISSNs: 2352-7102

Abstract

Abstract: Heating, ventilation and air-conditioning (HVAC) systems are responsible for a large part of building energy consumption. Lowering HVAC's energy consumption without compromising occupants' thermal satisfaction has been the focus of recent research. Existing studies showed that this requirement could be met by optimizing HVAC setpoints or using demand-based control strategies. However, such strategies require coupling with dynamic information such as occupancy to be effective. This study proposes an integrated framework to achieve real-time optimization of HVAC setpoints. This framework has three core components: a vision-based camera for detection, predictive HVAC energy and thermal comfort models, and HVAC temperature setpoint optimizer. Vision-based cameras were used to identify activities of occupants and equipment usage and predict a real-time value of internal heat gains. By feeding the information into the predictive HVAC energy and thermal comfort models, the HVAC load (heating or cooling) and the level of occupants' thermal dissatisfaction, predicted percentage dissatisfied were obtained using developed models, based on shallow neural networks. Afterwards, the HVAC temperature setpoint optimizer determined an optimal temperature setpoint following two optimization rules. When occupants were absent, HVAC system would be turned off to save energy. When they were present, an optimal temperature setpoint was determined by using Pareto-front. Based on the initial findings, the utilization of this framework can potentially lead to a reduction of HVAC heating energy by up to 36.8% and occupants' thermal dissatisfaction by up to 5.26%. In the summer, the savings of HVAC cooling energy would range from 3.5% to 33.9%, whilst occupants' thermal dissatisfaction could be decreased by 0.17–2.89%. Overall, this paper paved a path to achieve a win-win on HVAC energy savings and comfort by combining artificial intelligence techniques and building services. Graphical abstract: Image 1 Highlights: Vision-based approach was used to detect and predict real-time internal heat gains. Predictive room HVAC loads and thermal comfort models were developed. A framework for optimizing the HVAC setpoint was proposed. The proposed framework can reduce room HVAC loads up to 36.8% in winter and 33.9% in summer. The proposed framework can increase occupants' thermal satisfaction.