Abstract
The electrical usage of air conditioner is highly dependent on the behavior of the user. Different user behavior patterns can yield different electricity consumption results. However, it remains poorly characterized and quantified. Today, pervasive sensing and the Internet of Things present opportunities to collect rich datasets on products use; that can better profile variation in use-phase parameters. A data-driven approach relies on measuring complex social world through sensor systems, and representing user behavior with probability distributions. Variation in user behavior patterns due to lifestyle, weather, etc. can be estimated more accurately. The case of air conditioning was examined to illustrate the approach. In Singapore’s tropical climate, air conditioning or building cooling systems consume 30% of the national household electricity. Lack of integrated design software considering the uncertainty and variation in user-phase interaction usually leads to the oversized designs. Thus, factoring user behavior patterns when assessing the environmental profile of air conditioning systems is essential. 5 months’ data in high frequency for user behavior of 15 rooms in SUTD Dover Campus test bed were collected and investigated. User behavior related terms such as air conditioner using time, occupant time were calculated. A stochastic model for daily electricity consumption of this system was built that considered uncertainty using sensor observations of user interactions with the building environment. Six probability states were defined in this stochastic model and their histograms were formed. Study for the six probability states was conducted in two parts; first analysis part used the total data collected while the second analysis part used the different user types’ data. Using Monte Carlo Simulation, by employing distributions for six probability states in stochastic model, the daily electricity consumption were reported. The contribution to electricity consumption by each probability state of user behavior was presented as well. Results showed that the user behavior took up about 80% of the total electricity consumption and had a great influential power towards the electricity consumption uncertainty of building cooling system.