Deployment of crowdsourced occupant data to support fault detection and diagnosis in buildings

Abstract

This proof-of-concept paper proposes a method that integrates occupants' feedback and personal data to enhance the performance of fault detection and diagnostic (FDD) technology in buildings. The proposed approach uses a smartwatch app called Cozie to collect occupants' subjective feedback on thermal comfort and air quality, and the data is transferred to a cloud database through a data stream. Similarly, an application programming interface is used to retrieve sensor readings from the building automation system. FDD rulesets then use the data to determine process variable errors and report alarms. The results of an experiment show that occupant input accurately prove/disproved alarms and allowed for a faster and more accurate FDD. The occupants' feedback also helped prompt rule-based FDD algorithms and support the alarms triggered. Two independent surveys were also conducted. According to the first survey, occupants tend not to report discomfort to building operators through conventional methods unless the indoor temperature is more than 3 °C warmer or cooler than their desired temperature. The proposed method helped occupants report discomfort issues earlier, and the percentage of occupants choosing 3 °C warmer or colder conditions decreased by 29%. The paper concludes that the integration of occupants' feedback and personal data can significantly enhance FDD performance and provide more comfortable environments for building occupants.