Performance benchmarking of residential air conditioning systems using smart thermostat data

Faulty residential Heating, Ventilation, and Air Conditioning (HVAC) systems waste billions of dollars of energy usage in the U.S. each year, meanwhile heavily affecting occupant comfort and causing complete operational failures. Fault detection and diagnosis (FDD) methods assist technicians in discovering and locating faults as early as possible, thereby optimizing system efficiency through predictive maintenance. However, the majority of FDD methods are developed for commercial HVAC systems. This is because FDD needs data acquired from sensors installed on the HVAC equipment, but in the residential sector, the initial cost of installing additional sensors is uneconomical. In recent years, the development of smart thermostats and the Internet of Things (IoT) enables large-scale system operational data to be recorded. The collection of smart thermostat data has nearly no cost, and FDD can be conducted using aggregated data obtained from a large number of systems using IoT technologies. Hence, this paper proposes a performance benchmarking method for residential systems that only use smart thermostats as the data source. The proposed method analyzes the system pseudo steady-state behavior and compares a critical feature termed the weighted average difference of cooling effort (Delta Ec) both between systems and within each system. Three applications of the proposed method are detailed: identification of inadequate capacity, quantification of active capacity degradation, and verification of repair after maintenance. Within each application, real faulty systems are illustrated to demonstrate the usefulness of this method.