Impact Metrics for Residential HVAC Systems using Cloud-Based Smart Thermostat Data

Abstract

The aggregation of data from connected smart thermostats installed in a huge number of residential buildings has expedited the remote detection and diagnosis of faults in Heating, Ventilation and Air-Conditioning (HVAC) systems. Upon identification of faults in air-conditioning systems, manufacturers and occupants are interested to know how severe the impact of the faults is on the energy consumption and the thermal comfort of the occupants. Several studies in literature have previously attempted to quantify an energy impact metric and a thermal comfort impact metric of faults in an HVAC system, but the metrics developed lack the ability to be used objectively to compare several systems at once. Furthermore, no study has yet tried to examine the coupled relationship between the energy consumption of the system and the thermal comfort of the occupants to estimate an aggregate fault severity index of a system.

The current study attempts to provide a paradigm shift in the calculation of the energy impact metric. The thesis, firstly, proposes a methodology to model the energy consumption of the average system in a dataset comprising of similarly sized system operating in the same climate region. The performance of each air-conditioning system is compared to the performance of the average system to estimate the amount of impact faults have on their energy consumption. Additionally, the current study also estimates the level of thermal discomfort felt by occupants of the house using the Predicted Mean Vote (PMV) of the indoor environment. The average level of discomfort felt by the occupants living in the house is then compared with a baseline to estimate impact on the thermal comfort of occupants.

The two impact metrics are then combined together into one index that represents the fault severity index of the system which can then be used to rank systems to prioritize them for repair. The severity index of the system is a representation of the relative energy consumption level of the system if it were to produce no thermal discomfort. Another metric that comes as a by-product of this derivation is the amount of change in energy consumed by the system in order to make the indoor environment comfortable. The coupled nature of the four metrics will be delineated so as to gain an insight into the characteristics of air-conditioning systems. Causes for faulty behavior of systems are examined and systems with mechanical faults are segregated from systems operating under ineffective operating conditions.