ThermoCoach

Abstract

Thermostats have the potential for tremendous impact on global energy consumption, but unfortunately they are often not used effectively. In this paper, we present a new system called ThermoCoach that improves thermostat usability by giving personalized and actionable recommendations for thermostat use. The system senses human occupancy patterns in a home and emails the household suggested setpoint schedules that can be modified or activated with the click of a button. We performed a randomized controlled trial by deploying over 600 devices in 40 homes from 12 weeks to compare ThermoCoach with a manually programmable thermostat and the Nest learning thermostat. Results indicate that ThermoCoach saves 4.7% more energy than a manually programmable thermostat and 12.4% more energy than the Nest learning thermostat while significantly improving comfort over both approaches.