TOWARDS BETTER MODELING OF RESIDENTIAL THERMOSTATS

Thermostats are notorious for being difficult to use. Consequently, heating is manually controlled by occupants in most homes, including those with programmable thermostats. Conventional modeling approaches fail at capturing behavioral elements by relying on fixed setpoint schedules. To accurately assess the energy impact of thermostat operation requires two changes to the way simulations are carried out. First, thermostat setpoint programs must be defined based on how thermostats are actually operated instead of the usually assumed ideal behavior; and second, realistic occupancy schedules must be adopted instead of average schedules. This may require layering of behavioral logic atop thermostat logic. In this paper we present an approach for doing this using Energy Plus and American Time Use Survey (ATUS) data, together with some examples.