Abstract

Wifi-enabled “smart” thermostats are thought to hold substantial potential for managing peak domestic electricity demand. But under what atmospheric conditions are consumers willing to accept energy suppliers’ remote control of their heating, ventilation, and air conditioning (HVAC) systems? We analysed historical data on 231,344 instances between wherein Octopus Energy U.S. (OEUS) remotely enacted time-limited changes (≈1-150 minutes) to the heating and cooling set points of the thermostats of 1,159 of its customers in Texas. Using survival analysis — a statistical framework for modelling the elapsed time until a terminating event occurs (e.g., death) — we estimated how the probability of override changes over the course of 150 minutes in relation to atmospheric conditions near to a customer’s home (above figure). Non-causal results suggest that internal temperature, external temperature, and time itself combine to produce a powerful gradient along which customers’ propensity to endure retailer control varies.

Key findings

① Overall, consumers are willing to cede some control
Overriding remote thermostat control appears rare overall: just 6% of all “energy conservation events” ended due to a customer rejecting OEUS control by overturning their modulated set points.

② Temperature is associated with consumers’ willingness to accept
External temperature is clearly associated with customers’ propensity to endure retailer control. When atmospheric conditions are severe for Texas (i.e., 10-30°F and 90-110°F), the probability of customers accepting the set points of their thermostats can fall to 71-72% over the course of a 150-minute event.

③ Longer events are associated with a higher chance of override
When outside temperature is not severe, time appears to be the salient factor. On average, for OEUS, the survival probability decreases from ~95% about 25 minutes into the event, to ~80% at about 150 minutes. Around 1 hour seems to be a “sweet spot”, keeping survival probability to ~90%.

④ Pre-cooling/-heating strategies could have a big impact
Pre-cooling/-heating could make even long periods of control viable during bouts of extreme temperature and thus times when demand and grid constraints are likely to be greatest. Our non-causal modelling suggests that changes to internal home temperature prior to an event could see survival probabilities under more severe external conditions remain as high as 90-97% over 150 minutes (above figure; furthest-left and furthest-right of the five panels).