Quantifying the Impact of Override Behavior on the Performance of a Summer Direct Load Control Program

Abstract

Demand response (DR) programs represent an important tool for mitigating electric grid reliability risks posed by extreme weather, heat events, and increased electrification. Residential direct load control (DLC) programs with behavioral elements, such as an override option, are often used during the summer to reduce load from air conditioners (AC). Although override options make DLC more attractive to consumers, the high likelihood of an imperfect response can significantly reduce the effectiveness of the program overall. Thus, utilities must understand the impact of these behavioral elements on load-shaving capabilities. To fill this gap, we design a regression-based, deterministic model to predict the behavior of thermostats participating in ecobee’s Donate Your Data initiative under various temperature conditions. The model is then used to quantify the impact of the override option on the performance of 403 ecobee thermostats participating in Southern California Edison’s 2019 Summer Smart Energy Program. Our analysis estimates that although the group’s participation in the DLC program led to a 41% reduction in AC demand, 48% of potential load reduction was lost to exercise of the override option. Similarly, most of the so-called DR events followed a pattern of near-perfect participation for a short duration of time preceded by a steady increase of overrides as the event progressed. DLC programs with an override option are effective at reducing demand in the aggregate, but designers must consider the savings lost to behavior.