EE Savings from Optimized Connected Thermostats

This assessment involves recruiting current customers to participate in a simple schedule optimization program via push notifications from connected thermostats. This optimization program will make small temperature adjustments one degree per week for up to three week to arrive in a new thermostat schedule. This assessment will test for EE potential and coincidental demand peak shaving.

The optimized connected thermostat project was a Southern California Edison (SCE) assessment implemented by a connected thermostat manufacturer toward the end of the summer of 2017. The goal of the optimization was to reduce household energy usage during utility peak hours by making small adjustments to the set point temperatures of connected thermostats throughout SCE’s service territory. SCE contracted with an evaluation firm to conduct an independent validation of the manufacturer’s calculation of the energy savings attributable to the optimization. The manufacturer administered the optimization algorithm to a randomly selected group of customers in SCE’s service territory who already had a connected thermostat by the manufacturer. Each customer in this treatment group received a message on their thermostat prompting them to opt into the project, allowing the manufacturer’s algorithm to adjust their temperature set points and thereby reduce their home’s cooling load. The manufacturer also maintained a control group in SCE’s service territory that was not given an option to participate in the project. These connected thermostats collect data on temperature set points as well as runtime of the connected HVAC equipment. The manufacturer conducted its own analysis with this proprietary data to estimate the energy savings impacts of the optimization algorithm from an estimated change in daily cooling runtime. In theory, reduced cooling runtime will result in reduced energy usage (i.e., energy savings). The manufacturer’s evaluation approach required assumptions about the size of the connected HVAC cooling equipment to convert the estimated reduction in cooling runtime from the regression model used by the manufacturer to kWh energy savings. The main benefit of this approach is that the manufacturer’s thermostat data are available for the full population of treatment and control customers. We also estimated the hourly impacts of the optimization algorithm utilizing advanced metering infrastructure (AMI) billing data collected by SCE and the AMI Customer Segmentation (AMICS) modeling approach. Unlike the manufacturer’s proprietary thermostat data, whole-home billing data measures household kWh energy usage directly, without the need for an assumption about HVAC size. The optimization algorithm was implemented without collecting utility account IDs or any contact information that could be used to link participants to their SCE billing data. This was due to the fact that these participants were already using their thermostats under an existing set of terms and conditions that the manufacturer said it was unable to modify in order to accommodate this assessment. To identify a sample of participants for billing analysis, The evaluation firm and SCE developed a short web survey to gather participant contact information and consent to share data. These survey respondents were then matched to SCE utility accounts by home address and the manufacturer’s customer user accounts by email address. The resulting database contained detailed thermostat usage data and utility billing data for a small sample (