📘 Resource Overview & Documentation Details

You are viewing the technical directory listing for Demand Response Load Impacts: Smart Thermostat Lessons. This engineering resource is curated under our Smart Energy division to support automation specialists, control systems engineers, and manufacturing technicians looking for verified reference materials.

🔍 Contextual Abstract & Technical Scope

Share Summary:LinkedInEmail

Demand response load impacts depend on more than the number of enrolled customers. A 2020 evaluation of Southern California Edison’s Smart Energy Program (SEP), a residential program using Wi-Fi-connected smart thermostats, illustrates how dispatch availability, event timing, weather and event duration shape the capacity a utility can count on. For energy planners and automation professionals, the findings offer practical guidance on turning connected devices into a dependable grid resource.

How smart thermostat demand response works

SEP provides participating households with bill credits in exchange for allowing their thermostat provider to adjust cooling set points during demand response events. During an event, the set point can be raised by as much as four degrees Fahrenheit, reducing air-conditioning use and, in turn, household electric demand. Participants receive a courtesy notification but are not expected to take action themselves.

The program can be dispatched for reliability, economic, system-contingency or evaluation purposes. In 2020, SEP was called on 14 days between July and September. The first eight events were economic dispatches; the remaining six were reliability events during a period of extreme heat. About 52,000 customers were active during the event season, and dispatch could be targeted by sub-load aggregation point rather than across the entire service territory.

Demand response load impacts vary over an event

The central operational finding was that an event’s position in its dispatch sequence mattered more than the time of day. Load reductions were generally largest in the first event hour, then declined in subsequent hours. The evaluation linked this pattern to thermostat operation: a set-point increase initially curbs cooling demand, but as indoor temperatures approach the new setting, air conditioners gradually resume operation.

For the average one-hour event window from 7 p.m. to 8 p.m., estimated reductions averaged 0.75 kW per customer and 38.6 MW in aggregate. For the average four-hour window from 5 p.m. to 9 p.m., the corresponding averages were 0.46 kW per customer and 23.6 MW. These figures describe average hourly impacts, not total energy savings across the full event. The longer event’s first-hour impact was stronger than its later-hour performance, but declining reductions across the dispatch period lowered its average.

That distinction matters when comparing event strategies. A short event can produce a higher average hourly reduction, while a longer event can provide demand relief across more hours. Planners need to match the event profile to the system need rather than treating a single impact figure as representative of every hour.

Availability is as important as enrollment

Enrollment counts do not automatically equal dispatchable capacity. Contractual issues with one thermostat provider meant that some enrolled SEP households were unavailable for dispatch during 2020. The evaluation estimated that about 15% of participants were unavailable on a given event day, and the utility could not identify which individual accounts were affected. As a result, the analysis included enrolled accounts whether or not their thermostats actually received the event signal.

This reduced the measured average impact per enrolled customer. The report’s comparison showed why availability-adjusted and enrollment-based results answer different questions: dividing an observed impact by the share of accounts available for dispatch raises the estimated impact per available account, while the aggregate impact remains unchanged. For program operators, that makes vendor agreements, dispatch-platform coordination and visibility into device availability essential parts of capacity planning—not simply administrative details.

In 2020, the evaluation also noted a separate thermostat-system issue that made some accounts unavailable during one event. These examples underline the need to distinguish among customers enrolled, devices available, and devices successfully dispatched when assessing program performance.

Weather and event timing affect measurement

SEP is weather-sensitive because its load reduction comes from limiting air-conditioning use. Its contribution is therefore greatest during hot conditions and may be small or absent when cooling systems are not running. During the August 2020 heat wave, the program was dispatched on six consecutive days. On August 18, the summer system peak day, the four-hour event delivered an estimated average reduction of 0.66 kW per customer and 33.5 MW in aggregate across the full event hours used in the analysis.

Measurement also depends on clean event boundaries. Residential meter data in the evaluation was hourly, so events starting or ending partway through an hour created mixed intervals: some minutes before dispatch, during demand response, or after the event. Partial hours could dilute estimated impacts or combine demand reduction with post-event increases. Where dispatch timing is controllable, starting and ending on the hour can make impact estimates easier to interpret. Emergency dispatch may not allow that flexibility, so evaluation methods must account for partial intervals.

After an event, the analysis observed a snapback effect: demand rose above the reference level in the hour immediately following dispatch, then diminished. Event planning should consider this post-event profile alongside reductions during dispatch, particularly as program enrollment grows.

What forecasting needs to account for

Ex post analysis estimates what happened; ex ante analysis forecasts what a program may deliver under specified future conditions. SEP’s forecast used customer enrollment projections, weather scenarios and estimated per-customer impacts. It considered both normal 1-in-2 weather and more extreme 1-in-10 weather, as well as SCE and CAISO weather forecasts. Because the program relies on air conditioning, projected impacts varied by season and weather conditions.

The forecast also treated COVID-related changes in residential load as a factor that would diminish over time. This illustrates a broader planning principle: historical demand and program performance may reflect unusual conditions that should not be carried forward unchanged. The evaluation projected that SCE enrollment could approach 155,000 households by 2031, but aggregate capacity depends on the assumptions behind enrollment, availability, weather and per-customer performance.

Practical implications for program operators

  • Track dispatchability, not enrollment alone. Reconcile enrolled accounts with devices that are available and successfully reached.
  • Plan around event-hour decay. Model reductions by hour within an event, rather than assuming a constant impact throughout dispatch.
  • Include post-event behavior. Snapback can affect the net operational value of an event window.
  • Improve measurement conditions where possible. On-the-hour dispatch supports cleaner comparisons, while partial-hour events require careful treatment.
  • Use weather- and scenario-specific forecasts. A weather-sensitive resource should not be assigned one fixed capacity value for every season and condition.

Conclusion

SEP’s 2020 results show that smart thermostat demand response can provide meaningful peak-demand relief, especially during hot, system-constrained periods. They also show why reliable estimates require more than a participant count: device availability, event duration, dispatch sequence, weather, interval-data limits and snapback all influence measured and forecast performance. For utilities and system planners, integrating these factors into operations and evaluation is key to treating residential demand response as a resource with a clearly understood, condition-specific capability.

Access Official Technical PDF