Load shedding — the controlled disconnection of distribution feeders to reduce load — remains a last-resort tool that carries real costs: customer harm, regulatory scrutiny, and reputational damage. Demand response is designed to reduce the same load through voluntary curtailment before those costs materialize. The gap between when DR could have worked and when operators reach for load shedding is usually not a technical limitation. It's an information problem.
The Decision Point That Gets Missed
Under-voltage load shedding (UVLS) and operator-initiated controlled load shedding serve different functions. UVLS is automatic — it fires when voltage falls below protection relay thresholds regardless of operator intent. Controlled load shedding is a deliberate operator decision to disconnect feeders, typically when reserves have been exhausted and other options have been ruled out.
The missed decision point is usually not at the moment of load shedding itself — it's 20–45 minutes earlier, when the load trajectory was becoming apparent and DR dispatch would have been operationally feasible. Post-event analysis of controlled load shedding events consistently finds that the qualifying conditions for DR dispatch — forecast exceedance of reserve margin, sufficient enrolled DR capacity, adequate response lead time — were present before the load shedding decision was made. The DR option wasn't chosen; in many cases it wasn't evaluated.
The reason is operator cognitive load under stress conditions. When ACE is deteriorating, reserve margin is dropping, and the system is approaching distress conditions, the operator's mental model doesn't have bandwidth for a multi-step assessment of DR availability, expected response, and timing. The decision flow under stress runs toward the most familiar, controllable tool. Load shedding is familiar — operators train for it. DR dispatch requires confidence in asset availability data that may not be immediately visible in the EMS display.
What "Available DR Capacity" Actually Means in Real Time
A DR program's nominal enrolled capacity — the number quoted in resource planning documents — is not the number that matters at the moment of a dispatch decision. What matters is the available capacity at that specific moment, accounting for: assets currently in a recovery period from a previous curtailment event within the last 24 hours, assets that have opted out of the current event window, assets in maintenance status with the aggregator, and assets whose telemetry indicates they're already operating near minimum load (AC already at setpoint, no further reduction available).
For a program with 8 MW nominal enrolled capacity, the real-time available capacity at any given moment may range from 3 MW to 7.5 MW depending on these factors. An operator who knows only the nominal number cannot make a reliable decision about whether DR can cover the required load reduction. An operator who can see real-time available capacity — updated by asset telemetry on every polling cycle — has the information needed to choose DR over load shedding with appropriate confidence.
This real-time capacity visibility is the gap that most utility DR program management systems don't fill. Post-event enrollment dashboards and day-ahead availability estimates exist in most program management systems. Real-time capacity derived from SCADA telemetry and asset status feeds, updated at the 5-minute interval that operators use for dispatch decisions, is much rarer.
The Forecast Lead Time Requirement
HVAC-based demand-response assets have a response latency: after receiving a curtailment signal via OpenADR, the typical commercial HVAC system begins reducing load over 10–20 minutes as the thermal mass of the building absorbs the setpoint change. The full contracted MW reduction is available 20–30 minutes after the dispatch signal, not immediately.
This latency means that DR dispatch must be initiated when the load projection exceeds the threshold — not when actual load has already exceeded it. By the time measured load has crossed the threshold and the operator has assessed the situation, the DR dispatch window that could have prevented load shedding has already closed. The 20–30 minute response delay means the curtailment won't materialize until 50–60 minutes after the decision point, at which point load shedding may be the only remaining option.
The technical requirement is a 15-minute-ahead load forecast with explicit presentation of the probability that load will exceed the reserve margin threshold within the DR response window. This isn't a point forecast ("load will be X MW in 15 minutes") — it's a probabilistic statement ("there is a 78% probability that load will exceed the reserve margin threshold within the next 25 minutes, when DR response would be available"). That framing enables an operator to make a justified dispatch decision with appropriate lead time.
Making DR the Default in High-Risk Periods
Some utilities have operationalized the DR-first decision by creating automatic pre-activation protocols for high-risk operating periods. When day-ahead forecast indicates load above the 90th percentile historical level and temperature forecast exceeds a defined threshold, the DR program issues a courtesy notification to enrolled assets that a curtailment event is possible in the next 12–24 hours — warming up the program without formally committing to dispatch.
This pre-activation reduces the operational surprise of DR dispatch during actual events, improves response reliability (assets have had time to prepare for setpoint changes), and provides an opportunity to identify assets that won't be available before the event rather than discovering unavailability during dispatch. It also gives operators an earlier signal that DR is "in play" for the operating day, making it more likely to be considered before controlled load shedding becomes necessary.
Ancillary Services as an Alternative to Both
For utilities operating in ISO/RTO markets, ancillary services — specifically spinning and non-spinning operating reserves — provide an alternative to both DR dispatch and controlled load shedding during reserve margin events. When the ISO/RTO activates emergency operating procedures and deploys operating reserves, the relief to the system serves the same function as local load reduction, but through the wholesale market mechanism rather than the utility's own DR program.
The interaction between local DR dispatch and ISO/RTO ancillary service deployment creates a sequencing question: should the utility dispatch its DR program before calling on ancillary services, or wait for the ISO/RTO to activate market reserves? The answer depends on the cost of DR dispatch (incentive payments to participants) relative to the cost of ancillary services (which may be higher during scarcity conditions), the reliability of DR response relative to the certainty of market reserve delivery, and whether the utility's DR program qualifies as an ancillary service provider that would be compensated for the same curtailment it's already dispatching.
Utilities that have aligned their DR programs with ISO/RTO ancillary service qualification requirements can dispatch their own enrolled assets into the wholesale market during shortage events, recovering program costs through wholesale settlement rather than paying participants from operational budgets. This alignment requires program design and telemetry infrastructure that meets the specific ISO/RTO qualifying requirements for the relevant ancillary service product.