Demand Response Load Shedding: Thermostat Cycling & Cold-Load Pickup

DR Event Control

Idle — normal thermostat operation

Program participation 55%

Curtailment duty-cycle 40%

Outdoor temperature 34 °C

Aggregate Load

Actual load Baseline (no DR)

Actual load

0 kW

Baseline load

0 kW

Load reduction vs. baseline

Avg. indoor deviation from setpoint

0.0 °C

How it works

Each of the houses below runs an independent thermostat with dead-band hysteresis around a 22°C setpoint, driven by a first-order thermal model:

dT/dt = k·(T_out - T_in) - Q·[AC is on]
AC turns ON  when T_in > setpoint + 0.5°C
AC turns OFF when T_in < setpoint - 0.5°C

During a demand-response event, the aggregator applies direct load control (DLC): enrolled air conditioners are duty-cycled — forced off for a fraction of every control window regardless of the thermostat call, e.g. 40% off / 60% free-run:

Aggregate load  P(t) = Σ P_unit · [AC_i is on and not curtailed]
Peak reduction  = 1 - P_actual(t) / P_baseline(t)

Rotating (staggered) — each house's off-window is time-shifted, so only a steady fraction of the fleet is off at any instant. Coincident peak drops smoothly and recovers smoothly.
Synchronized — every enrolled house shares the same off-window. All of them call for cooling again the instant curtailment lifts, producing a cold-load pickup (CLPU) spike that can briefly exceed the pre-event baseline — a well-documented effect in utility AC-cycling programs.

Real-world relevance: this is the mechanism behind utility "AC cycling" / direct-load-control demand-response programs (e.g. PeakSaver, Cool Rewards) — an aggregator sheds coincident peak load across thousands of enrolled thermostats without shutting anyone's power off, and grid operators must plan for the rebound when the event ends.

AC running (load on) AC idle

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.