Grid Frequency Response · Path 2 of 2

Supply > Demand

Frequency climbs. Here's what happens, second by second.

3.4M MWh Curtailed1

That's how much solar and wind power CAISO deliberately curtailed in 2024, just to keep the Western Interconnection's frequency from climbing too high — 93% of it solar, mostly on mild, sunny spring days when supply ran ahead of demand. Curtailment is the last, slowest lever available. Here's what happens in the seconds before it's ever needed.

Arresting
0Sec
Automatic · physics

Electromagnetic Lift

As supply outpaces demand, generators draw less current — and by Lenz's Law, less current means less opposing magnetic drag on the spinning turbine. With less resistance holding it back, the turbine speeds up on its own, before any machine or person can react.

<1Sec
Automatic · sensing

The Response Trigger

Various sensors detect the grid's frequency drifting away from its 60 Hz baseline — the same sensors, whichever direction it drifts.

1–2Sec
Automatic · reflex

Absorbing Power

Grid batteries switch to charging, soaking up the surplus within 100–200 milliseconds. Automated turbine valves ease closed to let in less steam or water.

2–10Sec
Secondary · AGC

AGC Springs Into Action

Audible alarms flash on the grid operator's screens. Every 2–6 seconds, AGC recalculates the exact size of the surplus and sends corrective signals to regulating reserve — telling online generation to back off — a continuous loop that keeps running all the way through Recovery.

2–10Sec
Automatic · reflex

Primary Response Holds the Line

The initial battery-charging and valve response keeps running automatically, holding output steady and keeping frequency from climbing any further while the rest of the system prepares its next move.

Apex Point2

Is the rise arrested before generators start tripping offline?

Yes

Frequency stabilizes below the threshold — recovery continues below.

Rebound
20–52 Sec Automatic · reflex

Primary Response Tapers

That automatic response is proportional to the remaining surplus — as AGC-commanded generation backs off further, there's less error left to correct, so the initial response recedes on its own.

20–52 Sec Secondary · AGC

AGC Takes Over

Grid software commands online, already-synchronized generation — called regulating reserve — to throttle back using its available headroom.

Point B
Stabilizing frequency
Recovery
1–10 Min Operator action

Curtailing Excess Generation

Grid operators issue curtailment orders to solar and wind plants — the same lever CAISO already pulls on sunny spring days1, cutting output at the source rather than adding more.

10 Min – End of Recovery Operator action

Recovery to Point A

Curtailment and reduced dispatch close the remaining gap — frequency returns to its scheduled 60 Hz.

No

Frequency crosses the over-frequency protection threshold.

Last resort

Generators Trip Offline

Automated protection relays disconnect generating units to cut supply and protect equipment from over-speed damage — within seconds, since secondary control (1–10 min away) can't arrive in time to help. Unlike Path 1's UFLS, no customer loses power here — it's generation coming offline, not load being shed.

Over-frequency protection is built backwards from UFLS. Nothing tells a generator when to trip — NERC instead defines a “no trip zone” every unit has to ride through. Below 60.6 Hz it must stay connected indefinitely; above that, the time it is required to hold on collapses.3

Below 60.6 HzStay connected — no limit
60.6 Hz and above180 sec
61.6 Hz and above30 sec
61.7 Hz and aboveMay trip instantly

0.1 Hz is all that separates “hold on for 30 seconds” from “may disconnect immediately.” And leaving the envelope is permission, not instruction — which is why this branch has no five-stage table the way Path 1's UFLS does.

Automatic · consequence

Recovery to Point A

Frequency returns to its scheduled 60 Hz — the immediate crisis is over, but the tripped generation is still offline.

Operator action

Resynchronizing Generation

Once frequency is confirmed stable, tripped units go through the normal startup and grid-synchronization process before resuming output — on their own timeline, not a fixed restoration schedule the way UFLS-shed customers get.

References & Notes

  1. U.S. Energy Information Administration, Today in Energy, "Solar and wind power curtailments are increasing in California" — CAISO curtailed 3.4 million MWh of utility-scale wind and solar in 2024, a 29% increase over 2023; solar was 93% of all curtailed energy. Curtailment peaks in spring, when solar output is relatively high and demand is relatively low (mild temperatures, less heating/cooling load) — the same seasonal pattern behind California's record 51% solar month (see the companion Factoid of the Day). Curtailment happens for two reasons: oversupply (generation exceeds demand) and congestion (transmission lines lack capacity to move available power).
  2. NERC, Balancing and Frequency Control reference document (May 11, 2021) — same Points A/B/C/D and Arresting/Rebound/Stabilizing/Recovery structure as Path 1, direction-reversed. The document itself notes the C-point terminology flips with direction: "a similar response calculation can be made from the A point to the C point (nadir, if a generation loss or apex, if a load loss)" — the source for calling this chart's peak the "Apex Point" rather than reusing Path 1's "Nadir."
  3. NERC Reliability Standard PRC-024-3 — Frequency and Voltage Protection Settings for Generating Resources, Attachment 1, Table 2 (Frequency Boundary Data Points – Western Interconnection), high-frequency side: continuous operation below 60.6 Hz, 180 sec at ≥60.6 Hz, 30 sec at ≥61.6 Hz, instantaneous permitted at ≥61.7 Hz. Requirement R1 obliges each Generator Owner to set frequency protection so the unit does not trip or cease injecting current inside that “no trip zone,” with documented regulatory or equipment-limitation exceptions allowed under Requirement R3. The standard states explicitly that “the area outside the ‘No Trip Zone’ is not a ‘Must Trip Zone.’” That is the structural reason this branch cannot mirror Path 1's UFLS stages: underfrequency load shedding is a coordinated regional plan with assigned percentages per stage, while over-frequency disconnection is each unit's own protection setting within a common boundary — there is no published schedule of which generation comes offline at which frequency, because no such coordination exists. Frequency here is calculated over a window of time, per the standard's own note on the boundary tables.
  4. Open flag, not yet resolved: Path 1's UFLS side has a fully sourced 5-stage shedding table (WECC Coordinated Plan thresholds, percentages, relay timing). We have not found an equivalent sourced stage-by-stage table for over-frequency generator tripping in the Western Interconnection — the "Generators Trip Offline" branch above is presented qualitatively, based on general protection-engineering practice, not a specific WECC standard. Needs real sourcing (likely NERC PRC-024 generator ride-through requirements, or WECC's own over-frequency protection guidance) before this branch matches Path 1's rigor.
  5. Regulating reserve, spinning/non-spinning/supplemental reserve definitions, and the Points A/B/C/D framework are shared with Path 1 — see that chart's References & Notes for the full citations (FERC Ancillary Services; NERC Balancing and Frequency Control) rather than duplicating them here.