Yang Yulong — The Full Spectrum
English · 简体 · 繁體

Why must the grid balance supply and demand every second?

Because electricity cannot be stored in bulk the way oil can — it must be produced and consumed in the same millisecond. Whether generation matches consumption is written directly into grid frequency: 60 Hz in North America, 50 Hz across most of the world. Frequency is an absolute indicator — if it falls, consumption has already exceeded production — and it is simultaneously the rotational speed of every generator rotor on the system.

What separates electricity from every other energy carrier is that it cannot be stockpiled. Oil goes into tanks and coal onto stockpiles, but electricity announces its existence only at the moment it fails: cities go dark, hospitals switch to backup, financial markets halt. That requirement — produced and consumed in the same millisecond — makes the grid the largest and most fragile machine humanity has built. Electricity currently drives global load at a scale of 5.6 TW, and if every high- and low-voltage cable on Earth were laid end to end the total would exceed 80 million kilometres. Yet this machine has no central controller and no master switch; it sustains spontaneous second-by-second coordination among thousands of operators and hundreds of millions of individual acts of consumption.

Balance shows up as frequency because most electricity still comes from rotating machinery. In a Rankine cycle plant, steam drives turbine blades that spin at 3,000 rpm on a 50 Hz grid and 3,600 rpm on a 60 Hz grid — rotational speed and frequency are rigidly locked together. When consumption suddenly exceeds generation, those rotors can only lend power out of their own kinetic energy, so they slow, and frequency slows with them. Frequency is therefore the direct measure of the balance between every spinning turbine blade and every person switching on a kettle. The modern grid rests on three non-negotiable physical defences: frequency (the beat), voltage (the pressure — too high burns out electronics, too low overheats and stalls motors), and inertia (the buffer, the enormous kinetic energy of spinning turbine rotors, which buys the system several precious seconds of response time after a generator trips).

To hold that frequency line, system operators run automatic generation control (AGC), a closed-loop system monitoring frequency deviation. The instant frequency drops, AGC sends millisecond-scale digital signals to the governor valves on the turbines, commanding them to open and admit more steam — the first line of defence. Above it sits a pyramid of ancillary services: primary frequency response acts within seconds through turbine governors, a localised, decentralised reaction; spinning reserve consists of generators already synchronised to the grid but running below full load, able to ramp within 10 minutes; non-spinning reserve is fast-start plant such as open cycle gas turbines (OCGT), able to reach full power from cold within 30 minutes. This hierarchy ensures that even if a 1 GW nuclear reactor trips, the rest of the system has enough kinetic buffer and dispatchable muscle to prevent a full collapse.

The historical premise behind all this was a twentieth-century grid that was 'all pipes and no tank' — large generators supplied abundant power, but the system held no stored energy at all, so every kilowatt-hour had to be consumed the second it was produced. Large-scale wind and solar are rewriting that premise: battery storage can deliver frequency response in milliseconds, demand-side management converts 'dumb demand' such as EV charging and smart water heaters into dispatchable flexible resources, and virtual power plants (VPPs) aggregate thousands of small flexible loads to act like one large generator. The physical requirement for balance has not changed; what has changed is the toolkit available to achieve it.

The response hierarchy that keeps grid frequency balanced (The Full Spectrum, Chapter 12.5)
LayerResponse timeMechanism
Inertia (buffer)Instantaneous to secondsKinetic energy of spinning turbine rotors slows the frequency decline, buying response time
Primary frequency responseSecondsTurbine governors react automatically to frequency change; local and decentralised
Automatic generation control (AGC)Millisecond-scale commandsClosed-loop monitoring of frequency deviation, signalling governor valves to open
Spinning reserveWithin 10 minutesGenerators already synchronised but below full load ramp up
Non-spinning reserveWithin 30 minutesFast-start plant such as OCGT reaching full power from cold

Frequency is the direct measure of the balance between every spinning turbine blade and every person switching on a kettle. It is an absolute indicator: once it falls, consumption has already exceeded production.

The Full Spectrum, Chapter 12.5 'The World's Largest Machine: Topology and Rhythm'

Sources

  • CAISO, Managing the Duck Curve: 2023 System Update (net load variability and flexible demand)
  • The Full Spectrum: Every Energy Source Explained — A Singapore Perspective, Chapter 12.2
  • The Full Spectrum: Every Energy Source Explained — A Singapore Perspective, Chapter 12.5

This question is covered in depth in The Full Spectrum Every Energy Source Explained — A Singapore Perspective,第十二章 12.5、12.2

Written by Yang Yulong, energy systems architect, Singapore.

Published 2026-07-26 · Last updated 2026-07-26