How does the power grid balance supply and demand every second?

Grid operators keep supply and demand matched by watching one signal: the frequency of the alternating current. It is 50 hertz in the UK and 60 hertz in North America. If demand outruns supply, frequency falls. If supply outruns demand, it rises. Operators forecast demand, hold backup capacity ready, and change generator output in response, with some of this done automatically and some by instruction. The sources here come from the UK, the United States and Ontario, so details differ by region but the principle is shared. [1] [2]

In Great Britain, the Electricity National Control Centre run by the National Energy System Operator (NESO) forecasts demand close to real time, minute by minute. Its models use weather patterns such as sunshine and wind to predict how much electricity will be needed, where and when. The forecast is set against the generation available, and the control room then monitors frequency in real time. According to [neso.energy](https://www.neso.energy/what-we-do/systems-operations/what-does-electricity-national-control-centre-do), if demand rises, operators ask generators to ramp up or frequency falls, and if demand drops they ask for reductions or frequency rises. [1]

Sudden losses are handled with reserve. If a generator or interconnector trips, generation disappears immediately and frequency drops, so the operator activates contingency or backup generation to stabilise it. [1]

The US Department of Energy, citing the North American Electric Reliability Corporation, describes three essential reliability services: frequency response, balancing, and voltage control. Frequency response combines the natural physical reaction of other generators, controls built into generators and loads, and operator control. These responses unfold over hundredths of a second, seconds and minutes. Balancing services are the longer-term adjustments, made over several minutes to hours, including plans to cover planned outages of large plants such as nuclear units being refuelled. Voltage control is a separate task, managed by adjusting reactive power. See [energy.gov](https://www.energy.gov/cmei/articles/keeping-lights-essential-reliability-services). [2]

Ontario's system operator, the IESO, describes an even finer layer. It dispatches in 5-minute intervals, but supply and demand cannot be predicted exactly, so imbalances arise second by second. Generators with automatic generation control receive a regulation signal by telemetry and adjust output up or down, as often as every 4 seconds. The IESO tracks a measure called Area Control Error to see how well Ontario is balanced. Large or prolonged frequency deviations can cause instability and potentially blackouts. See [ieso.ca](https://www.ieso.ca/-/media/Files/IESO/Document-Library/training/QT-Regulation-Services.pdf). [3]

NESO's main tool is the Balancing Mechanism, a continuously open online auction with 30-minute trading periods and thousands of trades daily. Participants submit bids (prices to consume more or generate less) and offers (prices to consume less or generate more). The window opens 60 to 90 minutes before real time and closes at gate closure. NESO then checks what each participant is physically able to deliver and generally accepts the most competitively priced bids and offers, though operational and locational factors can also matter. [4]

Accepted bids and offers become Bid Offer Acceptances, which are instructions to change output. Participants must act on them, and the balancing team reviews the data and issues instructions around the clock. The Balancing Mechanism sits alongside other services such as dynamic response, reserve and short term operating reserve. [4] [1]