Electricity since 1920

The generation figures on this page come from Historical electricity data: 1920 to 2025, published by the Department for Energy Security and Net Zero (DESNZ) as part of the Digest of UK Energy Statistics (DUKES) — Accredited Official Statistics. The charts are GB Electricity's tabulation of that series and are not themselves official statistics. The station list is a separate DESNZ database that DESNZ itself calls experimental; it is corrected here from other open sources (below), so the stations chart and the map are not official statistics either. Nothing is fetched live: the page is built from the spreadsheet itself, and rebuilt when a new edition appears.

Published

Thursday 30th July 2026

Next edition

Thursday 29th July 2027

Open to revision

2017 to 2024 — DESNZ may still change these years

Generation charts

the sheet Estimated Historical Generation, in gigawatt hours, 1920–2025

Stations and map

the sheet List of Electricity Generators — 1,291 rows, 1,288 sites once 3 duplicates are left out, compiled for DESNZ's March 2024 article on where generation has been located — and 105 sites from DUKES 5.11 (2026 edition) that it lacks, marked as such: 57 sites opened 2023–2026 (7.0 GW), which a list that ends in 2022 cannot include, and 48 of 5 MW or more that opened earlier and that it omits (2.7 GW)

Area

the United Kingdom, including Northern Ireland — unlike the live pages on this site, which cover Great Britain

Coastline

Natural Earth, public domain — the same outline the interconnectors map uses

Corrected positions

Wikidata (CC0) and DESNZ's Renewable Energy Planning Database, Q2 2026, and DUKES 5.11 (both OGL v3) — 287 sites; see below

Licence

Contains public sector information licensed under the Open Government Licence v3.0

How it is made. From 1996 the figures are DESNZ's measured returns (DUKES table 5.6). Before that they are estimated: DESNZ scaled up the published supply figures to estimate generation, then shared it between fuels by the amount of each fuel burned. The method is set out in the June 2023 Energy Trends article UK electricity capacity and generation by fuel between 1920 and 2020.

What this page changed. Values are rounded to 0.1 GWh. Shares are of the eight fuels added together, not of the published total (see below for why they differ). In the station list, spellings of the same fuel or region are merged (solar and Solar, North west and North West). Grid references are converted to latitude and longitude; 269 that were wrong are replaced from Wikidata, the planning database or DUKES 5.11, 63 with no usable point are placed from DUKES, Wikidata or the planning database, 54 are placed approximately at their town, and 17 that could not be placed are left off the map. Another 26 rows with the wrong fuel, region or years are corrected from DUKES 5.11, DESNZ's planning database, Wikidata and Wikipedia, and 3 duplicate rows are left out — each is listed, with its evidence, on the station register. Nothing else is altered.

What these figures can’t tell you

Most of the century is estimated. Only 30 of the 106 years — 1996 onward — are measured. Before 1987 the figures cover the main generating companies, transport undertakings and industrial hydro and nuclear stations only, so they are not the whole country, and 1987–1989 involve a high degree of estimation.

The measured years don't quite add up. In 10 years, all between 1996–2025, the eight fuels fall short of the published total — at worst by 2.2% in 2020. The estimated decades balance exactly, because they were built to. The tooltip shows both figures whenever they differ by more than half a percent.

Wind, wave, solar and hydro are one band. The series cannot separate them before 1996, so the category that matters most today is the one it cannot split.

Biomass is in “other fuels”, not in the renewables band. That band also holds blast-furnace gas and waste; it was 16.1% of generation in 2025, more than nuclear, and most of it is biomass.

Pumped storage is gross. It counts what the stations generated before subtracting the electricity used to pump the water uphill, so it is not a net contribution.

Coke and breeze is nearly invisible. It never exceeds 1,928 GWh in a year — less than a pixel on the chart — but it is a separate line in the source, so it is kept.

Where the stations are

DESNZ warns about this list. Its own note on the sheet says: “It is an experimental database and therefore may contain errors.”

The grid references are the least reliable part of the file. As published, 184 of the 1,205 sites that have one convert to a point outside their own region — often hundreds of kilometres away, and among them some of the largest stations in the list — and 132 more share their exact point with a site in another region, because coordinates have been copied between rows.

This map corrects 192 of them and confirms 54 where DESNZ put them. Positions come from two open sources: Wikidata for 149 and DESNZ's own Renewable Energy Planning Database for 43. A site is matched by name and a compatible kind of plant, inside its own region unless its name belongs to only one site in the UK; a one-letter slip or a shortened name also needs the capacities to agree, and anything ambiguous is left alone. Every tooltip says where its position came from.

DUKES 5.11 corrects more. DESNZ's own station table, DUKES 5.11, gives a grid reference for most sites operating since 2004. It places 18 sites the list gives no usable point, and moves 62 whose point was more than 3 km from DUKES's — often hundreds, in Scotland, where a wrong point can still fall inside the region. 28 of these 80 are confirmed by Wikidata or the planning database; the other 52 rely on DUKES alone, and say so. Where an independent source backs the list's point instead, it stays.

And more are placed. Of the sites that have no grid reference in the list, 45 are placed by the same matching rules, from Wikidata or the planning database. And every offshore wind farm is checked against the coastline: Galloper and Aberdeen had a point on land — Galloper's in Cambridgeshire — and are moved to a point at sea from the sources. DUKES's own earlier editions, and the planning database, sometimes give an offshore wind farm the place its cable comes ashore. 13 sites had another site's point — the same grid reference, in the same region, so no region check could see it (Roosecote A, in Barrow, had Kearsley's, near Bolton) — and are moved to their own from Wikidata or the planning database. Last, 54 sites that no source places — mostly early municipal stations, closed long before DUKES begins — are put at the town they are named after, from Wikidata: approximate, drawn as hollow rings on the map, and labelled so.

17 still cannot be placed. They are left off the map rather than drawn in the wrong place, and 18 are shown where DESNZ put them and say position doubtful. The station CSV keeps every site, with where its position came from.

Positions are converted, not surveyed. The references are British National Grid (OSGB36), converted with the Ordnance Survey's own formulae; the datum difference of about 100 m is ignored at this scale.

“Operating” means between the commissioning and closure years. A site with no closure year is treated as still open, and some closure years are planned rather than actual. The exception is coal: the series records no coal generation after 2024, so the 3 sites listed as coal with no closure year are not shown after it. Some rows carried the wrong fuel as well as the wrong place; the ones found are corrected, but others may remain, so treat any single dot with care. The list covers major generators only, and its newest commissioning year is 2022; the 105 sites added from DUKES carry no closure year, so every one counts as operating.

To cite it

Department for Energy Security and Net Zero (2026). Historical electricity data: 1920 to 2025. Charted by GB Electricity, gbelectricity.uk.