Grows where the grid hasn’t.
From power plant to battery site
When a coal, gas or oil plant closes, it leaves behind its most valuable part: a high-voltage connection, a substation, land and permits. Batteries and solar can reuse all of it. We mapped 313 thermal plant sites across Africa, worked out when each one frees up, and tested what reuse does for carbon emissions, hour by hour in South Africa.
Feasible, and most of it is in South Africa before 2040
- Coal goes first. Africa runs 52 GW of coal plants with a median start year of 1986 and 122 GW of gas plants with a median of 2013. South Africa's coal stations set the pace: 13 GW free by 2030 on Eskom's official dates, 30 GW by 2040. North Africa's gas fleet mostly frees up between 2045 and 2060.
- On a coal site, build both. Our hourly model of South Africa's grid shows the best use of a freed coal connection is a battery at half the station's size plus solar where land allows. That saves 13 to 14 Mt of CO2 a year under the national plan, and 15 to 19 Mt if wind and solar are built at half its pace. A battery alone only helps once there is spare wind and solar to store; solar alone runs into a full grid at midday.
- Closing saves the most, reuse adds a third or more. Closing these plants avoids about 1,480 Mt of CO2 between 2026 and 2050 compared with keeping them running. Reusing the connections adds 290 to 430 Mt with batteries alone, or 420 to 620 Mt with solar and batteries.
- The carbon is worth more than the build at coal sites. At the US EPA's 2023 social cost of carbon (US$190 a tonne in 2020 dollars), the CO2 saved at coal sites is worth about 2.5 times the cost of the solar and batteries; at gas sites about the same as the cost. At lower carbon values, electricity sales have to cover most of the cost.
- Delay is the main risk. African grids are short of power, so old plants keep getting extended. South Africa's first gas tender produced no preferred bidders in October 2026, and Eskom has said some closures need careful thought. If South Africa's dates slip five years, only 6.5 GW is free by 2030.
Coal now, gas after 2045
Grid connection freed by thermal plants of 50 MW and above (one connection per site), counted from when each unit closes. Central timing uses Eskom's official dates in South Africa (five stations in 2030, Duvha and Matla in 2034), Morocco's coal exit by the 2040s, and otherwise 50 years for coal, 40 for gas and 35 for oil. The line on each bar runs from early timing (shorter lives) to late timing (five years later in South Africa, 60 years for coal and 50 for gas). Plants already mothballed or closed since 2015, such as Komati, count from today.
Where, by 2050 (central timing)
| Country | By 2030 GW | 2031-40 GW | 2041-50 GW | Main fuel |
|---|---|---|---|---|
| South Africa | 13.2 | 16.8 | 10.9 | coal |
| Egypt | 5.1 | 5.1 | 10.6 | gas |
| Algeria | 2.8 | 1.2 | 5.9 | gas |
| Libya | 3.8 | 0.0 | 4.8 | gas, oil |
| Nigeria | 2.8 | 0.6 | 2.0 | gas |
| Morocco | 0.5 | 0.3 | 5.4 | coal |
| Tunisia | 1.5 | 1.1 | 2.2 | gas |
| Ghana | 0.5 | 0.4 | 0.2 | gas |
| Sudan | 0.9 | 0.0 | 0.4 | oil |
| Zimbabwe | 0.2 | 0.9 | 0.0 | coal |
Reading the timing
Of the 32 GW freed by 2030, about 14 GW has an announced closure date (9 GW of it South African coal), 6.5 GW is already mothballed or closed, and 11.5 GW is plant at or past its usual age limit in Nigeria, Libya, Egypt and Algeria that could just as easily keep running.
Another 56 GW of gas plants and 11 GW of coal are being built or planned. Anything built now becomes a site for reuse only after 2060, and locks in emissions until then.
South Africa, hour by hour
A battery stores energy; it does not make any. It cuts emissions only when it charges from wind and solar that would otherwise be thrown away. To see when that happens, we ran South Africa's grid for every hour of 2030, 2035 and 2040, with the wind, solar, batteries and gas in the national plan (IRP 2025), coal stations closing on Eskom's dates, and three ways to reuse their connections:
Why the cases differ
- The grid fills up at midday. With the plan's 34 GW of wind, 25 GW of utility solar and 16 GW on rooftops, 22 TWh of wind and solar is already curtailed in 2035 and 41 TWh in 2040, before any site is converted. Extra solar on the old sites mostly adds to that, so case B falls from 12 Mt a year in 2030 to 8 Mt in 2040.
- A battery soaks up the surplus. Batteries on the old sites cycle 270 to 300 times a year and displace 0.6 to 0.8 tonnes of CO2 for every MWh they deliver in 2035 and 2040.
- If wind and solar lag, a battery stores coal. At half the plan's pace there is little surplus, so case A saves almost nothing. It still keeps the lights on: closing the stations with nothing in their place leaves 18 GWh of demand unserved in 2040.
What this means
B+ is the best case in every year and under both build paces, so it is the choice that does not depend on getting the rest of the grid right. It is also close to what Eskom plans: its 5 GW repowering pipeline puts solar, batteries and some wind and gas on six coal sites, starting with 122 MW of solar and 150 MW of batteries at Komati, closed in 2022.
Closing a station saves less at first than a simple count suggests. In 2030 the remaining coal fleet takes up much of the closed stations' output, so closure saves about 5 Mt that year; by 2040 the saving is 42 Mt as wind, solar and gas replace them.
What reuse is worth
Across Africa, 2026 to 2050, central timing. South African sites use the hourly results; elsewhere case B is a range from South African-style midday curtailment (low) to none (high), because no hourly model exists for those grids yet. The social cost of carbon is the damage one more tonne of CO2 does worldwide, so a tonne avoided in Africa is worth the same as anywhere.
| CO2 avoided, 2026-2050 | Mt |
|---|---|
| Closing the plants (vs keeping them running) | 1,480 |
| Added by reuse, case A: battery only | 290-430 |
| Added by reuse, case B: solar and battery | 420-620 |
| Added by B+, South Africa only | 310 |
| Value of the CO2 saved, US$ billion today | at US$51/t | at US$190/t | Build cost |
|---|---|---|---|
| Case A: battery only | 25-38 | 113-172 | 46 |
| Case B: solar and battery | 32-52 | 139-229 | 135 |
| B+, South Africa only | 21 | 90 | 31 |
Present value in 2026 dollars, counting 20 years of savings per site, discounted at the rate that goes with each social cost (3% for the older US$51, 2% for the EPA's US$190; the EPA's 1.5% value, US$340, roughly doubles the middle column). South Africa's carbon tax is R308 a tonne in 2026 (about US$17), and allowances bring the rate paid on power to about R46 (US$2.60). Build costs: batteries US$190 per kWh in 2026, falling 3% a year to 2035; solar US$900 per kW, in line with recent African plants. Reusing a connection also saves about US$75 per kW of new lines and substations, and years of waiting.
Find a site
313 thermal plant sites with at least 50 MW of grid connection, including plants already mothballed or closed since 2015. Click a column heading to sort. "Frees up" runs from the first to the last unit closing in our central timing; "Late timing" is the same if closures slip.
| Site | Country | Fuel | MW | Frees up | Late timing | Solar kWh/kWp | Case B Mt saved 2026-50 | Value / cost |
|---|
MW is the site's connection: units operating, mothballed or closed since 2015. "Announced" marks a published closure date; "Closed" a site with no unit running. Solar yield from PVGIS for a fixed panel at that spot. Case B and the value ratio use the v0.2 numbers above (US$190 a tonne); sites closing after 2060 show no saving. Download all fields as CSV (CC BY 4.0, credit Aloe Energy and Global Energy Monitor); a blank year there means after 2060.
Why it can work
- The site comes ready-made. Substation, high-voltage lines, land, water and permits are in place. Australia is building its largest batteries on coal sites for this reason: 500 MW at Liddell, Eraring next to a 2,880 MW coal plant, and the Waratah Super Battery on the old Munmorah site.
- Money is there for coal sites. The World Bank lent US$497 million for Komati, and the Climate Investment Funds approved a US$2.6 billion coal transition plan for South Africa in 2025.
- Batteries keep getting cheaper. A complete battery system averaged US$117 per kWh worldwide in 2025, 31% less than a year earlier.
- Grid space is scarce elsewhere. South Africa's Eastern and Western Cape have no connection capacity left, so the coal connections in Mpumalanga are among the few large blocks that will open up.
Why it might not
- Shortage keeps old plants running. Eskom kept five stations past their dates to 2030, Egypt is extending 1,250 MW of gas turbines by 13 to 15 years, and Zimbabwe has applied to license Hwange's first six units for 25 more years.
- A battery is not baseload. A 4-hour battery covers the evening peak, not a coal plant's output day and night. That energy has to come from new wind, solar or other plants somewhere.
- Land and grid strength. Solar needs about 2 hectares per MW, and coal sites lose the spinning machines that steady the grid, which is why Komati's plan includes a synchronous condenser. City gas plants have room for batteries only.
- Contracts and jobs. Some plants are tied to long contracts; Morocco extended Jorf Lasfar's to 2044. Komati closed before its replacement was ready, and the town paid for it. Eskom now plans its repowering ahead of each closure.
Method, sources and updates
How we calculate it
Plants. Coal, gas and oil units from Global Energy Monitor's Global Integrated Power Tracker (September 2026): operating, under construction, mothballed, or closed since 2015. Industry-only captive plants are left out. Units are grouped by site, and sites under 50 MW are dropped.
Timing. Announced dates where they exist: Eskom and NTCSA for South Africa, Morocco's coal pledge, and planned retirement years in the tracker. Otherwise plant age, with three sets of lifetimes. A plant already past its age limit cannot close before 2028, 2030 or 2033 in the three timings.
Emissions. Each unit's output from Ember's 2024 capacity factors by country and fuel, and IPCC fuel factors with typical efficiencies (about 1.03 t of CO2 per MWh for older coal, 0.40 for a combined-cycle gas plant). When a plant closes, the rest of the grid covers its output at the grid's emission factor: Aloe's future grid factors south of the Sahara, the 2024 average in North Africa, falling 2% a year after the last data year.
Hourly check. South Africa's 2019 hourly demand and wind and solar profiles from the Eskom Data Portal, grown 1.5% a year, less rooftop solar; the IRP 2025 build with gas four years late; coal stations at 66% availability and at least 40% load; Koeberg, diesel turbines and pumped storage. A least-cost dispatch for every hour, one month at a time. A run for 2025 gives 175 TWh of coal against 166 TWh in 2023 in CSIR's statistics.
Screening, not planning. No transmission limits inside the grid, one weather year, and costs that leave out running costs, battery replacement and site clean-up.
Sources
Global Energy Monitor, Global Integrated Power Tracker, September 2026 (CC BY 4.0); Ember yearly electricity data (CC BY 4.0); PVGIS 5.3, European Commission Joint Research Centre; Eskom Data Portal hourly series 2018 to 2023, as bundled in the open PyPSA-RSA model; Integrated Resource Plan 2025 (gazetted 28 October 2025); NTCSA Medium-Term System Adequacy Outlook 2027-2036 and Generation Connection Capacity Assessment 2025; Eskom repowering pipeline as reported by Engineering News (2025); World Bank Komati project; US EPA, Report on the Social Cost of Greenhouse Gases (2023); IWG interim estimates (2021); South African Carbon Tax Act rate path; BloombergNEF Energy Storage System Cost Survey 2025, as reported by Energy-Storage.news; Morocco's NDC 3.0 coal pledge (Powering Past Coal Alliance); IPCC 2006 Guidelines; Aloe Energy grid emission factors and power prices studies.
Public data only, indicative, not for system planning or investment decisions. The full source list with links is in the PDF note.
What changed
We rerun this twice a year, after Global Energy Monitor updates its tracker, and whenever South Africa publishes a new resource plan or adequacy outlook.
- November 2026 - First edition: 313 sites from the September 2026 tracker, three retirement timings, and an hourly check of South Africa's grid for 2030, 2035 and 2040.
Quote or embed it
Free to quote and embed with credit (see our data and terms). To embed the page:
Cite it
APA, with the date you viewed it:
BibTeX:
Common questions
Why put batteries where a power plant used to be?
Because the expensive, slow part of a new project is often the grid connection: the substation, the high-voltage line and the permits. A closed plant has all of them, usually with land and water too. Reusing them saves money (we use about US$75 per kW) and, more often, years of waiting for a new connection.
Does a battery cut emissions on its own?
Only when it stores clean energy that would otherwise be wasted. In South Africa that is true once the planned wind and solar are built: a battery on an old coal site then displaces 0.6 to 0.8 tonnes of CO2 per MWh it delivers. If wind and solar lag, the battery mostly stores coal power and saves almost nothing, though it still helps keep the lights on.
Which plants close first?
South Africa's Camden, Grootvlei, Hendrina, Arnot and Kriel, whose emission exemptions run to 31 March 2030, then Duvha and Matla in 2034. Komati closed in 2022 and is already being repowered. Elsewhere, the first sites are mothballed or very old gas and oil plants in Egypt, Libya, Nigeria, Algeria and Sudan.
Is this a forecast?
No. It is a screening with three timings. Closures in Africa often slip because grids are short of power, so the late timing is a real possibility, especially in South Africa.
Can I use the data?
Yes. The site table is free to download under CC BY 4.0, including for commercial work. Please credit Aloe Energy and Global Energy Monitor, whose tracker the plant list comes from.