How South Africa Makes Its Electricity
Three ways of turning something into current, and the honest trade-offs between them. Coal, nuclear and solar, measured on the same terms — because measured on different terms, any of them can be made to look like the obvious answer.
The trick of comparing power stations
It is remarkably easy to win an argument about electricity by choosing which question to ask. Ask about cost per hectare and solar looks bad. Ask about jobs created and coal looks unbeatable. Ask about carbon and nothing beats nuclear. Ask about how long it takes to build and nothing beats solar. Every one of those numbers is true, and every one of them is chosen.
So the only useful comparison is one where the questions are fixed in advance and the basis is identical. Everything below is quoted for one megawatt, run for a year, at the capacity factor that source really achieves in South Africa — not its rated output, which is a marketing number, but what it actually delivers.
Coal: burn rock, boil water, spin a turbine
About 80% of South Africa's electricity, and the reason the air over Mpumalanga is what it is. A coal station mills coal to a talcum-powder fineness, blows it into a furnace, and uses the heat to raise steam at around 540 °C and 165 bar. The steam drives a turbine, the turbine drives a generator.
It eats roughly 3 000 tonnes of coal a year for every megawatt — about ninety truckloads. A station the size of Medupi gets through a full 200-wagon train every few hours. That is why coal stations are built at the mine wherever possible: a mine-mouth station like Kendal, Matimba or Medupi has its fuel delivered by conveyor rather than by rail, and the saving is enormous.
The thermodynamics set everything else. A modern station converts 35–42% of the chemical energy in the coal into electricity; the rest leaves as waste heat, and that waste heat has to go somewhere. Wet-cooled stations evaporate 1.2–2 litres of water for every unit on your bill. Medupi and Matimba are dry-cooled and use about a tenth of that — a deliberate choice, because the Waterberg has coal and almost no water. The price of that choice is 2–3% of the station's output to run the fans, and a further loss on hot days, which is a worsening bet in a warming climate.
CO₂ is the one thing that cannot be filtered. Roughly 1 kg for every unit of electricity — a 2 kW geyser running for an hour is about 2 kg. Scrubbers catch sulphur, bag filters catch ash, but carbon dioxide is not a contaminant in the exhaust; it is the product of burning carbon. There is no version of coal that does not make it.
Nuclear: split atoms, boil water, spin a turbine
Koeberg, north of Cape Town, is the only nuclear power station in Africa: two pressurised-water reactors of about 970 MW each, running since 1984 and now licensed to 2044 and 2045. It has supplied roughly 5% of the country's electricity for forty years and produces the cheapest electricity Eskom makes, per unit, now that it is built.
The fuel is the startling part. About 27 kg of enriched uranium per megawatt per year — you could carry a megawatt-year of it in a rucksack, against 3 000 tonnes of coal for the same electricity. A single fuel pellet weighs about 7 g and holds roughly the energy of a tonne of coal.
Because the primary coolant must stay liquid at 155 bar, the steam is cooler than a coal plant's, and nuclear is therefore less thermally efficient — 33–34% against coal's 37%. It wins on fuel density and loses on capital cost and construction time. Every serious argument about nuclear is, underneath, an argument about discount rates: it is enormously expensive to build, takes a decade or more, and is then very cheap to run for sixty years.
Lifetime emissions come to about 12 g of CO₂ per unit, including mining, enrichment and construction — roughly a hundredth of coal. Koeberg cools with sea water and consumes essentially none; an inland station would need cooling towers and would then use more water per unit than coal does, because it rejects more heat.
The genuine problem is at the far end: spent fuel is 96% uranium, 1% plutonium and 3% fission products, and the long-lived fraction has to be looked after for longer than any institution in human history has lasted.
Solar: no fire, no steam, nothing spinning
A photovoltaic panel is not a heat engine. Light knocks electrons loose and current flows — no combustion, no boiler, no turbine, no moving parts at all. It is bound by the Shockley-Queisser limit rather than by Carnot, which caps a silicon cell around 29% and puts commercial panels at 21–23%.
The Northern Cape has among the best solar resource on Earth — roughly twice the annual yield of Germany for an identical panel. A megawatt needs about two hectares and around 2 000 panels, produces electricity at 35–45 g of CO₂ per unit (all of it from manufacturing, none from operating), and drinks about 0.02 litres per unit for washing dust off. In a water-scarce country that last figure is two orders of magnitude better than any thermal station, and it is systematically underweighted in public argument.
The catch is not efficiency and never was. It is that the output is a fixed function of where the sun is. A panel produces nothing at 8pm, whatever you are willing to pay for it, and the grid has to be balanced anyway. That is a storage question, and storage is priced per kilowatt-hour rather than per kilowatt — which is why a home solar quote and a home backup quote are such different numbers.
Side by side, on the same basis
| Coal | Nuclear | Solar PV | |
|---|---|---|---|
| Fuel per MW-year | ~3 000 t of coal | ~27 kg of enriched uranium | None |
| Capacity factor | 65–75% | 85–92% | 25–32% |
| CO₂ per unit | ~1 000 g | ~12 g | ~40 g |
| Water per unit | 0.1–2 L | ~0 (sea) / 2.5 L (inland) | ~0.02 L |
| Staff per MW | ~0.25 | ~0.7 | ~0.05 |
| Time to build | 6–10 years | 8–15 years | 8–14 months |
| Runs at night | Yes | Yes | Only with storage |
| In South Africa | ~80% of supply | Koeberg, ~5% | 6 GW+ and climbing |
Read down the columns and no source wins outright, which is the honest answer. Coal is dispatchable and filthy. Nuclear is clean, reliable and glacially slow to build. Solar is fast, cheap and absent for half of every day. A grid is an argument about how to combine them, not a choice between them.
Want this explained at a different level? The Generation Information Panel asks these same three sources the same six questions — fuel, people, technologies, materials, CO₂ and water — and lets you choose the depth: Grade 1, around the house, DIY, college or university. The facts do not change between them; only the language does.
Why any of this is in a city-building game
BraaiVille models all three, and models them differently on purpose. A solar farm can be built from nothing on your first day. A coal station cannot be built until you have a coal mine and a road between the two, because coal has to physically arrive. Nuclear needs a school built first and an existing station, because it runs on skilled people and nobody builds one as their first power station.
And every one of them is subject to the rule that catches players out most: a station only counts toward your grid once it is connected to somebody who can use the electricity. Build a solar farm alone in the veld and it supplies nothing, while you pay its upkeep every month. Generating is the easy half. Getting it to people is the half that decides whether it was worth building.