How Does a Nuclear Power Station Work? A Kettle With Complications

A uranium-235 nucleus struck by a slow neutron becomes unstable and splits, producing two lighter nuclei, energy, and two or three further neutrons. If on average exactly one of those neutrons goes on to split another nucleus, the reaction sustains itself at a constant rate, which is what a reactor at steady power is doing. Natural uranium is only 0.7 percent uranium-235, the rest being uranium-238, which does not sustain a chain reaction, so most reactor fuel is enriched to between three and five percent, far below weapons grade. The neutrons released by fission are too fast to be efficiently captured, so a moderator slows them, usually ordinary water, sometimes heavy water or graphite. Control rods made of a neutron-absorbing material such as boron or cadmium are inserted or withdrawn to adjust how many neutrons continue the chain, which is how power is regulated, and dropping them fully shuts the reaction down within seconds.

Most reactors in commercial operation are variations on a few schemes:

• •Pressurised water reactors, the most common, keep the primary water under enough pressure that it does not boil, and pass its heat through a heat exchanger to a separate secondary loop that makes the steam, so the turbine never sees radioactive water
• •Boiling water reactors let the primary water boil and drive the turbine directly, which is simpler and means the turbine hall is part of the controlled area
• •Heavy water reactors, including the Canadian design, use deuterium oxide as moderator, which absorbs fewer neutrons and allows natural unenriched uranium as fuel
• •Gas-cooled reactors, used extensively in Britain, use graphite as moderator and carbon dioxide or helium as coolant, running at higher temperatures
• •Fast reactors use no moderator and can consume plutonium and long-lived waste, and have proved difficult and expensive to operate reliably
• •Small modular reactors, the current focus of development, aim to reduce cost by building identical units in a factory rather than constructing each one on site

Two features distinguish a reactor from any other heat source. The first is decay heat: even after the chain reaction stops completely, the radioactive fission products continue generating several percent of full power for hours and declining amounts for weeks, so a reactor must be cooled after shutdown or it will damage itself. This is what destroyed three reactors at Fukushima in 2011, where the shutdown worked correctly and the tsunami disabled the backup generators needed to keep the cooling running. The second is that the material is dangerous if released, so the whole design is organised around containment in layers: the fuel is in ceramic pellets, the pellets are in sealed metal rods, the rods are in a steel pressure vessel, and the vessel is inside a reinforced concrete containment building. Modern designs add passive safety, meaning systems that work without power or operator action, using gravity-fed water and natural circulation, which addresses exactly the failure mode Fukushima demonstrated.

Spent fuel is the issue with the least technical difficulty and the most political weight. It is intensely radioactive on removal and is stored under water in pools at the station for several years, which cools it and blocks the radiation, then moved into dry casks of steel and concrete. The volume is small: the entire spent fuel output of a country's civil programme over decades typically occupies an area measured in hectares rather than square kilometres. The problem is duration, since some components remain hazardous for tens of thousands of years, longer than any institution has ever lasted, which is why deep geological disposal is the accepted solution and why almost nobody has built one. Finland is the exception, with a repository at Onkalo excavated into granite and expected to begin operating in the 2020s, and the American attempt at Yucca Mountain was cancelled after decades of work and billions of dollars, for reasons that were political rather than geological. Reprocessing, which extracts usable plutonium and uranium from spent fuel, reduces the volume and creates proliferation concerns, which is why the United States abandoned it and France, Russia and Japan did not.

Nuclear power produces electricity with lifecycle carbon emissions comparable to wind and lower than solar, runs continuously regardless of weather, and uses very little land. Its difficulty is cost, and specifically the cost of construction rather than of operation: recent Western projects have run years late and multiples over budget, with Hinkley Point C in Britain and Vogtle in the United States both exceeding their original estimates by large factors, while South Korea and China have built more cheaply and faster by constructing the same design repeatedly. The industry's own explanation is that the West lost the supply chains and skilled workforce during a thirty-year pause in construction and has been relearning at the customer's expense. The safety record, measured as deaths per unit of electricity generated, is among the best of any source including renewables, a fact that sits uncomfortably with public perception shaped by three accidents, of which Chernobyl in 1986 was by far the most damaging and involved a reactor design with a positive feedback characteristic that no Western regulator would have licensed.

A reactor splits uranium-235 with slow neutrons in a self-sustaining chain reaction, moderated to slow the neutrons and regulated by inserting neutron-absorbing control rods, and the heat boils water to drive an ordinary steam turbine. The engineering is dominated by decay heat, which continues after shutdown and must be cooled, and by layered containment. Waste is small in volume and hazardous for tens of thousands of years, with Finland the only country near a permanent repository, and the technology's main problem is construction cost rather than safety.