The Fukushima Daiichi accident, step by step

Fukushima Daiichi had six boiling water reactors commissioned between 1971 and 1979. Units 1 to 5 used the Mark I containment: a steel, light-bulb-shaped primary containment around the vessel, connected to a water-filled torus — the suppression pool — that condenses steam if the circuit breaks. All of it inside a concrete building whose top floor is light, meant for moving equipment, not for holding pressure.

The original ground, some thirty-five metres above sea level, had been cut down to ten metres to found the buildings on rock and shorten the lift for the cooling-water pumps. The design-basis tsunami height was 3.1 metres, revised to 5.7 metres in 2002. The emergency diesel generators, their switchgear and the batteries sat in the basements of the turbine buildings.

The magnitude 9.0 Tōhoku earthquake struck the coast at 14:46 on 11 March 2011. Units 1, 2 and 3 were operating and tripped automatically: the rods went in, fission stopped. Units 4, 5 and 6 were down for refuelling or maintenance, with unit 4's core fully unloaded into its pool.

Stopping fission does not stop the reactor. Fission products keep decaying and produce residual heat: around 6% of rated power in the first second, still close to 1% an hour later. A 1,380 MWt reactor is still making more than ten megawatts of heat an hour after shutdown, and that heat has to be removed for days.

The quake brought down the transmission lines and the plant lost offsite power. The thirteen emergency diesel generators started, the emergency cooling systems came in and everything worked as designed. For forty minutes, Fukushima Daiichi was a plant that had survived an earthquake larger than its design basis.

The first wave arrived at 15:27 and did no damage. The second, minutes later, was of the order of fifteen metres: five metres above the terrace the reactors stand on, and nearly three times the revised design-basis height. Water came in through the doors and ducts of the turbine buildings and filled the basements.

Twelve of the thirteen diesel generators were knocked out. With them went the electrical switchgear, the seawater pumps that dump heat to the ocean, and much of the battery capacity. By 15:37 units 1 and 2 had no AC power at all: station blackout, the condition safety analyses treat as the limiting scenario.

The contrast sits twelve kilometres away: Fukushima Daini, with four units, took smaller waves, kept one offsite line and brought all four reactors to cold shutdown. Onagawa, closer still to the epicentre, is built on a terrace of nearly fifteen metres and rode it out. Elevation, flood protection, available electrical supplies and heat removal capability were decisive.

A boiling water reactor has systems meant for exactly this: they run on the steam the reactor itself is still making. Unit 1 had isolation condensers, which condense steam against a water tank and return the condensate by gravity. Units 2 and 3 had RCIC and HPCI, steam-driven turbine pumps that inject water into the vessel.

These systems do not depend on large electrically driven pumps to move water, but their valves and instrumentation require power and control. Units 1 and 2 lost DC power; unit 3 initially retained it. On unit 1, valve isolation and missing indications made it difficult to establish whether the condenser was removing heat. On unit 2, RCIC continued operating for roughly seventy hours despite the loss of DC power: the sequence was not identical at all three units.

Unit 1 lost the water above its fuel that same afternoon. Units 2 and 3 held out longer, with different turbine-pump operating sequences: unit 3 lost cooling in the early hours of the 13th and unit 2 on the 14th. In all three, once the water fell below the top of the fuel, what came next was the same.

The cladding that holds the fuel is a zirconium alloy. Above about 1,200 °C, zirconium strips the oxygen out of steam: it oxidises and releases hydrogen. The reaction is strongly exothermic, so as well as making gas it accelerates the heating of the core that produced it.

In all three units the fuel overheated, broke up and ended up melting to the bottom of the vessel; in unit 1 there is evidence that much of it went through the vessel and reached the concrete of the primary containment. Every kilo of oxidised zirconium leaves behind tens of grams of hydrogen, and that hydrogen, together with steam, pushed containment pressure far above its design value.

Saving the containment meant venting it, and venting demands valves that need power and compressed air, and a decision taken in the knowledge that radioactive material is being released to the atmosphere. The vents were late and hard-won. The hydrogen that left the containment did not all go outside: it collected under the light roofs of the reactor buildings.

At 15:36 on 12 March the top floor of unit 1's building exploded. The images of that gutted building went around the world and are, for most people, the Fukushima accident. At 11:01 on the 14th unit 3's building went, more violently. In the early hours of the 15th unit 4's building exploded, with its reactor empty: the hydrogen had reached it through a vent line shared with unit 3.

The explosions occurred in the reactor buildings, outside primary containment, and wrecked their upper structures; injured workers, scattered highly active debris across the site and for days made it harder to connect a pump or a cable. Unit 2, where the upper floor did not explode, produced the largest release instead: its containment lost integrity another way.

Estimates of the atmospheric release run between 570 and 630 PBq iodine-131 equivalent — the first official figure, in June 2011, was 770 PBq — around 15% of what Chernobyl released. Much of it went out over the Pacific with the prevailing wind. The accident was first rated level 5 and a month later raised to level 7 for units 1 to 3 together.

Evacuation grew in rings: two kilometres on the night of the 11th, then three, ten and twenty kilometres at 18:25 on the 12th. More than a hundred thousand people left their homes. No acute radiation syndrome was recorded. UNSCEAR 2020/2021 documented no adverse effects in residents directly attributable to radiation exposure and does not expect statistically discernible cancer increases. This does not establish that every individual risk was exactly zero.

Evacuation and prolonged displacement caused serious health and social harm: transfers of vulnerable patients, interrupted treatment and loss of community ties. Disaster-related death records are not a count of radiation deaths, nor can every case be assigned to a single cause. Protecting people requires accounting for both exposure and the risks of emergency protective measures.

The four damaged units were written off for good and units 5 and 6 were closed as well. Spent fuel was removed from unit 4's pool in December 2014 and from unit 3 in February 2021. The first samples of melted fuel debris were taken from unit 2 in late 2024, with a telescopic arm and in quantities of grams: large-scale retrieval is aimed at the first half of the 2030s and full dismantling at thirty to forty years.

The water still used to cool the fuel debris, plus what seeps in, is treated in the ALPS system and, since August 2023, discharged diluted to the sea under IAEA review. Off site, the accident changed regulation across half the world: mobile and protected backup power, filtered containment vents, hydrogen recombiners, and an obligation to review external hazards — tsunami, flooding, seismic — against data rather than habit.