Mark ở vũ trụ khácVIMark học AI
Mark ở vũ trụ khác · Another Universe Anatomy of a Disaster · Nuclear · Risk management

The Fukushima Daiichi Disaster

15.7 m 11 Mar 2011

11 March 2011 · Fukushima Daiichi Nuclear Power Station, Fukushima Prefecture, Japan

What happeneda tsunami cut all power, and three reactors melted down
Lossabout 150,000 people evacuated; INES level 7 (the highest)
Root causea high tsunami warning was shelved; the regulator was "captured"
Main sourcesJapanese Diet investigation commission (NAIIC), July 2012; IAEA report, 2015
Reread in 1 minute

Essence

A wave higher than the wallThe plant was designed for a tsunami of at most 5.7 m. The real one was about 13–15 m.
The 15.7 m figure from 2008TEPCO itself calculated this wave height three years earlier, then sent it off for "further study".
The regulated held the penThe Diet commission called it "regulatory capture": the operator led the agency meant to supervise it.
Three kinds of deathThe tsunami killed nearly 16,000 people. Evacuation and displacement brought over 2,000 more. Acute radiation sickness: 0.
"Simply unthinkable"IAEA: the whole system believed an accident this size could not happen, so it did not prepare.

Sources: NAIIC (2012), pp. 16 and 20; TEPCO (24 Apr 2012, the 15.7 m calculation); TEPCO (8 Jul 2011, wave about 13 m); World Nuclear Association (wave about 15 m); Japan's National Police Agency (Mar 2026); Reconstruction Agency (data to 31 Dec 2025); IAEA, Director General's report (2015).

Part 1

The question of the case

A major power company had calculated, three years in advance, a wave high enough to swamp its plant. Why did that number stay in a drawer? Because an expensive warning is very easy to call "just one estimate".

2002–2011The warningAn M8 quake forecast along the coast. TEPCO calculates a 15.7 m wave. Nothing more is built.
→
11 Mar 2011Total blackoutAn M9 quake, then a tsunami over the 10 m site grade. Backup generators flooded.
→
12–15 Mar 2011Meltdown and evacuationThree cores melt, three buildings blown apart by hydrogen. The evacuation zone grows from 2 km to 20 km.

Sources: TEPCO (24 Apr 2012); World Nuclear Association (timeline, based on NAIIC).

Part 2

What happened

One chart tells most of the story: the wave had been calculated, but the wall was never raised.

Heights in metres above sea level. Original design tsunami: 3.1 m. After the 2002 reassessment: 5.7 m. Site grade of reactor buildings 1 to 4: 10 m. Tsunami of 11 March 2011: about 13 to 15 m. TEPCO's 2008 trial calculation: 15.7 m. 5 m 10 m 15 m 0 3.1 moriginal design(Chile 1960 wave) 5.7 mreassessed 2002 10 msite grade(units 1–4) ~13–15 mwave, 11 Mar 2011 15.7 mTEPCO's own2008 estimate water over the 10 m grade → floods below
The fifth column is the painful one. The design tsunami, reassessed in 2002, was still only 5.7 m. In 2008 TEPCO ran a trial wave-source model off Fukushima and got 15.7 m, but did not adopt the number. Sources give slightly different heights for the real wave: TEPCO measured about 13 m at its tide gauge; WNA says about 15 m coming ashore.Sources: IAEA, Technical Volume 2 (2015), pp. 5–6, 26, 30 (10 m grade, 3.1 m and 5.7 m design bases); TEPCO (24 Apr 2012, the 15.7 m calculation); TEPCO (8 Jul 2011, ~13 m); World Nuclear Association (~15 m).
11 Mar 2011, 14:46 · M9.0 earthquakeThe three operating reactors (1, 2, 3) shut down automatically. But a shut-down reactor still makes heat and must still be cooled.
About 15:42 · Total blackoutThe tsunami overtopped the site. Almost an hour after the quake the plant lost power: 12 of 13 backup generators failed, and switchgear and batteries flooded.
12 Mar, 15:36 · Explosion, Unit 1 buildingHydrogen produced by overheated fuel leaked into the building, mixed with air and ignited. Not a nuclear explosion.
14 Mar, about 11:00 · Explosion, Unit 3 buildingAn even larger hydrogen explosion.
15 Mar, about 6:00 · Explosion, Unit 4 buildingUnit 4 was not operating at the time of the quake, yet the explosion blew apart the top of the building.
About a month later · Level 7The top of the international INES scale, the same level as Chernobyl. Japan first rated it level 5.

Sources: World Nuclear Association, "Fukushima Daiichi Accident" (quake time, blackout, 12/13 generators, explosions, INES rating); NAIIC (2012), p. 16. WNA is an industry secondary source; these milestones match the IAEA report.

Evacuation rings widening within 24 hours. 20:50 on 11 March: 2 km radius. 21:23: 3 km. 05:44 on 12 March: 10 km. 18:25 on 12 March: 20 km. 15 March: residents at 20 to 30 km told to shelter indoors. About 150,000 people evacuated. 30 km20 km10 km 11 Mar · 20:50evacuate 2 km (prefecture) 11 Mar · 21:23evacuate 3 km (government) 12 Mar · 05:44evacuate 10 km 12 Mar · 18:25evacuate 20 km 15 Mar20–30 km: shelter indoors ~150,000 people evacuated each time the zone grew, residents moved again
Four evacuation orders in under 24 hours. The Diet commission wrote that many residents "had to flee with only the barest necessities and were forced to move multiple times". In hospitals and nursing homes inside the 20 km zone, "60 patients died in March from complications related to the evacuation".Sources: NAIIC (2012), pp. 19 and 38 (150,000 people, quotations, 60 patients); times of the orders: World Nuclear Association, citing NAIIC. The IAEA (2015) gives "more than 100,000".
Three death counts to keep apart. Earthquake and tsunami across Japan: 15,901 dead and 2,519 missing. Disaster-related deaths in Fukushima Prefecture, including those from evacuation and displacement: 2,350. Deaths from acute radiation syndrome: 0. earthquake + tsunamiall of Japan 15,901 dead · 2,519 missing disaster-related deathsFukushima Prefecture 2,350 acute radiationworkers and public 0
Do not add these three numbers together. Nearly 16,000 people died in the earthquake and tsunami, not because of the plant. 2,350 is the count of "disaster-related deaths" in Fukushima: people who died of exhaustion or worsening illness during evacuation and displacement. It includes the wider hardship of the earthquake, but it is more than double Miyagi's (932), although Miyagi was hit far harder by the tsunami. Nobody died of acute radiation sickness.Sources: National Police Agency, figures released 9 Mar 2026 (via Jiji); Reconstruction Agency, disaster-related deaths to 31 Dec 2025; IAEA (2015), p. 16 and UNSCEAR (2013) on acute radiation effects.
Episode 1

What happened

M9.013–15 mtotal blackoutthree kinds of death
Episode 2

Why

a lowered hill15.7 mregulatory capture"Made in Japan"
Episode 3

Work lessons

expensive warningsindependent checkersplanning for the "impossible"
Part 3

Why: five layers of cause

The terrain layer: the plant was lowered towards the sea

site grade · 10 mIAEA, Technical Volume 2, pp. 5–7
Cross-section of the terrain, not to exact scale. The natural ground at the plant site was about 30 to 35 m high. TEPCO dug it down and placed reactor buildings 1 to 4 at 10 m. Seawater pumps sat by the water at about 4 m. The 13 to 15 m tsunami flowed over the 10 m grade. sea natural ground ~30–35 m (dug away) pumps ~4 m units 1–4 wave ~13–15 m 10 m grade
A decision from the 1960s. The IAEA notes that the natural ground was about 30–35 m, but "TEPCO decided to build the plant at OP +10.00 m, i.e. much closer to the sea water level." In its words, "a site that was considered as a dry site in the original design basis became a wet site".Source: IAEA, The Fukushima Daiichi Accident, Technical Volume 2 (2015), pp. 5–7; seawater pumps ~4 m: World Nuclear Association.

The warning layer: the 15.7 m figure stayed in a drawer

ignored estimateTEPCO 2012; NAIIC pp. 16, 27
A staircase of warnings. July 2002: the government's earthquake research body says an M8 quake could occur anywhere along the trench from Sanriku to Boso. 2006: TEPCO and the regulator know a wave reaching site grade would cause a total blackout. 2007: a disaster council decides not to consider an M8 quake off Fukushima. June 2008: a TEPCO team reports a maximum wave of 15.7 m. July 2008: management does not adopt the figure. 2009: TEPCO itself moves the seismic back-check deadline to 2016. 7 March 2011: TEPCO tells the regulator in writing that a wave could exceed 10 m. Jul 2002"M8 anywhere" 2006blackout risk known 2007"not needed" Jun 200815.7 m calculated Jul 2008not adopted 2009deadline → 2016 7 Mar 2011told regulator
Nobody hid the number. TEPCO published the 15.7 m calculation itself in 2012, explaining that "the validity of this wave source model had been in question." The Diet commission wrote that since 2006 the regulators and TEPCO "were aware of the risk that a total outage of electricity" could occur if a wave reached site level, and that the regulator knew TEPCO had prepared no measures. Only four days before the disaster did TEPCO tell the regulator in writing that a wave could exceed 10 m.Sources: TEPCO, 24 Apr 2012 (HERP 2002, the 15.7 m calculation); IAEA, Technical Volume 2, pp. 38–39 (2007 decision); NAIIC (2012), pp. 16 and 27; criminal-trial testimony of 10 Apr 2018 (Asahi) and 19 Oct 2018 (Mainichi), secondary, on June and July 2008 and the request for an expert re-evaluation; the 7 Mar 2011 notice: Asahi (Aug 2011, secondary) quoting the safety agency NISA.

The oversight layer: the supervised led the supervisor

regulatory captureNAIIC pp. 9, 20
Before 2011: the same state bureaucracy both promoted nuclear power and regulated its safety. Lacking expertise, the regulator relied on the operator: regulatory capture. From 2012: Japan set up a new Nuclear Regulation Authority, separating promotion from regulation. Before 2011 promotenuclear power regulatesafety one bureaucracy the operator knows more than the checker From 19 Sep 2012 promote(ministries) NRAregulates alone functions separated
"Its regulation was entrusted to the same government bureaucracy responsible for its promotion." The Diet commission concluded that "the lack of expertise resulted in 'regulatory capture'", and that the regulators' independence was "a mockery". One example: the seismic back-check was due by June 2009; "TEPCO made the decision internally and unilaterally to reschedule the deadline to January 2016", and the regulator did not follow up.Sources: NAIIC (2012), pp. 9, 20 and 27; Nuclear Regulation Authority (NRA) annual report, founded 19 Sep 2012.

The belief layer: "simply unthinkable"

safety mythIAEA 2015, foreword
Two plants on the same coast. Tohoku Electric's Onagawa, closer to the epicentre: buildings at 14.7 m, wave 14.3 m, safe shutdown. TEPCO's Fukushima Daiichi: buildings at 10 m, wave 13.1 m, three meltdowns. 14.7 msite 14.3 mwave Onagawa · safe shutdown 10 msite 13.1 mwave Fukushima Daiichi · 3 meltdowns
Same country, same wave, different outcome. The IAEA Director General wrote that a major factor was "the widespread assumption in Japan that its nuclear power plants were so safe that an accident of this magnitude was simply unthinkable". The assumption was accepted by operators "and was not challenged by regulators or by the government". Onagawa, closer to the epicentre, stood higher than the wave.Sources: IAEA, Director General's report (2015), foreword, as quoted by World Nuclear News (2 Sep 2015); Onagawa figures: A. Ryu and N. Meshkati, Bulletin of the Atomic Scientists (10 Mar 2014), an independent analysis.

Believing it was safe

Diet commission: they "had firmly committed themselves to the idea that nuclear power plants were safe", so preparing for a severe accident looked unnecessary.

An expensive warning

Accepting 15.7 m meant a seawall, flood-proofing and permits. The defendants later called the figure "nothing more than one estimate".

Delay instead of refusal

Nobody said "no". The warning went off for more study and deadlines slipped, until the wave arrived first.

Sources: NAIIC (2012), pp. 17 and 27; "nothing more than one estimate": the defence at the criminal trial, per Asahi (Apr 2018, secondary).

The response layer: once the power went, every script expired

station blackoutNAIIC pp. 16–18

No script for a long blackout

The commission wrote that, given the deficiencies in training and preparation, once the total blackout including DC power occurred, "it was impossible to change the course of events".

A tangled chain of command

The Prime Minister went to the site in person to direct workers. The commission called it a direct intervention that confused the chain of command.

Residents told too late

The commission traced the confusion of the evacuation to the regulators' negligence in emergency preparation.

Sources: NAIIC (2012), pp. 16–19; the blackout quotation via World Nuclear News (5 Jul 2012).

Part 4

Fact check: what is true and what is myth

13 things often said about Fukushima, checked against the primary reports. Click each line for the reasoning.

SOLIDA tsunami above the design basis caused the total blackout

Design basis 5.7 m, site grade 10 m, wave about 13–15 m. IAEA: the tsunami levels far exceeded the hazard assumptions. Diet commission: the total blackout was the point of no return.

SOLIDTEPCO had a 15.7 m tsunami estimate in 2008

TEPCO published the calculation itself (24 Apr 2012), with the defence that the wave-source model was "in question". Court testimony says management decided not to adopt it in July 2008.

SOLIDNobody died of acute radiation sickness

IAEA (2015): no early radiation-induced health effects were observed among workers or the public. Six workers received over 250 mSv, but below levels that cause radiation sickness (WNA).

SOLIDThe evacuation itself killed people

Diet commission: 60 patients from hospitals and nursing homes in the 20 km zone died in March from complications related to the evacuation. Fukushima Prefecture had 2,350 disaster-related deaths by the end of 2025 (Reconstruction Agency), though that figure does not separate out the nuclear cause.

CAUTION"A disaster Made in Japan", caused by Japanese culture

These are the words of the commission's own chairman, Kiyoshi Kurokawa, in the English edition: the root causes were "our reflexive obedience; our reluctance to question authority … our groupism; and our insularity". But according to critics cited by AP, the passage is missing from the Japanese text, and the English edition itself says the Japanese version prevails where they differ. Professor Gerald Curtis (Columbia University) wrote: "To pin the blame on culture is the ultimate cop-out." The IAEA wrote that the contributing factors "were not unique to Japan". And Onagawa, same culture, survived. The body of the report names specific, fixable failures: a captured regulator, slipped deadlines, no preparation for a blackout.

CAUTION"Nobody died from Fukushima radiation"

Nobody died of acute radiation sickness. But in 2018 Japan's health ministry accepted that a worker's lung cancer (about 195 mSv cumulative) was caused by radiation, and his family was to be compensated (NPR). UNSCEAR says any increase in radiation-related cancer is "unlikely to be discernible", meaning too small to measure, which is not the same as certainly zero.

CAUTION"Nobody could have foreseen that wave"

The official sources disagree. Diet commission: the direct causes "were all foreseeable". Tokyo District Court, civil case (2022): the accident "could have been avoided" by making the main buildings watertight. But the criminal courts (2019, 2023, Supreme Court 2025) acquitted three executives, and the Tokyo High Court (June 2025) overturned the civil ruling, finding long-term forecasts were not data that demanded immediate action.

CAUTION"Only the tsunami; the earthquake did no damage"

The Diet commission said Unit 1 may have been damaged by the quake, possibly a small loss of coolant, before the wave hit. According to WNA, the Nuclear Regulation Authority concluded in 2013 that the quake did not damage Unit 1. Unresolved.

CAUTION"Child thyroid cancer soared because of radiation"

Detected cases did rise, but UNSCEAR (2021) attributes that to "ultrasensitive screening procedures" applied to all children, and says radiation-induced thyroid cancer is "not likely to be discernible, in any of the age groups considered, including to children".

CAUTION"Fukushima was as bad as Chernobyl"

Same level 7 on the INES scale. But the estimated release was about 10–15% of Chernobyl's (Japanese agency estimates from 2011, per WNA), and UNSCEAR (2013) found the health risks "far lower than those for Chernobyl".

FALSE"Fukushima radiation killed thousands, tens of thousands"

This confuses the earthquake and tsunami deaths (15,901, National Police Agency) with the nuclear accident. UNSCEAR: no adverse health effects among Fukushima residents have been documented that are directly attributable to radiation exposure.

FALSE"The reactors exploded like a nuclear bomb"

The explosions in buildings 1, 3 and 4 were hydrogen explosions: hydrogen produced by overheated fuel cladding leaked into the buildings, mixed with air and ignited. They blew off roofs and walls; there was no nuclear chain-reaction explosion.

FALSE"The treated-water release poisoned the Pacific"

IAEA (comprehensive report, 4 Jul 2023): TEPCO's approach is "consistent with relevant international safety standards" and the discharges "would have a negligible radiological impact on people and the environment". Concerns about trust and fisheries are real, but "poisoning the ocean" has no measured basis.

Sources: NAIIC (2012), pp. 9, 16–17, 38; IAEA (2015); TEPCO (2012); UNSCEAR, press release of 9 Mar 2021 on the 2020/2021 report; AP (13 Sep 2012) on the "Made in Japan" debate; AFP (Mar 2025) and AP (Jun 2025) on the rulings; NPR (5 Sep 2018); IAEA on treated water (4 Jul 2023), as quoted by Japan's Ministry of the Environment.

Part 5

Assessing the case

Scores are Mark's assessment based on the sources in Part 8, not figures from a study.

Avoidable
4/5
Warnings in advance
5/5
Transparent inquiry
4/5
Lessons retained
3/5
Done right after the disaster
  • The Diet set up its first independent investigation commission, with public hearings
  • A new nuclear regulator, separated from promotion (2012)
  • IAEA and UNSCEAR published international assessments, updated over years
Not done right
  • No executive convicted; the courts disagree on responsibility
  • About 27,000 people still unable to return home (Nov 2025, per WNA)
  • The "culture" explanation blurred the specific failures

Not simply a natural disaster: nature brought the wave, but people decided how high the wall would be.

More: why "avoidable" scores 4/5 and not 5/5

An M9 quake and a 13–15 m wave are genuinely extreme. Nobody can say for sure whether flood-proofing begun in 2008 would have been finished before March 2011. The 2022 civil court found that making the main buildings watertight would have taken about two years and been enough to avoid the accident; the High Court in 2025 disagreed that the executives were obliged to act immediately.

But Onagawa shows that better preparation was possible on the very same coast. So Mark gives 4: largely avoidable, with some uncertainty.

Sources: Tokyo District Court civil ruling, 13 Jul 2022 (Asahi, secondary); Tokyo High Court, 6 Jun 2025 (AP); 27,000 people: World Nuclear Association.

Part 6

Work lessons: don't dismiss a warning because it is expensive

Use this when someone brings a risk that would cost money, time or schedule to address: security, backups, safety, legal.

Four steps when you receive an expensive warning: log it in the risk register with the name of whoever raised it and the date; ask what you lose if it is right, against what the fix costs; put a deadline on further study; make the checker independent of whoever pays. 1 · log itwho raised it, whennot buried in email 2 · what if it's right?loss vs cost of fixnot just "low odds" 3 · set a deadlinefor "further study"and interim action 4 · independent checknot paid by the checkedand able to stop "It's just one estimate" is not an answer.
Each step patches one failure at Fukushima. Nobody owned the 15.7 m figure, "what if it's right" was replaced with "the model is in question", "further study" had no deadline, and the checker depended on the checked.
Compare with the worst credible scenarioThe 5.7 m design was based on one known wave. The right question is "what is the worst case the experts have already written down?"
Backups must not sit with the riskThe backup generators flooded with the same wave. At work: backups on the same server, a plan B that depends on the same person.
Find an "Onagawa" to compare withHow do competitors, the team next door or another industry handle the same risk? The difference is evidence, not "different circumstances".
Count the cost of the response tooA rushed evacuation also kills. Emergency plans must be rehearsed with the most vulnerable people, not only on paper.
Continuous questioning and openness to learning from experience are key to safety culture.Yukiya Amano, IAEA Director General, foreword to the report on Fukushima Daiichi (2015), via World Nuclear News.
Part 7

Further reading

National Diet of Japan commission (NAIIC), English executive summary (2012)Read the Conclusions (from p. 16) to see why it called this a "manmade" disaster.
IAEA, The Fukushima Daiichi Accident: Report by the Director General (2015)The international synthesis, with five technical volumes. Volume 2 has the full history of the tsunami reassessments.
UNSCEAR, 2020/2021 Report, Annex BThe scientific assessment of radiation doses and health effects; read it to separate fact from fear.
Anatomy of a Disaster #10: Space Shuttle Columbia (2003)The series finale: 17 years after Challenger, the same organisation, the same kind of mistake.
Part 8

Sources

Primary documents first. Where only a secondary source was available, it is marked next to the figure.

  1. The National Diet of Japan, NAIIC, The official report of The Fukushima Nuclear Accident Independent Investigation Commission: Executive summary, July 2012. copy at nirs.org
  2. IAEA, The Fukushima Daiichi Accident: Report by the Director General, GC(59)/14, 2015, and Technical Volume 2. iaea.org
  3. TEPCO, explanation of the trial tsunami calculations, 24 Apr 2012; press release 8 Jul 2011. tepco.co.jp
  4. UNSCEAR, 2020/2021 Report, Annex B; UN press release, 9 Mar 2021. unis.unvienna.org
  5. Reconstruction Agency (復興庁), disaster-related deaths, as of 31 Dec 2025. reconstruction.go.jp
  6. National Police Agency, dead and missing, released 9 Mar 2026, via Jiji Press. adnkronos.com
  7. Nuclear Regulation Authority (NRA), annual report (founded 19 Sep 2012). nra.go.jp
  8. A. Ryu and N. Meshkati, "Onagawa: The Japanese nuclear power plant that didn't melt down on 3/11", Bulletin of the Atomic Scientists, 10 Mar 2014. thebulletin.org
  9. World Nuclear Association, "Fukushima Daiichi Accident" (updated Sep 2026), an industry secondary source. world-nuclear.org
  10. Press: World Nuclear News (5 Jul 2012, 2 Sep 2015); AP, "Japan nuke probe head defends report" (13 Sep 2012); Asahi Shimbun (Aug 2011; Apr 2018; Jul 2022); AFP (Mar 2025); AP (6 Jun 2025); NPR (5 Sep 2018).