The tsunami that caused the Fukushima accident could not have been predicted

The 2011 Tohoku earthquake was of exceptional intensity, and the tsunami far exceeded the protections of the Fukushima plant. Nevertheless, new methods of assessing the seismic hazard had identified the possibility of such a tsunami. Unfortunately it was too late to take the necessary measures and avoid the accident.

This article is almost entirely from a twitter thread published by Mickael Mangeon (@mangeon4).

History of the power plant construction

After the atomic bombings of Hiroshima and Nagazaki (1945) and the end of the US occupation of Japan (1952), the Japanese embarked on civilian nuclear power as part of the US "Atoms for peace" program launched by President Eisenhower at the United Nations in 1953.

Following the development of research reactors, the first nuclear reactor to produce electricity was the Tokai-I reactor of English design (159 MWe magnox reactor - run on natural uranium with graphite moderator and carbon dioxide gas coolant), commissioned in 1966. In the mid-1960s, the Japanese quickly turned to US models: Pressurized water from Westinghouse at Mihama and Boiling water from General Electric at Tsuruga and Fukushima Daiichi.

In 1966 the Fukushima Daiichi site was chosen. It is located on the Pacific Ocean, on the east coast of Honshū Island (Japan's main island), about 250 kilometers north of Tokyo. Construction work began in 1967.

The waterfront consists of a cliff more than 30 meters high. To facilitate the pumping of cooling water from the ocean and improve the performance of the project, the Japanese operator, Tokyo Electric Power Company (TEPCO), decided to excavate the 20-meter cliff.

The plant is therefore located 10 meters above sea level for the first four nuclear reactors, and 13 meters for the last two reactors built a few years later. Some buildings and equipment are also located at lower levels.

Studies and assessment of the seismic hazard

The threat of an earthquake and tsunami was considered by Japanese engineers and experts from the design of the plant. The evaluation of the seismic hazard was based on a historical approach from a catalog of 500 years of data. The postulated scenario was an earthquake of magnitude 7, occurring a few tens of kilometers from the site.

According to Japanese experts, this earthquake was not likely to produce a significant tsunami.

However, at the design stage, a water level of 3.122 m above sea level was used as a reference level to protect the plant (by a dike, in particular). This very precise value of 3.122 m corresponds to the highest water level recorded in the region, in the port of Onahama near the site, on May 24, 1960. It was caused by a tsunami that crossed the Pacific Ocean after the Valdivia earthquake (Chile, magnitude 9.5).

This level was used as a reference value for the plant. The reactors, located on the platform at +10 m, are largely sheltered while the other buildings and equipment are also built, at least, above this reference level.

This figure of 3,122 m was based on a historical reality: in the "recent" history of Japan (500 years), the Fukushima region had not experienced any significant tsunamis! They are very common in Japan, but in an area much further north.

Caption: Map of historical tsunamis recorded along the coast of Tohoku between 800 AD and 1965.

Traces (oral history, stone markers, sediments, etc.) of ancient tsunamis do exist, however. Later, it would be understood that a major tsunami had already taken place in this region following the Jogan earthquake in 869, but without clear information on the impact in Fukushima.

After the terrible earthquake in Kobe in 1995 (magnitude 7, killing 6,437 people and injuring 43,792), the Japanese authorities launched a redesign of seismic risk prevention and created the Earthquake Research Committee of the Headquarters for Earthquake Research Promotion.

As part of this program, a new seismic hazard map for Japan was produced, using a new methodology at the forefront of scientific knowledge.

Using the in-depth knowledge and instrumentation of faults, Japanese experts were able to accurately measure/calculate the energy loading of the different portions of faults and define a characteristic earthquake for each portion (of reduced size).

With this new methodology, the reference earthquake for the Fukushima region was re-evaluated to a magnitude of 7.4, and 7.9 for the Fukushima plant in application of a new guide published by the Japanese safety authority (Nuclear Safety Commission) in 2006.

Following the redesign of seismic hazard prevention, tsunami hazard was also re-evaluated for all of Japan based on new standardized methods from the Japan Society of Civil Engineers (JSCE) in 2002, leading to new studies.

The first study assessed the maximum water level at 5.7 meters and the second in 2009 estimated a level of 6.1 meters. This value was then retained by TEPCO and elevation work (in particular of the pumping station) was carried out.

The JSCE method was then the standard. It would be used for all Japanese nuclear power plants until the time of the accident, on plants near Fukushima Daiichi, for example: Fukushima Dani, Tokai, or Onagawa.

At the same time, 3 exploratory methods were tested and provided much higher values: one study showed water levels of up to 15.7 m at some parts of the site!

To reach this water level, this study conservatively estimated that an earthquake of magnitude 8.2 could occur anywhere on the subduction fault off the island of Honshu, regardless of historical data and characteristic earthquakes.

Another probabilistic study by TEPCO indicated that a 10 m tsunami was possible, but very unlikely (one chance in 100,000 to 1,000,000 years). Finally, the integration of the Jogan tsunami (year 869) into a calculation, gave a result of 9.2 m.

Compared to previous estimates (3.122 m, then 6.1 m), these new values seem incredible. Considering these new methods which were not yet stabilized, in August 2010, TEPCO set up a working group dedicated to the re-evaluation of the tsunami hazard. Too late!

On March 11, 2011, following a magnitude 9.1 earthquake, waves of 11.5 to 15 m crossed the dikes of the Fukushima Daiichi power plant. Deprived of cooling capacities, reactors 1, 2, and 3 melted down and lead to a major nuclear accident.

Management of a nuclear accident therefore compounded the awful effects of the earthquake and tsunami: 15,897 dead, 2,534 missing, and 6,152 injured. We will not discuss the sequence of events or consequences of the accident here, as they have been discussed at length elsewhere.

Conclusion

While the water level on the site was close to the latest exploratory estimates, the overall scenario of March 11, 2011 had not been foreseen by any Japanese specialists, as explained by the director of the Fukushima plant during a hearing.

Instead of seismic scenarios for a fault portion of a few tens of kilometers, there were 500 km of fault that collapsed that day. After the fact, scientific studies and the IAEA noted several shortcomings in the assessment of hazards performed by the Japanese:

• A significant portion (80%) of the energy loaded on some fault portions over the past century had not been released by earthquakes. Japanese experts considered that it must have dissipated by slip, without earthquake.
• Until recently, ancient events had not been taken into consideration even if they were known about, but sometimes data was missing (the Jogan earthquake of 869, for example).
• An excess of confidence in a scientific paradigm, that of seismic rhythms or characteristic earthquakes. This paradigm was dominant internationally at the time, but other paradigms existed and unforeseen scenarios had already occurred: for example, the Sumatra earthquake in 2004 (video) that caused a tsunami (at least 250,000 deaths) is the typical example of an unforeseen mega-earthquake in the region for reasons similar to the Japanese case.

Obviously, this is not meant to be a “cheap” critique of Japanese experts and scientists. However, this history does teach us that the methodological foundations of risk assessment should never be taken for granted.

This is the difficulty of building and maintaining high-risk structures using advanced scientific knowledge that is constantly growing and changing. Fukushima made errors in assessments visible, but they would probably have gone unnoticed if not for the accident.