Today Greenpeace issued the 52-page report "Lessons from Fukushima". In it the Japanese nuclear catastrophe is analyzed in detail, and its causes and consequences exposed. The report correctly focuses on a few crucial issues: the lack of accountability for the disastrous consequences of nuclear incidents, the lack of a correct approach to the potential risks involved in the production of nuclear energy, and the failure of proper emergency planning.

The document is very instructive to read. I found appalling the description of the many cover-ups of which TEPCO, the company running the Fukushima-Daiichi nuclear plant, is responsible. These cover-ups, which attempted to avoid public concern over the claimed safety of the plants, have now been disclosed, but of course it is too late. The Japanese have already learned the lesson the hard way on March 11th last year.

Of special interest is reading about the fact that the tsunami danger had been largely predicted -I would say announced: one reads, for instance,

 In its annual report, which have been made public since 2001, the Japan Energy Safety Organization (JNES) had predicted possible damage that a tsunami could cause to Mark 1 nuclear reactors at the Fukushima plant. One report said that if a breakwater expanding up to 13 m above sea level was hit by a 15 m high tsunami, all power sources would be knocked out - including outside electricity and emergency power generators. In such a situation, the repoirt said, cooling functions would be lost and the reactor core would be 100% damaged -a meltdown, in other words. The breakwater at the Fukushima n.1 plant was 5.5m high.

Even more ironic -if one can still have the power of appreciating irony when dealing with such dramatic and recklessly profit-driven risk management- is the following paragraph:

In a unfortunate twist of fate, TEPCO informed NISA [the Nuclear and Industry Safety Agency in Japan] that the Fukushima-Daiichi nuclear power plant could be hit by a tsunami exceeding 10 meters while the plant was only designed to withstand a tsunami of 5.7 meters, just few days before the earthquake and tsunamy triggered the three meltdowns of the Fukushima-Daiichi nuclear station. After the accident, it was revelaed that the warning came from a in-house TEPCO 2008 study that company officials had dismissed as "unrealistic".

In the end, one is bound to ask whether we can still talk of nuclear safety, or rather of nuclear risk. The report takes a clear stand in favour of the latter, and justifies this by plain math:

By 2011 the world had accumulated just over 14,000 years of reactor operating experience. The International Atomic Energy Agency (IAEA) safety guidelines state that the frequency of actual core damage should be less than one in 100,000 years. Hence, with more than 400 reactors operating worldwide, a significant reactor accident would be expected to occur approximately once every 250 years.

But, as the report clarifies, in those 14,000 years we have witnessed five core accidents: the Three-Mile Island one in 1979, the Chernobyl disaster in 1986, and the three reactors in Fukushima in 2011. Even considering the latter as a single accident, one is left to compare an observed rate of 3/14000, or one disaster per 4700 year per plant (equivalent to about one disaster per 12 years in the current situation) to the prediceted one-per-250 years. So the real question to me is, when is the next disaster going to happen ? Probably not far in the future. There is a 10% chance that it will occur in the next year or so -and it might be the plant next door.

Another way to look at it is with Poisson statistics. If one were to take at face value the core damage rate of 1-in-100,000 years IAEA standard, allegedly fulfilled by modern-day plants, one might compute the probability of getting three or more accidents in 14,000 years of operation by simple math: the predicted rate is 14000/100000=0.14 events, while the actual rate is 3 events. The probability that a rate of 0.14 produces 0, 1, or 2 events is then

P(0-2;μ=0.14) =
= P(0)+P(1)+P(2) =
= exp(-0.14) + 0.14*exp(-0.14) + 0.14^2*exp(-0.14)/2

which equals 0.999588: so the probability of getting three or more core incidents is 1-P=0.000412. In other words, the probability that the nuclear plants work by the stated standards of the IAEA is less than half a permille. The "null hypothesis" that the IAEA rate is what exists in the 400 operating plants is discarded by any of the typical test sizes -be it 0.10, 0.05, or even three sigma. Food for thought.