The accident at Fukushima Daiichi, Japan, occurred on 11 March 2011. This nuclear disaster, the third on such a scale, left a lasting mark in the minds of hundreds of millions of people. Much as Three Mile Island or Chernobyl, yet another place will be permanently associated with a nuclear power plant which went out of control. Fukushima Daiichi revived the issue of the hazards of civil nuclear power, stirring up all the associated passion and emotion.
The whole of this paper is devoted to the risk of a major nuclear accident. By this we mean a failure initiating core meltdown, a situation in which the fuel rods melt and mix with the metal in their cladding. Such accidents are classified as at least level 5 on the International Nuclear Event Scale. The Three Mile Island accident, which occurred in 1979 in the United States, reached this level of severity. The explosion of reactor 4 at the Chernobyl plant in Ukraine in 1986 and the recent accident in Japan were classified as class 7, the highest grade on this logarithmic scale. The main difference between the top two levels and level 5 relates to a significant or major release of radioactive material to the environment. In the event of a level-5 accident, damage is restricted to the inside of the plant, whereas, in the case of level-7 accidents, huge areas of land, above or below the surface, and/or sea may be contaminated.
Before the meltdown of reactors 1, 2 and 3 at Fukushima Daiichi, eight major accidents affecting nuclear power plants had occurred worldwide4. This is a high figure compared with the one calculated by the experts. Observations in the field do not appear to fit the results of the probabilistic models of nuclear accidents produced since the 1970s. Oddly enough the number of major accidents is closer to the risk as perceived by the general public. In general we tend to overestimate any risk relating to rare, fearsome accidents.
What are we to make of this divergence? How are we to reconcile observations of the real world, the objective probability of an accident and the subjective assessment of risks? Did the experts err on the side of optimism? How should risk and its perception be measured?