Spent fuel transportation risk assessment: transportation accident analysis1

Abstract The US Nuclear Regulatory Commission (NRC) has recently completed an updated Spent Fuel Transportation Risk Assessment, NUREG-2125. This assessment considered four types of accidents that could interfere with routine transportation of spent nuclear fuel: those in which the spent fuel cask is not affected, those in which there is loss of lead gamma shielding, those in which radioactive material is released and those that could result in a criticality event. The probability of a particular type of accident is the product of the probability that the vehicle carrying the spent fuel cask will be in an accident and the conditional probability that the accident will be of a certain type. An accident in which the spent fuel cask is not damaged or affected at all is the most probable: 99·95% of vehicle accidents are less severe than the regulatory hypothetical accident, and most accidents that are more severe than this still do not lead to loss of shielding or release, which occur in fewer than one accident in a billion. If a lead shielded cask is involved in one of these impacts, the lead shield can slump, and a small section of the spent fuel in the cask will be shielded only by the steel shells. The resulting external doses are significant but would result in neither acute illness nor death. The collective dose risks are vanishingly small. Consequences and risks of an accidental release of radioactive material are similar, since only very small amounts of material would be released, and only through damaged cask seals. The study also examined the probabilities and risks associated with several possible fire scenarios previously analysed by the NRC, and showed that even such events do not result in significant risks. Inclusion of such events increases the estimated risk by only a small fraction. Another accident type that is of potential concern is one that leads to a criticality event. This study has shown that the combination of factors necessary to produce such an event is so unlikely that the event is not credible.