Quantitative Safety Goals

Abstract

An extensive heritage of quantitative safety knowledge now makes it feasible to consider quantitative safety goals for nuclear power plants. Comparative risk is the foundation for some of the deliberations, but most of the discussion is based on knowledge of probabilistic risk assessment (PRA) studies. The licensing notion of design basis accidents is retained for those families of events having mean frequencies above 10-4/reactor-year with PRA being used to screen the events. The consequence limit discussed, for individual risk, is that of no identifiable public injury for design basis accidents. This approach yields a dose limit of 5 rem at 10-4/reactor-year. A limit line is derived for the individual risk having a slope of -1.3. Societal risks are measured by latent cancers which could result from accidents. For safety policies which allow balancing of risks, more latent cancers can be tolerated because of the large and variable spontaneous cancer rate from non-nuclear causes. This may become a suitable and feasible policy for the present generation of light water reactors. If, on the other hand, safety policies were to emphasize nuclear risk in an attempt to lower it, a much smaller number of latent cancers might become an appropriate goal. A future generation of nuclear power plants might make such an approach more feasible. Risks from damage to public property are considered on a first cut basis.