Reactor Accident Risks in Perspective

Abstract

According to the PRAs, reactor accidents may be expected to cause about 4 premature deaths per year in the United States. By comparison, coal burning air pollution kills thousands of times that number, and oil and gas, through air pollution, explosions, fires, and asphyxiation, kill hundreds of times that number. It is sometimes argued that deaths from air pollution are less important because they are undetectable, but the same is true for the cancer deaths from nuclear accidents. If we are interested only in detectable deaths, the reactor accident toll is only a few deaths per century. The worst accident considered in the Reactor Safety Study, expected once in 10 million years, causes 3500 detectable deaths, a number already equalled by an air pollution episode from coal burning (London-1 952). Since large consequence nuclear accidents are hypothetical (in LWRs), they should really be compared with hypothetical accidents from other energy sources. There are many examples of these that are very much worse than the worst reactor accidents that have been discussed. The biggest problem in obtaining public understanding of reactor accident risks is that it is irrational to treat risk questions without considering probabilities, and the public does not understand probability. One solution is to adopt a cut-off probability, such as not considering events expected less than once in 500 years. Buildings, bridges, dams, etc are designed to withstand a once-in-500-years earthquake, tornado, flood, etc, and smaller probability events are not even considered. My preferred approach is to convert all risks to loss of life expectancy, LLE. The LLE for U.S. energy sources is: reactor accidents- 0.01 days (= 15 minutes), coal - 13 days, oil - 5 days, gas - 3 days. But by far the most dangerous energy strategy is energy conservation, with an LLE well over 100 days. If over-zealous energy conservation substantially reduces our wealth, the consequences could be very much greater. If one believes that effects of radiation are uncertain, it is useful to compare reactor accident risks with other radiation problems. Radon in homes provides a 3000 times larger radiation risk, but arouses far less concern. Other risks that give us an LLE of 15 minutes are listed, such as an overweight person increasing his weight by 0.005 ounces, or driving an extra half mile per year. A bar diagram is provided comparing a wide variety of risks in terms of the LLE they cause. It demonstrates that the nuclear risk is truly trivial.