Evidence indicates nuclear plant risk is smaller than assumed
Public jeopardy from accidents at nuclear power plants may be significantly smaller than has been assumed. Nuclear safety research conducted in the last five years clearly suggests such a conclusion. ``There is no doubt that the preponderance of evidence -- for almost all reactor types -- now suggests that the risk of severe accidents is much less than has been portrayed in past studies and far, far less than the public perception,'' summarizes Robert Bernero, head of the system engineering division of the Nuclear Regulatory Commission (NRC).
New knowledge is not altering the possibility that a major accident will occur at a nuclear reactor, but it is significantly reducing a quantity called the source term, an estimate of the amount of radioactive material likely to be littered about the environment should such an accident occur.
In recent months an ad-hoc nuclear industry group that has been studying this issue reported that ``the risks and consequences to the public of severe nuclear accidents are significantly below those predicted by previous studies.'' And a press release by the American Nuclear Society announcing a report on this work included the more sweeping assertion that ``an accident in a nuclear power plant that would release enough radiation to endanger the public is virtually impossible. . . . ''
While refusing to confirm such an absolute conclusion, the first independent review of this research, conducted by a committee of the American Physical Society (APS) and released late last week, has largely validated these efforts.
``New calculations indicate that for many of the ways in which a severe nuclear accident might occur,'' the amount of radioactivity that would reach the environment ``is much lower than that predicted 10 years ago . . . ,'' the group's cautiously worded press release began.
``This work is very important, and it should be done,'' acknowledges Steve Sholly of the anti-nuclear Union of Concerned Scientists. He argues, however, that a number of uncertainties remain and that it will require several more years of research before its overall significance comes clear.
In the past, source-term estimates have been very crude. Because of the great uncertainties involved, they have also purposely overstated, for safety's sake, the amount of radioactivity that might escape in an accident. For instance, a 1957 Atomic Energy Commission (AEC) report (WASH-740) estimated that as many as 3,400 people might be killed and 43,000 injured in a major accident. This classified report had a dramatic impact when leaked to the public and was the basis of some dialogue in the anti-nuclear movie, ``The China Syndrome.''
But WASH-740 was based on the highly pessimistic and clearly unrealistic assumption that half of the radioactive inventory in a reactor core would escape in a severe accident. In an effort to provide a more realistic basis for judging the probabilities and consequences of reactor accidents, the AEC commissioned a $2 million, two-year safety study in 1973.
Known as WASH-1400, this borrowed an aerospace method called probability risk assessment (PRA). The previous approach had been to rely on a specific set of hypothetical accidents, comparable to the use of 50- and 100-year floods in civil engineering. But this method was attacked because it failed to provide positive assurance of adequate protection against other types of accidents.
By contrast, PRA is an open-ended approach. It starts with the basic reactor design with its thousands of valves, dozens of pumps, and miles of piping. From this, experts construct thousands of possible accident sequences. Then they try to calculate the likely consequences of these scenarios and identify those that pose the most risk.
WASH-1400 proved highly controversial. It reduced estimates of the consequences of a reactor accident comparable to that considered in WASH-740 to 92 fatalities and 200 injuries. In its executive summary, the study equated the chance of being killed by an accident at a nuclear reactor to that of being struck by a meteorite. This was brandished by nuclear supporters and attacked by critics. Such comparisons were ultimately disavowed by the NRC, so the PRA approach was not widely accepted until after the 1979 Three Mile Island (TMI) accident, when analysts realized that it estimated the incident's odds at a credible one incident per 300 reactor years.
Based on the study's conclusions, more radioactivity would have escaped to the environment at TMI than actually did -- particularly radioactive iodine, which is considered the most harmful of the substances involved. Realization of this touched off the current round of research, which has been going on for five years now.
So far this work has led to reductions in source-term estimates for three basic reasons: The massive, concrete domes over the nuclear reactors have proven stronger than expected; volatile radioactive materials like iodine and cesium combine chemically to form less-mobile compounds than had been thought; and the surfaces in the reactor buildings and the large volumes of water present trap radioactive material more effectively than calculated.
``When you think through the series of physical and chemical processes you have to go through [to release large amounts of radioactivity to the environment], it looks less and less likely that this can happen in reality,'' maintains Richard Vogel, who heads source-term research at the Electrical Power Research Institute.
Along with the nuclear critics, however, the APS committee points out that significant gaps and uncertainties in the research remain.
In some cases, source terms have increased, rather than decreased. Only a limited number of plant designs have been analyzed. Efforts to determine the consequences of dropping large amounts of molten fuel on the reactor's concrete base mat have reached contradictory conclusions.
Research now under way may resolve these and other shortcomings in the current source-term research and should definitely be continued, the APS report urged. -- 30 --