UCS has obtained a preliminary analysis by the Nuclear Regulatory Commission (NRC) of a hypothetical severe accident at a nuclear power plant in Pennsylvania very similar to the one at Fukushima Daiichi. The NRC analysis finds that—even assuming early evacuation of the area—the accident could cause nearly 1,000 cancer deaths among the population within 50 miles of the plant, on average. Under unfavorable weather conditions, that number could be much higher.
The October 2010 draft report, which UCS obtained under the Freedom of Information Act (FOIA), contains some of the results of a long-delayed NRC study known as the State of the Art Reactor Consequence Analyses (SOARCA) project. The NRC initiated SOARCA in 2005 to provide “updated and more realistic analyses of severe reactor accidents.”(Links to the report are given at the end of this post.)
In part, the purpose of SOARCA was to show that such accidents would not be as bad as previous NRC studies had indicated. However, the new study shows that, at least for the 50-mile population, the average predicted risk of cancer deaths is only a factor of 3 lower than if one assumes the worst-case radiation release characteristics used in the 1982 study known as “CRAC2.” Given the large uncertainties associated with studies of this kind, a difference of a factor of 3 is not significant, so one can conclude that SOARCA essentially confirms, rather than refutes, the results of past studies.
The SOARCA report analyzes the consequences of an earthquake-induced “station blackout”— the total loss of off-site and on-site alternating current (AC) power—at the Peach Bottom nuclear power station in southeast Pennsylvania. This is exactly what happened at the Fukushima nuclear plant. Moreover, Peach Bottom consists of two General Electric Mark I boiling-water reactors. These reactors, known as BWR/4’s, are similar to Fukushima Daiichi units 2, 3, 4 and 5. Thus the event chosen by the NRC to analyze in detail several years ago was uncannily close to the Fukushima Daiichi accident.
The NRC report evaluates two different station blackout (SBO) scenarios. In one of them, known as a long-term SBO, station batteries, which provide direct current (DC) power, continue to operate until they are exhausted, assumed to occur after 4 hours. The batteries cannot power the main coolant systems, but they can provide enough electricity to maintain the instrumentation systems and operate certain valves, allowing operators to manually operate an auxiliary cooling system (known as the Reactor Core Isolation Cooling System, or RCIC) for some period of time.
If additional sources of electrical power are not brought online before the batteries are exhausted, however, operators lose control of the RCIC system, and eventually the system will fail, the reactor core will overheat, the water in the core will boil away, the nuclear fuel in the core will melt, hydrogen gas will accumulate and cause explosions, and radioactive materials will be released into the environment. This is what occurred at Fukushima Daiichi units 2 and 3.
Another scenario SOARCA analyzes is known as a short-term SBO. In this case, DC power is unavailable, operators are not able to use any auxiliary cooling systems, and coolant loss and core melt occur sooner than in the case of the long-term SBO. The event at Fukushima Daiichi unit 1 was essentially a short-term SBO (although because it was a different design than Peach Bottom, the NRC’s analysis is not directly applicable).
The NRC report also evaluated the possibility that operators might be able to use additional mitigative measures to compensate for the loss of all normal sources of AC and DC power. These measures include emergency diesel generators, diesel-driven firewater pumps, and procedures to manually operate the RCIC system and containment venting system without batteries available to power valves and instrumentation. These measures, often called B.5.b measures, were introduced after the 9/11 attacks raised questions about the ability of nuclear plants to recover from the loss of large areas of the plant due to fires and explosions resulting from a malevolent aircraft attack.
However, information revealed by UCS and by the NRC over the last several months has called into question the feasibility and efficacy of these B.5.b measures. So these cannot be assumed to help.
In the long-term SBO analysis with no mitigative measures assumed, the NRC finds that the water level in the reactor core will drop below the top of the nuclear fuel 9 hours after the total loss of power occurs, and the core will be completely uncovered at 11.6 hours. At 19 hours, the nuclear fuel in the core will have completely melted and dropped to the bottom of the reactor vessel, and 30 minutes later the molten fuel will have melted through the reactor vessel and leaked onto the floor of the containment.
The model predicts that hydrogen generated by the reaction of steam with the overheated zirconium alloy encasing the uranium fuel will cause explosions in the reactor building soon after that. These explosions and other phenomena will cause containment failure and significant radiological releases to the environment after 20 hours. (The model predicts that roughly 2% of the cesium and 4% of the iodine in the reactor core is ultimately released to the environment.)
The NRC study assumed that 99.5% of the population within the 10-mile emergency planning zone around Peach Bottom is evacuated within 5 hours and 15 minutes after notification of the accident, and that 50% took potassium iodide tablets. The study also optimistically (and unrealistically) assumed that 20% of the public within 10 to 20 miles away also evacuate within that same time period, even though, as the report admits, “there is no warning or the notification for the public residing in this area, which is not under an evacuation order.”
These two assumptions mean that in the SOARCA study much of the population within 20 miles of the reactor had evacuated long before any radiological releases occurred at the plant. With these assumptions, the study predicts that the average risk to the public within 50 miles of the plant of contracting a fatal cancer is 70 per million. Combining this with the results of a recent report by the Associated Press that found that the population within 50 miles of the Peach Bottom plant is 5.5 million, the NRC predicts that 385 cancer deaths would occur among this population. Much of the radiation exposure causing these illnesses results from long-term occupation of areas that are contaminated but would still be habitable under Pennsylvania laws.
If the people within 20 miles had not finished evacuating in this time, the number of projected cancer deaths could have been larger, as some of the evacuating people could have been exposed to the plume.
This can be seen from the analysis of the of the short-term SBO, in which radiological releases are predicted to occur after about 8 hours, and are somewhat higher than for the long-term SBO (2.3% of the cesium inventory and 10% of the iodine inventory). However, due to the rapid evacuation assumed, even in this case the consequences to nearby residents are limited. Still, the study finds the average cancer fatality risk among the 50-mile population in this case is 160 per million; for a population of 5.5 million, this corresponds to 880 cancer deaths, with a greater proportion due to exposure during the early phase of the accident than in the case of the long-term SBO.
The report goes on to note that the result it obtains for the short-term SBO, 160 deaths per million, is only about a factor of 3 smaller than the worst-case result obtained by the using the worst-case radiological release characteristics of the 1982 Sandia CRAC2 study, even though the CRAC2 worst-case was for a much more serious accident that resulted in a much larger radiological release occurring much sooner (and before most of the population in the 10-mile zone was able to evacuate).
Given the very large uncertainties associated with these kinds of analyses, a factor of 3 difference is not meaningful. Also, this is not a true apples-to-apples comparison, because SOARCA did not even do its own analysis of the type of event represented by the worse case in CRAC2, which was a large-break loss of coolant accident with early containment failure. If SOARCA had done such an analysis, it would have found more severe consequences than its analysis for short-term SBO, and the difference between the SOARCA and CRAC2 worst case would have been even smaller. Thus the SOARCA study, which was intended to show that the CRAC2 results were unrealistically high, has instead essentially confirmed the CRAC2 results.
It’s important to note that the NRC’s results in the SOARCA study are averaged over a large number of different weather conditions. The actual number of cancer deaths resulting from such an accident could be much greater for unfavorable meteorology. The computer codes used by NRC to conduct these analyses typically show that the worst-case results are 10 times the average value, or even larger.
In fact, when the CRAC2 study was first released, it only provided the average values of the results of the health consequences of accidents at U.S. reactors. However, Congressman Ed Markey of Massachusetts obtained and (together with UCS) released a file containing the worst-case results obtained by the Sandia analyses, which did not appear in the NRC report. The fact that the worst-case results were much higher than the average results caused quite a scandal at the time (see, eg, “Nuclear Study Raises Estimates Of Accident Tolls,” Washington Post, Nov. 1, 1982).
For instance, the CRAC2 average number of latent cancer deaths within 500 miles for Peach Bottom was 2,800, but the peak number turned out to be 37,000—thirteen times greater. A similar factor will likely apply to the SOARCA study’s results, so that one might expect as many as 10,000 cancer deaths within 50 miles from the short-term SBO. But just as in the days of CRAC2, the NRC has not provided the maximum values of its analyses.
In summary, the NRC has just spent more than five years and likely a considerable sum of money to essentially reconfirm the validity of the radiological consequence analyses it has been carrying out since the 1980s. This should be a disappointment to nuclear power advocates who were expecting the study to show that the public health consequences of severe accidents were far less than previous studies had indicated. And in fact, the Fukushima accident has provided a real-world demonstration of how large a radioactive release such accidents can cause.
In responding to the UCS FOIA request, the NRC posted the draft SOARCA report on July 14, 2011. It appears in three parts:
SOARCA draft, part 1
SOARCA draft, part 2
SOARCA draft, part 3
The NYT posted a story about this study today.