Skip to Main Content
Skip Nav Destination

Appendix: Specific accidents

12 December 2011

Chernobyl, 1986

The accident at Chernobyl, in Ukraine, remains the most catastrophic event of the past 60 years. Its magnitude of 8.0 is a more useful discriminator than International Nuclear Event Scale (INES) level 7.

Kyshtym, 1957

The event at the Soviet Union’s Kyshtym plant of INES level 6 is poorly quantified, with estimates ranging from 74 to 1850 petabecquerels, or magnitudes 6.2 to 7.6. I have used an intermediate figure of 1000 PBq, or magnitude 7.3.

Sellafield, 1955 and 1957

The UK’s Sellafield complex, which includes Windscale and Calder Hall, has long been one of the world’s largest nuclear installations. The frequent occurrence of Sellafield in the incident database[1] may be explained in several ways:

  • 1.  There has been an abnormally meticulous record of incident reporting.
  • 2.  Sellafield is much more accident prone than other nuclear installations.
  • 3.  Nuclear installations in general are underreporting events of level 4 and below.

The truth is probably a combination of all three.

Geoffrey Webb and colleagues classified the magnitude of the 25 March 1955 Sellafield release as INES level 4 and estimated that “up to a few tenths of a [terabecquerel] could have been released to atmosphere”[1]—say 0.1 TBq, multiplied by the equivalence factor of 10 000 and based on the assumption that the “beta/gamma radioactivity” comprised plutonium.

The Sellafield events mostly turn out to have a lower magnitude than the INES level to which they were assigned. The 1957 Windscale fire, for example, is classed at INES level 5 but has M = 4.6. On the other hand, the 14 July 1955 fire released some (unspecified) alpha radioactivity. The 8 December 1955 fire is said to be similar to the 25 March one. So although I have used Webb and colleagues’ estimates of activity released,[1] I have been unable to apply any equivalence factor to their release estimates. That means that the magnitudes of −2.4 and −2.7, respectively, for the two accidents may be too low by orders of magnitude if, for example, uranium was burnt.

Three Mile Island, 1979

Determining the releases from Pennsylvania’s Three Mile Island, an INES level 5 accident, has been contentious. There is convincing evidence that the releases were underreported by a minimum of one order of magnitude. The official Nuclear Regulatory Commission (NRC) figure is 10 megacuries (370 PBq).[2] Joy Thompson, Randall Thompson, and David Bear quote 22 MCi (814 PBq),[3] whereas Arnold Gundersen estimates anywhere between 100 and 1000 times the NRC figure.[4] Gundersen also points out that the sum of the NRC releases yields 36 MCi. I use a conservatively low figure of 100 MCi (3 700 000 TBq). Lastly, epidemiological studies, which are themselves controversial,[5–13] point to a significant epidemic of cancer that is clearly related to Three Mile Island and that would not have occurred if the NRC figure were correct. However, even if we use the incredible NRC figure of 10 MCi, it would yield M = 6.9. The accident clearly belongs in the catastrophic group.

Fukushima Daiichi, 2011

The INES level of the Fukushima Daiichi accident in Japan has been revised upward several times from level 4 to the current level 7. In April 2011 the independent Nuclear Safety Commission estimated 630 000 TBq of atmospheric release,[14] as did the Tokyo Electric Power Company.[15] That is very close to the estimate of 670 000 TBq by Masamichi Chino and colleagues,[16] but in June 2011 the Nuclear and Industrial Safety Agency gave an estimate of 770 000 TBq,[14, 17–18] used herein. That yields magnitude 7.2 on the revised scale. The latest estimate (October 2011) by Andreas Stohl and colleagues[19] is 1 592 000 TBq, yielding a magnitude of 7.5.

Rocky Flats, 1957 and 1969

For the Rocky Flats accidents in Colorado, John Till and colleagues estimate that the median total quantity of Pu released in the 1957 fire was 780 GBq and in the 1969 fire was a range of 0.37 – 2.2 GBq.[20] For the latter I assume 1 GBq. With the radiological equivalence of 10 000 applied,[21] the iodine-131-equivalent releases are 7800 TBq in 1957 and 10 TBq in 1969. Those quantities correspond to INES levels 5 and 3, respectively, and magnitudes 5.2 and 2.3, respectively.

Chelyabinsk, Lake Karachay, 1967

The incident at Chelyabinsk in the Soviet Union was due to the desiccation of the polluted Lake Karachay during a very dry summer. Christopher Winter assigned it INES level 6, based on a claimed release of 190 PBq.[22] Another website estimated 22 TBq.[23] Dmitriy Burmistrov and colleagues estimated about 20 TBq.[24] I have calculated the release of 5600 TBq using the data of Peremyslova and colleagues.[25] That value corresponds to INES level 5 and a magnitude of 5.0.

Seversk (formerly Tomsk-7), 1993

The Seversk accident in Russia is classed as an INES level 4. The atmospheric release of 4.3 TBq of long-lived isotopes[26] has been divided into uranium and plutonium in the same proportion as the solution in the tank that exploded. That yields a total atmospheric release of about 3500 TBq, 131I equivalent, or magnitude 4.8.

Paks, 2003

The releases due to the accident at the Paks nuclear power plant on 10 April 2003 are found in the Hungarian Atomic Energy Commission’s report.[27] It is an INES level-3 event but a NAMS (nuclear accident magnitude scale) magnitude 3.9.

Tokaimura, 1999 (criticality accident)

During the 20-hour duration incident in Japan, some 2500 PBq (2.5 × 1018 fissions) of activity, greater than the amount released at Fukushima, were generated in a mixing tank. The INES guide cites it as an example of a level-4 accident.[21] However, because the fission activity was confined in the tank, there were only local (near-field) effects from the direct neutron, beta, and gamma radiation. The World Nuclear Association states, “While 160 TBq of noble gases and 2 TBq of gaseous iodine were apparently released, little escaped from the building itself.”[28] That assertion directly contradicts the estimate of Chino and colleagues of 8 × 1012 Bq/h (which, summed over 20 hours would yield 160 TBq) for radioactivity released to the atmosphere; they made the estimation by back-calculating from observations taken over a wide region around the site.[29] If their figure is correct, and the release contained 99% noble gases as claimed by the World Nuclear Association, then the total of released 131I may be around 2 TBq—a NAMS magnitude-1.6 event. So although it is classified as an INES level-4 event, it has a very small magnitude M.

Unquantifiable releases

The failure of the NRX reactor at Chalk River in Ottawa, Canada,[30] on 12 December 1952 is classed as an INES level-5 event because of the off-site release. About 10 KCi (370 TBq) of fission products were released into the cooling water and hence off-site; however, there are conflicting accounts of atmospheric release. Peter Jedicke claims there was no atmospheric release,[31] whereas the German Wikipedia page on nuclear accidents asserts that 100 TBq of fission products were released to the atmosphere in addition to 400 TBq into the water. I have omitted that event from the main list until an off-site atmospheric release can be confirmed.

Two accidents of INES level 4 occurred at the Saint Laurent nuclear power plant on the River Loire in France in 1969 and 1980. Between 535 and 740 MBq of 239Pu and 240Pu activity are said to have been deposited in the river sediments as a result of the 1980 accident.[32] Many French-language websites mention that the Institute of Marine Biogeochemistry of the École Normale Supérieure in Paris carried out a study that proves the existence of Pu in the sediments of the River Loire from the reactor site to the estuary and ascribes its presence to one or both of the accidents. The study was written by Jean-Marie Martin and Alain Thomas, but neither a title nor a date of publication are available. The institute is presumably now part of the department of biochemistry and ecology of continental environments, which publishes its research papers online (but with a seemingly incomplete record) back to 1984. The online pollution monitoring program run by the Institut de Radioprotection et de Sûreté Nucléaire shows ongoing 239Pu and 240Pu pollution of the Loire by the Saint Laurent site.[33]

In November 1975 the core of one of the reactors at the Soviet Union’s Leningrad nuclear power plant was partially destroyed. Radioactive gases were vented to the exterior over a period of a month as part of the emergency cleaning.[34] The German Wikipedia page on nuclear accidents states that it was an INES level 4–5 event, but it gives an incorrect date of October 1974. The Greenworld website (Russian, with English translation), in contrast, assigns it to INES level 3. The estimates of releases vary from 0.5 – 55 PBq, but no sources for those figures are available. A paper on the Bellona website cites the same INES and release figures.[35] Greenworld states that there was a release of “uranium fission products (137Cs, 134Cs, 144Ce, 90Sr, etc.) and transuranics (238Pu, 239Pu, 241Am, etc.) into the reactor graphite cladding,” whereas the nuclear accident compilation in the Proposition One website[36] states that the atmospheric release was mainly 131I.The accident was evidently severe, but neither the INES level nor the NAMS magnitude can be estimated to better than an order of magnitude, mainly because of the uncertainty about what fission products and how much of them were released to the atmosphere.

The INES level-4 accident in the A-1 reactor at Jaslovské Bohunice, Slovakia, in 1977 caused it to be shut down and subsequently decommissioned. There was also an accident the previous year. Releases are not stated; however, Jozef Kuruc and L’ubomír Mátel ascribe the presence of 90Sr, 239Pu, 240Pu, and 241Am contamination in soils of the surrounding region as due in part to the A-1 accidents.[37]

The November 1982 accident at Chernobyl, classed as INES level 5, undoubtedly released radioactivity into the atmosphere,[38] but the available individual dose rates cannot be integrated into an overall release estimate.

The November 1983 discharge at Sellafield resulted in around 50 TBq entering the sea and thence as particulate matter onto local beaches, which were temporarily closed.[1] That is an example of discharge into water, followed by transfer of the activity back to land, but it is difficult to quantify in terms of radiological equivalence.

In December 1972 there was a fire and two explosions at the Gulf United Nuclear Corporation fabrication plant near Pawling, New York, where Pu fuel was being manufactured for fast breeder reactors. An undetermined amount of Pu was dispersed off-site,[39] so the event can hardly be less than INES level 4. A NAMS magnitude-4.0 event would be produced by the release of the order of just 10 g of 239Pu and 240Pu to the atmosphere; given that the fire and explosions were serious enough for the plant to be closed down, it is likely that the release could have been one or two orders of magnitude above that weight of Pu. Furthermore, the incidence of chronic myelogenous leukemia (CML) in Pawling is apparently 3 in a town of 5000,[40] when the expected value would be 1 – 2 per 100 000 population. The CML Wikipedia webpage states, “The only well-described risk factor for CML is exposure to ionizing radiation.” So the CML cluster at Pawling suggests that at least one serious release occurred from the plant.


Close Modal

or Create an Account

Close Modal
Close Modal