Supplementary Information/Analysis to MEPC39/INF.15:
"The Sea Transport of Vitrified High-Level Wastes:
Unresolved Safety Issues"
Edwin S. Lyman, PhD
Nuclear Control Institute
Submitted to the 40th Session of the Marine Environment Protection Committee
of the International Maritime Organization
July 11, 1997
In December 1996, the Nuclear Control Institute (NCI) released a report entitled "The Sea Transport of Vitrified High-Level Radioactive Wastes (VHLW): Unresolved Safety Issues." (NCI ). The report analyzed the potential consequences of a severe accident in coastal waters that resulted in damage to the transport ship and the VHLW shipping cask, causing the cargo to sink to a depth of 200 meters (typical of the maximum depth of the continental shelf).
The report's chief finding was that such an event could result in significant health consequences to consumers of marine products from the region. This contradicts the claims of the companies in charge of shipping the waste, as well as the results of a highly publicized study by the Central Research Institute of the Electric Power Industry of Japan (CRIEPI), which found that radiation doses to humans would be extremely small. NCI (1996) showed that the CRIEPI results were not consistent with other studies that analyzed similar accidents and found that doses could be hundreds of thousands of times larger.
NCI (1996) was submitted to the Marine Environment Protection Committee (MEPC) of the International Maritime Organization (IMO) by Greenpeace International. The International Atomic Energy Agency (IAEA) submitted a highly critical commentary [IAEA (1997)] on the report to the Maritime Safety Committee (MSC) of the IMO. The IAEA conceded that NCI (1996) "...in general does a mathematically correct analysis of the conditions it analyzes," but it called the report "fundamentally flawed and ... easily misleading" because (a) it considers a highly improbable scenario but does not attempt to calculate the probability of the scenario; and (b) it does not "establish the sequencing of events necessary to lead to the conditions that are analysed."
A detailed review of IAEA (1997), prepared by NCI, finds that the IAEA report provides little specific information to support its conclusions, and contains numerous misleading statements as well as several outright errors. Space limitations restrict what can be included here; the interested reader is referred to the longer paper for additional details.1
Overall, the IAEA's critique of NCI (1996) can be characterized as "muddying the waters": instead of trying to increase understanding and reduce uncertainties, IAEA (1997) overcomplicates the issues at hand without attempting to resolve them.
However, the IAEA does not dispute the fundamental conclusions of NCI (1996); namely, that the consequences of the specified accident could be severe. In fact, an IAEA Chairman's Report which became public after NCI (1996) was released acknowledged that "if a large irradiated fuel package were to be lost on the continental shelf, some large exposures could result"2 (a comment which applies equally well to a VHLW package). Yet the IAEA remains conspicuously silent about the patently false claims circulated by British Nuclear Fuels Limited (BNFL) that even if VHLW became "directly exposed to the sea ... the effect of such a scenario would be negligible."
Since the IAEA admits that severe accidents can result in large exposures, it is incumbent on them to demonstrate that the probabilities of such accidents are acceptably small. However, the IAEA has not done this convincingly or comprehensively. Little quantitative information exists on the frequency of sea accidents that could create environments more severe than those simulated by Type B tests, or on the resistance of Type B packages to such accidents. The IAEA invokes the principle of "graceful failure" --- that Type B packages are so robust that they can tolerate much more severe accidents --- but this is largely accepted on faith, and has not been verified for the package used to ship VHLW by sea.
In an attempt to resolve these issues, the IAEA is now undertaking a Coordinated Research Program (CRP) on the issue of sea transport. According to the aforementioned Chairman's Report, "the question of whether [a] sea-mode dependent package requirement might be needed ... appears to depend upon the final results of the CRP." The present lack of information undermines the IAEA's assurance that such accidents are unworthy of regulatory consideration. For emergency planning purposes, it is prudent (and consistent with IAEA's own policy) to assume that severe accidents are possible and evaluate their consequences.
Contrary to the IAEA's claim, the following postulated chain of events resulting in a VHLW cask being damaged and lost at sea is plausible. For each event, outstanding technical questions are discussed that illustrate why quantification of their probabilities is such an uncertain undertaking.
1) Involvement of the VHLW transport vessel in a beyond-design-basis collision.
The Pacific Nuclear Transport Limited (PNTL) ships used to transport VHLW are designed so that if struck by a vessel traveling at 15 knots with a displacement of 23,000 tonnes, the striking vessel will not penetrate the cargo area. It is unclear how they would fare against larger, faster ships that are common on the high seas. A ship with 50,000 tonnes displacement traveling at 24 knots would have a kinetic energy more than five times greater than, and could penetrate twice as deeply as, the vessel the VHLW transport ship is designed to resist.
IAEA (1997) incorrectly implies that the safety features of PNTL ships are comparable to those of ships which an OECD report3 judged could reduce the probability of occurrence of a severe accident to a negligible level. The reference ship considered in detail in the OECD report, which was designed "specifically to reduce the risks from transportation accidents," is considerably more robust than the PNTL ships. The OECD reference vessel is strong enough to prevent penetration into the cargo hold by a vessel of any size with a speed less than 24 knots. In addition, the ship has a large quantity of urethane foam insulation that would provide protection from an engulfing, 982 C fire for three days and would prevent the ship from sinking even if it were cut in two.
The OECD report briefly considered another type of ship which resembled the PNTL vessels (purpose-built but essentially conventional in design), but did not analyze it in detail because it was recognized that it would provide a lesser degree of safety than the ship described above. The OECD estimated that the probability that one of these ships could lose its cargo as a result of an accident was between 0.12% and 0.16% per year --- a hardly negligible value.
2) A breach of the cargo area, subjecting the VHLW casks to an accident environment more severe than the IAEA Type B test, followed by sinking of the vessel or expulsion of a cask into the sea.
IAEA (1997) asserts that analyses sponsored by the U.S. Department of Energy show that the forces imparted to a cask during a ship collision would be "about the same order of magnitude as those produced by the regulatory tests." However, a Sandia National Laboratories (SNL) study indicates that the impact environments that a cask can encounter during a marine collision may be more severe than the Type B impact test. For example, it was found that following the initial collision, the shipping cask was subjected to a series of more than ten closely spaced impacts (as a result of collisions alternating between the two hulls), at speeds of up to 33.6 meters per second, and ultimately was crushed between the two hulls.4 In contrast, the Type B test involves only one impact test, at a speed of 13.3 meters per second.
Even if all the impacts were less severe than the Type B impact, the cumulative effect of the sequence could stress the package beyond the failure point. Thus the conclusion of the SNL paper that "these impact forces are likely less than would be seen during a regulatory drop test" does not appear to be supported by their results. In more energetic collisions than the one analyzed by SNL, impact energies would be even greater and regulatory conditions could be exceeded by even greater margins.
3) Breaching of the VHLW package as a result of excessive stress.
The IAEA's assumption that a Type B package is capable of withstanding accidents greatly exceeding the regulatory accident is not supported by the available evidence.
Collision: The IAEA cites a recent SNL study that finds "cask-like" structures can survive "forces greatly exceeding those imparted by the regulatory tests" without gross failure --- in particular, that they can survive impacts of 4 times the energy of the regulatory tests (an impact speed of 26.8 meters per second [m/s], about twice the 13.3 m/s regulatory speed). However, it neglects to mention that the same study found that these casks exhibited "significant leakage from both seals" following an impact of only 5 times the energy (2.25 times the velocity, or 30 m/s).5 (Note that this is less than the 33.6 m/s cask velocity predicted in the SNL study of ship collisions described above.)
Fire: NCI (1996) has cited the use of elastomer seals in VHLW shipping casks as an example of non-conservative design. Because of the low temperature failure threshold (250-300 C) of these materials, an 800 C fire of a modestly greater duration than the 30-minute regulatory fire could cause breach of containment. The IAEA does not dispute this fact, but chooses to question whether the occurrence of such a fire aboard ship is "credible," ignoring the documented record of shipboard fires that last from days to weeks. NCI (1996) has argued that elastomer seals burden the cask with an abrupt failure mode should it encounter beyond-design-basis thermal conditions.
The IAEA cites the "good history of performance" of elastomer seals, as demonstrated by a series of elevated-temperature tests done at SNL, which found very few failures. However, these test results are an exercise in circular reasoning, because for the series, "the intent was to select [target test] temperatures for which there were high probabilities for success, rather than testing for failure."6
A more interesting observation from the cited SNL study is that the manufacturer's recommended groove widths for seal mountings do not take into account differential thermal expansion of metal and seal, so that seal failure due to overfilling of the grooves during a fire can occur. In the test series, this problem was solved by increasing the widths of the grooves. However, to evaluate the containment performance in a fire of an actual shipping cask, which presumably has grooves of the standard width, the effect of overfilling on seal performance must be taken into account. Thus the results of the SNL test series, which did not include "specific controlled tests ... to compare performance of `standard' groove width fixtures to the modified ... fixtures," are not useful for evaluating seal performance in real situations.
Immersion: The IAEA argues that Type B shipping casks must "withstand an immersion test of 200 meters without rupture of the containment." They do not point out that this test need be carried out only for one hour, and that an undamaged specimen can be used. This test is not relevant to a realistic accident situation in which a cask, perhaps initially damaged in a collision or fire, then remains at a depth of 200 meters for a period of weeks to months or longer. The IMO should note that COGEMA, BNFL and FEPC provided erroneous information on this point in an paper submitted to the Special Consultative Meeting in March 1996, which claimed that Type B packages are "after these [collision and fire] tests, put in at least 200 meters of water for 8 hours."7 While the IAEA mandates that the immersion test at 200 meters must be carried out for "not less than one hour," VHLW casks have only been tested to the minimum standard of one hour.8 Also, there is no indication that the immersion test has been performed on packages previously subjected to impact and fire tests.
(4) Rapid corrosion by seawater of the sensitized stainless steel VHLW canisters.
IAEA (1997) challenges the hypothesis of NCI (1996) that the VHLW stainless steel canisters are "sensitized" (made more vulnerable to corrosion) because of an incompatibility of the selected steel with the VHLW production process. IAEA (1997) does not argue that sensitization does not occur (and provides no evidence to support such a conclusion), but only that NCI (1996) has not provided definitive proof. As noted below, the weight of the evidence clearly indicates that the VHLW canisters undergo significant sensitization. It should be the responsibility of the producers of VHLW to ensure that sensitization, which has implications for the safety of transport and storage, does not occur.
A series of papers issued by the Central Research Institute of the Electric Power Industry of Japan (CRIEPI) has analyzed the sensitization and corrosion behavior of a variety of stainless steels, including Type 309S, which COGEMA and BNFL now say that they use. The CRIEPI reports show that Type 309S will sensitize under typical VHLW cooling conditions.9 Moreover, cooling curves specific to the production of COGEMA glass in publicly available literature show that the canister temperature does remain in the sensitization range for several hours.10 According to a SNL report, under these conditions "sensitization of the VHLW canisters is likely."11
To conclusively resolve this issue, COGEMA and BNFL should perform chemical analysis on samples of stainless steel from actual VHLW canisters in storage and make the results and methodology available for public review. This course of action was recommended by NCI in 1994, and, to our knowledge, never carried out.
IAEA (1997) describes several prerequisites for intergranular stress-corrosion cracking (IGSCC) to take place. These would all be present in the event that a VHLW canister were exposed to sea water following the sinking of a ship.
IAEA (1997) says that IGSCC is unlikely because "industry sources have stated that very little operational stress is present in the canisters as presented for transport." However, it is not "operational stress" that is of concern, but "residual stress." VHLW canisters are frozen into a state of high tensile stress because of the differential thermal contraction of the stainless steel and glass during cooling. In addition, the thermal stresses induced by the welding of the canister lid can also be significant.12
For sensitized stainless steels, IGSCC has been observed for temperatures as low as 20 C,13 which could be experienced by a severely damaged VHLW cask lost in coastal waters. If the shipping cask were to remain largely intact, the temperature of the water that infiltrates into the cask cavity would be quickly heated to 100 C or greater (as is pointed out in the CRIEPI study), at which rapid IGSCC can take place. The intense gamma radiation emitted by the canisters will also promote IGSCC because of radiolysis effects. Thus, the environmental conditions encountered in coastal waters will be severe enough to promote IGSCC of the sensitized VHLW canisters in a sunken cask.
5) Leaching of radionuclides from the VHLW into the marine environment.
Contrary to the contention of IAEA (1997), NCI (1996) does not assume that the shipping cask "disappears," or that the entire waste inventory is instantly released into the ocean. Instead, it conservatively assumes that the cask is damaged to the extent that there is a moderate flow of water through the cask cavity, so that the leach rate of the glass is not suppressed by saturation effects.
In contrast, the CRIEPI study assumes that the VHLW cask is undamaged except for failure of the O-ring seal. The low dose rates it obtains can be largely attributed to the assumption of an unrealistically small width for the gap between the cask and lid, resulting in an underestimate of the flow rate. The release rate depends strongly (a third-power dependence) on this width, and thus small variations can lead to large variations in the consequences. The CRIEPI report does not discuss the sensitivity of their results to variations in the input parameters.
However, the anomalously low results in the CRIEPI report cannot be entirely attributed to this factor --- other assumptions in their model must also play a role. It is difficult to understand why the radiological consequences of such an accident in Japanese coastal waters would be so much smaller than those elsewhere, especially since the major exposure pathway is through fish consumption, which forms a large part of the typical Japanese diet.
IAEA (1997) expresses the concern that it is misleading to directly compare different studies because of their sensitivity to local geographic patterns. How then can they justify allowing BNFL to use the results of the CRIEPI study, specific to a sinking off the coast of Japan, to reassure en-route states all along the route? Their point only underscores one that the Nuclear Control Institute and Greenpeace have made repeatedly --- that site-specific environmental impact statements need to be carried out along the entire shipping route.
End Notes1. Edwin S. Lyman, PhD, "Response of the Nuclear Control Institute to IAEA Comments on 'The Sea Transport of Vitrified High-Level Wastes: Unresolved Safety Issues," May 1997. Available on request from the Nuclear Control Institute, 1000 Connecticut Ave. NW, Suite 804, Washington, DC, 20036. E-mail: email@example.com. Back to document
2. C. Young, Chairman's Report, Advisory Group Meeting on "Modal Issues in the Safe Transport of Radioactive Material," International Atomic Energy Agency, Vienna, 4-6 November 1996, p. 76. Back to document
3. Nuclear Energy Agency, Feasibility of Disposal of High-Level Radioactive Waste into the Seabed, Volume 2, OECD, Paris, 1988, p. 144. Back to document
4. V. Porter and D. Ammerman, "Analysis of a Ship-to-Ship Collision," Proceedings of the PATRAM '95 Conference, Volume III, p. 1083. This study computed the average force as a function of time on a cargo of shipping casks during a collision with a 16,750-tonne ship traveling at 30 knots. It was found that the magnitude of the initial average compressive force on the cargo was 370 meganewtons (MN), corresponding to an average acceleration of 343 g for a 110 tonne VHLW cask, which is 15% greater than the 300 g average acceleration measured during 9-meter drop testing of the cask. Integration of the impulse vs. time curve shows that this impact would send the cargo toward the opposite hull at a speed of 33.6 meters per second. Although the hulls are not "unyielding surfaces," as are required in the regulatory tests, they have been stiffened to provide resistance to collisions. Even if half the energy were lost to hull deformation, the energy imparted to the package would still be more than twice that associated with the regulatory test. Back to document
5. D. Ammerman and J. Bobbe, "Testing of the Structural Evaluation Test Unit," Proceedings of PATRAM '95, Volume III, p. 1123. Back to document
6. D. Brownowski and P. McConnell, "Performance Characteristics of O-Ring Seals for Radioactive Material Packages When Subjected to Extreme Temperatures," Proceedings of the 11th International Conference on the Packaging and Transport of Radioactive Materials (PATRAM '95), Volume IV, p. 1791. Back to document
7. BNFL, Cogema, and FEPC, "Information Paper Submitted to the Special Consultative Meeting of the IMO by BNFL, Cogema, and FEPC," undated, p. 16. Back to document
8. Y. Gomi et al., "Demonstration Test for Transporting Vitrified High-Level Waste: Immersion Test," in the Proceedings of PATRAM '95, Volume III, p. 1145. Back to document
9. M. Mayuzumi, "Effect of Relative Humidity on Stress Corrosion Cracking Susceptibility of Candidate Canister Materials," Komae Research Laboratory Rep. No. T86050, p. 20, Table 4. Back to document
10. R. Simon, Commission of the European Communities, Brussels, "The Qualification of Waste Forms and Engineered Barriers," in Radioactive Waste Management and Disposal (L. Cecille, ed.,), Elsevier Applied Science, London, 1991, p. 159, Fig. 1. Back to document
11. J. Sprung et al., "Comments on a Paper Titled "The Sea Transport of Vitrified High-Level Radioactive Wastes: Unresolved Safety Issues," SNL, SAND97-1130, May 1997. A detailed NCI analysis of this paper will be available soon. Back to document
12. According to the Aomori Experts' Advisory Group Report, this residual stress is estimated to be 22-24 kg/mm2 (216-235 MPa). This value is 79-85% of the room-temperature yield strength of Type 309S stainless steel (275 MPa), a stress level at which IGSCC can readily occur. Back to document
13. A. John Sedriks, Corrosion of Stainless Steels,, 2nd ed., John Wiley & Sons, New York, 1996, p. 287. Back to document
Sea Transport Page What's New Page Home Page