Accidents at sea

CRIEPI estimated the consequences of a radioactive release from irradiated nuclear fuel, high level waste, and plutonium due to the loss of the transport package into the deep ocean and into shallow seas off the north-east coast of Japan. In both cases it was assumed that no action was taken to recover the lost package. Submergence of the flask to a depth of 2500 m was assumed after loss into the deep ocean. Loss into shallow coastal waters was assumed to result in flask submergence to a depth of 200 m. Radioactive release into the deep ocean was conservatively modelled assuming that the release rate was controlled solely by leaching of radionuclides from the bulk material matrix with no credit taken for retardation of release by fuel rods, canisters, and/or the radioactive material package. For a radioactive release after a package had been submerged into shallow waters, any retarding effect of fuel cladding or canisters was neglected. Instead, leaching of radionuclides was assumed to cause the water in the package to become saturated by the radionuclides in the radioactive material being shipped; the release of radionuclides-saturated water from the flask was controlled by the buoyancy-driven flow of water that ran through the gap in the failed o-ring seal of the package.

Once released into the ocean, the concentration of radionuclides was estimated using a multi-compartment flow model [31] for deep ocean release and ocean current data [32] for near shore release. The maximum calculated surface concentrations of radionuclides were then used as input to a marine food pathway model [33, 34], which in turn provided doses for individuals whose diet followed a Japanese market basket developed by the Nuclear Safety Committee of Japan and who ate only maximally contaminated marine foods that had become contaminated due to the hypothetical loss of the package into the ocean. Table XXV presents the ‘maximally exposed individual doses’ estimated with the aid of these calculations.

TABLE XXV. CRIEPI ESTIMATES OF ‘MAXIMALLY EXPOSED INDIVIDUAL DOSE’

RESULTING FROM THE LOSS OF A RAM PACKAGE INTO THE OCEAN

Nuclear material Quantity Accident

location Submergence ‘Maximally ^(a)

High level waste 1 flask: 28 canisters Near shore At sea (a) “Maximally” means here that all seafood that has been ingested is assumed to be contaminated.

7.4.2. IPSN-CEPN study

IPSN-CEPN used the POSEIDON code [21, 22] to estimate the ‘maximally exposed individual doses’ that might result if 1 kg of plutonium powder containing about 4 × 10¹⁴ Bq of Pu nuclides and Am-241 was released into the western English Channel during a shipping accident. The compartment model implemented in the POSEIDON code models flows between well-mixed compartments and within each compartment adsorption and scavenging of radionuclides by sediments, sediment resuspension, dissolution of adsorbed radionuclides, and entry of radionuclides into marine food chains due to uptake of contaminated water and sediments by marine plants and organisms. Consumption of contaminated marine foods is compared against a market basket for reference population groups that allows doses to be calculated for individuals in the groups who eat seafood caught only from specific ocean regions (ocean compartments). Table XXVI presents the results of these calculations using the POSEIDON code.

TABLE XXVI. CONSEQUENCES OF LOSS OF 1 KG OF PLUTONIUM INTO THE WESTERN ENGLISH CHANNEL

(a) Maximal here means that all seafood ingested is assumed to be contaminated. Market basket values reflect critical groups in all areas of the world according to current known dietary habits.

7.4.3. Sandia study

SNL used the MARINRAD code to estimate the ingestion doses that might result from the loss into the ocean of a TN-12 irradiated nuclear fuel flask while traversing the Grand Banks fishing region. The MARINRAD code models transport of radionuclides between ocean compartments by ocean currents, deposition of radionuclides onto compartment sediments, uptake of radionuclides from these sediments and/or ingestion of suspended radionuclides by seaweed, plankton, crustaceans, molluscs, and larval fish, bioaccumulation of radioactivity due to predation in marine food chains, and radiological exposures caused by ingestion of marine foods and desalinized sea water, inhalation of seas pray, swimming in contaminated sea water, and exposure to contaminated sediments.

The calculation assumed that the ship collision caused the TN-12 flask to be lost into the sea and that the entire flask inventory was released into ocean waters over time periods ranging from 3 to 300 years. The results of the calculation indicate that radiological exposures are largely determined by the ingestion pathway and were largest for individuals who consumed seafood taken exclusively from the Top Labrador compartment of the 19-compartment ocean model, the compartment that contains the Grand Banks. Near-term yearly individual doses for individuals who consumed seafood harvested exclusively from this compartment increase as the radionuclide release time decreases. When release takes place over three years, yearly individual doses reach a maximum value of about 18 mSv/a five years after the sinking of the RAM transport ship and then fall to 10 mSv/a 100 years after the sinking. When release takes place over 300 years, average yearly individual doses throughout the first 100 years are about 0.4 mSv/a.

7.5. Discussion

The illustrative consequence calculations described above are quite conservative. The MACCS and RADTRAN calculations are conservative for at least three reasons: first, because the likely result of deep penetration into the RAM hold by the bow of a striking ship is not flask failure but instead is the pushing of the ‘unfailed’ flask through the far shell of the ship into the ocean which would mean that radioactivity would be released into the oceans rather than to the atmosphere as was assumed for these accident analyses; second because, at least for port accidents and probably for coastal accidents near a developed coastlines, fire fighting equipment would be deployed to fight the ship fire and thus the enhanced atmospheric release hypothesized for the fire accident would either not occur or would be substantially decreased;

and third because recovery of a flask lost into a harbour channel or into the ocean at a distance of a few tens of kilometres from shore would be routine and would normally be accomplished long before any significant release of radioactivity would take place.

Similarly, for the reasons set forth below, the individual yearly doses estimated for loss of Type B packages into the ocean are also very conservative. They are conservative for irradiated nuclear fuel because CRIEPI studies [35] show that the flask failure does not occur when submergence depths are less than 3000 m after loss of the flask into the ocean. Thus, flask and rod failure will usually have to occur by corrosion. But an ISPN review [36] of steel corrosion rates in sea water suggests that perforation failure of Type B packages is likely to take at least several years if not several decades. Thus, the rapid release of radioactivity assumed in the SNL ocean loss calculations leads to a substantial overestimate of release rate and thus also of the concentrations of radioactivity in marine foods. Moreover, given modern

deep ocean salvage capabilities, flask recovery is likely before these flask failure time periods are exceeded. But mainly, the individual yearly doses estimated are conservative because contaminated seafood reaches individuals in the general population through the commercial food distribution system, which means that the individual doses caused by consumption of this contaminated seafood will always be substantially smaller than the ‘maximally exposed individual doses’ estimated by the ISPN, SNL, and CRIEPI calculations. Thus, the real ingestion doses that might be received by members of the general public following the loss of a Type B package into the ocean will always be very small, much smaller than normal background radiation exposures, and thus of little significance.

In conclusion, even if these conservative assumptions are ignored, the illustrative consequence calculations described in this section have one result in common. They all predict doses that are very small when compared to the average annual dose normally incurred by individuals due to exposure to natural (e.g. cosmic rays, radon, terrestrial radionuclides) or routine man-made sources of radiation (e.g. medical X rays). Thus, these illustrative calculations suggest that the radiological consequences that might result, if a ship transporting a Type B package were involved in a severe maritime accident, are not of great concern.

8. DISCUSSION