Accidental Melting of Radioactive Sources

Accidental Melting of Radioactive Sources K. Baldry 1, D. S. Harvey 2, A. Bishop 3 1 RWE NUKEM Limited, Windscale, Cumbria CA20 1PF, UK Email: keith.baldry@rwenukem.co.uk 2 Corus RD&T, Rotherham, S60 3AR, UK Email: david.s.harvey@corusgroup.com 3 RWE NUKEM Limited, Windscale, Cumbria CA20 1PF, UK Abstract. The problem of accidental melting of radioactive sources during metal production is well known. It is, however, impossible to prevent all incidents, since some radioactive sources cannot be detected, even by the best detector systems. There continues to be a need to consider the consequences of accidental melting. Because these incidents are rare, the plant workforce generally has limited experience and instrumentation to cope with them. Melting events have been analysed and scenarios developed. Probable source sizes and nuclides are presented with the fractions that partition to each phase of the melting process. Consequences of events are discussed and the risks to each group of melting plant workers and to the off site public assessed. Guidance to assist plant operators has been developed. This gives initial actions that can be taken by workers without expert knowledge, and follow up actions once experienced staff are available. This study has been sponsored by the European Commission Directorate General Environment. 1. Introduction When a radioactive source is melted, there is a need for the melting plant operators to make a number of important decisions in a short time. The staff of metal production plants have to make the necessary decisions based on limited information, but they are not radiation specialists, and very rarely have they had experience of such an incident. The decisions taken could affect the health and safety of the workforce, and the general public. These decisions also affect the cost of making the situation safe, and returning the plant to normal working. There is a potential conflict between ensuring health and safety of the workforce, and minimizing the costs of lost production. Any melting incident causes great anxiety among those involved. The costs of such incidents are typically measured in millions of euros. The purpose of this project, undertaken with the support of the European Commission, was to make available existing experience with a view to minimizing the risk of exposure, direct or indirect, to ionizing radiation of workers at metal plants, of workers at ancillary plants and of members of the public following the accidental melting of a radioactive source. 2. Review of Previous Incidents When a radioactive source is melted the three possible outcomes are that the radioactivity partitions mainly to either the metal, the slag, or the off-gas dust. The first stage of the work was to examine some examples of previous melting incidents of each kind. Well-documented examples of each type of incident were selected. These were studied with the generous cooperation of the people at melting plants where the incidents occurred and are documented by the authors [1]. 3. Melting processes Most of the recorded incidents have occurred in the production of steel in arc furnaces due to the nature of the scrap source material. The practices of arc furnace steel production are similar at all sites. It is normal to use large volumes of oxygen which cause turbulence in the melt, and create large volumes of off-gas. The turbulence cause some of the slag, and metal to be entrained in the off gases. 1

In basic oxygen steel (BOS) production there is again strong turbulence, and entrainment of material in the off gas. The induction furnace process for steel production does not usually entail gas blowing, and is a very much more quiescent process. The quantity of off-gas dust is minimal, and largely composed of volatiles from the melt rather than slag, and metal entrained in the off gases. Hence any volatile radioactivity will be in a concentrated form in the off gas dust. Similar furnaces are used for the melting of copper and aluminium. In the production of copper there may be some gas blowing and turbulence, causing entrainment of slag and metal in the off gases. In the production of aluminium the use of oxygen is avoided, but it is common to use halide-containing gases, in the refining process and these cause some turbulence. Salts that also may be halide, are used in the refining process for aluminium. 4. Scenarios 4.1. Source Sizes The IAEA report Methods to Identify and Locate Sealed Disused Sources [2] lists source activity for various uses and is shown in Table I. It represents worst case sources that might be melted accidentally. Table I: Disused Sources Maximum Size of Each Radionuclide [2] Radionuclide Activity (GBq) Comment Americium 241 800 Am-241/Be neutron well logging sources Caesium 137 600,000 Sterilization Cobalt 60 1,000,000 Teletherapy Yusko [3] lists known accidental melting events that have occurred across the world. This gives another approach as shown in Table II. The numbers for Co-60 and Cs-137 reflect their widespread use in industry. Am-241 is probably underrepresented for reasons described in section 4.2.2. The source sizes quoted in this list are approximate because it can be difficult to estimate the size of the source that was melted after the event. Table II: Known Accidental Melting Events Maximum Size of Each Radionuclide [3] Radionuclide Activity (GBq) Americium 241 1.7 Caesium 137 1000 Cobalt 60 15,000 Radium 226 0.7 The very largest sources are generally under producer control as well as user control. In addition they are more readily detected when scrap is monitored for radioactivity. For these reasons they are less likely to be accidentally melted. The very low likelihood of encountering such a source is confirmed by looking at known melting events (i.e. it has not occurred). The risk of melting these worst-case sources is not quantified but considered very low. The consequences of the worst-case sources are included in this report to provide bounding conditions. Angus [4] and Crumpton [5] discuss the frequency of loss of control of sources. Angus concludes that the most significant risk from the sources under consideration in this report comes from large Cs-137 sources (in excess of 400 GBq). These have a medium frequency of control being lost, and a low frequency of being melted. In terms of potential radiation doses, these are considered the highest risk melting events. 2

The more typical sources under consideration are generally not under producer control and hence can more readily be lost (for example a company goes bankrupt and its radioactive sources are forgotten). The sizes of the typical sources considered in this report are taken from known melting events. The likelihood of a particular plant encountering such a source is low, but it is likely that similar sources will be found at some plant in the future. It is therefore reasonable to prepare responses to the typical rather than worst-case source sizes. 4.2. Behaviour of radionuclides 4.2.1. Radioactivity partitioning to the off gas dust There have been many incidents in which caesium 137 has been melted in steel production. This volatile radioisotope always partitions to the off-gas dust and so the radioactivity is almost completely confined to the gas cleaning system and very little is retained in either the steel or the slag. The amounts of Cs-137 radioactivity retained in the steel and the slag are not readily detectable. Hence the experience has been that the radioactivity is almost wholly extracted by the gas cleaning system. The result is that there can be high levels of radioactivity throughout the system, and there can be significant levels of radioactive contamination emitted by the gas cleaning plant. Some radioactivity will pass through the gas cleaning plant and be emitted to atmosphere. A typical plant removes 99% of the dust from the gas stream [6], and since the volatiles will have condensed by the filtration stage it is suggested that the amount of radioactivity captured will be the same. Hence the amount emitted to atmosphere will be of the order of 1% of that melted in the furnace. 4.2.2. Radioactivity partitioning to the slag The radioisotopes that partition to the slag include radium, and Am-241 and the other actinide elements. These emit alpha radiation, and so the main hazard is from radioactive dust absorbed in to the body. There have been few melting incidents reported in which radioactivity partitions to the slag. It is believed that many incidents of this kind may go undetected because slag is not usually checked for radioactivity. If the slag remains in bulk form the exposure of the workforce to radiation is likely to be very low. Exposure can be higher if the slag is broken down in to a dusty form, as can occur during slag processing. The dust containing the radioactivity can then be inhaled and result in internal radiation exposure. For any radioactivity that is absorbed by the slag some will be retained in the melting furnace, and will cause some contamination of the slag on the subsequent melt unless the furnace is decontaminated. Some slag will also pass in to the gas cleaning system, since there is always some physical entrainment of slag in the off-gases. Hence there might be contamination of the gas cleaning system. The basis for such a scenario is the melting of an actinide element such as Am-241 in steel production. This radioisotope is representative of the actinides (e.g., thorium, uranium, plutonium, and americium), all of which behave similarly in metals production. 4.2.3. Radioactivity partitioning to the metal Most of the incidents in which radioactivity is absorbed by the metal have involved the radioisotope cobalt 60. This radioisotope always partitions to metal and is distributed throughout it. There is then some radiation hazard if people spend time near large amounts of the metal, such as a store of slabs. For any radioactivity that is absorbed by the metal some will be retained in the melting furnace, and will cause some contamination of the metal on the subsequent melt unless the furnace is decontaminated. Cobalt 60 does not become chemically combined with the slag. In practice, however, the slag will contain metal that has become physically entrained. Hence when Co-60 is melted there will be some contained in particles of metal in the slag. The amount in the slag will depend on the steelmaking process, the properties of the slag, and the care with which slag and metal are separated in the process. 3

For any radioactivity that is absorbed by the metal some will also pass in to the gas cleaning system, since there is always some physical entrainment of metal, and metal oxides in the off-gases. Hence there might be contamination of the gas cleaning system. 4.2.4. Partition percentages Kopsick [7] reports the percentages of each type of radionuclide to each partition (arc furnace). Table III: Partition of Radionuclides in an Arc Furnace [7] Steel Slag Off gas Type of radionuclide Metal seeking 99.2% 0.5% 0.3% Slag seeking 1.0% 94.5% 4.5% Off gas seeking 0.001% 0.001% 99.98% 4.3. Dose model methodologies External radiation from steel product, slag and off gas dusts is calculated by taking the source sizes in Table II, partitioning them according to the percentages in Table III and then using proprietary software Microshield to calculate the external dose rates. The dispersal and inhalation methodology for internal doses discussed by Ford [8] is used: The calculated dose d = Q C V R e (1) where: Q Source size (Bq) C Dispersion coefficient (s/m 3 ) V Reference man breathing rate (m 3 /s) R Release fraction e Dose coefficient (Sv/Bq) [9] The dispersion coefficient C (s/m 3 ) is the integrated air activity concentration per unit release of activity. It is a measure of the total air activity concentration a person would be exposed to if a unit of activity (1 Bq) was released and the resulting cloud of activity passed by that person. Atmospheric conditions (e.g. wind, turbulence) provide the mechanisms for dispersion. They have been calculated using the Gaussian Plume dispersion model. Source sizes Q from Table II are used, partitioned to off gas, slag and product in accordance with the percentages in Table III. In the summary Table V, doses from the worst case source sizes (Table I) are included for comparison, though it is noted that events of this scale have not been recorded. Calculated exposures for the following exposure scenarios are considered: External radiation from metal product. Radioactivity released early in the melt in an arc furnace as fugitive emissions Radioactivity released from the melt into the melting shop environment with dusts Off gas dusts in the ventilation system and bag house inhaled during handling maintenance Dust raised from the slag as it is handled Dust raised by flame cutting and other treatments of contaminated product Radioactive contamination emitted from the plant to the areas surrounding the plant. The hazard from unmelted scrap awaiting entry to the process was outside the scope of this study. 4

4.4. External radiations Geometry affects the dose rates. The following is assumed. The geometries are typical of slab product, slag piles and dust collection bags. This approach is justified as including the likely worst cases. It is noted that individual situations will vary, but it is believed that alternative credible geometries will tend to produce lower dose rates: For the metal product, 120 tonne (t), density of iron (7.86 g/cm 3 ), slab size 1.5 m 0.3 m 35 m For the slag, 10 t, density ~50% iron (3 g/cm 3 ), slag pile dimensions 1 m 1.6 m 1.6 m 1 t off gas dust, density (1 g/cm 3 ) diluted in 10 t other (non-contaminated) dusts, dimensions 1.1 m 1.6 m 1.6 m Table IV: Calculated dose rates (msv/h at 1 m) for realistic source sizes Radionuclide Size (Bq) Steel Slag Off gas dust Co-60 1.5 E13 7.5 1.1 0.72 Am-241 1 E10 negligible negligible negligible Cs-137 1 E12 negligible negligible 3.0 A Co-60 source of 1 E15 Bq could theoretically produce dose rates from the product of 500 msv/h at 1 m, and a Cs-137 source greater than this from the off gas dusts. It is noted that this extreme level of radiation has not been measured following a melting event, and the risk is considered low, however it indicates the importance of monitoring and access control in the early stages following detection. 4.5. Internal radiations 4.5.1. Internal doses - fugitive emissions Fugitive emissions could be significant if a source is volatilised early in the melting process. This will occur only with volatile radionuclides such as Cs-137. Assume that Cs-137 melts in an arc furnace and that 1% is released with fugitive emissions. The majority will be in a buoyant plume of gas and goes directly upwards and is taken out with the melting shop ventilation system. Empirical data are not available for such a scenario indeed the outcome would vary considerably depending on source placement and time of melting but it is considered reasonable to assume pessimistically that 1% of this emission remains in the operating area and that a crane driver or other worker is exposed. A dispersion coefficient of 6.7 E-3 s/m 3, release fraction of 1 E-04 and a dose coefficient of 6.7 E-9 Sv/Bq gives a scenario exposure to a melting shop worker of 1.5 µsv. 4.5.2. Internal doses - dust in operating areas For internal doses resulting from dust in the operating areas, Cs-137 is addressed, as it generates the highest workplace concentrations. In a 100 te melt, 1 te of off gas dust is released. The measured quantities of inhalable dust in an arc furnace melting shop are 1.5 mg/m 3 [6]. Assume that the Cs-137 is uniformly distributed amongst the dust. The activity concentration would be 3 kbq/m 3 from one melt, which forms, say, 20% of the dust in the shop. A 30 minute exposure to this Cs-137 would lead to a dose of 1 µsv. The same scenario with Am-241 would lead to a dose of 2 µsv. Mobbs [6] addresses continual processing of low activity scrap over a 12 month, 1800 hour working year, with such scrap forming 1% of the total. A simple comparison between the methodologies applied by Mobbs and by this report can be made by attempting to extend this report s assumptions out to 12 months. This is done by standardizing the activity concentration, decreasing the percentage of contaminated scrap as part of the total from 20% to 1%, and increasing the exposure time from one to 1800 hours. The dose is then reduced by a factor of 9 E-03. The dose over 12 months is 1.8 E-08 Sv/y, in comparison to Mobbs 1.5 E-08 Sv/y. 4.5.3. Internal doses - dust in gas cleaning plant If 1 E+12 Bq Cs-137 is distributed in 10 te dust in the bags of the gas cleaning plant, the activity concentration is 1 E05 Bq/g. The highest levels of dust during handling operations can be defined by 5

the maximum levels that would be physically credible to breathe. A figure of 10 mg/m 3 is used and the consequent activity concentration would be 1000 Bq/m 3. A half hour exposure would result in a dose of 4 µsv, with no respiratory protection being worn. Taking Am-241, with 4.5% going to the off gas, the dose would be 155 µsv in half an hour. It is noted that dust handling operations are generally undertaken using respiratory protection, and the protection factors are not accounted for in this calculation. 4.5.4. Internal doses - dust raised by flame cutting Flame cutting is the technique that will raise the most particulate and is considered the worst case for contamination raised by product treatment. Take the slab discussed in section 4.4. A 1000 GBq Co-60 source results in activity levels of 8 kbq/g in the product. The surface area of the end of a slab being cut is 1.5 m x 0.3 m, or approximately 5000 cm 2. Say the cut is 0.5 cm thick and 10% of the metal resuspended in inhalable form by the cutting. The resulting dose to the operator would be 350 µsv, assuming that no respiratory protection is worn. 4.5.5. Internal dose from handling slag Taking an Am-241 source of 10 GBq, a melt size of 100 te and consequent slag mass of 10 te, the activity concentration is 950 Bq/g. If a dust concentration of 3 mg/m 3 is raised during slag handling, then the local activity concentration is 2.85 Bq/m 3. An exposure of 46 µsv would result in half an hour. 4.5.6. Internal doses Off site doses to members of the public Assume that a Cs-137 source melts and that 1% is released [6] to the environment. Assuming a ground level discharge, H=0, worst case weather conditions (very stable), and the distance from source, x is 200 m, the dispersion coefficient C is 1.7 E-03 s/m 3. The resultant exposure is 38 µsv. Taking the same scenario, with 4.5% of a 10 GBq Am-241 source going to the ventilation system and 1% being discharged to the environment. The resultant dose would be 68 µsv. The dispersion coefficient for mean weather conditions (an example location in south central UK is used) and a 10 m stack release height is 1.1 E-04 s/m 3. The resultant dose for the most exposed member of the public would be 2.4µSv. The risk presented by such events must consider the frequency that such events are likely to occur. Angus [4] indicates that the maximum typical size of Cs-137 sources available to be lost from regulatory control and then accidentally melted is only 100 GBq. However, there is a possible exposure route for members of the public, and environmental monitoring must form part of the follow up actions in the event of an off-gas or slag seeking radionuclide melting event. 4.5.7. On site doses to members of the workforce A member of the workforce is likely to be necessarily nearer the point of discharge. For a ground level discharge, the dispersion coefficient for a distance, x of 30 m is 4.3E-02 s/m 3 [8]. With a 1 E12 Bq source the dose would be 950 µsv. A 10 m stack height reduces this by a factor of 4. 4.6. Induction furnace Induction furnaces melting steel produce higher internal doses to workers than arc furnaces, though as has been noted the arc furnace is more likely to accidentally melt a source due to the nature of its source material 4.7. Copper and aluminium smelting The report has addressed the more commonly encountered situation of sources involved in steel melting accidents. Copper and aluminium events will be similar in outcome, with slightly modified dose outcomes. Data from Mobbs compares directly the consequences for steel, copper and aluminium. External doses are similar for steel and aluminium, but less for copper. 6

4.8 Summary of dose consequences The study has estimated the likely exposure of the workforce during the accidental melting of a range of radioactive sources. The main results are shown in Tables V and VI. Working times are likely to be less if the event has been detected. Table V: Potential external exposure Task (30 minute working time) Dose (msv) Radionuclide Product handling 3.3 Co-60 Slag handling 0.6 Cs-137 Off gas dust handling 1.5 Cs-137 Table VI: Potential internal exposure Source of exposure (30 minute working time where applicable) Dose (msv) Radionuclide Fugitive emissions 0.002 Cs-137 Melting shop general environment 0.001 Cs-137 Dust in gas cleaning plant (assuming no respiratory protection) 0.004 Cs-137 Dust raised from the slag 0.046 Am-241 Dust raised from flame cutting (assuming no respiratory protection) 0.35 Co-60 External to the plant - public exposure (worst case weather conditions) 0.04 Cs-137 External to the plant - public exposure (typical weather conditions) 0.002 Cs-137 External to the plant - worker exposure (ground level discharge) 0.95 Cs-137 The theoretical worst case sources (Table I) are larger than the source sizes used in the above calculations (Table II); by a factor of 60 for Co-60 a factor of 600 for Cs-137. The risk of the worst case events is extremely small and no such event has been recorded. The largest Am-241 can generate higher theoretical internal doses than the Cs-137 values quoted, however they are very much less likely to be encountered. 5. Guidance for Responding to a Melting Incident 5.1. Emergency arrangements The melting plant emergency arrangements (disaster plan) needs to address the likely sources to be encountered and the probable outcomes. The authors give guidance for such plans [1], which is summarized below. 5.2. Workers affected by each scenario Taking the external and internal dose consequences, the groups of workers that are potentially affected by a melting event are those who undertake hands-on work with the product and by-products. Such activities include: Working close to the product Flame cutting or other aggressive treatment of the product Slag handling Off gas dust handling Working in or close to ventilation and bag house systems Working outside in the path of the plume Controls should focus on these activities. Other personnel such as crane drivers, general shop workers, and other personnel on site are most unlikely to be affected 7

5.3. Emergency Actions (initial response) Alarms should be verified by repeat measurements and gamma spectrometry of product samples. On confirmation, a first response is required that can be undertaken by personnel on plant without any expert knowledge and prior to assessment of analytical results: Keeping ventilation systems running to maximise extraction from operating areas. Minimise personnel access to the product Segregate and minimise handling of product and slag Do not handle the off gas dusts prior to the arrival of expert advice. Continue pour and processing of product 5.4. Management of the incident after initial emergency actions Once technical or management staff familiar with radiation protection are available, further controls, assessment and monitoring can be implemented as detailed below and developed in the contingency plan. Action levels are proposed to enable radiological safety, legislation and plant operational requirement to be best met: If radioactive content does not exceed 0.3 Bq/g then product can be processed normally [10]. If dose rates do not exceed 10 µsv/h then no action is required to restrict external dose. If dose rates are between 10 µsv/h and 1 msv/h then access should be restricted so far as possible in order to minimise exposure. If dose rates exceed 1 msv/h then access should be carefully managed. Dose control levels of 1 msv for any employee would be appropriate. 5.5. Other considerations A melting plant would want to establish relations with an expert body that can provide detailed monitoring assistance and advice. Expert advice should be sought before process materials known to have radioactive content are disposed of. This is because of the expertise required to accurately assess radioactive content, and because of possible specific national legal requirements. Guidance is given for the use of contaminated metals by European Commission report recommendations [10]. The levels for the major nuclides discussed in this report are 1 Bq/g. Contaminated liquors are more difficult to assess and should be quarantined so far as possible pending expert advice. 6. Conclusions Information has been gathered on a number of incidents in which radioactive sources have been melted. The radioisotope involved affects the outcome of an incident. Some radioisotopes partition mainly to the off-gas dust, some to the slag, and some to the metal. In all cases the radiation exposure of the people involved in the melting and casting of the metal is likely to be below 1 millisievert if the event has been detected. (1 millisievert is the annual maximum exposure allowed from work with radioactivity for a member of the public in the European Union). The radiation exposure of members of the public has been negligible in all melting incidents, though the theoretical highest doses can be significant. Contingency arrangements would not need to address evacuation of homes and other facilities in the vicinity of the works as doses are likely to be much less 8

than evacuation reference levels. Note that undertaking radiation and contamination surveys of the areas surrounding the melting shop and bag house following a melting event will still be required. The work required for decontamination can be minimised if action is taken as soon as possible after the melting has occurred, but other factors must also be considered. For example, if the radioactivity is mainly in the metal then further processing of the metal will tend to spread the contamination. It may, however, be more practical to cast the steel in the normal manner than to leave it as in a ladle. If contaminated material is moved off the melting plant site before the radioactivity is detected then the work required for decontamination is likely to be greatly increased. The decontamination of a melting plant can be time-consuming, and the radiation exposure of some of the people involved might exceed 1 millisievert. 7. Recommendations Contingency plans need to be prepared by plant management in advance of a melting incident. These need to be based on knowledge of both the melting process and the practice of radiation safety. Individuals need to be trained in the principles of radiation safety, and instruments for monitoring of radiation levels should be available. External contacts should be identified who can offer specialist expertise in event of a melting incident. The role of government authorities, and the level of assistance they provide varies from state to state, and should be explored. 8. Acknowledgements The authors wish to thank the people and companies who have been willing to assist in providing help and information. The information, derived from actual incidents and experiences, has been fundamental to the completion of this report. The work has been undertaken with the support of the European Commission (Directorate General Environment; Radiation Protection) and this paper reflects the findings in the report submitted to the Commission. 9. References 1. Baldry, K., Harvey D.S., Bishop, A., Handbook for Radiation Safety Interventions Following Accidental Melting of Radioactive Sources at Metal Plants, European Commission Directorate General Environment (2003) 2. IAEA, Methods to Identify and Locate Sealed Disused Sources, IAEA - TECDOC - 804, July 1985 (quoted in European Commission proposed Council Directive on the Control of High Activity Sealed Radioactive Sources, COM(2002) 130, 18/03/02) 3. Yusko, J., Pennsylvania Department of Environmental Protection, USA 4. Angus, M.J., Crumpton, C., McHugh, G., Moreton, A.D., Roberts, P.T., Management and Disposal of Disused Sealed Radioactive Sources in the European Union, EUR 18186 5. Crumpton, C., Management of Spent Radiation Sources in the European Union: Quantities, Storage, Recycling and Disposal (1996), EUR 16960 6. Mobbs, S.F., Harvey, M.P., Methodology and models used to calculate individual and collective doses from the recycling of metals from the dismantling of nuclear installations, European Commission, Radiation Protection 117 7. Kopsick, D., Potential recycling of scrap metal from nuclear facilities; EPA contract No. 1W- 2603-LTNX; Technical support document prepared for the US Environmental Protection Agency (Sept 2001) 8. Ford, Harrison, Potts, UKAEA Safety Assessment Handbook, UKAEA/SAH (2001), utilizing: a. Morris, B.W., Darby, W.P., Jones, G.P., Radiological Consequence Models for Workers on a Nuclear Plant, AEA/CS/RNUP/47820021/Z/1 (1995) b. Holloway, N., Models for Operator Dose Assessment in Radioactive Material Handling Accidents, SRD/CLM(93) P47 (1993) 9

c. Morris, B.W, Review of In-building Worker Dose Models for use in AEA Safety Cases - Part 1: Inhalation Dose. SDG/TA/Tech Note 93/1 (1993). d. Clarke, R.H., A Model for Short and Medium Range Dispersion of Radionuclides Released to the Atmosphere, NRPB-R91 (1979). e. Cooper, P.J., Underwood, B.Y. Guidance on Calculation of Doses Close to the Release Point Arising from Accidental Atmospheric Releases, SRD/94852110/92/R1 (1992). 9. ICRP, Dose Coefficients for Intakes of Radionuclides by Workers, ICRP Publication 68, 1994 10. European Commission, Recommended radiological protection criteria for the recycling of metals from the dismantling of nuclear installations, Radiation Protection 89 (1998) 10