(last updated 3 Nov 1996)
Radioactive Waste Management Associates
March 30, 1992
(Updated November 14, 1995)
(reproduced here with permission)
Radioactive Waste Management Associates
526 W. 26th Street Rm.517, New York, NY 10001, USA
Tel. +1-212-620-0526, Fax: +1-212-620-0518, E-Mail: firstname.lastname@example.org
> For figures and references, please contact Radioactive Waste Management Associates.
In a series of detailed technical reports for the Atomic Energy Control Board , IEC Beak Consultants Ltd (Beak83) calculated the radiation doses to the local and global population due to one hypothetical Elliot Lake tailings site. In this report for Northwatch, we have used these results to project the dose due to all tailings at Elliot Lake. We then used two estimates of the expected fatal cancers resulting from unit doses to project low and high estimates of the resulting health effects. Following Beak, we have projected doses and consequent health effects both for the local population and for the globe as a whole. Also following Beak, we have broken these out according to the contributions of aquatic releases, radon gas and wind suspension of particulates. The results of these calculations are presented in Tables 1 through 3 and discussed in the Summary section immediately below.
We then provide a comparison of Beak's base case and ours and a derivation and assessment of our scaling factors. This is followed by a description of Beak's modelling of Aquatic Releases and then the derivation of our projections of doses and health impacts resulting from aquatic releases. The next four sections provide a similar analysis of Beak's modelling of atmospheric releases of radon and of particulates and our projections from them. We then offer our assessment of where their model may have omitted plausible chains of events leading to significant releases of radioactive material, indicating that the results we present, although uncertain, represent a low estimate of potential health impacts. Because of high concentrations of iron in Elliot Lake tailings and snow cover during part of the year, the average radon exhalation rate shown here are less than averages presented in UNSCEAR.
In a series of report since 1983, NUTP has improved the Beak model. Modelling of the uranium tailings pile has been refined, additional radionuclides and dose pathways incorporated, and in-growth of daughter products in downstream water bodies accounted for. The effect of these refinements has been to increase the projected doses. However, major episodic events, such as floods, dam breaks, and major erosion, continue to be ignored by the NUTP studies. For long time periods, the probability of such events approaches unity and could cause major environmental impacts. We discuss these events in qualitative terms. Since these subsequent NUTP studies pertain primarily to critical receptors and not to regional and global population groups, we did not quantitatively incorporate the results here.
Finally we discuss the impact of various alternative methods of containing the mill tailings, including both Beak's proposals and our own suggestions for long term containment with the smallest possible health effects. A section is devoted to the economic costs of remediation measures.
2. System definitions and the development of scaling factors
2.1 Assumptions and uncertainties
2.2 Chemical processes
2.3 Physical characteristics of the actual Elliot Lake tailings
2.4 Estimation of Health Effects
3. Aquatic Releases
3.1 Modelling of Aquatic Releases
3.2 Projection of Doses and Health Effects From Aquatic Releases
4. Air-borne releases - Radon
4.1 Modelling of Air-borne Releases - Radon
4.2 Projection of Doses and Health Effects From Radon Releases
5. Air-borne releases - Particulates
5.1 Modelling of Air-borne Releases - Particulates
5.2 Projection of Doses and Health Effects From Particulate Releases
6. Radiation Release Pathways Omitted from Consideration
7. Alternative tailings disposal methods
7.1 Quality assessment
7.2 Health risk assessment
7.3 Cost assessment
In projecting complex events and circumstances far into the future, large uncertainties are unavoidable, and some of these uncertainties are indicated by the range we present for each physical quantity. For the radiation dose commitments, the uncertainty is in large part due to the uncertain physical future of the piles: will they be drained and barren, vegetated or saturated and swampy? We develop estimates for each scenario in the appropriate sections, and in this summary we bracket the possibilities by giving the lowest and the highest results of these separate calculations, as explained in Section 2.3. In addition, as explained in Section 2.4, the range of responsible opinion on how many deaths will result from a given dose gives rise to further variation in expected deaths within each column. Although not presented here, genetic effects and non-fatal cancers are each expected to be of about the same magnitude as the fatal cancers.
Aquatic releases from Elliot Lake tailings result in the projected health effects shown in Table 1-1. The projected health effects are separately listed for residents of the Serpent River Basin and the global community, which includes the entire North Atlantic region. The Serpent River Basin (which includes the North Channel) is assumed to have a constant population of 10,000. As seen, the fatal cancers due to all Elliot Lake tailings, range between 170 and 2200 over a 500-year period. This does not include fatal cancers incurred by uranium miners and other persons directly exposed by walking on the tailings piles. The global health effects range between 1,600 and 22,200, calculated over a 1000-year period, again assuming a constant population density. As discussed below, the range of responsible opinion on how many deaths will result from a given dose is then the source of the variation in expected deaths within each column. Genetic effects and non-fatal cancers are each expected to be about the same magnitude as the fatal cancers.
As we discuss later, because the Beak model does not correctly model the ingrowth of radium due to the decay of thorium downstream, the likely health effects due to aquatic releases are expected to be greater.
Table 1-1. Aquatic Releases from Elliot Lake Uranium Tailings. Projected Radiation Dose Commitments and Health Effects - Base Case Serpent River Basin Global ------------------------------------------------------------------ Radiation Dose Commitment (million person-rems) 0.74 - 0.86 6.8 - 8.4 Fatal Cancers 170 - 2200 1600 - 22,000
RADON, a decay product of radium, is an inert, radioactive gas which escapes the tailings pile. Though radon has a half-life of only 3.8 days, it will distribute itself throughout the Northern Hemisphere. Assuming 4 billion inhabitants, Beak estimates can be projected to give the radiation doses shown in Table 1-2, and from them we estimate the global health effects, primarily fatal lung cancer. Again, the uncertainty presented arises both from including the largest and smallest results from the three scenarios and from different estimates of how many deaths will be produced by a given dose commitment. The first column shows that per year, the number of health effects are quite low, with the saturated case producing very low rates of radon release, while the vegetated case produced the greatest rate. By "saturated" tailings, we mean the tailings are completely covered with water; "vegetated" means grass and other vegetation covers the tailings pile. Beak's calculations only extend for 1000 years, at the end of which all three scenarios indicate continued high levels of radon being released. For the barren and vegetated cases the decline in radon release was exponential, and we integrated this declining rate to obtain the amounts indicated under "Integrated Total". For the saturated case we could not make an estimate, since although low, the release rate was still increasing at the end of 1000 years. Also, all of Beak's calculations assume the piles retain their physical integrity. If erosion or other events allow some redistribution of the material, the releases will be considerably larger. The column titled "Maximum Release" attempts to estimate an upper limit for these possibilities by combining the first year's release rate with the half-life of the parent thorium as described in Section 4. These estimates of millions of deaths would result only if the piles are substantially redistributed.
Table 1-2. Radon Releases from Elliot Lake Uranium Tailings, Projected Global Health Effects Integrated Maximum First Year Total Release -------------------------------------------------------------------------- Radon Released (1 - 110) x 10^3 (84 - 110) x 10^6 1.2 x 10^10 (Curies) Global Radiation 85 - 10,000 (8 - 10) x 10^6 1.1 x 10^9 Dose (person-rems) Global Fatal Cancers 0 - 26 (2.3 - 26) x 10^3 2.8 x 10^6
Airborne radioactive particulates and radon from the tailings pile suspended by the wind produce the dose commitments and health effects to local residents shown in Table 1-3, assuming a constant population of 50,000 persons within 80 km of Elliot Lake and looking at the first 1000 years after creation of the tailings. As discussed below, we include two estimates, one ("Partial Shielding") based on Beak's assumption that radionuclides are largely contained within the tailings pile, and one ("Maximum Release") which assumes that erosion and other effects allow the radionuclides to escape. In addition a modelling parameter, the "small particle enhancement factor", gives rise to substantial uncertainty as shown by including results from both ends of its normal range of values. The overall expectation then ranges from 1 to 2200 fatalities among the local population. The very small numbers arise from the "saturated" case where the radionuclides are trapped beneath water in the piles. If the piles lose their structural integrity during the several hundred-thousand year lifetime of the radionuclides, the effects could be closer to those indicated in the "Maximum Release" column.
Table 1-3. Particulate Releases from Elliot Lake Uranium Tailings: Projected Doses and Health Effects Within 80 Km of Tailings, Over 1000 years. Normal Release Maximum Release SPEF = 1* SPEF = 2.5 SPEF = 1 SPEF = 2.5 --------------------------------------------------------------------------- Radiation Dose (person-rems) 1100 - 72,300 1100 - 120,000 400,000 860,000 Fatal Cancers 1 - 190 1 - 310 1050 2200 * SPEF = small particle enhancement factor; see text.
Comparing the three primary pathways of radioactivity from the uranium mill tailings pile to humans, we note that the health effects within 80 km are somewhat greater from aquatic releases than from wind suspension of particulates. However, when the global health effects are compared, radon produces by far the largest number of health effects.
An important finding of our study is that, for the disposal possibilities considered by Beak, the total dose to Serpent River Basin residents and to the global population is almost independent of the method of disposing of mill tailings, although the time period over which doses are received will depend on the method of tailings disposition. That is, the range of different decommissioning methods considered by Beak and others only serve to spread or narrow the dose over time. We discuss this issue and offer other possible waste disposal methods in the section on Comparative Methods of Tailings Disposition, below.
If one assigns an undiscounted dollar value to human life (the numbers we have seen range from $20,000 to $10 million), the nuclear option becomes quite pricey. But also, the range of remediation methods that are cost-effective is quite large. However, it is imperative that these costs be included in the price of nuclear electricity now, rather than leaving it to future generations to clean up or suffer the consequences.
Two decommissioning methods have been examined. The benefits achieved in terms of reducing hazardous release as well as the difficulties and uncertainties pertaining to those options are summarized below and detailed in Section 7. Comparative costs are also estimated. The two methods considered and compared with the base case, are earth cover and vegetation, and flooding the tailings.
Earth cover and vegetation: The cover could be achieved with various layers. First rock, or peat or any material available locally would provide the necessary isolation; second a layer of top soil would allow vegetation to grow. Flooding: The option presented by Beak consultants refers to the construction of an impervious dam, maintaining the water table above the tailing surface.
Table 1-4. Comparison of Decommissioning Options Objectives Earth cover and vegetating Flooding ------------------------------------------------------------------------ Tailings stabilization and isolation intrusion and misuse prevention can be achieved if cover is at good solution if flooding can be least 10 feet maintained and lake is not used in the future as a recreational area prevention of tailing spreading by wind erosion and surface thick cover necessary. Vegetation wind erosion avoided as long as would protect the earth cover from flooding maintained; before water runoff erosion flooding, clay or bentonite could placed on the tailings in case of lake drain. Surface water runoff will erode the watershed and provide the tailings with a naturally protecting layer of sediments prevent spreading by flood barriers should be constructed can dams and spillways be properly built and maintained ? Control of radon and gamma emission radon earth cover provides control if greatly reduced, if maintained thick and compact enough; but forever veget. might enhance radon release gamma same conditions as for radon control ? Prevention of leachate contamination of surface water or groundwater Since oxygen is necessary for the production of pyrite-generated acid, and oxygen transport is via both diffusion and water percolating through the tailings, the most promising approach to the problem is to minimize those two factors: Oxygen penetration is reduced No more unsaturated zone due to the earth cover. Vegetating enhances transpiration Oxygen penetration is limited due and thus reduces the hydraulic to reduced diffusion of O2 in input by 32%. water, thus pyrite oxidation is greatly reduced.
An important finding of our study is that the total dose to Serpent River Basin residents and to the global population is almost independent of the method of disposing of mill tailings, although the time period over which doses are received will depend on the method of tailings disposition. That is, the range of different decommissioning methods considered by Beak and others only serve to spread or narrow the dose over time. We discuss this issue and offer other possible waste disposal methods in the section on Comparative Methods of Tailings Disposition, below.
Table 1-5. Collective dose commitment for different management options for the Serpent River area t integration 1,000 5,000 10,000 time (yrs) --------------------------------------------------- Base case 610,463 * * Cover +Veget. 599,513 * * Flooding 77,471 514,650 692,588 * no change from previous value; t - radiation dose in person-rems
The study by IEC Beak Consultants was selected because it is the most comprehensive we have seen on the subject. The Beak studies calculate radiation doses to the local, regional and global populations. Later studies by NUTP, while refining and improving the model, focus on critical receptors and are therefore not useful for our purpose of calculating the total health effects due to Elliot Lake tailings. Similar studies are not available in the U.S. since uranium mining takes place in the arid Southwest, while in the Elliot Lake region, the role of water in the dispersal of radioactive material is paramount. The study consists of a summary (Beak83) and two detailed technical appendices, one studying waterborne releases (Beak83W) and one studying atmospheric releases (Beak83A), including both radon and wind-borne particulates. Where possible we provide summaries of the mechanisms considered and models used in Beak's study, but it is impossible and inefficient to reproduce it in detail, so detailed verification of our calculations will require that the reader have access to these studies.
IEC Beak Consultants are quite cautious about the predictions of their model, pointing out (Beak83W, p. i-ii) that many unlikely assumptions (such as constant population density or the absence of major changes in the pile's confinement) and many areas of scientific uncertainty (the lack of empirical verification of many of the simulation model's predictions) combine to create a situation involving substantial uncertainty. In addition, their study was done for one "generic" North Ontario tailings pile, and clearly a direct study of the existing piles would be much more desirable. Desirable as more studies and greater certainty might be, however, decisions must be made now, and whatever the shortcomings of these studies and our extrapolations of them, they are the best information currently available. The only safe alternative is to ignore all modelling and assume the entire inventory of radioactive nuclides in the tailings pile escapes to the larger environment and that people are exposed to it. This would certainly give a maximum possible damage estimate, and would result in health effects far in excess of those we present here, but this scenario is so unlikely to occur that using such estimates would constitute an unreasonably pessimistic assessment of the situation. Accordingly, we present our extrapolations from Beak's work as the best currently available estimates, with the caveat that substantial inaccuracies could turn up and in the hope that funds will be found to do the validation field studies and model refinements that are so sorely needed.
The results by Beak used initial formulations of the UTAP computer model which was subsequently improved1, particularly with regard to radioactive releases to water. Under Phase II of the NUTP program, the quasi-steady-state model was altered to incorporate time varying behavior of the natural system, and thorium, calcium, sulphate, iron and pH were explicitly modelled. Under Phase III, the simulation model was expanded to include major radionuclides and water erosion of tailings cover materials, additional pathways, further distance for air dispersion and water dispersion over the entire watershed.
The Beak model did not calculate the in-growth of radium-226 from the decay of thorium-230 in receiving water bodies. As seen in Fig. 2-1, radium-226 concentrations decline to zero between 405 and 505 years. In reality, thorium-230 continues to generate radium-226, and the concentration continue to rise with time, as shown in Fig. 2-2. The effect is the radiation doses to persons due to aquatic releases is greatly underestimated by Beak.
Another major omission by Beak and subsequent NUTP studies is the exclusion of episodic events, such as fires, earthquakes, dam breaks, droughts and major floods. Over the long time periods over which the tailings remain hazardous, such events are likely and could have major consequences. In addition, erosion is poorly modelled, using the Universal Soil Loss Equation. Our physical examination of Elliot Lake tailings shows that acidic conditions are causing concrete spillways and sluices to deteriorate rapidly, making future dam breaks and major erosion a realistic possibility. Without inclusion of major erosion probabilities and consequences in the model, projected doses are greatly underestimated. Nevertheless, as we mentioned earlier, because the Beak model more comprehensively examines health effects for all affected population groups, we have elected to use the model here.
The mobility of those radionuclides when present in a soluble form or their immobilization when precipitated or adsorbed command their potential release in the environment at levels of concentrations of eventual public health concern. It is therefore of extreme importance to understand the basic mechanisms of this radiochemistry in order first to assess the extent of the hazard to which the public might be exposed and second to identify the chemical steps of strategical interest when thinking of impeding this hazard.
We will describe the two sequences of events during which radionuclides undergo equally relevant chemical transformations: first during the milling process when uranium is extracted from its host ore and second after deposition of the tailings slurry in a natural basin. At each step, to the extent of the current knowledge, we will try to identify the species present, their chemical form and their relative concentration.
This succession of events described below and the subsequent dose of radioactivity that will be projected assume that there is no massive erosion of the tailings pile, or failure of the dam or surrounding elevated land. Also, no vegetation grows on the pile. It is referred to, in the Beak study, as the 'base case'.
Radioisotopes still solubilized are carried with the leach liquor to the final neutralization tank where they precipitate out of the solution. At this point the solid waste (tailings) is separated from the solution and the uranium is removed from the clarified solution by IX. The barren solution from which tailings and uranium have been stripped, is then mixed with the tailings to produce a residual slurry which is neutralized with lime to a pH of 8 to 11 and pumped to a tailings basin. This pH neutralization results in the formation of more gypsum which coprecipitates with all the remaining dissolved thorium-230, lead-210 and polonium-210 and about 50% of the radium-226 present in the leach liquor.
At Elliot Lake, the tailings are sand-like in consistency and contain a large quantity of pyrite (4 to 7% by weight) which oxidizes in the oxygen-rich unsaturated zone (i.e above the water table) of the tailings. The formation of acid and iron sulphates occurs by the following chemical steps:
FeS2 + 3O2 + 2H2O --> Fe2+ + 2SO42- + 4H+ pyrite 2FeS2 + 2H2O + 7O2 + 2CaCO3 --> 2FeSO4 + 2CaSO4.2H2O + 2CO2 pyrite ferrous gypsum sulphate 4FeSO4 + O2 + 2H2SO4 --> 2Fe2(SO4)3 + 2H2O sulphuric ferric acid sulphate Fe2(SO4)3 + 6H2O --> 2Fe(OH)3 + 3H2SO4 ferric sulphuric hydroxide acid (geothite)
Then the tailings pile undergoes a complex chemical evolution. In effect, the tailings porewater that occurs deep within the tailings impoundments has a chemical composition similar to that of the original neutralized liquid effluent that was discharged from the mill at the time of tailings deposition. This so called 'process water' not yet substantially altered by chemical processes in the tailings, has a neutral pH, low concentrations of iron and heavy metal, and moderate concentrations of sulfate. But the tailing pore water that has already recharged into the tailings since tailings deposition is affected by pyrite oxidation in the unsaturated zone. This 'recharged water' is characterized by low pH, and high concentrations of iron, sulphate, heavy metals and radionuclides. Consequently, chemical profiles in the tailings indicate the presence of a lower zone of original pore water and an upper zone of recharged pore water with an abrupt interface between the two zone. The recharged water from rain and snow is gradually replacing the original pore water. A significant percentage of residual calcite in the tailings prevents much of the zone of recharged water from becoming strongly acidic as a result of pyrite oxidation but it appears that as the calcite is consumed or becomes non-reactive due to coating with iron hydroxide, the zone of acidic pore water will gradually deepen and that it will eventually occupy all the tailings pile. This chemical evolution and the downward movement of the acidic front command the release of radionuclides and heavy metals.
Computer models of the tailings pile distinguish four distinct phases. This is shown in Fig. 2-3.
PHASE 1: basic leaching; tailings are still neutral but becoming more acidic. 160 yrs
The production of hydrogen ions (H+) causes a progressive decline of the pH of the upper zone pore water. As this acidic infiltrating water moves further down it attains a basic pH, being neutralized by the excess lime added during the neutralization steps. As water infiltrates, it also dissolves existing gypsum but gypsum dissolution proceeds slowly being impeded by the precipitation of additional gypsum in the zone between lower and higher pH waters where calcium and sulphate ion dissolved from solid calcium and oxidized pyrite are present. Consequently, radionuclides trapped in the gypsum are released very slowly but then adsorb reversibly with new minerals in the tailings. Over a period of a few months following deposition, radium-226 shows a decreased activity in the tailings porewater probably due to strong adsorption coefficient when pH is still high or coprecipitation of radium onto tailings solids, such as newly formed gypsum. A key parameter is the concentration of sulphate ions, since radium will co-precipitate with barium suplhate. At a pH of 7 to 8.5, thorium remains and/or co-precipitates as a hydroxide (Th(OH)4) in tailings.
In the saturated zone where basic conditions prevail, jarosite is transformed into geothite, which has the net effect to immobilize the radionuclides present in the pore water.
PHASE 2: acid leaching, pyrite is being consumed, tailings are acid. 35 yrs.
When the excess lime is totally neutralized, the tailing pore water becomes acidic (pH 2) and gypsum dissolution is no more slowed by the formation of additional gypsum due to depletion of soluble calcium. Dissolution of gypsum occurring during those acidic conditions, release sulfates and more radionuclides which, with the exception of thorium, adsorb irreversibly to the tailings. At the same time pyrite oxidation continues to produce sulfate until pyrite depletion leaves gypsum dissolution as nearly the only source of sulfate.
At a pH below 4.5 thorium hydroxide is assumed to solubilize as a thorium sulfate complex and be released from the tailings pile in a short time pulse (Beak estimates one year). It will reprecipitate when entering environment with a higher pH such as a receiving lake. Then almost all the thorium will settle to lake sediments. Thorium in the tailings pile or lake sediments will decay to radium, under the chemical conditions a much more soluble radionuclide than thorium.
Geothite formed from jerosite in saturated layers is now unstable when pore water becomes acidic and changes back to jerosite with a consequent release of radium-226.
PHASE 3: all the pyrite has been oxidized, maximum gypsum dissolution 200 yrs.
The pH returns to a 6-7 range and dissolved gypsum saturates pore water. More radionuclides dissolve at the gypsum dissolution rate and adsorb reversibly to tailings. Jarosite continues to transform to geothite, but this geothite and its radium remains immobilized since tailings never reach sufficiently acidic conditions again.
PHASE 4: No more gypsum
Uranium, thorium, radium, polonium and lead migrate only at the rate of the diluting radionuclides adsorbed to tailings surfaces; no more nuclides arise from gypsum dissolution since no more gypsum is present.
The chemical process summarized above is common and relevant to both water-borne and air-borne releases from the pile.
The actual Elliot Lake tailings masses are characterized in Table 2-1, taken from an AECL report . The total volume derived in Table 4 is 1.15 x 108 m3, giving a volumetric scaling factor of 33.8, while the total area is 1095 hectares, giving an areal scaling factor of 27.4. The mass of tailings is not well known, since they are delivered to the lake by pipeline and are not valuable. If we assume the (dry weight) density is the same as Beak's (1240 kg/m3), the total dry tailings mass at Elliot Lake would be 142 million tonnes. The authors of the data in Table 2-1 used a density of 1350 kg/m3, which would give a total mass of 156 million tonnes. Where needed, we will use the average, 149 million tonnes, where the five per cent uncertainty is quite small compared with that of many other quantities used in these studies.
Table 2-1. Deposits of Uranium Mine Tailings at Elliot Lake a Location Volumeb Surface Date of Current Status (10^6 m^3) Area (ha) Operations ------------------------------------------------------------------- Lacnor 2.00 24 1957-60 Mainly re-vegetated; portion under water. Nordic 8.10 101 1957-68 Re-vegetated; water collection & treatment. Stanrock- 4.20 52 1958-64 One dam failed; some Can-Met water treatment; some re-vegetation. Spanish- 0.33 5 1958-59 Re-vegetated or under American water. Pronto 1.55 47 1955-66 Re-vegetated. Millikan- 5.6 33 1958-64 Re-activated. Stanleigh Stanleigh 8.2c 400 1983-96 Operating Quirk 31.1c,d 192 1956-61, 1968-90 Operating Panel 11.9c,d 123 1958-61, 1979-90 Operating Denison 46.9c,d 280 1957-92 Operating ------- ------ TOTALS: 120 1257 a. From Atomic Energy of Canada, Ltd., Inventory of Waste Quantities: Report to the Siting Process Task Force, Energy, Mines and Resources, Canada, Ottawa, ON, 1987, Rio Algom Ltd, `EIS with respect to Decommissioning the Quirke and Panel Waste Management Areas' and Denison Mines Ltd, `EIS, Decommissioning of the Denison and Stanrock Tailings Management Areas,' 1995. b. Some volumes estimated from weight data using 1350 kg/m3 in Ref a. c. Volume as of the end of 1985. d. Includes waste rock mass.
In projecting future impacts the state of the tailings piles is of course of prime importance. Beak considered three significant cases: a "Base Case" with a porous dam and no significant vegetation, a "Vegetated Case" where the surface of the tailings, still held by a porous dam, became covered with self-sustaining native vegetation and finally, a "Low Permeability Dam Case", where a dam having permeability comparable to bedrock keeps the entire pile saturated with nearly stagnant water. The "Vegetated Case" results in near zero dust losses and greater radon losses compared to the "Base Case"; the saturation from the "Impermeable Dam" slows chemical processes and migration of radionuclides through the water table. In the sections that follow, we present tabulations of our extrapolations and calculated health effects for all three cases, but it is also prudent to ask what relation they bear to the actual future of the Elliot Lake tailings piles. One hint from Table 2-1 is that the piles which have been closed for some years all have at least some natural vegetation. Another factor is the degree of expected saturation, and Table 2-1 indicates that wide variability will be the rule here, with one dam having failed (guaranteeing a low water table and excessive dispersal of radioactive silt) while other areas have poor drainage and are apparently "naturally" in a saturated condition. The historical record presented in Table 2-1 indicates that the future of Elliot Lake can certainly be expected to incorporate elements of all three scenarios, and that our estimates for future health effects should be bounded by the largest and smallest numbers obtained from the three scenarios. Accordingly, the results for each scenario are developed in the appropriate section below, while the summary tables in Section 1 display the largest and the smallest values obtained when the three tables produced for each case are compared.
Beak consultants included in their study three cases (Cases 4,5 and 6) in which they assumed that 90% of the leachable Radium and Thorium were removed from the tailings before discharge from the mill. These cases are irrelevant to our study, since we are examining the impact of existing piles for which no attempt at removal was made.
The presence of one broken dam (Stanroc) indicates, however, that all of Beak Associates's estimates may be conservative with respect to emissions if large amounts of the tailings break free from the piles and are swept downstream. We will deal with this possibility in more detail in Section 6, but the total radioactivity of the pile is in this respect a quantity worth knowing. Beak gives radioisotope concentrations for their prototypical tailings pile (Beak83W, Table 2.3) which we incorporate into Table 2-2. For each isotope we show the atomic mass, half-life, the number of short-lived daughters not explicitly listed (CRC80 - CRC Handbook of Chemistry and Physics, Weast ed. 1980) and then Beak's estimates of the grams per tonne of tailings for each nuclide present in the host mineral, in the gypsum or jarosite and in the pore water. Beak derived the concentrations from the assumption of secular equilibrium in the freshly mined ore and the subsequent disruptions described in their model of the milling process. A standard calculation gives the activities shown in the right column, which sum over the two decay chains to give a total activity of 3.9 x 10-3 Ci per tonne. Thus to the extent that Beak's characterization of Elliot Lake tailings is accurate, they contain approximately 554,000 Ci of long-lived radionuclides; 90% of this activity is due to the U-238 series.
Table 2-2: Activity of Tailings Pile Nuclide Half- Number ---- Concentration in ---- Activity life of fast Mineral Gypsum/ Pore Daughters Jarosite Water (microCi (years) --- (g/tonne) --- (g/m3) /tonne) ------------------------------------------------------------------------- U 238 4.46E+09 2 53 21 1 76 U 234 245000 2.8E-03 1.1E-03 5.5E-05 25 Th 230 75400 8.6E-04 1.7E-02 5E-05 359 Ra 226 1600 5 1.7E-05 3.2E-04 1E-06 2019 Pb 210 22.3 1 2.2E-07 4.2E-06 1.2E-08 677 Po 210 0.379 3.8E-09 7.2E-08 2.2E-10 342 End series Th 232 1.4E+10 190 178 1.2 41 Ra 228 5.75 1 7.5E-08 7.2E-08 1.6E-10 80 Th 228 1.91 6 2.5E-08 2.4E-08 5E-10 284 End series ------ Total activity: 3902
An alternate estimate of radioactivity in Elliot Lake tailings shows that 1 tonne of tailings has a radioactivity of at least 1.85 x 108 Bq, or 5.0 x 10-3 Ci of radioactivity. Thus, Elliot Lake contains approximately 710,000 Curies of long-lived radionuclides. About 90% of this activity is due to the U-238 series.
The processes by which these radionuclides are dispersed and the amounts to which the various populations are exposed are discussed in the individual sections to follow.
Slightly tighter bounds could be obtained using the "90% confidence limits" of a recent National Research Council study, which found risks of 500 to 1200 cancer deaths per million person-Rems for males and 600 to 1200 for females (BEIR V, 1990, p.6). The "confidence limits", however, represent the Council's statistical limits according to their analysis, while Gofman's estimate represents a different opinion of how the coefficient should be obtained from available data. This field is in a state of flux, with recent upward revision of human sensitivity to radiation included in the studies by Gofman and the National Research Council, and we feel that using Gofman's estimate at the high end reflects genuine uncertainty among experts as to the risks of low level ionizing radiation.
In several cases where our calculations indicate a large number of fatal cancers can be expected the individual doses are quite low. Some may argue that "threshold effects" make such doses harmless to individuals, but with the International Commission on Radiation Protection, the National Research Council (BEIR V, 1990, p.4) and other regulatory bodies we reject these assertions as unproven hypotheses, unsatisfactory as a basis for policy decisions affecting millions of future lives. In particular, we ignore the various "dose cut-off" cases developed by Beak (Beak83A, Section 4.2) and use only data developed with zero dose rate cut-off.
The proportions of each radionuclide which are taken up from the water by sediments and by plants, fish and animals are highly uncertain. In part this is because the uptake is assumed proportional to concentration, with the proportionality constants (the "Kd's "for sediments, the "concentration factors" and "transfer coefficients" for various exposed and edible organisms) obtained experimentally under laboratory conditions or in very different environments. The Beak report itself stresses the great uncertainty involved in applying these coefficients in a new and complex environment, and calls for field validation of the modelling. Nonetheless, given the radioactive concentrations, the food consumption patterns and lifestyles of Serpent River Basin residents and regional and global populations are taken into account to finally estimate the total radioactivity ingested in a particular year in what is called the "Population Exposure Model".
Once the distribution of radioactive materials has been calculated, the "Dose Commitment Model" is used to calculate the actual radiation doses received by the population under analysis. This changes from case to case: sometimes it is only local residents, the "critical population" who receive the largest exposure. In other cases the entire population of the Northern Hemisphere will be examined. In addition to ingestion, the direct radiation dose from walking along Great Lakes beaches and swimming is also accounted for.
The quantity calculated in the Dose Commitment Model is called the "incomplete collective effective dose equivalent commitment" for the population in question. The "effective dose equivalent" part has to do with accounting for differences in sensitivity of different human tissues and organs to radiation to derive an overall dose for the individual. "Collective" simply refers to summing over the exposed population. The idea of a "dose commitment" is to integrate the expected rate of exposure over future time to determine the total "person-rems" of exposure that will occur as a result of the continued (albeit slowing) release of radionuclides. This quantity is called "incomplete" if the calculation is terminated at some finite future time, which may occur either for technical reasons, because predictions based on events so far in the future are regarded as not meaningful or because the individual doses have become very small. We do not always accept Beak's truncation of the integration, and will discuss each case in detail when we vary from their procedure.
These results are then scaled up to include the effects of all the tailings in the Elliot Lake area for the three relevant scenarios as described in the following section.
As derived from Table 2.1, the volumetric scaling factor is 33.8 from Beak's generic tailings pile up to the actual volume of tailings in the Elliot Lake area. Taking into account the comparable depths of the actual piles and Beak's generic pile, we assume that the amount of radioactive nuclides released to the water system will simply scale with the volume of the piles, and accept Beak's estimates for the number of persons exposed to these nuclides. Thus the radiation dose in person-rems, shown in the top lines of Tables 6-8, are simply scaled up by a factor of 33.8 from Beak's numbers.
We look at two population groups, the "local" population of the Serpent River Basin, which includes the North Channel but not the Great Lakes in general, and the "Global" population, which is exposed as the waters of the Great Lakes are swept out to sea and taken up by the global fish catch and other pathways. The local population is assumed to be a constant 17,000 (including averaged seasonal residents).
For the "base case" Beak's Table 6.1 shows that the dose commitment is established after an integration time of 500 years. That is, the dose commitment does not continue to rise, indicating that leachable material has left the tailings pile and has already exposed the local population. Over the next 500 years the global dose rises by about 1.5%, indicating that some of the nuclides had not yet been distributed to the oceans, but an integration time of 1000 years clearly includes all the dose commitment to be expected under this scenario. Accordingly, the "Global" and "Local Watershed" entries for case 1 of Table 6.1c (BEAKW83) are scaled up by 33.8 to give the top row of numbers in Table 3-1.
As pointed out earlier, Beak's original analysis does not account for ingrowth of radium-226 from thorium-230 in downstream water bodies. As a result, the health effect calculations have a sharp cut-off at 500 years. Total doses due to water pathways do not rise after 500 or 1000 years, depending on the proximity of the water body considered. However, since thorium-230 in lake sediments continually give rise to radium-226, which goes into solution in lake bodies, radiation doses due to this water pathway must continue for tens of thousands of years into the future. Thus, the Beak numbers are serious underestimates.
Beak made no predictions of human health effects resulting from these exposures. Following the discussion in Section 2-4, we have calculated a high and low estimate of the number of fatal cancers to be expected under each scenario and for the local and global populations. Again, the wide range of possible fatal cancers reflects the uncertainty in the radiation health community on the potential of low levels of radiation to induce cancers in humans. The total number of cancers would be about twice the number of fatal cancers.
Table 3-1: Aquatic Releases from Elliot Lake Uranium Tailings. Projected Radiation Dose 1000 Year Commitments and Health Effects - Unsaturated, Unvegetated Piles ("Base Case"). Serpent River Basin Global ---------------------------------------------------------- Radiation Dose Commitment (person-rems) 754,000 7,000,000 Fatal Cancers 170 - 2000 1600 - 18,000
We now work up the same set of numbers for Beak's Case 2, where a vegetated cover provides stability and prevents erosion. For aquatic releases this cover has little effect and again using a 1000 year integration period we find slightly lower numbers of fatal cancers resulting, shown in Table 3-2.
Table 3-2: Aquatic Releases from Elliot Lake Uranium Tailings. Projected Radiation Dose 1000 Year Commitments and Health Effects - Vegetated Cover, Unsaturated Pile. Serpent River Basin Global ---------------------------------------------------------- Radiation Dose Commitment 740,000 6,800,000 (person-rems) Fatal Cancers 170 - 1900 1600 - 18,000
The case of the saturated pile, studied in Beak's case with an impermeable dam, is more uncertain. If the same 1000 year integration period is used, the doses and cancer fatalities appear to be greatly reduced. However, examination of Beaks' Figures 6.15 and 6.16 reveals that what has happened is that the dose has been postponed and slowed, but that it still gets out eventually, reaching full release after about 6200 years. Using the 10,000 year integration period, from Table 6.1e, gives results, presented in Table 3-3, that are actually a little worse than the base and vegetated cases.
Table 3-3: Aquatic Releases from Elliot Lake Uranium Tailings. Projected Radiation Dose 1000 year Commitments and Health Effects - Saturated Pile (Impermeable Dam). Serpent River Basin Global ---------------------------------------------------------- Radiation Dose Commitment 855,000 8,450,000 (person-rems) Fatal Cancers 200 - 2200 2000 - 22,000
These results make clear that for the aquatic releases, there is really no significant difference between the three relevant cases, especially when compared to the uncertainty in the sensitivity of the human organism to minute doses of radioactivity.
Most of the radon is born within the mineral matrix comprising the bulk of the tailings. It must diffuse out through the mineral to pores in the tailings pile, then diffuse upward through the water films and air of the unsaturated pores or the water of pores in saturated areas to reach the atmosphere, where it is readily mixed and spreads through the northern hemisphere in a matter of weeks. The calculations representing this complex behavior are based on mass balance equations for the radon in and out of each of the various media, finally giving rise to a surface flux from the tailings pile. The results are clearly dependent on a large number of parameters such as diffusion coefficients, void fractions, emanation coefficients and so on.
One result of the evolution of the tailings pile is that as the radium is washed deeper into the pile, the radon is born deeper below the surface and less of it reaches the atmosphere before decaying. The result is a continuous decrease in the surface flux of radon from the pile. Of course this result depends on the stability of the tailings piles, and if dams were to fail, allowing the piles to be washed downstream and dispersed, far more radon would be emitted. We examine the consequences of this possibility below.
BEAK modeled regional exposures to the released radon using the UDAD model developed by Argonne National Laboratory (ANL79). This complex model is key to the results for local atmospheric releases discussed in Section 5; for radon, it is used only to predict regional exposures, and since these are small compared to global exposures, the regional exposures are not very significant in the total health effects due to radon releases and are omitted from the tables in this section. (They are described and included in Section 5 on local exposures through atmospheric pathways.) For the global exposures, the method is much simpler: the emitted radon is simply rapidly mixed into the entire northern hemisphere atmosphere, and the resulting dose to the four billion hypothesized inhabitants is calculated over the next thousand years.
There is a serious underestimate of the doses and cancers due to radon in Beak's assessment. The problem is that their model was only run for the first thousand years, and in fact radon will continue to emanate from the pile for thousands of years thereafter. We attempt to fill in this analytic gap in two ways: first, where possible, we extrapolate their results out over much longer time intervals. Second, we develop a simple model giving an upper limit on the possible radon flux and attendant cancers and present it as an independent estimate. These calculations are detailed in the following section.
In the second line doses are calculated using Beak's estimated dose conversion factor of 0.09 person-rems/Ci released (Beak83A, p. 3-14). This is based on complete dispersion of the radon during a three to four week global air mass circulation period so it can easily reach the 4 billion inhabitants of the Northern Hemisphere; according to Beak, the factor includes all radiation pathways. Finally, the third line gives the cancers expected from our two estimates of the potency of ionizing radiation.
For all three cases, Beak's Table 2.5-1 and Figure 2.3-1 both indicate that substantial quantities of radon are still being emitted at the end of the 1000 year integration period, and it is for this reason that the doses are referred to as "incomplete". We nonetheless tabulate these releases, incomplete doses and expected health effects in the column titled "1000-year Total"; these estimates, for all three cases, are based firmly on Beak's modelling.
But clearly, much radon will be released after the end of the 1000 year integration period. Beak's Table 2.5-1 and Figure 2.3-1 indicate that as material is washed deeper into the pile, the radon flux declines with time for the "base" case and the "vegetated" case in a way that is approximately exponential after the first 300 years. (This "exponential" decay is based in the chemical evolution of the pile and has nothing to do with the radioactive decay of the radon precursors, which is a much slower process.) Assuming that the radon flux continues to decline exponentially, it is a simple matter to fit an exponential and integrate it over all time to obtain an expected total emission, a "complete" global collective dose and the resultant fatal cancers. The results are all tabulated in the "Integrated Total" column of Tables 4-1 and 4-2. These are almost certainly minimal estimates, since there is no fundamental reason to expect the release of radon to continue to slow down at the same rate.
Table 4-1: Radon Releases from Elliot Lake Uranium Tailings, Projected Global Health Effects - Unsaturated, Unvegetated Piles ("Base Case"). 1000-Year Integrated First Year Total Total ----------------------------------------------------------------------- Radon Released (Curies) 8.9x10^4 4.0x10^7 8.4x10^7 Global Radiation 8.0x10^3 3.6x10^6 7.6x10^6 Dose (person-rems) Global Fatal Cancers 2 - 21 820 - 9300 1700 - 20,000
Table 4-2: Radon Releases from Elliot Lake Uranium Tailings, Projected Global Health Effects - Vegetated Cover, Unsaturated Pile. 1000-Year Integrated First Year Total Total ----------------------------------------------------------------------- Radon Released (Curies) 1.11x10^5 5.5x10^7 1.1x10^8 Global Radiation 1.0x10^4 4.9x10^6 1.0x10^7 Dose (person-rems) Global Fatal Cancers 2 - 26 1100 - 13,000 2300 - 26,000
For the "saturated" case, the initial release is much smaller, since the diffusion of radon is much slower through the water, but it actually increases as time goes by, and is still increasing at the end of the 1000 year period covered by the simulation. The mechanisms leading to this are obscured in the modelling, making it impossible to extrapolate this case out past the 1000 year limit.
Table 4-3: Radon Releases from Elliot Lake Uranium Tailings, Projected Global Health Effects - Saturated Pile (Impermeable Dam). 1000-Year Integrated First Year Total Total ----------------------------------------------------------------------- Radon Released (Curies) 9.4x10^2 1.1x10^6 ? Global Radiation 85 9.7x10^4 ? Dose (person-rems) Global Fatal Cancers 0 - 0.3 22 - 360 ?
Table 4-1, 2 and 3 are troubling because they are based on the assumption that the piles will retain their physical integrity forever, and Table 2-1 reveals that one dam has broken already. We saw in Section 2-3 that the Elliot Lake piles will contain about 554,000 Ci of radioactivity, much of it Thorium and Radium which will eventually release Radon. Prudence demands that we make another estimate, one consistent with a less certain future for the physical containment. The proximate source of the radon is the thorium-230 left in the tailings by the extraction process. Clearly this thorium will decay with a half-life of 7.54 x 104 years. If we assume that radon exhalation begins in the first year at the rate predicted by Beak and used above, but that thereafter it declines in accord with the exponential decay of the thorium, we will be effecting a compromise: we are not assuming, as Beak does, that the radionuclides are "naturally" hidden away more and more effectively as years go by, but we are not assuming that the dams break and the tailings get strewn out in a completely exposed configuration, either. The results are presented in Table 4-4 for each of the three scenarios. The assumed situation can be described as one where unanticipated erosion maintains the pile in approximately its initial geochemical state with respect to the ease with which radon can escape. The results are presented in Table 4-4 for each of the three scenarios.
It is important to add that the effects of wind erosion of particulates on radon release need to be included. Wind is expected to distribute tailings particulates within an 80 km radium of the tailings pile. These particulates would include uranium, thorium and radium, all precursors of radon gas. Thus, the emission of radon is not solely a function of pile dynamics, but also a function of tailings dispersal.
Table 4-4: Radon Releases from Elliot Lake Uranium Tailings, Projected Global Health Effects - Maximum Release. "Base Case" Vegetated, Saturated (Unsaturated, Unsaturated (Case 3: unvegetated) (Case 2) "Impermeable Dam") ------------------------------------------------------------------------ Radon Released (Curies) 9.7x10^9 1.2x10^10 1.0x10^8 Global Radiation 8.7x10^8 1.1x10^9 9.2x10^6 Dose (person-rems) Global Fatal Cancers (.20 - 3.2)x10^6 (.25 - 4.0)x10^6 2100 - 34,000
The question of including deaths numbering in the millions but extended over many millennia is a source of great controversy. The arguments for truncating radiation doses in time are the following:
But AECB consultants have argued that radiation doses should neither be truncated in space nor time . The ICRP has argued similarly . We also have not truncated radon radiation doses in space or time. Despite the recommendations of its consultants and the ICRP, in granting licenses to uranium mines, the AECB focuses only on short-term doses to critical receptors and does not consider global radiation doses. Environmental Impact Statements recently prepared for mining in Saskatchewan does not even mention global radiation doses or total radiation dose commitments due to radon releases.
We remind the reader we have still omitted from consideration all the radon produced by the 7% of uranium remaining in the tailings.
For the unvegetated, unsaturated ("base") case, considerable wind-driven suspension of particulates takes place, modelled by a method due to Travis which incorporates effects of wind speed, surface moisture, particle size and so on. As a result of the removal of fine particles by the wind over the first few decades, the erosion rate slows, and in the end only material in the top 20 centimeters of the piles is disturbed. Snow cover prevents erosion during winter, so the results are scaled down by 33% (Beak83A, p.2-19). The authors do note that due to the omission of a "small particle enhancement factor", the results may be underestimated by a factor 2.0 to 2.5 (Beak83A, p. 2-18). Finally, both the wind erosion and radon evolution models incorporate the model of geochemical dynamics of the tailings pile developed in the aquatic analysis (Beak83W). Essentially, the oxidation of pyrite produces a strong acid, which together with rain water which dissolves gypsum, washes radionuclides further down into the tailings pile. The result is that suspended particulates become less radioactive over time, and releases by this mode are greatly reduced after about 100 years.
The mathematical models employed (Beak83A, p. 2-14 ff.) to calculate both the wind erosion and its subsequent atmospheric transport is the Uranium Dispersion and Dosimetry (UDAD) model, a computer code developed by the US Nuclear Regulatory Commission (NRC79). This is a "Gaussian plume" model which describes the transport of particulates away from the source (the tailings pile) as settle, disperse and are deposited on land and water. Only the 80 km region surrounding Elliot Lake, within which all particulates fall, is modelled. Using the 1981 Census, the specific local population is factored in, including the influx of summer residents. This population is estimated to be about 50,000 persons, including 17,250 at Elliot Lake. Due to a fall of in uranium mining operations, the local population may have declined since 1981.
While the particulates are suspended, people can breath them in and have them deposited in the lung tissue, giving rise to a potent internal exposure through inhalation. In addition, the particulates are deposited on the earth's surface and on plants. Some are temporarily resuspended. When deposited on plants or soil, they may be incorporated into plants, which may in turn by eaten by humans, fish or animals. Fish or animals may be eaten by humans. Thus, humans may also be exposed to radiation internally by ingestion of contaminated plant, animal or water, or may be exposed externally, due to air immersion and exposure to radiation from particulates deposited on the ground. All these calculations were repeated both for a hypothetical, most severely exposed individual, assumed to be living on Quirke Lake, and for the total population of 50,000 described above, the results we will make use of in the next section.
Radon emissions occur for all three cases as described in Section 4 just above; its distribution over the local area is included in the UDAD modelling and contributes about one-half of the local doses in the unvegetated, unsaturated ("base") case. For the vegetated and saturated cases, the absence of particulate suspension leads radon and its daughters to be the only contributors to local doses. The results of our projections are given in the next section.
Beak's calculations for local cumulative collective doses are repeated for several "cutoffs", where they (implicitly) argue that below some thresh-hold there are no health effects. As discussed in Section 2.4, we reject this notion and use only data for the case of no cut off level, or "cutoff = 0 rem/year", which at 2640 rem (Beak83A, Table 4.2-1a) scales so give the dose in the left- hand column of our Table 5-1 just below.
Beak's data go out for 1000 years, and it is clear from their tables that this is an "incomplete" dose (that is, the dose is still rising at the end of the 1000 year period). However, numerical analysis of their tables indicates that the complete collective dose is either less or much less than 10% greater than their 1000 year dose (depending on how the curve-fitting is done), so we will use those values as an adequate approximation to the complete dose.
As discussed above, Beak set the "small particle enhancement factor" ("SPEF") equal to one in their calculations, while saying the results could easily be 2.5 times larger had this effect been included (Beak83A, p. 2-18). We bring this factor back in in the second column of Table 5-1, but we cannot simply scale the previous results up by 2.5, since the SPEF only affects particulates, and not the flux of radon from the pile. We therefore take the cumulative collective dose and subtract out the dose due to Rn-222 and its daughters Pb-210 and Po-210, scale the resultant up by 2.5 and add the dose due to radon and daughters back in. The areal scaling factor then gives the result in the second column of Table 5-1.
Table 5-1. Particulate and Radon Releases from Elliot Lake Uranium Tailings: Projected Doses and Health Effects Within 80 Km of Tailings, Over 1000 years. Unvegetated, Unsaturated ("base case") Piles. ------ Normal Release --- ----- Maximum Release ----- SPEF = 1.0 SPEF = 2.5 SPEF = 1.0 SPEF = 2.5 -------------------------------------------------------------------------- Radiation Dose (person-rems) 72,300 120,000 400,000 860,000 Fatal Cancers 17 - 190 28 - 310 93 - 1050 200 - 2200 SPEF = "Small particle enhancement factor"; see text.
Finally, we consider the possibility that broken dams, floods or other catastrophic events make Beak's assumption that radionuclides migrate to isolated locations deep within the pile inappropriate. If such events occurred, spreading the tailings over a large area, the releases seen during the first five years would be more appropriate for the long term activity of the tailings. We show in columns three and four, labelled "Maximum Release", the 1000 year incomplete collective doses derived by multiplying the 5-year dose (Beak83A, Table 4.2-1) by 1000/5 and by the areal scaling factor for both values of the SPEF. Note that the SPEF has a larger effect here, since particulates are a larger fraction of the dose if the surface is fresher.
Since the vegetated and saturated cases give rise to little or no particulate suspension, they also give rise to few or no cancers or other health effects attributable to particulates. Radon is evolved in the vegetative cover case (Beak83A, Table 4.3-1a). The cumulative collective doses to the local population from radon after 1000 years and for the maximum release case are shown in Table 5-2, having been calculated exactly as they were for Table 5-1 with the omission of the SPEF, since there are no particulates.
Table 5-2. Particulate and Radon Releases from Elliot Lake Uranium Tailings: Projected Doses and Health Effects Within 80 Km of Tailings, Over 1000 years. Vegetated Piles. Normal Release Maximum Release ----------------------------------------------------------- Radiation Dose 53,000 110,000 (person-rems) Fatal Cancers 12 - 140 25 - 280
For the "impermeable dam" or saturated case, Beak's data (Beak83A, Table 4.4-1) implies a normal release of 1100 rem and from 0 to 3 fatal cancers, too small a number to merit a table. The "maximum release" calculation described just above leads to a comparable small number. The problem is that as seen in earlier sections, in the saturated case the radioactivity is successfully trapped beneath the water, but then it simply remains there. If the piles retain their integrity for the hundreds of thousands of years it takes the thorium and its daughters to decay, releases will be small. If not, releases will be large and the effect on the local population may be large as well. This will be discussed further in Section 6.
The version of the UTAP model employed by Beak to study Elliot Lake tailings was an earlier and cruder phase of model development. Later refinements have greatly improved the physical model, and have led to higher estimates of radioactive releases to the water and air. As discussed earlier, the Beak model assumes that pyrite oxidizes, pore water in the tailings pile becomes increasingly acidic, causing gypsum to dissolve, along with thorium, uranium and lead. These radionuclides re-precipitate further down in the pile, below the advancing acidic front. Eventually, when the saturated lower portion of the tailings become extremely acidic, with a pH near two, these radionuclides are released from the pile in a short term pulse. Radium dissolution seems to follow a different course - its release more closely follows the generation of sulphate ions. With excess sulphate ions, radium appears to co-precipitate with barium sulphate, independent of the acidity of the pile. Not until phase III of UTAP development , when pH, iron and sulphate were explicitly modelled, was radium release correctly modelled. In the phase III model, radium releases into downstream lakes are estimated to rise, then level off and remain constant after 1000 years. The model is still not satisfactory in that it does not account for the in-growth of radium-226 from thorium-230 in down stream waters , as shown in Fig. 6-1. This ingrowth of radium-226 implies a rising radium-226 concentration at 1000 years, where the models are cut off.
An upshot of this revised modelling of the tailings pile is that radon releases are underestimated by Beak. If radium-226 does not wash down to the bottom of the tailings pile as rapidly as originally estimated by Beak, then more radium-226 will be present nearer the surface. Thus, more radon gas will be released. These effects are different from the concerns raised in Section 4.2 where any of several mechanisms could allow greater radon release than was found by Beak's modelling. There, in Table 4-4, we examined the effects of various physical or chemical processes that could keep radon production closer to the surface.
Episodic events are not incorporated into the UTAP model. By episodic events we mean flood, dam break, drought, earthquake, tornado and human intrusion. This is an important omission for the case of an impenetrable dam, where all radionuclides, except for small seepage, are held up. Over the time periods when tailings radionuclides are hazardous, the probability is near certainty for many episodic events.
In particular, the modelling calculations above assume the hydrological flow rates are constant over time. Catastrophic events are not considered and the tailings mass is assumed to remain intact. However, this does not appear to be the case. For example, the Nordic tailings pile is eroding; a river channel appears to be forming within it. In addition, some tailings fines have blown from the pile, like shifting sands. Further, we expect the spring runoff to cause river sediments to be disturbed and wash further downstream. The Beak model and subsequent refinements all assume well behaved erosion. The model employs the Universal Soil Loss Equation for erosion. But in addition to evidence of erosion from the Nordic tailings pile, frost could cause unpredictable erosion patterns, that is, part of the tailings mass could become completely frozen and water runoff could form unpredictable channels. The net effect would be to increase the expected radionuclide releases. We have included some of these possibilities in our "maximum release" calculations in Sections 4 and 5.
We have reason to believe that the wind suspension health effects are underestimates because several factors which increase the dose were not taken into account. For example, the aquatic and air models are sharply circumscribed - what gets into the air does not provide a subsequent dose from water and visa versa. Resuspended radium and thorium particulates will give rise to radon gas. Suspension of particulates effectively increases the surface area of the tailings pile. Suspended particulates which redeposit themselves into local streams will increase the water dose and should thus be incorporated into the aquatic model. Radioactive material in water which washes onto beaches will increase radon evolution and should be incorporated into the air model. None of these flows of material are taken into account in Beak's model.
Much of the radioactive material washed to the bottom of a tailings pile will eventually be released from the pile when the pH of the pile drops to 2. Once washed from the tailings pile, radioactive thorium-230, the precursor of radium and radon gas, will become a particulate and fall to the sediment in the neutral receiving stream or lake. This thorium may be washed downstream in the spring runoff and then accumulate on a beach, giving rise to radon gas. This movement of radionuclides from the tailings pile into receiving waters, and the evolution of radon gas further downstream, is not taken into account by the atmospheric model.
Beak's radon model assumes 1/3 of radon is trapped in snow during the winter months. This radon will decay to other radionuclides in the uranium decay chain which will become trapped in the snow. During spring runoff, these decay products should either re-enter the tailings or be washed downstream. Beak's model omits this pathway, thus dropping 1/3 of the radon from further consideration without justification.
In this section, we will compare two decommissioning methods suggested by Beak, case 2, earth cover and revegetation, case 3, flooding achieved through control of dam permeability. We will summarize the benefits achieved in terms of reducing hazardous radioactive exposure and compare them to the difficulties and weaknesses inherent to those options. Also we will give a rough estimates of the costs involved. Beak consultants also included in their study three other cases (cases 4, 5, and 6) in which they assumed that 90% of the leachable radium and thorium were removed from the tailings before discharge from the mill. These cases are irrelevant to our study, since we are examining the impact of existing piles, which contain substantial amount of radium and thorium, for which no attempt at removal was made.
Other options have appeared in the technical literature. Underground mines have been proposed. The idea appears attractive in that it would remove the tailings from public sight and access, and would partly solve the problem of erosion and air emissions. However, one major obstacle comes from the fact that the original volume of the mine would be insufficient since after milling the volume of the tailing has expanded to more than twice the size of the extracted ore. At Beaverlodge, 40% of the original tailings were placed back in underground mines . If the full volume cannot be accommodated, then the tailings should be classified. The finer section, which contains the highest concentration of radionuclides, would be delivered to the mine and the less hazardous coarse one would adopt a surface storage solution. This would involve a physical separation process, that could create dust in the process, though a hydraulic suspension method could be employed.
Another alternative assumes that a 3 m thick cap of pyrite-reduced tailings would be applied over the entire tailings surface In this scenario, a portion of the underlying pyritic tailings remain available for oxidation, albeit at reduced rates. Because of the effects of the pyrite-reduced cap, perpetual treatment of leachate would be required, but the rate of acid generation would be reduced. An improvement to this scenario would consist of making the dam impermeable so that the water table would rise into the pyrite-reduced cap and thereby more effectively halt acid production.
1. to provide effective long-term stabilization and isolation in order to:
1.1- reduce the chance of human intrusion so as to prevent the use of tailings as a construction material, as backfill around structures, and as landfill
1.2- protect the piles from natural spreading by wind erosion and surface water runoff
1.3- prevent spreading by flood damage to the piles; and
1.4- prevent tailings from contaminating surface and groundwaters
2. to control radon and gamma emissions from the tailings,
3. to protect water quality in order to prevent contamination of water through leaching of radioactive or other hazardous materials from the tailings into surface of water, or groundwater aquifer.
While the above objectives are desirable, specific numerical guidance is not codified in Canadian regulations. In particular, AECB regulations have no specific requirements for radon control or attenuation of gamma fields. The requirements for specific mills are based on the dose to critical individual receptors, which because of the remoteness of most Canadian mining operations, are not demanding. Requirements are not based on minimizing global or regional collective doses. Also, because of the remoteness of mining locations, human intrusion on the tailings pile is not considered. Generally, it is considered technically feasible to achieve stability of tailings structures for a 200-year period. Radiation doses are evaluated for a 1000 to 2000 year time period. Since tailings will remain hazardous for hundreds of thousands of years, this time cut-off sharply reduces the potential number of health effects.
We will now review how the various options could help achieve the objectives mentioned above
Tailings are a high grade sand "ideal" for construction or as fill. A passive barrier such as a 10 foot thick earthen cover would prevent unintentional intrusion since a variety of human activities, unless a home with basement were constructed on the tailings pile itself.
- Preventing erosion
Thick earth covers will prevent tailings from becoming windborne or waterborne, therefore eliminating the potential for surface particulates to resuspend. Furthermore vegetation will protect the earth covers from erosion and stabilize them. Shallow-rooted vegetative covers provides the best protection. Native local plants requiring minimum maintenance have to be identified.
- Flood protection
Construction of barriers designed to withstand floods will be necessary. In particular, diversion of upstream water away from the tailings pile, and sluice pipes to prevent water from washing away the tailings dam.
Natural cover materials such as earth can control radon emissions to the atmosphere. Thickness of the cover, moisture content and permeability affect the efficiency of the control. Whether or not vegetating actually enhances radon exhalation is controversial. Research by NUTP shows that under Canadian conditions simply re-vegetating the surface of the tailings made almost no difference to the radon exhalation rate.
BeakA83 shows that radon flux associated with a vegetated site is higher than the flux associated with the bare site at any time because of greater diffusion, emanation and transpiration. In effect, root penetration and transpiration result in an increase of the emanation coefficient and the transpired water causing a reduction of the hydraulic input is also expected to carry some soluble radon gas thereby enhancing the surface flux rate of radon. However, for the vegetated area, the total dose rate to an individual is still slightly lower than for the reference case since the dose contributions result from radon and particulates. While radon emanations may be increased by vegetation, suspended particulates will be reduced. Not until the tailings mass has become acidic and the radium has leached through the mass does a significant variation between the reference case and the vegetated exists. A 38% reduction in total dose rate occurs after 500 years and the total collective dose commitment is 20 to 25% lower with a zero cut-off. Of course, if a considerable thickness of soil cover has been added, the radon emissions would be reduced. - Gamma Radiation Gamma radiation is the most significant exposure which would result from casual access to the site. Modellings to date have not included this pathway but a recent pathways analysis submitted to the AECB regarding decommissioning of Quirk Lake tailings was requested to include gamma radiation resulting from 200 hours per year casual access to the site. Even exposures of this short a duration dominated the pathways analysis as the prime component of dose.
Earth thick enough to sustain vegetation will significantly reduce gamma radiation; and earth thick enough to reduce radon emissions will also reduce gamma radiation to significant levels.
If flooding can be maintained, this method prevents access to the tailings but the lake should not in the future be used as a recreational area.
- Preventing erosion
The saturation of the tailings surface eliminates the possibility of surface wind erosion of particulate matter. Concerns are raised about the importance of having a deficit of particulate or solute transport out of the lake in the short term, and the fact that erosion of the outlet might cause this to change in the long term. In this connection, most uranium mining areas in Canada are located on the Canadian shield with a igneous granite bedrock outlet which should erode at a much slower rate than sedimentary rock or unconsolidated materials (in the absence of any tectonic event).
- Flood protection
Flood protection does not apply to this option since the concern is rather whether or not sufficient water can be available under average precipitation conditions to maintain the tailings in a saturated state. On the other hand dam failure or a change in hydrological or conditions could cause the lake to drain. Unfortunately, episodic events, such as dam breaks or severe drought, are not evaluated by mining companies. Over the long run, the probability of such events becomes more certain.
In order to reduce radon emission from uncovered area, it is important to limit tailing beaches. Rio Algom has determined that it is practical to flatten the tailings surface and has achieved an effective water depth of 1 to 2 feet over 75% of the tailing area. Radon emission could then be reduced to one fourth of the base case values. Over the 25% of the surface not covered by water, pyrite oxidation could proceed more rapidly and acid could be generated.
Assuming that water cover is maintained to a sufficient depth, direct gamma exposure from tailings should result from residual exposure to uncovered tailings as well as waste rock in dykes, dams and roadways.
Table 7.1 Projected Health Effects from Elliot Lake Tailings Region of Concern Base case Vegetated Flooded cover tailings --------------------------------------------------------------------- Regional, Serpent River Aquatic (1000 yrs) 170 - 2,000 170 - 2,000 200 - 2,200 Local Rn and particulates (1) 17 - 2,200 12 - 280 000000 --------------------------------------------------------------------- Total 187 - 4,200 182 - 2280 200 - 2,200 Global (x10,000) Aquatic (1,000yrs) 0.16 - 1.8 0.16 - 1.8 0.2 - 2.2 Radon, Max release (2) 20 - 230 25 - 280 0.21 - 2.4 --------------------------------------------------------------------- Total 20 - 222 25 - 282 0.4 - 4.6 (1) lowest value of the range :Normal release SPEF 1 highest value of the range: Maximum release SPEF 2.5 (2) Table 4.4
A study of the Beaverlodge area of Northern Saskatchewan allows to compare the relative costs of the various options. In that case, 10.1 million tonnes of mill tailings were produced of which 40% was placed back in the mine and 60% placed on surface. The total drainage area is 14.2 km2 = 1420 ha.
The cost of the flooding option for the Beaverlodge site obtained by dredging of the tailings beaches or construction of dams at the outlets of the lakes is as follows:
Dredge $1,779,000 (1983 Cnd. $) Dams $3,127,000 Combination $2,500,000 Vegetation between $834,00 and 1,365,000, according to the process chosen
The capital cost given for the decommissioning of the Quirke mine tailings basin (1987 Cnd. $) is as follows:
Surface area: 190 ha, watershed area: 275 ha
Surface grading and vegetation $2,000,000 Dam construction (flooding) between $4,295,000 and $5,400,000
An AECB report for a generic site, with surface area of 50 ha. drainage basin 150 ha.
Soil cover (1982 Cnd.$) $9,8000,000 Tailings preparation and vegetation between $1,350,000 and $2,000,000
While it is difficult to generalize from the above costs to the potential costs of decommissioning the Elliot Lake complex, certain generic costs can be approximated. Obviously, each tailings area in the Elliot Lake complex must be examined on an individual basis. Dam construction and sluiceways will be specific to each tailings configuration. However, if the decomissioning method is soil cover and vegetation, we can determine the costs of covering 1095 ha with 1 meter of soil. Assuming Means data, the cost of moving topsoil, assuming one could find such a large volume of topsoil locally, is $7.50 per cubic meter or $82 million. The cost of landscaping and vegetating this large expanse is estimated to be $0.85 a square meter or $9.3 million. In addition to these costs, additional costs of engineered structures and fencing would have to be factored in. Follow-up monitoring and maintenance would have to continue for the indefinite future. These costs should be estimated and money should be set aside for a perpetual care fund. We have not made these estimates here, except to note that they are small compared to the projected health costs.
Beak83W An Approach to the Calculation of Dose Commitment Arising from Different Methods for the Long-Term Management of Uranium Mill Tailings Through Aquatic Pathways, Technical Appendix, Atomic Energy Control Board, INFO-0097, 1983.
Beak83A Atmospheric Dispersion of Radionuclides from Uranium Mill Tailings Disposal Sites, Technical Appendix, Atomic Energy Control Board, INFO-0097 (App), 1983.
1 Senes Consultants, "Probabilistic Model Development for the Assessment of the Long-Term Effects of Uranium Mill Tailings in Canada: phase II," OSQ 84-00207, prepared for the National Uranium Tailings Program, October 1985.
Senes Consultants, "Uranium Tailings Assessment Program (UTAP.3): User's Manual," 61728-01-SQ, prepared for the National Uranium Tailings Program, April 1987.
Senes Consultants, "Approaches to Risk Assessment for Canadian Mill Tailings: phase II," OSQ 85-00225, prepared for the National Uranium Tailings Program, March 1986.
Senes Consultants, "Probabilistic Model Development for the Assessment of the Long-Term Effects of Uranium Mill Tailings in Canada: phase III," OSQ 85-00182, prepared for the National Uranium Tailings Program, July 1986.
Atomic Energy of Canada, Ltd., Inventory of Waste Quantities: Report to the Siting Process Task Force, Energy, Mines and Resources, Canada, Ottowa, ON, 1987.
R=dN/dT=C*ln(2)/(A*u*T), where R is the decay rate in sec-1, C the concentration, A the atomic mass, u the atomic mass unit and T the half life. For T given in years and R in Curies, this becomes R=3.58x105C/(A*T). The similar calculation for pore water gives Ci/m3, which are converted to Ci/dry tonne using Beak's dry density of 1.24 tonne/m3.
Moffett, D and M Tellier, "Uptake of Radioisotopes by Vegetation Growing on Uranium Tailings," Canadian Journal of Soil Science, 57 (4), p. 417-424, 1977.
J Gofman, Radiation-Induced Cancer from Low-Dose Exposure: An Independent Analysis, Committee for Nuclear Responsibility, San Francisco, 1990. Select and View (HTML)
J Gofman, Radiation and Health, Sierra Club Press, San Francisco, 1981.
National Research Council, Health Effects of Exposure to Low Levels of Ionizing Radiation, BEIR V, National Academy Press, Washington, D.C., 1990.
Radon Releases from Uranium Mining and Milling and Their Calculated Health Effects, Nuclear Regulatory Commission, NUREG-0757, February 1981.
. In particular, we fitted F=a*exp(-bt) to the last two data points of Beak's Table 2.5-1 for cases 1 and 2 and made the scale corrections described above. We used temporal midpoints 400 and 750 years to fit the exponential, and integrated from t=1000 years on out.
Integrating the exponential gives a total release equal to the intial release rate multiplied by the half-life and divided by the natural logarithm of two.
Maclaren Plansearch Inc., "Optimization in the Decommissioning of Uranium Tailings," prepared for the Atomic Energy Control Board, INFO-0321, June 1987.
International Commission on Radiological Protection, "Radiation Protection Principles for the Disposal of Solid Radioactive Waste," Annals of the ICRP 15 4, 1985 (Publication 46).
Midwest Joint Venture, Denison Mines Ltd, Operator, Environmental Impact Statement, August 1991.
Senes Consultants, "Probabilistic Model Development for the Assessment of the Long-Term Effects of Uranium Mill Tailings in Canada: phase III," OSQ 85-00182, prepared for the National Uranium Tailings Program, July 1986.
"Peer Review of the National Uranium Tailings Program's probabilistic system model, by Atomic Energy Research Company for NUTP, 61735-01-GR, 1987.
Steffan, Roberston & Kirsten, "Supplementary Report on Frost Action in Uranium Tailings," prepared for the National Uranium Tailings program, 61730-01-SQ-Suppl,1988.
AW Ashbrook, "Decommissioning and reclamation of Eldorado's Beaverlodge mine/mill operations," in Proceedings of the 9th International Symposium, 5-7 Sept 1974, by London Uranium Institute, 1985.
DSMA ATCON, Ltd, "A Study to Measure and Evaluate the Effect that Vegetation has on the Emanation Rate of Radon from a Uranium Tailings Pile," prepared for the National Uranium Tailings program, OSQ-83-00273, March 1985.
SENES Consultants Ltd, "Preliminary Environmental and Radiological Pathways Analysis for the Quirke and Panel Waste Management Areas," prepared for Rio Algom Ltd, April 1991.
Maclaren Plansearch Inc., "Optimization in the Decommissioning of Uranium Tailings," prepared for the Atomic Energy Control Board, INFO-0321, June 1987.
WISE Uranium Project (home)