Dose Reconstruction for the Fernald Facility: A Review of Task 4 (1994)

1. As indicated in the report, the ICRP respiratory model (International Commission on Radiological Protection, 1979) will be replaced with a more elaborate age-specific version, but no commitment is made to use this more sophisticated model. The new model has been approved by ICRP, and a comprehensive report describing it should be available soon. Serious consideration should be given to using the new ICRP model in the final dose estimation or the new NCRP model when it is available. However, because radon dosimetry is evolving rapidly, and the authors of the report are advised to follow the literature in that field closely. In any case, exposures from radon progeny should be added to those for other radionuclides for the appropriate tissues. The dose coefficients for radon progeny will vary according to travel time between the silos and the receptors; near the silos, there will be much more radon gas than radon progeny and the doses from radon gas could be estimated.

2. If the authors of the report find it appropriate to estimate dose rates rather than committed doses, they should be aware that the National Radiological Protection Board has published a report on that topic (Phipps et al., 1991).

3. The authors of the report should consider using the report by Eckerman and Ryman (1993) in which external irradiation doses are calculated for many situations.

4. A glossary and a list of symbols should be included. Many equations are used and some of the symbols are used twice with different meanings. For example, T stands for “time” in Appendix A and for “tissue” in Appendix T.

5. Uncertainties are treated thoroughly throughout the document except for the dose-conversion factors (Appendixes A and T). Because the dose conversions recommended by ICRP are prepared for regulatory purposes, they are by definition conservative. Body

organs were assigned somewhat arbitrary tissue weighting factors by the ICRP. For example, in order to have a 0.25 contribution, five organs were each assigned a tissue weighting factor of 0.05 even though the calculated value for the thyroid was 0.02. This leads to considerable uncertainty that depends on the organ in question. Also, because of biologic variability among individuals, the dose-conversion factors present a large variability. The uncertainties should be considered carefully, especially for the various physical and chemical forms of uranium that were released into the environment.

6. The validation exercises show that the environmental transfer models are adequate. However, it its disturbing to see that month-to-month increases and decreases in the air-monitoring data, and, to some extent, in the gummed-film data, do not match the variation in source term. This discrepancy should be discussed in detail.

7. Appendix A: The epidemiologists will need the dose distributions that the authors plan to determine. It also might be useful to validate the highest estimated doses; this would require identifying a limited number of persons (5 to 10) who are thought to be members of the critical group and providing detailed estimates of their individual doses.

8. Appendix C: This review of the radionuclides released from the FMPC between 1951 and 1988 is a very useful component of the document. It shows the importance of the doses that result from atmospheric release of uranium. The committee's only criticism is that the doses from radon and its progeny should have been included in the screening exercise.

9. Appendix T: This reference for the dosimetric factors and related information that will be used to convert predicted exposure of individuals into dose is an important component of the report. However, it is not clear what the rationale is for separating the dose from radon progeny and the dose from the long-lived radionuclides. It does not seem that the ICRP-60 tissue-weighting factors presented on page T-6 will be used in the dosimetry.

10. Executive Summary: The major deficiency of the executive summary is that it does not emphasize which environmental pathways are most likely to contribute the largest doses. Although the authors state “atmospheric pathways will likely dominate the total dose from FMPC releases” (page vi), the exact meaning of this statement is unclear. If, for example, the atmospheric pathways account for 99% of the dose, consideration of the other pathways could be greatly simplified.

11. On page x, surface water transport is identified as an “important exposure pathway for liquid effluents,” even though it is probably not a major exposure pathway when the total dose is considered. The executive summary should include an analysis of error to prepare the reader for “reasonable” dose estimates that vary by orders of magnitude. It is not clear what expectations the authors have about the range of reasonable estimates and whether the error is primarily due to the choice of environmental model or to miscalculating the source term. The authors should identify the potential sources and magnitude of error so they can concentrate their efforts on areas with the greatest uncertainty.

12. Page v, para. 3, ln. 10: What is the definition of “generally been quite good”?

13. Page vi, para. 1, ln. 2: “Note” should be “not.”

14. Page vii, para. 3, ln. 5: “About 49%” should be rounded to “about 50%”.

15. Page viii, para. 7, ln. 8: “Reasonably good” average agreement should be defined.

16. Page ix, para. 2, ln. 7: What does the total deposition estimate of 2 × 10⁵ kg to 9 × 10⁵ kg of uranium represent? Is this a 95% confidence interval? Why does this differ significantly from the values reported by Stevenson and Hardy (1993)?

17. Page ix, para. 3, ln. 1: What is the definition of “a significant component of the dose”? 1%? 5%? 20%?

18. Page ix, para. 4, ln. 5: Define “most.”

19. Page x, para. 9, ln. 3: “Three private wells have been contaminated by uranium.” What is the definition of “contaminated”?

20. Page 2, para. 1, ln. 1: “anouther” should be “another.”

21. Page 6, para. 6, ln. 3: The details of how the authors estimate that 50% or more of the particulate material is likely to have deposited within a radius of 5 mi is unclear.

22. Page 10, para. 3: The authors state the relative contribution of uranium for atmospheric releases and the relative contribution of radium for the surface-water releases. They should also provide information about the relative contribution of atmospheric releases versus water releases.

23. Page 33, para. 1, ln. 1: What is the definition of a “significant direct exposure”?

24. Page 37, para. 2, ln. 4: What is the definition of a “significant contributor”?

25. Page 37, para. 4, ln. 5: What is the definition of “reasonable”?

26. Page 37, para. 4, ln. 8: What is the definition of “good agreement”?

27. Page 37, para. 4, last line: What is “reasonable agreement”?

28. In particular, the exposure pathway analysis is not complete unless primary and secondary source streams are included or are excluded deliberately, in accord with some identifiable criterion. On this basis, although the report is elaborate and covers most of the area in considerable detail, in many cases it deals with some modeling subjects generically without ever focusing on specific areas of concern that are highly site-specific. This applies particularly to the issue of uranium dust as the principal material released from the plant.

Comments on “Exposure Pathways and Dose Calculations” (pp. 38-44) and relatedAppendixes Q, T, and U.

29. The arrangement of this section is awkward. It emphasizes the agricultural pathway (food chain) but does not specifically include the air-exposure pathway or the inhalation model, except to the extent that Appendix T describes the lung model and provides some dose-conversion factors for inhalation. However, that treatment is directed more specifically to radon progeny, and there is no discussion in the main text of the consequences of inhaling uranium-bearing dust.

30. Some discussion is needed to link the transport chapter with the inhalation and ingestion scenarios, for all streams. This also would allow the formulation of more meaningful and comprehensive conclusions.

31. Apart from the RAGTIME model, which deals specifically with the food chain, it is not clear how the comparison of this intake with the inhalation and drinking-water pathways will be accomplished.

32. Although p. 36 lists drinking water and the monthly dilution (MD) model, there is no further reference to them in this section and to their inclusion in the final dose calculation. It is not clear to what extent the MD model has been validated by Steva's measurements (Steva, 1988) of radioactivity in drinking water in the vicinity of the FMPC. Otherwise Appendix R indicates unresolved appreciable discrepancies between measured and calculated values (Tables R-6 and R-7).

33. Are there any measurements on record to validate the ingestion scenario (p. 39)? Where in Appendix Q, and how, is this specific scenario described?

34. Appendix Q is a detailed and thorough description of the food pathway and of the crop census of the Fernald district. However, given the low solubility of some

uranium oxides and the small transfer coefficients for uranium (Tables Q-1-3), it would be useful to do some screening calculation for the most sensitive products (e.g., eggs and lettuce) to see whether this pathway merits further detailed calculation for uranium releases.

35. The discussion on dose-conversion factors (DCFs) in Appendix T is uneven and unnecessarily extended in some parts (Table T-1 should probably be placed somewhere else). It is not clear, for instance, on what basis the airborne dust particles, principally uranium products, are going to be categorized according to clearance time from the lung after onsite depletion of the coarser sizes. This deficiency in turn results in the presentation of a generic discussion of DCFs but insufficient focus on uranium as the principal source material. Thus Tables T-4 and T-5 are somewhat misleading because they avoid any estimate of biologic half-life or clearance time for uranium, whether soluble or insoluble, for episodic or chronic inhalation.

36. A large amount of information has been published about dose calculation models for attached and unattached radon progeny; Raabe's paper (Raabe, 1969), the only one cited here, is certainly not the most up to date. The discussion on pp. T-10-12 dodges the issue of what procedure will be adopted for this dose estimate.

37. For cancer induction estimates it is important to have more of an age profile than the proportion of adolescents to adults in the population.

38. The demographic section (p. 40 and Appendix U) indicates a very slowly growing population. Figure U-3 shows the population doubling from 1950 to 1960 and increasing by 25% to 1970. This does not seem to match the growth-rate coefficients in Tables U-1A and U-1B.

39. Given the uncertainties and magnitude of this total dose assessment and the very small total populations exposed, 5,300-6,300 in 1960 within 5 mi, it would be useful to discuss in the Conclusions section whether this is a large enough population on which to base statistically valid epidemiologic conclusions.

40. Overall, some editorial action is required to eliminate material that is only marginally relevant to Fernald. The purpose of this report is not to present a textbook on all possible pathway scenarios.

41. The committee is not clear whether this report covers any source terms added to the study since the original draft report.

42. Previously, there was some suggestion of using a puff model for airborne releases; this report seems to assume that releases from roof vents predominated at all times and that a wake model is more appropriate. It is not clear to the committee that this is necessarily so for the major episodic releases in the early years?

43. Page ix refers to confining the analysis to 1960-1962. This might be reasonable for validating any model, as the data are more complete for that period than for others. However, once it is adopted the model will have to be usable for the whole period of plant operations and it is important to discuss the appropriateness of doing so.

44. In Appendix N, K_d values are intended to indicate soil adsorption of dissolved ions. If most of the uranium releases were in oxide form and as particulates, Eq. N-1 does not apply, because the particulates will move through topsoil layers by entrainment and will be trapped in interstices rather than adsorbed. Subsequent dissolution of uranium as the uranyl anion would depend on local pH, which is given as 5.1-6.0 in Table N-5. This is not very acidic in a moderately clayed soil, and so it is questionable whether the procedure illustrated in Table N-7 for deriving cumulative deposition from measurements by assuming a value for K_d is valid. This, incidentally, underlines the importance of indicating the chemical form of releases in the source term report.

45. Because early releases were mostly episodic, the low soil concentrations of uranium offsite could be used to argue in favor of surface runoff into Paddy's Run for most of that material, if deposition occurred in dust form on a relatively dry surface. This

assumption would have to be incorporated into the surface water source term. Resolving the discrepancy in the source-term estimates should be a priority.

46. The soil measurements might have little bearing on validation of the airborne model, since they could be of questionable value if wake effects predominated. Appendix G illustrates this problem as it oscillates between gravitational settling for coarse particles from a steady plume (Fig. G-3) and a dry-deposition velocity for very fine particles from a plume (Eq. G-3). The discussion is good, but it leaves open the question of source depletion of coarse particles on site from puff releases for low-elevation release points, which could represent the largest release fraction during episodic events. This is confirmed by the authors on page F-11, but it should be developed further.

47. The treatment of radon releases from the silos, in Appendix P, is greatly improved over the draft versions of the Task 2 and Task 3 reports, which will need to be updated. Unfortunately, there still is no correlation between the radium-and-radon inventory in the stored waste and the airborne releases, and hence there is no stated justification for the release terms quoted on p. P-7 of 140 Ci/yr (continuous) and 810 Ci/yr (daytime only). Also, it is not clear whether a separate source-term-and-release scenarios were adopted for conditions before and after the dome was sealed. The later monitoring measurements deal only with radon seepage after the cracks in the dome were sealed and probably have little bearing on previous release conditions and their radon-progeny equilibrium. If it is argued that merely the magnitude of the radon release was changed by sealing the cracks, then this should be stated and justified. The low P/O ratios in Table P-8 are surprising. The model probably overestimates radon concentrations because it ignores the inhibitory effects of rainfall and snow cover on the release of radon from the ground. The committee must conclude that the source-inventory-and-release term was consistently underestimated. Otherwise, just for validation of the model, any agreement to within an order of magnitude should probably be considered acceptable.

48. Page x and Appendix R: Assuming appreciable on-site contamination (largely from anecdotal evidence) and on-site fallout, surface runoff to Paddy's Run might have represented a significant source stream, especially in the earlier years. It is not clear from Fig. R-4 to what extent this assumption can be verified or to what extent the data for 1977 and 1978 can be related to accidental uranium dust releases. The agreement between the GENII model and measured concentrations in Fig. R-6 could make the origin of the source stream unimportant in any case, except that the solubility characteristics of the measured uranium impinge on any derived dose values either for drinking water or for irrigation. Exposure parameters for the maximum individual seem unrealistically high, although in view of the very low final dose estimate in Table R-5 this might not matter. This topic might require revision before estimates for the early years, with their high release rates, are included and the dose values are made final. A table, such as Table R-5, listing only the maximum dose value, however improbable, can be misleading and can cause considerable alarm among those who do not appreciate the distinction. The final report should list the average exposure to a given population group, with some range given for a 95% confidence limit.

49. The uncertainty analysis in different parts of the report is highly inconsistent and, for various reasons, expressed in different terms (e.g., compare Tables O-1, Q-9, and RS-3). It is not immediately clear how such variations would be reconciled in the final report. A consistent procedure should be outlined for propagating uncertainties and, in particular, for taking into account the substantial systematic errors in many of the early sampling and analysis data. There also will be an appreciable uncertainty associated with early episodic releases, and it will be important to conduct a sensitivity analysis to eliminate minor, but highly uncertain, contributions to the total dose.

50. Appendix T on dose-conversion factors is a good general review of the lung model for exposure to airborne radionuclides. However, because most of the airborne matter consisted of uranium oxides or some dust-attached fission products, this discussion really should focus on those materials. Appendix M gives no indication of any transuranics being released; hence the discussion on p. T-13 could be irrelevant.