The Hanford Environmental Dose Reconstruction Project: A Review of Four Documents (1994)

The two volumes of report PNWD 2033, by C. M. Heeb, provide an analysis of iodine-131 releases from the Hanford site in 1944 to 1947. The committee was impressed by the two documents, which summarize iodine-131 releases for the critical early years of operation. The release of 685 kCi of iodine-131 constitutes the most important airborne source term. The approach taken in the documents is to rely primarily on releases calculated from daily measurements of operational conditions that are directly related to iodine-131 and other fission-product production. The operational measurements include power levels, fuel loading and unloading, and cooling times between fuel discharges and processing and are comprehensive and complete for calculation of iodine-131 generation and release. However, stack-monitor measurements to document the releases were not referred to in this document.

The committee delved into whether the release of iodine-131 was estimated completely or underestimated because of lack of data. That issue was also raised by members of the public during the open meeting in Richland, Washington, during which considerable concern was voiced that DOE was hiding data on releases from events that they had kept secret from the public and the HEDR project. Although some data on stack measurements, stack flows, and meteorologic conditions are missing, comprehensive data sets on daily operations were available for use in calculating releases. To the best of this committee's knowledge, the operational data are complete and summarized in report PNWD 2033; it is known exactly when each reactor started, what the power level was in each fuel panel, and when each batch of fuel from each reactor was processed. The daily operational logs for the reactors and the fuel-reprocessing plants were found and corroborated by the dose-reconstruction team; a detailed review of report PNWD 2033 revealed no gaps in operational data.

The operational parameters from the daily logs can be reliably used to determine iodine-131 inventories in the reactor fuel, and these inventories and a conservative-release scenario can be used to establish the amounts released and the periods (roughly several hours per batch) over which such amounts were released. Those are reliable calculations because considerable care was taken to log all the daily operational parameters at each reactor to establish the amount of plutonium produced in each batch of fuel. The iodine-131 production is directly a function of power level which is directly related to the total fissions that occurred. The total fissions can be derived from the amount of heat produced in each fuel channel, which was recorded by continuous measurement of the inlet temperature and outlet temperature of coolant water flowing through each fuel channel in the reactors. Figure 2 shows the relationship between those measured values and the reactor design and operation in simplified form. The measurements are simple, their accuracy in the 1940s was as good as it would be with current techniques, and the committee believes that the daily data were logged dutifully and without bias; i.e., they are complete and an excellent source of information generated by Hanford's workers for determining the source term.

A fuel charge was typically left in the reactor for about a month, after which it was discharged from the fuel channel as a total batch after the desired burnup was achieved (as indicated by the operations log). The measured heat output (from the temperature rise in the coolant water in each channel) is directly related to the number of fissions, which produced new atoms of fission products; for each 1,000 fissions in this type of reactor, 29 atoms of iodine-131 are produced (Rose and Burrows, 1976). Once the amount of heat produced in each channel is determined, the number of fissions in each batch of fuel can be calculated, and that yields the number of iodine-131 atoms produced. The number of atoms of iodine-131 determines the activity, which is usually expressed in curies. This is known as the iodine-131 fuel inventory.

Once the iodine-131 inventory is known for a batch of fuel, the amount released from the inventory is determined from the fuel-reprocessing protocol. Figure 3 is a simplified

schematic of how iodine is released from fuel reprocessing. The two most important characteristics that determine the amount released are cooling time after removal from the reactor (to permit the decay of short-lived radionuclides) and off-gas holdup and treatment. In the first few years of Hanford operations, the urgency of recovering plutonium caused the fuel to be processed after only a few days of cooling, when substantial amounts of iodine-131 were still in the fuel. That is the main reason why most of the iodine-131 releases occurred in 1944 to 1947; cooling times were later extended to about 6 months to allow most of the iodine-131 to decay before processing began.

The times between discharge from the reactor and initiation of fuel reprocessing are also in the operations logs for each batch of fuel. Decay of the iodine-131 inventory for each cooling time is a straightforward calculation, and it is well established how much iodine-131 was available for release when the fuel was initially dissolved in acid, a step that releases all the iodine-131 inventory from the fuel to the process system (see Figure 3). The amount released to the environment after dissolution is the fuel inventory minus the amount retained (and unavailable for release) in the processing chemicals, the amount removed by plateout in the processing and off-gas systems, and the amount removed from the off-gas by scrubbers and filters. The amount of iodine-131 released to the environment is very dependent on the fraction of iodine-131 inventory released and the chemical form of the iodine. The fraction evolved from the dissolution step was established to be 85% of the fuel inventory by subtracting the amount of iodine-131 bound to organic molecules measured in the processing chemicals; no plateout or off-gas cleaning was assumed. These determinations therefore tend to overestimate rather than underestimate the release; i.e., they do not substantially underestimate the release.

The overall estimate of iodine-131 releases for 1944 to 1947 is, in the committee's judgment, reasonably complete. Because off-gas losses of iodine-131 were not taken into consideration, report PNWD 2033 does not underestimate the overall iodine-131 release in 1944 to 1947. The estimated release of 685 kCi of iodine-131 exceeds an earlier estimate of 400 kCi provided by Heeb and Morgan (1991). The 685-kCi value is based on plant

records that were not known to Heeb and Morgan, because these records had been retired and stored. The committee believes that the 1992 (or more recent) estimate is supportable, although episodic releases from plant accidents, spills, or filter failures might have occurred, and they are not clearly accounted for in the report. If or when episodic releases did occur, their contribution to the total releases was probably small.

In several experiments conducted at the Hanford site, inventories of iodine-131 in small batches of fuel were released to determine migration, deposition, and behavior of iodine-131 in the environment. Those releases appear to be well documented, and even though the releases were large, they were considerably lower (because smaller amounts of fuel were involved) than those associated with the reprocessing of total reactor-fuel inventories. These experimental amounts could have dominated the annual releases of iodine-131 in later years (the 1950s and 1960s) because at that time fuel was cooled to allow iodine-131 to decay before processing. The special releases were made close to the ground (unlike stack releases from fuel reprocessing), so much of the iodine-131 was deposited on vegetation and soil on the Hanford site where it was measured. Although the special-release experiments seem to have been documented, the committee cannot provide complete assurance that all releases were documented. The dose-reconstruction team appears to have followed up on all potentially missing events suggested by the public and employees at the Hanford site; to our knowledge, no important events of this type have escaped examination, and doses resulting from unaccounted-for releases are not likely to produce a total dose which exceeds the HEDR staff's estimate of the total dose.

On the basis of fuel-burnup records and the volatility of iodine during fuel reprocessing, and the volatility of iodine during fuel reprocessing, the authors assumed a constant fraction of iodine losses in the source term. The author of report PNWD 2033, C. M. Heeb, assumed conservatively that the only form released was elemental iodine, I₂; this gaseous form would yield the highest calculated offsite doses. Few data are available to establish the proportions of I₂and organic iodine actually released. Appendix B in Volume 1 lists some evaluated release factors but seems to refer only to losses of volatile iodine

from solution. The validity of the assumption of volatile I₂ as the only species of iodine released needs justification.

Report PNWD 2033 deals extensively with the question of missing data and the uncertainty of estimated releases. It claims that even finding more data would not substantially change the findings and estimations. If that is accurate, there might be little to be gained scientifically by further efforts to uncover documents, which might or might not exist, especially if the uncertainties in the remainder of the dose reconstruction are even larger than those in the source term. But the public's impression that not all appropriate efforts to find pertinent data have been made could undermine the credibility of the project.

Report PNWD 2033 includes data that support the distribution function used in the model (see the first paragraph on page 3.3 in Volume 1); however, no attention seems to have been paid to how the result might have been different if other, equally plausible distribution functions had been used. For example, page 4.18 states that triangular distribution functions were used for two major variables “because these are peaked distributions.” Other peaked distributions, such as a Gaussian distribution, do not appear to have been considered or compared for fits to the reference data. A sensitivity analysis of how different the results would be if plausible alternative distribution functions were used would be most helpful.

The committee has some concern about the absence of documentation of daily releases of iodine-131 from the reactors themselves. An analysis of fuel technology at the time suggests that such releases should have been measurable, although they were no doubt considerably smaller than those due to fuel reprocessing. If available, data on iodine-131 concentrations in the reactor cooling water might be useful in determining associated airborne releases.

A more important concern involves iodine-129 releases, which were not addressed. Iodine-129 should have been present because the number of iodine-129 atoms produced would be about one-third the number of iodine-131 atoms produced. Because of the long

half-life of iodine-129, the cooling time would not affect the release of this radioisotope. Although iodine-129 was not an important source of human exposure, its long half-life should afford an opportunity to measure environmental iodine-129 and thus a means to validate the iodine-131 modeling. Releases after about 1950 contained minimal iodine-131, because fuel was cooled before processing, but substantial amounts of iodine-129 would have been released. For completeness, the source-term analysis should address total radioactive-iodine releases during reactor operations and iodine-129 releases associated with fuel reprocessing. Population doses and risks would then need to be calculated for the entire period that the reactors were operated.

In summary, the reconstruction of estimated releases of 685 kCi of iodine-131 is thorough. The committee is impressed that the data were available and in a form for obtaining power levels and their daily fluctuations. Once such data are available on each batch of discharged fuel and the cooling times, it is relatively straightforward to calculate the iodine-131 inventory and releases during fuel reprocessing. The releases are not likely to be underestimated, inasmuch as the authors took care to account for all fuel changes, operational parameters including measured power levels, fuel cooling times, and residual fractions in process chemicals. The only concern that the committee has is in the paucity of information about the releases of iodine-131 from the reactor themselves although it realizes these are likely to be minor compared with those from reprocessing. The doses calculated on the basis of this assumption are conservative (i.e., likely overestimated), in that the iodine-131 released was probably not all in the elemental form as was assumed.

The modeling techniques chosen for the environmental transport pathways (atmospheric, surface water, and groundwater) are briefly discussed in report PNWD 2023, by D. B. Shipler and B. A. Napier. The atmospheric pathways are dealt with in two coupled computer codes: RATCHET is used to model the atmospheric transport and deposition on the ground of the released iodine-131, and DESCARTES is used with the results from

RATCHET to estimate the concentrations of iodine-131 in terrestrial environments (soil, vegetation, and foodstuffs). The equations used in DESCARTES and the parameter values used for iodine-131 are presented in report PNWD 1983, by S. F. Snyder and colleagues.

Those documents give the surface-water pathways and the groundwater pathways less attention than the atmospheric pathways. That is probably appropriate, but the basis for it should be explained in the reports. More important, major decisions concerning the models to be used either have not been made or at least were not presented in the report. The general impression of the committee is that the conceptual approach being considered for the modeling of environmental pathways is sound. However, the impression is based much more on discussions with experts at Hanford than on the contents of the four reports provided to the committee for review; important information is missing from reports PNWD 2023 and PNWD 1983, such as validations, uncertainties, episodic releases, and meteorology.

The atmospheric pathways of exposure considered for people onsite are inhalation, air submersion, and groundshine (radiation emitted by radionuclides on the ground). Those pathways are also considered for people offsite, in addition to ingestion of milk, meat, eggs, poultry, leafy and other vegetables, fruits, and grains, all of which contain radionuclides.

In the HEDR project, the decision was made to expend more effort on the earlier periods (mid-1940s), when the largest releases of radionuclides occurred and reliable monitoring data were scarce, than on the later periods, when releases were relatively small and reliable monitoring data were more abundant. Three periods are considered: 1944-1949, 1950-1972, and 1973-1991. Very fine temporal, spatial, and agricultural data resolutions have been selected for the estimation of radionuclide concentrations during the first period; resolution of the calculations became increasingly coarse for the second and the third periods.

The committee is concerned about omissions in the documents that should have been addressed. For example,

• Pathways other than those mentioned above should be included if specified criteria are met. The nature of those pathways, and criteria which are not covered in DESCARTES were not described.

• DESCARTES itself, which is used to calculate the environmental concentrations of the radionuclides of interest, is not adequately described. The only information available is a list of equations in report PNWD 1983.

• Although up to 12 radionuclides are considered in the atmospheric pathways, the radionuclide-specific parameter values used in DESCARTES are given only for iodine-131.

  Quality control is important and needs further discussion in the documents. For example:

• During 1944-1949, very few monitoring data were available. The report did not state how DESCARTES was validated for all the radionuclides considered.

• For the other two periods, 1950-1972 and 1973-1991, it seems that the available monitoring data are accepted without question. The amount of effort expended to review the validity of those data should be described.

• The methodology for the uncertainty analysis conducted for modeling iodine-131 dose is impressive. However, biases that tend to increase the dose estimates have been accepted (for example, in the case of the chemical form to be considered), and triangular distributions have often been used when lognormal distributions might be more appropriate. The deposition velocity and the immersion-dose rate factor, which differ substantially from usual literature values, need to be corrected or justified.

With respect to the issue of clarity and detail, report PNWD 1983 is difficult to read by itself without detailed knowledge of the entire project. Some sections, such as Section 2.6, cannot be understood by people unfamiliar with the contents of the codes RATCHET, DESCARTES, and CIDER. Table 3, presenting the modeling techniques anticipated for use for each of the pathway-time combinations, was missing in the copies that were originally provided to this committee. The committee strongly believes that this document should be “self-standing” so that any scientist or administrator interested in the HEDR

project can understand and use it. This would require, at the very least, short descriptions of the computer codes and of the various tasks.

Public involvement is not mentioned in the documents under review, although it is known that the TSP makes every effort possible to ensure that the public is involved in the HEDR project. The committee believes that such efforts should be mentioned in the documents.

Pathways of exposure to contaminated river water include: direct consumption; swimming; boating; shoreline exposures; consumption of resident fish, anadromous fish, and other seafood; and possibly consumption of irrigated crops. The exact level of detail for modeling of the river and its pathways has not been determined.

Very little information on aquatic pathways is provided in report PNWD 1983. The committee believes that the modeling approach is conceptually sound, but the quality of the work will depend on how the modeling approach will be implemented and documented with respect to the completeness of the data. The committee recommends that because pathways other than those mentioned in the document are to be included in the future by HEDR if specified criteria are met, the nature of those pathways should be indicated in the document. In addition, onsite groundwater pathways, including surface runoff, are not considered; the committee recommends that this decision be justified.

As previously mentioned, quality control is important and is discussed for water pathways more than for the atmospheric pathways. An extensive database is available of measurements made of various radionuclides at various times and at various locations of various media associated with the river. This database will be used to validate the models. The committee recommends that the uncertainty analysis for the radionuclides considered be as thorough as that conducted for iodine-131 for the atmospheric pathways. With respect

to the issues of clarity and detail, the committee feels that the information given in the documents regarding water pathways is sparse, but that what is provided is quite clear.

Doses will be assessed on the basis of the environmental concentrations of radionuclides in air, soil, and terrestrial foodstuffs for the atmospheric pathways and in water, sediments, and aquatic foodstuffs for the surface-water and groundwater pathways. The method to be followed is briefly described in report PNWD 1983.

For the atmospheric pathways, the environmental concentrations obtained in the model DESCARTES are used in the model CIDER to calculate individual doses. The equations used in CIDER are presented in report PNWD 2023. Little information is provided on the surface-water and groundwater pathways. The models are not fully explained, and there is little discussion how they have been or can be validated.

With respect to the issues of uncertainty and reliability, the committee is concerned about several important gaps in the four documents. For example,

• The nature of radiation doses and the number of people exposed should be specified. Approaches to be used in calculating radiation doses throughout the various

periods of operation at the Hanford site are described. However, the radionuclides that will be considered in the dose calculations are not identified. Also, it is not indicated whether the persons whose doses are to be calculated will be specified (i.e., identified by name) or unspecified (i.e., represented by hypothetical persons). The estimation of doses to specified persons requires much more work than the estimation of doses to unspecified persons, and it would be useful to indicate how many individual doses will be calculated. In addition, the type of annual dose that will be calculated should be made clear (are these effective doses or doses to particular organs or tissues?), as well as the dose coefficients that will be used.

• The decision to exclude the ingestion pathway in the doses received by the populations on site should be justified.

• Additional pathways will be included in the assessments if it is determined that such pathways have “the potential to add more than 5% of the total annual dose for any individual at a time when the dose exceeds the TSP guidelines.” But those TSP guidelines are not specified in the documents provided to the committee, and it should be indicated whether only the effective dose or the dose to any organ or tissue of the body is considered, as well as how the effective dose was estimated.

• The distribution systems of agricultural products are taken into account for the 1944-1949 period but not for the periods that follow. That difference should be justified, especially because the dairies are not uniformly distributed around the site and their operation changed in the postwar period.

• How the individual doses are to be validated and the expected uncertainty estimates associated with them should be specified.