Review of Methods for Estimating Radiation Doses to Workers at Hanford: Letter Report (2001)

Page iii

Footnote 1: This description of effective dose needs to include mention of the radiation weighting factor as well as the tissue/organ weighting factor.

Page iv

The discussion would be much clearer if it were pointed out that the two sets of particle contamination discussed (the rusting of the duct work and the Ru releases) were from separate facilities. Some discussion of the facility names would be useful before the discussion of emissions.

“released primarily from the REDOX reprocessing facility.”: Is there any indication that these came from anywhere else?

“twelve release points, 5 processing facilities in the 200 Areas and 7 reactors.”: Is that a total of 24 locations, or are there just 12? At this point of the text, most readers would have no idea what the “200 Areas” are. Some geographic background is needed to introduce the locations described.

Page vii

It would have been helpful to include a column in Table S-1 for the direct radiation dose so that the reader can check to see whether the effective doses add up appropriately.

Page ix

For the “worst-case” particles in Table S-2, the radioactivity and the likely minimal particle size necessary to achieve the activity are not specified. The particle size for the supposedly respirable particle in this table is indicated elsewhere (p. 3–46) as 20 micrometers; however, almost all of the particles that reach the deep lung are 10 micrometers or less. This table summarizes the calculated results of inhalation and ingestion of two general categories of an effective Hanford particle. For the first part of the table, the comparison is made for a “1940 Active Corrosion Particle.” This is a clear comparison based on inhalation (Table 3–11) or ingestion (Table 3–9) of a particle containing 5 µCi of a defined mixture of beta-emitting radionuclides. Observed differences in the dosimetric results reflect the differences produced by two different routes of exposure. The second part of this table, dealing with a “1950s Active Ruthenium Particle” is not as clear. Here, the comparison is made using a 20-micrometer particle containing 2.2 µCi ¹⁰⁶Ru from the Redox plant (Table 3–12) for the inhalation exposure and ingestion of an active particle

containing 300 µCi (Table 3–10). Here the dosimetric differences between the inhalation and ingestion routes are heavily influenced by differences in the particles considered. Particle characteristics should be included in the table to emphasize these differences, but as they are not, the results for the “1950s particle” are very confusing.

Page ix

“It is unlikely that there would be any records reflecting internal body contamination for the early years.”: On the contrary, sampling of feces of four men whose work was primarily in the stack area is mentioned in HW-10261, dated 6/11/1948, indicating that some records might be available. HW-12869, dated 3/1/1949, discusses the negative results of 24 sputum samples taken from Health Instrument volunteers working for various fractions of time outdoors in the 200 area. HW-7920, dated 10/30/1947, indicated that urine testing for ¹⁴⁴Ce was in progress and that feces testing for ¹⁴⁴Ce was proposed. Of course, any such records might have been destroyed or, as indicated in the first meeting of the committee, might be in medical files that are confidential.

Comments on Section 2 (Radionuclide Releases to the Atmosphere)

Page 2–3, Section 2.2.1, first paragraph

It is difficult to reconcile this paragraph with the summary of HW-7865 provided in the bibliography. The latter does not give the beta-to-alpha ratio, nor does it give the stated activity ordering for elements (Ce, Y, Sr). Examination of the source document indicates that the statements are accurate, but it would be helpful if the information in the text were included in the bibliographic summaries. Similarly, the list of additional investigations said to be provided by Parker in HW-7920 is not given in the bibliographic summary. In that case, however, no distinction has been drawn (as was drawn in the source document) between tests underway and tests proposed. For example, assays in animals were proposed but were not in progress.

Page 2–3

In the bibliography for the active particles, one would expect to see references to HW-9259 and HW-11082.

Page 2–4

“Assuming that this range reflects the 1st and 99th percentile of the distribution, we find that a consistent lognormal distribution of particle diameters would have a median of about 180 µm and a geometric standard deviation (GSD) of about 2.5.”: What is meant by “size” in this paragraph? What does “consistent” mean in this context? There is no need to make such assumptions about percentiles of the distribution. Figures 1, 1A, 2 and 2A of HW-10261 provide histograms for the distribution of sizes of 111 particles “picked at random”, allowing better estimates for the statistics of a lognormal distribution. The “sizes” provided in these figures are volume (Figures 1 and 2) or cross-sectional area (Figures 1A and 2A), either of which could be related to an aerodynamic diameter or to a mass-based “size”.

What is an “effective density”? The phrase “smaller fraction of the distribution was not included in the sample studied in detail” is unclear.

“The average mass of particles segregated for analysis was reported to be 1 mg, with…”: The bibliographic summary of HW-7920 states that “particles that could be segregated have a mass of 0.1 to 1 mg”, a substantially different statement; but the bibliographic summary is incorrect.

Page 2–5

“If the mean diameter were 3 µm and the GSD of the distribution were 2.5 (as above), the geometric mean particle diameter would be about 2.0 µm. The range of diameters corresponding to the 1st and 99th percentile value would be 0.2–20 µm. The upper tail of this distribution would just reach the lower end of the distribution of the larger particle diameters described above.”: This appears to be speculation. The report should include a summary of the data that support these statements.

“The ratio of the surface area for the 95th percentile diameter of the distribution of large particles to the surface area for the 5th percentile diameter of the distribution of small particles is about 2×10⁷. This is broad enough to account for the observed variation in activity per particle of 2.5 pCi to 3.2 µCi (HW-10261).”: Was that range given for the large particles only (the 111 specks picked out for examination) or for the full distribution (the bibliographic summary is not clear)?

Page 2–6

The reader is referred to HW-28780 for particle-size distribution. This refers to particles in the 291-S stack (according to the bibliographic summary). The bibliographic summary for the later reference (HW-32209) talks about the Redox 200-ft stack. Are these the same? The bibliographic summary for HW-32319 discusses ventilation of building 202-S, stating that there is a 200-ft stack 300 ft NE of the building. The last also states that “all REDOX stacks are designed to discharge at 3000 feet per minute”, suggesting multiple stacks. The discussion and bibliographic entries are confusing.

“The distributions of particles…were broader.”: Does this mean that the variance was larger?

“The theoretical density of that material is about 7 g cm^-3, and the estimated effective density is about 4.9 g cm^-3.”: How were these estimates made?

Page 2–7

“In the absence of information, a GSD of ~3 is assumed for this particle distribution.”: How can one use a GSD of approximately 3? Why 3? What are the consequences of this choice? There should be an explanation of this choice.

Page 2–9

A discussion of potential releases from the reactors due to cladding failures suggests that releases from the reactors would be small. However, the failures were not necessarily as benign as indicated by this discussion. For example, HW-46726 indicates that the regular estimates of emissions from the reactors did not include specific events, in this case, “one sample not included in Tables IX, X, and XI was collected on September 17, 1956, from 100-C reactor stack during burning of a uranium slug on the rear face.” An estimate of a total of 0.6 Ci of beta emission is provided, but no estimate of U and Pu emissions appears to be given.

Other relatively large events are noted in the reports and omitted from the summary tables. Tabulating such events would add to the value of the report. Two reports attempting such a tabulation are HW-54636 and HW-84619, although the committee does not know whether they are complete in their tabulations even for the period they cover (1952–1964). HW-54636 summarizes four incidents involving ruptured or burning fuel elements in 1952–1957. In addition, there is a discussion of several contamination incidents involving overflow, break, or dry out of impoundment basins (for contaminated coolant water). Overflows and breaks resulted only in local contamination, but dry outs led to the spread of wind-blown contamination for several miles downwind. The Ru emissions from the 200 area are discussed in HW-54636, but that report also lists two other 200-area stack emission incidents and contamination incidents in the 200-area burial grounds. The 300 area suffered three burial ground fires and a laboratory fire. Although HW-54636 dismisses most incidents as causing only local ground contamination, there might have been some associated inhalation exposures at some distance. HW-84619 similarly summarizes environmental contamination incidents in 1958–1964. In the 100 areas, three instances are given of contamination spread by high winds when the 107 basins were dry, and two instances of burning fuel elements are mentioned. In the 200 areas, the excess emission of radioiodine due to an inadvertent dissolution of short-cooled fuel is mentioned (63 Ci of ¹³¹I emitted, September 2–3, 1963), as are eight other incidents.

Page 2–11

Figure 2–2 shows “estimated ¹³³Xe releases”, and the text gives a short sketch that allows us to hypothesize that they were estimated from the fuel-processing rates discussed in Section 2.1; but there is no direct statement that it is RAC estimating these. Some description of how they were estimated is in order.

Page 2–12

“In general, it is reasonable to expect that the isotopes did not selectively attach to particles independently and that release fractions for the several isotopes should be similar.”: What is the basis for this statement? Whether it is correct depends on the source of the particles and the mechanism of their formation. Some discussion of the source of the particles is in order. In addition, and to the extent possible with contemporary data, the hypothesis needs to be checked against empirical data.

“The ⁴¹Ar releases estimated by Heeb (1994) were adjusted for an operating fraction of 0.9 for these calculations.”: Why did Heeb (1994) or RAC not incorporate actual operating experience?

“Because the radionuclide releases changed dramatically with time as cooling times were increased and effluent treatment systems were added,”: This implies that the release fractions changed with time. That seems likely for the T and B plants after installation of sand, then glass fiber, filters. Changes were also made in the Redox plant air system (shutdown period, June 10– July 18, 1954). What happens if the measured release fractions are segregated by time and by plant?

Pages 2–9, 2–12, and 2–13 and Page 2–14, Tables 2–2 and 2–4

Where do the estimates shown in Tables 2–2 and 2–4 come from? Table 2–2 states no source. Page 2–9, states that “these calculations show that”; what calculations? Heeb (1994) is given as the source for estimates of ⁴¹Ar, ³H, and ¹⁴C from the production reactors (page 2–9), ¹³¹I from fuel processing facilities (page 2–11), and ¹⁰³Ru, ¹⁰⁶Ru, ¹⁴⁴Ce, 90Sr, and ²³⁹Pu (page 2–12). That leaves ⁸⁹Sr, ⁹¹Y, ⁹⁵Zr, ¹³⁷Cs, and ¹⁴¹Ce in Table 2–4 unaccounted for. Where do these estimates come from?

Page 2–13, Figure 2–3

This figure is misleading because it is shown as a continuous line. It should show individual points or distributions for individual nuclides. Where is the compilation of data on which it is based? During the public presentation at the first committee meeting, Mr. Voillequé indicated that there were relatively few data points and that they were mostly on Pu with a few each on some other radionuclides. As recommended above, the sources for all such data points should be documented, and the values themselves should be provided in the form in which they were derived. Moreover, there should be some attempt to determine whether the different radioisotopes conform to the same distribution. There is no way to determine whether, for example, all the estimates for Pu fall at one end of the distribution in Figure 2–3 and all the estimates for, say, Ce fall at the other. Similarly, there is no way to determine whether estimates for different plants are different whether estimates made at different times are different.

Page 2–13

“These estimates reflect provisional estimates of the effectiveness of effluent treatment equipment that was installed.”: What does this mean? How were the provisional estimates derived? What were they? Where are they listed?

Page 2–15

The ground irradiation screening value for ¹³⁷Cs does not appear to have been corrected according to the description in the first paragraph. The 30-year buildup value given in NCRP 123 (1996) is 9.9×10^-2 Sv per Bq m^-3. Correction to a 3-year buildup gives a screening factor of 1.33×10^-2 (see calculation below), not the 6.6×10^-3 given in Table 2–5. In fact, the value

given appears to be almost exactly half that for 3 years. It corresponds to approximately a 1.45-year buildup. A 3-year buildup would also make a 10% difference in ¹⁰⁶Ru.

The screening factors given in Table 2–5, but for 30-year buildup time, can be found in NCRP 123, Appendix B, Table B.1 (page 104 contains ¹³⁷Cs). Those for the ¹³⁷Cs−^137mBa chain are (in Sv per Bq m⁻³): inhalation, 6.3×10⁻⁵; plume immersion, 6.5×10⁻⁷; and ground shine 9.9×10−², for a total of 9.9×10−2 Sv per Bq m⁻³ to two significant digits. The method used is described in NCRP 123, Section 8.2.1 (see specifically Equation 8.2 on page 66), and includes a buildup factor given by Equation 5.4 on page 50; see the definition of GC(λ_i,t_b) below Equation 8.2 of NCRP 123. For the two-nuclide chain of ¹³⁷Cs, the decay rate of ^137mBa is so large (half-life of 2.55 minutes, compared with 30 years for ¹³⁷Cs) that it can be neglected, so the buildup factor is just (1−e^−λt)λ⁻¹ where λ is the effective decay constant for ¹³⁷Cs and t is the buildup time. If we use an effective decay constant corresponding to the half-life of ¹³⁷Cs (30 years), the buildup factor is 21.64 years for a 30-year buildup, and 2.90 years for a 3-year buildup; that leads to an adjusted screening factor of 9.9×10⁻²×2.90/21.64=1.33×10⁻² Sv per Bq m⁻³ (because inhalation and immersion contribute negligibly). For this calculation, the committee has neglected the effective physical removal half-life of ¹³⁷Cs from the soil surface due to leaching and harvesting. If a 70 year half-life for those processes is included (NCRP 123, Table 5.1, page 51), the effective half-life of ¹³⁷Cs is shorter than 30 years, and this leads to a larger discrepancy between the value in Table 2–5 and that calculated by adjusting the buildup time to 3 years.

Page 2–15, Table 2–6

Something like this table can be reproduced only if the screening factor for ⁹⁵Zr is set at approximately 1.5×10⁻⁴, not at the (correct) value of 3.4×10⁻³ given in Table 2–5. The rank of ⁹⁵Zr would be 2 if the correct value were used, whereas it was incorrectly calculated as 8 by RAC and dropped from further consideration.

Page 2–17, Table 2–8

This is supposed to contain the “limited information” available from Paas (1953). There are six such references (a, b, c, d, e, and f). Which one is intended? What were the standard deviation, the range, and so on, for the measurements? How were the measurements made? The emphasis on the concentrations should be justified. It seems more logical to look for relationships between the release rates and the power levels.

Page 2–17, Section 2.4.1

The distribution used seems to be inappropriate. The required estimates appear to be monthly average ⁴¹Ar emissions. One would expect steady ⁴¹Ar production at any particular reactor operating condition, not fluctuations from month to month. The uncertainty in that level would be reflected by the standard error of the mean of the measurements. Substantially more information about the measurements should be provided. At the public briefing at the first committee meeting, Mr. Voillequé stated that, to the best of his recollection, Table 2–8 contained all the information available in the original references. The data in Table 2–8 are cited as coming from Paas (1953), a reference that includes six entries in the bibliography, as noted in the

previous comment. The table comes from HW-28009, which also provides the number of measurements (three) for the 105-C stack. That document, however, also refers to HW-27641, where additional data are provided, the numbers of samples taken for 105-H (eight), 105-DR (eleven), and 105-D (four). The additional data permit an estimate of the standard error of the mean for at least three of the measurements given in Table 2–8. HW-28009 states that “spot measurements of the ⁴¹Ar activity density in reactor areas effluent gases will be maintained in the future.” In view of that statement, was a search for further records conducted?

Page 2–18

Top paragraph: No basis is given for the particular triangular distribution selected. In addition, some basis for the triangular (0.8, 0.9, 1.0) average on-line fraction should be provided. Why was operating experience not examined?

“The complete set of icosatiles is contained in an Excel® spreadsheet (Link to ⁴¹Ar Releases.xls)”: There is no such table in any of the spreadsheets that have been made available. There are, however, unexplained entries in cells a88.h136 of the sheet “⁴¹Ar releases” in Section_2.xls.

Page 2–19 to page 2–20

“The same decrease in input rate is seen for ¹⁰⁶Ru in Figure 2–5, although it is not as large.”: In fact, no such decrease appears for ¹⁰⁶Ru for the period cited (May) in Figure 2–5 nor does it appear in the spreadsheet data (in Section_2.xls).

Page 2–20

The values given in the “processing rates” sheet of Section 2.xls have transcription errors. The committee identified the following errors in the file Section 2.xls in the transcription from appendix tables B of PNWD-2222-HEDR (and the committee presumes that these data were the basis for the emission estimates):

Page 2–20

“The neutron capture cross section for ⁹⁰Sr is higher, 0.8 compared with 0.1.”: What are the units?

The inventories of ¹³⁷Cs and ⁹⁰Sr are assumed to be equal, apparently because Heeb (1994) did not calculate the ⁹⁰Sr inventories. Why were ORIGEN runs not used to confirm this rough equality for the burn up used in the Hanford piles? Such runs were performed to obtain the correlations used for the other fission products in the HEDR project and should be available to perform similar correlations for 90Sr. The failure to use already-generated HEDR data is not explained.

“In general, it is expected that a common process led to formation of particles that became airborne during fuel processing and were carried in the process vessel exhaust air streams.”: That might be true for the particles from each specific part of the plant. For the T and B plants, the particles came from at least two areas, and were handled differently (by sand filters and by glass wool filters). What are the implications of the different source locations? Some discussion of the process flow stream and particle source(s) is in order to tell the reader why “in general, it is expected…”.

Page 2–21

“Nonetheless, with this general principle in mind, the data on release fractions for the particulate radionuclides were reviewed.”: What data? No references to such data are provided.

“For the early years of operation of T Plant and B Plant, when there was no effluent treatment, the release factor was increased by an average factor of 150, the mean of a uniform distribution with bound of 100 and 200.”: Which release factor was “increased”? What was the basis for the factor of 150? What is the implication of the “uniform distribution with bounds of 100 and 200”? Was that uniform distribution used somewhere? What are the modeled dates for this change? Are they the same for T and B plants? How do the estimates of fission-product releases obtained with this method comport with the estimates given in HW-10758, which on the basis of direct measurement of stack gases, suggests that emissions were lower in 1945 and 1946 than in 1948?

“To provide a high upper bound, the geometric standard deviation of the distribution was taken to be three.”: Why was a “high upper bound” needed? Why is a GSD of 3 taken to be high, when the release GSD already discussed is larger? What was the basis for this estimate?

“Distributions of monthly total releases were also computed by summing the distributions of release estimates for the individual plants.”: First, what does this mean? Was the sum performed probabilistically, for example; and how were the releases from each plant correlated in the summation? Second, of what interest is it? Of what use is this sum of releases, inasmuch as each release has to be dispersed separately?

Page 2–21 and page 2–22

These pages give the release fraction estimates, but there is no indication of where the values come from. For the whole of Section 2.4.2, there is no reference to any measurement of releases, and the reader cannot tell where the estimates for a release fraction (parameters of lognormal distributions) come from. The only thing the reader can do is guess that the release fractions estimated by Heeb (1994) were used, as described on page 2–12, Section 2.3.3, first paragraph. But where are these data compiled by RAC to make estimates with?

Page 2–22

The fourth paragraph, describing distributions for unusual Ru releases, is incomprehensible. It seems fundamentally incorrect to use a “release fraction” for releases that are not proportional to the processing rate. When were these emissions supposed to occur? Some well-defined events and some not-so-well defined periods are described in the documentation. The total emissions from those events have been estimated. Why introduce this “release fraction” approach when measurements are available?

Page 2–27

Table 2–10 “operating facilities, lagged by two months, were used to estimate the Z Plant processing rates.”: This remark can be better explained.

“These monthly release data and surrogate monthly production amounts were used to estimate?”: How?

Table 2–10. How does one identify a distribution type with only four, five, or seven values? How does one distinguish between lognormal and a loguniform?

“Piecewise uniform distribution,”: What are the intervals or breakpoints? How can this type of distribution be justified with so few data points?

“returned an average value of 7.2×10^-8 was used”: What does this mean?

“The lognormal distribution given for 1957–1958 was assumed to be representative of early operations and was also used to estimate releases for those individual years.”: What are the “early years” and how was this distribution used “to estimate release for those individual years”? What is the justification for using the same distribution for both years and for the particular choice of the parameter values? It is not clear how the “information” contained in this table is to be used. Perhaps a random-number generator is used to sample one of the distributions to generate a release fraction. Is this fraction combined with the surrogate-processing rate to produce the Z plant release? If so, the report never mentions it. There is also a reference to a “cautious uncertainty factor”; what was this, and how was it used? Obviously, there is a great deal of uncertainty associated with the modeling of these distributions, none of which is reflected in later discussions of uncertainties.

Page 2–28

“after the time period of interest in this study.”: What is the period of interest in this study?

Page 2–29

There are repeated references to a “silver reactor”, but the reader is not told what this is or what it does. Other references are on pages 4–13 and 4–29.

“There is no simple way to reconcile these differences.” The committee disagrees with the statement. We note the following: Warren (1961), on page 7, gives his estimate of ¹³¹I content at reactor discharge as a function of power level These all correspond to 23,800 Ci/ton per MW/ton, so they do not account for variable burn up. In his tables, Warren gives “None” as the remaining ¹³¹I going into the dissolver for all cases with a lower end of cooling time over 155 days and most cases with cooling times over 139 days. Wherever he gives a value that is nonzero, however, it is straightforward to compute the ¹³¹I amount at reactor discharge, obtaining lower and upper bounds on the quantity from the range of cooling times he gives. Those bounds can be converted to tons of uranium, assuming power levels of 6.375 MW/ton in 1959 and 7.083 MW/ton (corresponding to the values used by Heeb), and using Warren’s conversion of 23,800 Ci/ton per MW/ton.

Accumulating the results to months yields a direct comparison with the quantity of uranium given in the HEDR documents (and used as the basis for the RAC report). In every case except December 1959, the quantities are consistent: the lower and upper estimates obtained from Warren bracket the quantity given in the HEDR document. Although that calculation cannot account for the fuel cooled over the very long term (which was assigned zero ¹³¹I content at dissolving by Warren), no inconsistency is apparent. In December 1959, the estimate obtained from Warren is 135.6–507 tons, whereas the value from HEDR is 107 tons. However, that discrepancy is relatively small compared with other uncertainties. It might arise from a mismatch on power density, for example, or from Warren’s use of a fixed value of Ci/ton per MW/ton, independent of burnup.

Furthermore, the range of cooling times given by Warren on a day-to-day basis is consistent with the monthly mean given in the HEDR documents, in that the minimum and maximum cooling times in every month bracket the mean value given in the HEDR document.

Thus, there appears to be a simple way to reconcile the differences cited.

The 3-month (and occasionally 1-month) environmental monitoring reports from Hanford contain monthly estimates of ¹³¹I and total filterable beta emissions from each plant stack from about July 1951 through at least 1959 (beyond which the committee did not look). In 1958, online gamma spectrometric measurement was instituted to characterize individual nuclides, and later monthly emission reports included this characterization. Ruthenium emissions from Redox are reported monthly from its startup. Those measurements have not been compiled in the RAC draft report, but they are direct measurements (for ¹³¹I) of the emissions required or place very strong constraints (for the filterable beta emissions and ruthenium) on the fission-product emissions.

By contrast, the measurements recorded in Warren (1961) and used as the basis for the emission fractions of iodine used by RAC for the post-1949 period are only a very limited subset of the measurements just mentioned. Examination of the monthly emission factors obtained from Warren (1961) also suggests that they are not adequate for use as a generic distribution applicable over long periods, in that plotting the release fraction against the input to the dissolvers shows an extremely good inverse correlation: log(release fraction)=-0.59609- 0.56888[log(input to dissolvers)]; r²=0.693. RAC should compile and report the available monthly emission measurements and use them directly for the emission estimates (for ¹³¹I), use them to place constraints on emissions (for other fission products, on the basis of the total filterable beta measurements), or carefully explain the reason for not using them. The committee was able to locate daily total ¹³¹I emissions (all plants combined) for 1959 (HW-64371), but does not know whether daily measurements are still available for other periods or for each plant separately; nor would the use of such daily records substantially improve RAC’s estimates.

For ¹³¹I, RAC has been inconsistent in the use of emission fractions for the post-1949 period. RAC’s emission fractions for ¹³¹I were based on daily measurements reported by Warren (1961). For the T and B plants, the emission-fraction distribution used was that obtained for the daily emission fraction obtained from Warren (1961); this was the method used by HEDR (Heeb, 1994). For the Redox and Purex plant, RAC rejected the daily emission-fraction distribution for the excellent reason that the daily emission-fraction estimates omitted a large fraction of the emissions. Instead, monthly emission-fractions were computed and used. Thus, different methods have been used for the T and B plants and for the Redox and Purex plants, on the basis of different interpretations of the same measurements. If RAC continues to use an emission-fraction approach (rather than the reported measurements themselves, as discussed in the previous paragraphs), the final report should justify the differences in approach.

Some discussion is also in order on the accuracy of the measurements of ¹³¹I emissions on which emission estimates are based, particularly on the potential for plating out in the sampling lines. Such a possibility was evidently known early because of the condensation observed during the “green run” (HW-17381-DEL-REV2). The committee noted a discussion of the stack sampling method used at Redox around 1961 (HW-69205) that would presumably overcome some sampling problems. Some idea of the accuracy of emission-rate estimates might also be gleaned from the verification sampling reported in HW-56993.

“generic uncertainty of 10% (one standard deviation) was applied to the estimates;”: How? What distribution was assumed?

“For later times, a release factor uncertainty with a median of one and a geometric standard deviation of two was applied.”: Why are the earlier releases from the T and B plants supposed to be known better than the later ones? Is it because it was all released? Do these distributions allow more release than input?

“subjective uncertainty factor, described by a uniform distribution with bounds of 1.0 and 1.8, was incorporated to reflect the failure to measure organic iodide releases and losses in sampling lines.”: Is there any chance that the lower end of this distribution would ever be realized? There must have been some organic iodide and some losses in sampling lines.

“On the basis of those estimates, three of which (two for REDOX Plant and one PUREX Plant) were considered extreme, distributions of release factors for the two plants were constructed.”: Where are they? Can the reader see them?

“For the REDOX Plant, a piecewise uniform distribution, which ranged between 0.0064 and 0.15 and returned a mean of 0.032, was used to estimate monthly releases between 1952 and 1961.”: This does not describe the distribution adequately.

“The reduced ranges were chosen because inclusion of the highest values would lead to large overestimates of the ¹³¹I releases in many months.” Reduced from what? How is it known this would lead to such large overestimates?

“These species would not have been included in the measurements used by Warren (1961).”: Some discussion of the methods is needed so that the reader can understand this. Did Heeb attempt to take these into account?

Page 2–29 to page 2–30

The discussion of organic iodide is insufficient. Is it possible to reconstruct the quantities of ¹³¹I emitted during the time these measurements were taken (do the data still exist)? Have there been further reports on the phenomenon?

Page 2–30

“The effects of elemental iodine deposition in the sampling system were not estimated during the years when the measurements of interest were performed.”: Has anyone examined this problem? Were the people involved then unaware of it? Was it a nonproblem—something that was thought about, investigated, and found to be negligible? If there was deposition, was it later scavenged (to appear, slightly decayed, in the next measurements)?

“Limited information suggests that losses of 30–40 percent could have occurred.” What information? How was this estimate obtained?

What was the basis for the use of Warren’s estimate over that of Heeb? The later discussion suggests that Warren had access to detailed information that apparently is not now available. Was it available to Heeb? Can it be made available?

“the assignment of little ¹³¹I activity to the PUREX Plant processing”: Whose assignment? There is no way to tell from the text.

“The release estimates cited do not validate the estimates of Warren (1961) for the amount of ¹³¹I in the fuel, but they do agree very well with his estimates of releases.”: Whose “release estimates cited”? Or are these the release measurements?

“That agreement suggests that Warren had access to both the detailed processing information and effluent release data. The cooling time estimates of Warren (1961) are also more consistent with

the release data.”: But it has just been stated that Warren apparently gets the ¹³¹I in the fuel wrong. If that is not what has just been stated, the previous statement is extremely confusing.

“In contrast, comparison of the predicted total releases for 1959–1960 with measurements (Junkins et al. 1960, Foster and Nelson 1961) shows that the observations, with revisions for sampling losses, lie within the range of estimated releases.”: Where do we see this agreement? What “revisions for sampling losses” were applied here?

Page 2–30 to page 2–31

It is not clear from the text whose calculations are referred to for Figure 2-11. Were these calculations performed by Heeb, by Warren, or by RAC? Whose “predicted values” are substantially exceeded by the measurements? [Mr. Voillequé clarified at the first committee meeting that these were RAC’s predicted values.] The distribution in Figure 2–11 suggests that these are the calculations of RAC. The committee tried the effect of imposing a reasonable distribution assumption on the range of decay times (uniform distribution with some common minimum) and found that the discrepancies largely go away: there is at most only a 100-fold difference between max and min. Was there a measurement covering month 168, the month with minimal processing? What periods did the measurements cover? Was there a measurement covering February 1959, the month with minimum estimated emissions?

“Releases in late 1960 appear to be residuals from earlier processing and that may be true for early 1959 as well.”: Would a better model for ¹³¹I emissions include a lagged, decayed version of the previous month, supplemented with known special emission episodes?

Page 2–31, Figure 2–11

Do the measurements shown represent all available measurements? Are these Junkins et al. (1960) and Foster and Nelson (1961), or are there others? What “estimated corrections for sampling losses” were applied to the measurements in Figure 2–11?

Why does Figure 2–11 show estimates only every 6 months? Are these 6-monthly averages or individual months every 6 months? What happened between them?

Comments on Section 3 (Dose Calculation Methods)

Page 3–5

The wind speed, u, varies with height above ground. The height that is considered should be specified.

The effective release height, h, is not explicitly considered in Equation 3.2.1–1.

Page 3–6

What value is assumed for the mixing height, H?

Is it correct to say that “Eqs (3.2.1–4) and (3.2.1–5) give the concentration C”?

“the breakpoint occurs”: There is no breakpoint; the curve is continuous. The only break is in the formula used to calculate it.

Is the reference (source) for Equations 3.2.1–4 and 3.2.1–5 EPA (1995)?

Page 3–8

Integration is implicit in Equation 3.2.1–6. Are the results presented taken from EPA (1995) or were they obtained by RAC? Last line of Section 3.2.1, reference for more-complex decay schemes, why are they not needed for the Hanford source term?

Page 3–9

Lines 1–2 say that the modeling of the processes is “similar” to that in EPA (1995). Why? If it is only “similar”, what are the differences and the reasons for them?

It is good to have some of the equations used to estimate dry deposition, but it would have been more helpful to the reader to include figures or tables with numerical values of v_d as a function of the variables considered: particle size, wind speed, atmospheric stability, and so on.

Page 3–10

The cited reference (Figure 1–11 of EPA 1995) gives scavenging rates for particle sizes from 0.1 to 10 µm, not 0.1 to 10 mm diameter. Furthermore, the values given in Table 3–3 are uniformly 10⁴ times higher than those given in Figure 1–11 of EPA (1995) for particle sizes from 0.1–10 µm diameter.

Line 7 says, “interpolates”, why and how? On the first line after Equation 3.2.2–8, which “first two terms”, and why are the other terms neglected?

The reference for the vertical term below Equation 3.2.2–6 is Equation 3.2.1–4, not –2.

The viewing link is incorrect in the .pdf file. It should be possible to integrate all the terms of the equation, not just the first two.

Equation 3.2.2–6

The upper limit of the integration is H if the rain falls from above H, and the height of the cloud if the cloud is below H (although then the scavenging coefficients have to be modified to take account of in-cloud scavenging).

Page 3–11

The second line after Equation 3.2.2–11 says “optionally apply”. Why? At whose “option”? Does this mean that the user of the program decides? If so, on what basis?

Page 3–13

How does “based on…” differ from “taken from…” Were the entries modified or reproduced verbatim?

Page 3–13, Table 3–4

Why is there a 30-year buildup time for radioactivity in soil? It was said earlier that this was modified to 3 years.

This table gives a single value of 1000 m/d for the deposition velocity. Does this include both wet and dry deposition? How is it derived from Equations 3.2.2–1 to 3.2.2–7?

Pages 3–14 and 3–15, Section 3.2.4, Item 4

The text indicates a 12-year buildup period, contradicting Table 3–4.

Are the values given in Table 3.5 on this page for Ba, Rh, and Pr used anywhere in the report?

Sections describing the models used for inhalation and for external irradiation seem to be missing.

Page 3–15, Section 3.3.1

Does the power curve used for the wind speed versus height calculation depend on stability class? If not, why not?

Page 3–16, Section 3.4, Line 5

“conservative approximations,”: What are they, where do they affect the codes, and in what sense are they conservative?

Page 3–29

Line 2 (after file illustration) uses “frequencies”, but these are counts and hence cannot be “normalized to add to 1.0”. Presumably, each instance of “frequency” or “frequencies” should be “relative frequency” or “relative frequencies”.

In the second paragraph, does the table from the source document not identify the category? What does “on the average” mean in this context? What is the justification for the claim that

“object oriented methods have been used in its preparation to improve the organization, to help avoid programming errors and to facilitate locating and correcting errors that do occur”?

Page 3–30

In what way was the code tested and validated, and by whom? What were the results?

Does spline.c, include or depend on the code available in Numerical Recipes? If not, what is the source? If so, it should be cited? Why use splines for interpolation? The spline is used when the objective is generally a maximally smooth interpolated surface. Is that appropriate in this application? What is being interpolated? Apparently, a number of assumptions are hard-coded into the program. There should be a list of them (parameter values, algorithm choices, and so on).

Page 3–31

Where is the “script” behind SURVEY accessible? The file “survey.xls” available on the CD-ROM does have the “script” imbedded in it; that is, once the file is loaded in EXCEL, one can read the formulas one cell at a time. That is not quite sufficient. The formulas should be included, perhaps in an appendix, in the report. In what way was the script tested and validated and, by whom? What were the results?

Page 3–33

“The wind-driven resuspension of radionuclides routinely released from the Hanford facilities is treated in our Hanford Calculator model using a mass loading approach.”: But this is not documented in the description of the model algorithms, although a “roughness length” for resuspension is shown in the input on page 3–23.

Page 3–38

There is a strange citation of EPA (1985) for details of distribution of breathing rate. The standard reference would be to EPA (1997) (Exposure Factors Handbook).

Page 3–39

“If the number of particles contacted is much less than one, then there is only a slight probability that contact would have occurred at all under those conditions,”: What frequency (count) is “much less than one”? Is zero “only a slight probability”?

Page 3–44

“Further investigation of the probability of inhalation as a function of location and particle size is planned. Results will be included in a subsequent version of this report.”: We cannot comment on this.

Page 3–46, Table 3–12

This table gives doses resulting from inhalation of an active particle but the ICRP report (publication 67) referred to in footnote a deals with exposure by ingestion. Should not ICRP publication 71 be the reference here?

References

All of the ICRP publications listed (56, 60, 66, 67, 68, 71, 72) have been published in the Annals of the ICRP. It would be helpful to the reader to use a consistent set of references to the Annals here as is done for ICRP 72.

Comments on Section 4 (Historical Environmental Monitoring Data)

Page 4–1 and following

“…we have relied most heavily upon the quarterly environmental reports published during 1945–1955.” A complete list of the reports that were relied on is in order, particularly including the quarterly reports mentioned here. The bibliography includes a few of them, but their explicit identification (among the more than 50,000 declassified documents available from Hanford) would greatly benefit the reader.

Page 4–9

“In summary, all available exposure data using M and S ionization chambers were compiled into electronic spreadsheets for the following 13 locations between July 1945 and December 1955.”: Where are these spreadsheets?

Page 4–20

“We compiled the data from 10 locations for 1946–1955 into an Excel spreadsheet.”: Where is it? Why this particular period? Was this the limit of available data, or was there some other consideration? What happened after 1955, for example?

Page 4–23

“lead pig”: The committee knows what this is. Will other readers?

“Stack samples and air filters were counted on the first shelf below the window, and vegetation samples were counted on the second shelf because of the bulky pellet.”: Without some further description of the geometry of the counting apparatus, this sentence is uninterpretable.

Page 4–29

“We compiled data for nonvolatile beta activity in vegetation samples from the quarterly reports covering 1949–1955 for nine locations.”: It is unclear what governed the timeframe for this and other compilations. Was this the limit of available data, or was there some other consideration? What happened after 1955, for example? Were these nine locations the only ones available, or were others omitted deliberately for some reason?

Page 4–31

“The decrease in contamination observed near REDOX in 1955 is believed to be due to radioactive decay and weathering of ruthenium released during 1953–1954.”: The maximum appears to occur in January 1955. That is not consistent with release in 1953–1954. It could have been released in December 1954. Is this consistent with the estimates for release of Ru?

Comments on Section 5 (Example Calculations)

How was the solubility of particles factored in the dose calculations? Are there data on solubility?

Page 5–5, Table 5–2

Ingestion of soil is a pathway that does not seem to have been covered in Section 3. How was it modeled? The total effective dose for ¹³¹I+ ^131mXe should be checked.

Page 5–8

The ingestion dose from ¹³¹I depends strongly on age. Were young children considered in the dose estimation? Was the inhalation dose received indoors considered for members of the public?

Page 5–9

“Foot” is not an SI unit

Pages 5–9 through 5–14

The reader expects to see some dosimetric results here, even if they are given elsewhere in the report.