National Center for Environmental Health Radiation Studies Branch Radiation Studies Branch | NCEH | Contact Us

Summary of Findings

Estimates of the releases of ¹³¹I to the atmosphere from the Hanford operations between 1944 and 1972 are documented in the HEDR document PNWD-2222 HEDR (Heeb, 1994). This document was reviewed to determine if the HEDR calculations were performed accurately and in a technically defensible manner, and if the estimates of the source terms from the HEDR calculations could be used for the estimation of air concentrations, deposition, and doses. Our review indicates that a number of errors exist in the HEDR source terms, particularly for the time period after 1951. The magnitudes of these errors are large enough so that an independent estimation of the source term should be conducted for this time period. Our review of an independent estimation of the source term conducted by Herrmann and Herrmann (1996) suggests that this document provides technically defensible estimates of the source terms for the 1948-1960 period.

A summary of the errors found in the HEDR calculations for the 1951-72 period is presented in Table 1. These errors are discussed in detail in Sections 2 through 7. A review of the approach used by Herrmann and Herrmann (1996) is presented in Section 8.

Table 1 Summary of errors found in the HEDR source term calculations (Heeb, 1994) for the 1951-72 period with the magnitude of underestimation and the time periods to which the errors apply.

Table 1 (Continued)

Table 1 (Continued)

1 Factor of underestimation of the true monthly release.

2 Magnitude of uncertainty is expressed as the ratio of the upper-bound value to the lower-bound value of the 95% confidence interval of the monthly source term.

3 The bias induced by using the median instead of the arithmetic mean is of secondary importance since the distribution of release factors in Heeb (1994) does not properly reflect uncertainty in monthly values.

1 Introduction

The Hanford Site was built in Washington State during World War II to provide plutonium for the United States weapons program. Plutonium produced in the nuclear reactors at Hanford was chemically processed to obtain weapons-grade plutonium. Four chemical separations Plants (T, B, REDOX, and PUREX) were used at the Hanford Site between 1944 and 1972. The T Plant was in operation between December 1944 and February 1956; the B plant was in operation between April 1945 and June 1952; the REDOX Plant was in operation between January 1952 and November 1966; and the PUREX Plant was in operation between January 1956 and December 1972. Spent uranium fuel slugs contained in aluminum cladding were dissolved in these plants first in a sodium hydroxide (caustic) solution, after which the bare uranium metal was dissolved in a concentrated nitric acid solution. A number of radionuclides, including ¹³¹I, were released to the atmosphere during the dissolution process.

Various radioiodine emissions control equipment were in place in each of the separation plants between 1948 and 1972. Figure 1 presents the emission controls that were in place on the T and B Plants and Figures 2 and 3 present the emission control equipment that were in place at the REDOX and PUREX Plants during their operation at Hanford. Figure 4 through 7 present the complete operation history of the T, B, REDOX, and PUREX Plants, respectively, with respect to the radioiodine emission controls in place at various times. Heeb (1994) discusses the HEDR methodologies that were used to estimate the amounts of ¹³¹I released from the Hanford separations plants between 1944 and 1972. The following sections present the specific findings of this report.

2 HEDR (Heeb, 1994) Estimates of the Amounts of Iodine-131 Processed and Released from the REDOX Plant During 1959-60 are Inconsistent With and Substantially Less Than Amounts Documented for This Time Period (Warren, 1961)

All HEDR release estimates after August 1951 (Heeb, 1994) were based on release factors reported in Warren (1961). Warren (1961) tabulates the amounts of ¹³¹I processed in the REDOX Plant and released from the REDOX Plant stack for the 1959-60 period. The curies of ¹³¹I processed and released, as reported in Warren (1961), were compared with the curies processed and released, as reported in Appendix B.1 of Heeb (1994). A major discrepancy was found between the numbers reported in these two documents, as shown in Table 2. It appears that the amounts of ¹³¹I processed and released during each month of 1959 and 1960 have been severely underestimated by Heeb (1994). On an annual basis, the amounts of ¹³¹I processed during 1959 and 1960 at the REDOX Plant have been underestimated by factors of 67 and 12, respectively, whereas the releases have been underestimated by factors of 61 and 30, respectively.

Figure 1 Iodine containing streams in T and B Plant [taken from Herrmann and Herrmann (1996)].

Figure 2 Iodine containing streams in the REDOX Plant [taken from Herrmann and Herrmann (1996)].

*200 Area Monthly Report 1954

Figure 3 Iodine containing streams in the PUREX Plant [taken from Herrmann and Herrmann (1996)].

Table 2 Comparisons of the amounts of ¹³¹I processed and released as reported in Heeb (1994) and as reported in Warren (1961). The reported releases from Warren (1961) represent measured values.

It is not at all clear as to what caused the discrepancy between Heeb (1994) and Warren (1961). Heeb (1994) does not provide any rationale for the differences between the reported releases in Warren (1961) and the estimated releases in Appendix B.1 for the REDOX Plant. It is also not clear as to how many years of releases are affected by this discrepancy. It should also be noted that, in light of the discrepancies in Warren (1961) and Heeb (1994) in the curies of iodine reported to have been processed in the REDOX Plant, it is probable that Heeb (1994) has severely underestimated the curies of ¹³¹I processed in other years within the 1951-1972 period.

The estimates of the curies processed in the REDOX Plant during 1959 and 1960 by Heeb (1994) and by Warren (1961) were also compared with those reported in another Hanford report by Beckman (1961) and in Herrmann and Herrmann (1996). The comparisons are presented in Table 3. For 1959, the estimates from Herrmann and Herrmann (1996), Backman (1961), and Warren (1961) are very nearly the same. For 1961, however, the estimates from Warren (1961) and Backman (1961) are a factor of 1.8 above the estimates of Herrmann and Herrmann (1996).

Table 3 Comparison of the curies of ¹³¹I processed in the REDOX Plant in 1959 and 1960 as reported in various documents.

Since the entire analysis of Heeb (1994) is based on the measured releases and release factors reported in Warren (1961), the defensibility of the source terms in Heeb (1994) for the entire period after August 1951 is suspect for these reasons: (1) the source terms from the REDOX Plant reported in Heeb (1994) do not match the measured and documented releases in Warren (1961) for the 1959-60 period; (2) the curies of ¹³¹I processed in the REDOX plant as reported in Heeb (1994) do not match the reported quantities in Warren (1961); and (3) at the very least, as a calibration check, Heeb (1994) must be able to reproduce the results of Warren (1961) for the 1959-60 period because the entire analysis in Heeb (1994) is based on the release factors presented in Warren (1961). It should be noted that a similar calibration-check could not be performed on the same type of data reported in Warren (1961) for the PUREX Plant because the data were presented in graphical form and the ordinate axis could not be read.

3 HEDR Misapplied Measured Release Factor Data from 1959-60 to the 1951-57 Period When Fewer Emissions Control Equipment Were in Place

 

Release factors for HEDR (Heeb, 1994) were developed from data collected (Warren, 1961) between January 1959 and December 1960 at the REDOX and PUREX Plants. During 1957, several improvements were made in the ¹³¹I emission control equipment at the REDOX and PUREX Plants. These improvements resulted in an increased ¹³¹I removal efficiency and lower ¹³¹I releases to the atmosphere (Warren, 1961). The applicability of these lower release factors to the earlier period (1951-1957) is, therefore, not defensible.

As shown in Figures 2, 3, 6, and 7, process upgrades during 1957 for the REDOX and PUREX Plants included the addition of an acid absorber followed by a caustic scrubber (Warren, 1961). The acid absorber was believed to have provided additional capacity for the removal of radioiodine from stack effluents. Since the caustic scrubbers appeared to be unnecessary for radioiodine removal, they were converted, in September 1958 in the REDOX Plant and in April 1958 in the PUREX Plant, to nitric acid absorbers (Warren, 1961). Prior to the upgrade in 1957, effluents from the silver reactor "… occasionally contained significant quantities of radio-iodine for one reason or another…" (Warren, 1961).

Application 1: REDOX Plant, January 1952 – September 1957

Herrmann and Herrmann (1996) used a frequency distribution with a mean release factor of 0.05 for the silver reactors in the T, B, and REDOX Plants based on silver reactor release factors taken on the REDOX . This would apply to the entire control system prior to the addition of the acid absorber. This is a factor of 4 higher than the value of 0.0125 assumed by Heeb (1994). This indicates an underestimation of the monthly source terms during the January 1952 – September 1957 period by a factor of 4 for the REDOX Plant because of the use of data in Warren (1961) that are relevant to only the 1958-60 period to earlier release periods.

Underestimation by a factor of 10 for the 1952-57 period could result from using the additional observation by Warren (1961) that 90% of the iodine to the acid absorber was trapped in the liquid (from laboratory tests). If the generic release factor for the silver reactor and acid absorber was 0.0125 for the 1959-60 period, the release factor from the silver reactor alone during the January 1952 – late 1957 would be about 0.125. It should be noted that the releases of ¹³¹I from the vent line (connected to processes that the dissolver solution is subjected to after the dissolution by nitric acid) are neglected for this application.

Application 2: T Plant, January 1951 – February 1956

Acid absorbers were not added to T Plant between January 1951 and February 1956. Herrmann and Herrmann (1996) used a mean release factor of 0.05 for the T Plant based on a literature review and T Plant data taken in 1954-55. As in the case of the REDOX Plant, the source terms in Heeb (1994) for the T Plant potentially represent underestimations by at least a factor of 4 because of the use of data in Warren (1961) that are relevant to only 1958-60 and earlier release periods.

Application 3: B Plant, January 1951 – June 1952

Using the same information that was used for the T Plant, the source terms for the B Plant presented in Heeb (1994) potentially represent underestimations by at least a factor of 4 because of the use of data in Warren (1961) that are relevant to only the 1958-60 period.

Application 4: PUREX Plant, January 1956 – September 1957

As in the case of the REDOX Plant, process upgrades in the PUREX Plant in 1957 resulted in decreased release factors for ¹³¹I during the 1959-60 period. If the acid absorber captures 90% of the iodine entering it (Warren, 1961) and the release factor of the combined silver reactor and acid absorber system is 0.002 (Heeb, 1994), the release factor for the silver reactor alone must be 0.02. This is 10 times higher than the value used in HEDR (Heeb, 1994) during this period. Therefore, because of the use of data in Warren (1961) as the basis to estimate the source terms, an underestimation by an order of magnitude could have occurred for the HEDR source terms for the PUREX Plant for the period of January 1956 to September 1957.

4 HEDR Incorrectly Accounted for the Inexperienced Operation of the Silver Reactors in B and T Plants During the Early Periods of Their Operation

The silver reactors became operational in January 1951, coincident with a period of sharp increase in iodine inventory to the dissolver. The new system and high load combined to cause operational problems reflected in high release factors reported for these early months (Heeb, 1994). Heeb (1994) uses monthly data through July 1951, but then the data stop, and Heeb reverts to the use of generic release factor of 0.0125 for August and thereafter. The last measured release factor in July 1951 was 0.125. There is no evidence in Heeb (1994) or in other supporting documents (e.g., Paas and Soldat, 1951; Warren, 1961) that the operational problems ended in July 1951. The average value of the release factor measured between January through June 1951 was 0.058, approximately a factor of 4.6 above the release factor used in Heeb (1994). The release factors were also seen to be directly proportional to the throughputs of ¹³¹I into the dissolver. The throughput for the T and B Plants went down by a factor of 5 after April 1952 and February 1952, respectively. Therefore, to account for both inexperienced operation as well as the effect of throughput on the release factor, it seems reasonable to use the value of 0.058 as representative of August 1951 to April 1952 period for the T Plant and as representative of August 1951 to February 1952 period for the B Plant. Consequently, the monthly estimates of the source terms reported in Heeb (1994) underestimates the true releases by an average factor of 4.6 for both T and B Plants during the early period of the operation of the silver reactor.

The methodologies used to estimate the releases of ¹³¹I and the associated uncertainties for the 1944-49 period has been documented in Heeb (1993) and Heeb (1994). We have not completed our review of the HEDR source term for 1944-1950. The focus of the review presented in this document is the post-1951 period. The corresponding methodologies for the 1950-1972 period have been documented in Heeb (1994). Heeb (1994) developed an analytical estimate for the uncertainty of the monthly source terms for this period. The 95% confidence level factor of the source term, which represents the ratio of the upper bound value of the 95% confidence interval for the source term to the median or 50% value of the source term (referred as UF_97.5/50 in the rest of this document), was estimated by Heeb (1994) to be 4.8. The uncertainty in the source term was dominated by a generic release factor that was treated as being equally applicable to each of the four reprocessing plants - the T, B, REDOX, and PUREX Plants. The release factor was defined as the ratio of the amount of ¹³¹I released to the atmosphere from the stack to the amount of iodine-131 processed in a plant.

For the 1950-1972 period, Heeb (1994) estimated the geometric mean or median values for the monthly source terms from T, B, REDOX, and PUREX facilities. These source terms are presented in Appendix B.1 in Heeb (1994). The problem associated with the use of the geometric mean source term for the calculation of the atmospheric concentrations of ¹³¹I is discussed in Section 6. In this section, the primary focus is on the derivation and use of uncertainty estimates.

5.1 HEDR did not propagate the source term uncertainties estimated by Heeb (1994) to air concentrations, ground deposition, and doses

The estimates of uncertainty in monthly source terms for the 1950-1972 period (Heeb, 1994) should have been propagated through to the estimates of the air concentrations and ground deposition. This was not done. Only the single-value estimates (geometric means) of the monthly source terms, as presented in Appendix B.1 of Heeb (1994), were used in the calculation of the air concentrations and ground deposition. Therefore, the uncertainty in the predicted air concentrations and ground deposition represents primarily the uncertainty in meteorological parameters. The rationale for not propagating the monthly source term through to the estimate of air concentrations and ground deposition is provided in the following response provided by Battelle Pacific Northwest Laboratories (Hoewing, 1999a):

"A single average value of the 1950-1957 releases was used, in conjunction with the average of the 1945-1949 releases. In this way, the uncertainty of the release fraction in the later years is assumed to be described by the distribution of release fraction in earlier years. While this procedure does not utilize the uncertainty information presented in PNWD-2222 HEDR, HEDR did not believe this was a problem. For the time period after 1949, HEDR did not and does not have sufficient information to reconstruct either the hourly meteorology or the hourly releases. The process HEDR used to describe uncertainty in the dose calculations was to allow the variability of the earlier period to act as a surrogate for the later period. By randomly selecting the annual releases from earlier period, HEDR expects that this provides an increase in the ranges of the resultant concentration/deposition distributions over any of the previous years, but maintains the internal correlations within any single realization. HEDR cannot easily quantify this increase in the estimated uncertainty range, but because the releases in the later years (1950-1972) totaled only 5.8% of the cumulative release, HEDR believes that the variability that this adds to the dose of anyone present during the earlier period is small."

There are several problems with this rationale, as discussed in the following:

1. The use of uncertainty from the 1945-49 period to estimate uncertainty for the 1950-72 period is not defensible because the processes that led to the releases of ¹³¹I in the two time periods are quite different. The high releases in the 1945-49 period were due to (a) the lack of any filters for most of the time period (1944-1948) and (2) the use of generally shorter cooling times prior to the dissolution of the irradiated fuel. For both periods, the source term was defined as the product of the amount of ¹³¹I processed and the release factor. Although the uncertainty in the estimate of the amount of ¹³¹I processed in the dissolver plants (T, B, REDOX, and PUREX) are not expected to be different in the two periods, the uncertainty in release factors are expected to be different. It is also shown to be different by Heeb (1994).

Heeb (1993) indicates that the 95% confidence interval of the release factor for the 1944-47 period lies between 0.84 and 0.97, with a UF_97.5/50 of 1.15. The mean release factor was reported in Heeb (1993) as 0.905. The release factors changed slightly during the 1948-49 period, as described in Heeb (1994), with the addition of water scrubbers and sand filters. While the mean release factor was reduced during 1948-49, there is no evidence from the documentation provided in Heeb (1994) that suggests that the UF_97.5/50 value deviated significantly from 1.15.

For the 1950-72 period, when the silver reactors were in place, Heeb (1994) estimated the uncertainty in the release factor to be represented by a lognormal distribution with a mean of 0.00567 and variance of 0.000021359. The UF_97.5/50 for this distribution has a value of 4.1. Thus, assuming that the methodology used by Heeb (1994) was correct, the 95% confidence interval for the release factor for the 1950-72 period spans over an order of magnitude, compared to the almost negligible uncertainty for the 1945-49 period. Therefore, the observation from Hoewing (1999a) that "…The (sic) process HEDR used to describe uncertainty in the dose calculations was to allow the variability of the earlier period to act as a surrogate for the later period," is incorrect.

It should also be noted that Heeb (1994) substantially underestimated the uncertainties for the T, B, and REDOX Plants for the 1950-72 period. This issue is addressed later in Section 5.2.

2. According to Hoewing (1999a), "..By randomly selecting the annual releases from earlier period, HEDR expects that this provides an increase in the ranges of the resultant concentration/deposition distributions over any of the previous years…." This process may (or may not) increase the uncertainty in the later years (after 1950) from those in earlier years (1945-49). However, whether this process is adequate to address the uncertainties from components used to retain gases and aerosols emitted from the dissolver that were installed during the later years (which were not part of the analysis for earlier years) is highly debatable.
3. According to Hoewing (1999a), "..For the time period after 1949, HEDR did not and does not have sufficient information to reconstruct either the hourly meteorology or the hourly releases… HEDR cannot easily quantify this increase in the estimated uncertainty range, but because the releases in the later years (1950-1972) totaled only 5.8% of the cumulative release, HEDR believes that the variability that this adds to the dose of anyone present during the earlier period is small." Because of the absence of any emission control equipment between 1944 and 1947 and the presence of very limited, but well-understood emission control equipment between 1948 through 1950 (water scrubber and sand filter), the uncertainties associated with the availability of information regarding the performance of the equipment for the earlier years (1945-49) are quite small. Uncertainty for the later years, however, is dominated by the lack of a defensible knowledge-base about the performance of individual equipment (e.g., the silver reactor and the acid absorber). Uncertainty stemming from reasonable knowledge in the 1945-1949 period cannot be extrapolated in any defensible manner to address large uncertainty arising from lack of knowledge after 1950.

Although it has been assumed in HEDR that the releases in the later years (1950-1972) totaled only 5.8% of the cumulative release, there are reasons to believe that the releases for this period have been severely underestimated, as discussed in Sections 2, 3, 4, and 6. Biases resulting from underestimation of the releases during 1950-72 and correspondingly expanded 95% confidence intervals would mean that the releases in the later years could have contributed significantly more than the value of 5.8% of total releases. For a person born in 1950 or thereafter, these releases may result in significantly higher doses and risks than estimated using the original assumptions made in HEDR.

5.2 HEDR substantially underestimates the source term uncertainty for the T, B, and REDOX Plants

As presented in Section 5.1, the uncertainty in the source term (Heeb, 1994) was dominated by a generic release factor that was treated as being equally applicable to each of the four dissolver plants - T, B, REDOX, and PUREX Plants. Uncertainty in the release factors was developed by Heeb (1994) on the basis of the efficiencies of the emission control equipment reported by Warren (1961). Warren (1961) reported the results of daily measurements conducted between January 1959 and December 1960 on the REDOX and PUREX Plant emission control equipment that consisted of silver reactors and acid absorbers. These measurements (158 for REDOX Plant and 230 for PUREX Plants) yielded the uncertainty distributions for the REDOX and PUREX Plants used to obtain uncertainty estimates for the release factor. The median value for the release factor for the REDOX Plant was estimated to be 0.0125 and it was estimated to be 0.002 for the PUREX Plant. The amount of curies released each month from each of the four plants (T, B, PUREX, and REDOX) is reported in Appendix B.1 of Heeb (1994). The ratio of the amount released to the amount processed during a given month yields a value of 0.0125 for the T, B, and REDOX Plants (after August 1951) and a value of 0.002 for the PUREX Plant. Therefore, the amounts of ¹³¹I released, as reported in Appendix B.1, represent the median of the distribution.

Uncertainty on the generic release factor applied to all four plants was estimated by combining the available release measurements from the PUREX and REDOX Plants from Warren (1961) and using an analytical error propagation method described in Heeb (1994). The geometric mean of the distribution of the generic release factor was reported as 0.0044 [(= exp(-5.43)], and was assumed to be applicable to all releases. This assumption results in a severe underestimation of the release factors for the REDOX, T, and B Plants, and a minor overestimation of the release factor for the PUREX Plant.

Therefore, a bigger problem exists with the HEDR estimate of UF_97.5/50 for the release factor for the T, B, and REDOX plants. For the REDOX Plant alone, a UF_97.5/50 for the release factor was estimated as 16.4 from the data provided by Warren (1961). For the PUREX Plant, a UF_97.5/50 for the release factor was estimated as 3.4. Heeb (1994) estimated the UF_97.5/50 for the generic release factor, applied to all plants, as 4.1. Therefore, the UF_97.5/50 developed by Heeb (1994) severely underestimates the uncertainty for the T, B, and REDOX Plants. For the PUREX Plant, the resulting overestimation is minor.

In the final step of the calculation of curies released to the atmosphere, the uncertainty in the curies processed is multiplied by the uncertainty in the release factor (with the combined REDOX and PUREX data) to yield the overall uncertainty in the source term. The overall uncertainty in Heeb (1994) is represented by a UF_97.5/50 of 4.8. If the uncertainty factor for the curies processed in the REDOX Plant is multiplied with the uncertainty factor for the release factor for the REDOX Plant [with the REDOX Plant data from Warren (1991)], the overall UF_97.5/50 for the REDOX Plant source term is 16.5. This value is also applicable to the T and B Plants. Similarly, for the PUREX Plant, the overall UF_97.5/50 is 4.2.

This analysis clearly shows that the uncertainties in the source terms for the REDOX and PUREX Plants were incorrectly estimated by Heeb (1994). More specifically, the uncertainty in the estimate of REDOX Plant source term has been severely underestimated. The underestimation of the UF_97.5/50 is a factor of 3.4 (= 16.5/4.8). For the PUREX Plant, the overestimation is, however, minor, a factor of 1.14 (= 4.8/4.2).

Miley et al. (1994) documents that the arithmetic mean of the source term was used to calculate the downwind air concentrations and ground deposition from the Hanford releases. This means that the uncertainty in the source term has not been propagated to the estimated air concentrations and deposition for the post-1949 time period. Therefore, the overall source term UF_97.5/50 of 16.5 needs to be propagated to the downwind air concentrations and ground deposition (for the REDOX, T, and B Plant releases); which would suggest that the upper bound air concentrations and depositions will increase by at least an order of magnitude for these releases. Similarly, the overall source term UF_97.5/50 of 4.2 needs to be propagated to the air concentrations and deposition as a multiplication factor for the PUREX Plant releases; which would suggest that the upper bound value for the downwind air concentrations and ground deposition will increase by a factor of at least 2.

We believe that the uncertainty in the estimates of release factors should be calculated using a methodology that is appropriate to the scale of the problem. The source term is estimated on a monthly time scale. Accordingly, the uncertainty in the release factors must be estimated on a monthly scale. Therefore, the endpoint for determining the uncertainty in the release factors on a monthly basis is the uncertainty associated with the average monthly release factor based on the data collected for the 24 consecutive months between 1959 and 1960. This uncertainty is expected to be lower than the UF_97.5/50 of 16.5 obtained for the T, B, and REDOX Plant and the UF_97.5/50 of 4.2 for the PUREX Plant as determined from the corrected analysis of Heeb (1994, as documented in the previous paragraphs). The uncertainty in the monthly release factors have not been estimated here because it was decided to use the source terms from Herrmann and Herrmann (1996) for the estimation of doses, risks, and probabilities of causation.

6 HEDR Uses the Medians Instead of Arithmetic Means of the Source Terms for the Air Concentration and Ground Deposition Calculations

Miley et al. (1994) states that arithmetic means of the monthly source terms were used to obtain the predictions of downwind air concentrations and ground deposition from Hanford for all years after 1950. For the months following August 1951, Heeb (1994) used the release factor for the REDOX and PUREX plants that are presented in a graphical form in Heeb (1994) to compute the source terms. These figures show the release factors to be lognormal distributions. In the narration on pages 4.11, 4.12, and 4.13 of Heeb (1994), it is stated that the median release factors for the REDOX and PUREX Plants were obtained as 0.0125 and 0.002, respectively. The reported median value for the REDOX Plant was independently confirmed from the data points provided in Warren (1961). The arithmetic mean of the release factor was estimated to be 0.0339 for the REDOX Plant from the same set of data points (from Warren, 1961) and as 0.0025 for the PUREX Plant. On a log-probability plot, the data points for the REDOX Plant fall in a straight line, indicating that the distribution for the REDOX Plant release factor is log-normal and not normal. In other words, the logarithm of the release factor is represented by a normal distribution. Therefore, the median (or geometric mean) of the distribution of release factor is expected to be significantly different from its arithmetic mean.

This fact is confirmed by Heeb (1994) in two places. First, in the last paragraph on page 4.16, it is stated that "The (sic) generic uncertainty factor for I-131 released from the PUREX and REDOX plants were based on the measurements provided in Warren (1961). Inspection of the data showed the release factors to be lognormally distributed. The mean and the standard deviations for the two plants are shown in Table 4.6." On page 4.17, the title of Table 4.6 reads: "Mean and Standard Deviation of the Natural Logarithm of the Iodine-131 Release Factors for the REDOX and PUREX Plants."

In Appendix B.1, for releases after August 1951, the ratio of ¹³¹I released to ¹³¹I processed for any month consistently yields a value of 0.0125 for T, B, and REDOX Plants, and a value of 0.002 for the PUREX Plant. In other words, the releases were estimated by multiplying the monthly estimate of ¹³¹I processed with the geometric mean of the release factors. This indicates that Appendix B.1 contains median values of releases for all months after August 1951. To obtain the arithmetic mean values these monthly values of releases must be multiplied by a factor of 2.71 (= 0.0339/0.0125) for the T, B, and REDOX Plants and by a factor of 1.25 (= 0.0025/0.002) for the PUREX Plant. This was not done. Thus, the source terms used in air dispersion modeling do not represent monthly averages after August 1951; they represent monthly medians. This includes a bias towards underestimation of the actual ¹³¹I releases.

Battelle Pacific Northwest Laboratories (Hoewing, 1999b) confirmed that the values in Appendix B.1 of Heeb (1994) represented median values. These values were directly used in the estimation of air concentrations, deposition, and doses.

To summarize, the source terms used to estimate the air concentrations, deposition, and doses for the months following August 1951, were misrepresented as arithmetic means. The resulting underestimation is represented by a factor of 2.71 for the REDOX (between January 1952 and November 1966), T (between August 1951 and February 1956), and B Plants (between January 1951 and June 1952) and by a factor of 1.25 for the PUREX Plant (between January 1956 and March 1972).

If the uncertainties in the source term are ignored for the months after August 1951, as was done by HEDR calculations, the underestimation in source term should translate directly as increases in the estimated central values of the reported air concentrations, deposition, and doses.

If the uncertainties in the source term are correctly taken into account (per discussion of the corrected uncertainty analysis of Heeb, 1994, presented in Section 5.2) for the months following August 1951 (with or without the bias factor of 2.71), the increases in the upper bounds of the 95% confidence intervals can be more than an order of magnitude for releases from the REDOX, T, and B Plants; for releases from the PUREX Plant, the upward bias correction in the upper bounds can be more than a factor of 2. However, if the source terms from HEDR are to be used for further analysis, the overall uncertainty must be reevaluated to reflect the uncertainty in the monthly estimates of release factors, as discussed in Section 5.2; the overall confidence interval of the uncertainty is, therefore, expected to be lower than the corrected estimates from Heeb (1994). As stated earlier, the source terms from Herrmann and Herrmann (1996) have been used for the estimation of airc concentrations, deposition, doses, risks, and probabilities of causation in this report.

7 HEDR Documents Do Not Establish the Accuracy and Reliability of the ¹³¹I Stack Measurements Presented in Warren (1961)

As discussed earlier, the entire analysis reported in Heeb (1994) for the period between August 1951 and 1972 is based on the measurements of ¹³¹I emitted from the REDOX and PUREX stacks during 1959 and 1969, as reported in Warren (1961). The stack sampling for ¹³¹I consisted of absorbing ¹³¹I from the stack sampling line into a caustic scrubber. An early description of the sampling method used at Hanford is described in Miller (1954) as follows:

1. A bypass flow of stack gas (about 0.5 scfm out of a total stack flow of about 200,000 cfm) was passed through a caustic scrubber column. The iodine scrubbing efficiency was estimated as 75% for this early version. The sampling flow is assumed to have continued for the entire dissolver run.
2. The quantity of ¹³¹I captured in the sample was determined by passing the solution through a counting unit (scintillation counter) with an efficiency of 3% (later versions would undoubtedly have used more accurate gamma spectroscopy equipment).
3. Corrections were applied to account for ¹³¹I decay during sampling, for less than full capture by caustic solution, for the counting efficiency, and for background.
4. The relative flows through the absorber and stack were measured and used to calculate the total release to air.

Miller (1954) stated that the greatest error occurred from the estimation of sampler flow. This is an early date for the iodine sampler, which must have undergone improvements because of later developments. The determination of the release of minute quantities of ¹³¹I in 200,000 cfm air flow by means of sampling is a technically challenging endeavor. There are several ways in which the determinations may become biased. For example, the 0.5 scfm sampler flow may not be representative of the 200,000 cfm stack flow, or iodine may become trapped in lines or adsorbed on surfaces and hence may not reach the sampler. However, none of the available documents provides any assessment of the accuracy or reliability of the sampling, monitoring, and counting system or discusses how the accuracy improved/changed as a function of time.

Thus, the release factors reported in Warren (1961) and used by Heeb (1994) have not been established to be reliable and accurate.

Herrmann and Herrmann (1996) present an independent assessment of the HEDR source terms for the 1948-1960 period. The primary difference between their analysis and that of Heeb (1994) is that Heeb (1994) relies completely on the measurements of ¹³¹I releases reported in Warren (1961) for the August 1951 - 1972 period, and Herrmann and Herrmann (1961) rely on the operating information on individual process units used to control the emissions of ¹³¹I to the atmosphere. Drawbacks of the analysis conducted by Heeb (1994) have already been addressed in Sections 2 through 7. In this section, a brief summary of the approach used by Herrmann and Herrmann (1996) is presented.

The primary features of the approach used by Herrmann and Herrmann (1996) can be summarized in the following manner:

1. Herrmann and Herrmann (1996) conduct the assessment in a chronological manner. Their overall methodology to estimate the source terms and associated uncertainty is similar to that of Heeb (1994). In other words, they also estimate the source term as a product of the amount of ¹³¹I processed and a release factor. However, their release factors change as a function of time, depending on the emission control units in place at any given time. Thus, they build a history of operations for each of the reprocessing plants in operation (T, B, REDOX, and PUREX) at any given time and develop release factors separately for each plant.

2. The release factors are based on available information on the operation of individual units. Their primary source of information is the European experience from German and French reprocessing plants. However, they also use information available in HEDR reports, from open literature publications, and from first-hand experience to develop a defensible set of release factors for each plant as a function of time. They provide supporting rationale for the choice of almost all parameter values. This makes their report transparent and reproducible. Most of the values of ¹³¹I removal efficiencies used in Herrmann and Herrmann (1996) are substantiated with references and seem to stand the test of face-validity.

The approach used by Herrmann and Herrmann (1996) for propagating uncertainties is defensible; however, more specific concerns on their approach still need to be looked into, particularly regarding the establishment of month-to-month correlations among parameters and the propagation of uncertainty in the presence of correlations. It should be noted that if correlations are not correctly established and propagated, the potential for artificially reducing the confidence interval of predicted releases cannot be ruled out. Nevertheless, the 95% confidence intervals of source terms from Herrmann and Herrmann (1996) span over an order of magnitude for several months during the August 1951 to 1972 period. Source term estimates from HEDR (Heeb, 1994) and Herrmann and Herrmann (1996) are presented in Table 4.

If it is assumed that the discrepancy identified in Section 2 affects the estimated releases from Heeb (1994) for only 1959 and 1960 and only for releases from the REDOX Plant, it is then possible to apply the correction factors presented in this appendix on Heeb’s source terms and obtain the arithmetic mean estimates of corrected source terms. Heeb’s corrected source terms are also presented in Table 4. It must be noted that it is not known definitively whether the discrepancy identified in Section 2 affects any other year of release besides 1959 and 1960 and whether the releases from T, B, and PUREX Plants are also affected.

A review of the Herrmann and Herrmann (1996) report suggests that the source terms developed therein can be used to estimate air concentrations, doses, and risks for the 1948 to 1960 period. For the 1961-72 period, a new, independent analysis is underway.

For the analysis of dose, risk, and probability of causation conducted in this report, we have used the source terms presented in Heeb (1994) for the 1944-1950 period. For 1951 to 1957, we have used the monthly source terms developed by Herrmann and Herrmann (1996). For the 1951-57 period, we have propagated the uncertainties associated with the monthly source terms to the estimates of air concentration, ground deposition, dose, risk, and probability of causation. The methodology that was used to calculate the air concentration and deposition is discussed in Appendix B. The reasons for not conducting the calculations beyond 1957 at the present time are addressed in Appendix B.

Table 4 Comparison of annual source terms from HEDR (Heeb, 1994), Herrmann and Herrmann (1996), and HEDR corrected for the factors identified in Sections 2 through 6.

a HEDR source terms for 1944 to July 1951 represent the arithmetic mean of the annual releases (Heeb, 1994). HEDR source terms after 1951 represent the median values reported in Heeb (1994).

b These are arithmetic mean source terms developed from the monthly distributions presented in Herrmann and Herrmann (1996). Herrmann and Herrmann (1996) provide the following information regarding the spread of the confidence intervals for the annual source terms: The early years at the T and B Plants, which had higher release fractions, had on the order of 4:1 ratio of the upper confidence interval to the lower confidence interval. The lower release fractions have higher ratios. For example, the REDOX 1 had on the order of 10:1 ratio, and PUREX 3 had approximately a 25:1 ratio. The ratios were higher when looked at on a monthly basis.

c Corrections to the HEDR (Heeb, 1994) values are based on discussions in Sections 2 through 6. Years 1944-50 are not affected by these corrections.

d Estimate includes contributions to the source term from May to December 1948.

e Estimate includes a correction of the releases in 1959 from the REDOX Plant, as reported by Heeb (1994), by a factor of 61.

f Estimate includes a correction of the releases in 1960 from the REDOX Plant, as reported by Heeb (1994), by a factor of 29.5.

 

Based on the numerous problems identified with the source terms presented in Heeb (1994) for the August 1951 to 1972 period, the estimates of releases presented in Heeb (1994) for this period clearly represent severe underestimates of the actual releases.

A review of the Herrmann and Herrmann (1996) report suggests that the source terms developed therein are more defensible than those in Heeb (1994). We have, therefore, substituted the HEDR source term estimates for the 1951-1960 time period (Heeb, 1994) with those from Herrmann and Herrmann (1996) for the estimation of air concentrations, deposition, dose, and risk. We are currently conducting an independent analysis for the post-1960 period.

Backman, G.E. 1961. Iodine Problems in the 200 Areas. HW-89068.

Heeb, C. M. 1994. Radionuclide Releases to the Atmosphere from Hanford Operations, 1944-1972. Hanford Environmental Dose Reconstruction Project. PNWD-2222 HEDR, Battelle Pacific Northwest Laboratories. Richland, WA.

Heeb, C. M. 1993. Iodine-131 Releases from the Hanford Site, 1944 Through 1947. Volume 1 –Text. PNWD-2033 HEDR Vol. 1. Battelle Pacific Northwest Laboratories. Richland, WA.

Herrmann, B. and Herrmann, F. J. 1996. I-131 Release to Atmosphere from Hanford Separations Plants for the Period May 1948 through December 1960. Preliminary report prepared for lawyers in Hanford Litigation. Germany.

Hoewing, K. L. 1999a. Letter to Jim Thomas on January 14, 1999.

Hoewing, K. L. 1999b. Letter to Jim Thomas on January 26, 1999.

Miller, D. A. 1954. Calibration of Iodine Stack Monitor. SEP-W2-2.

Miley, T. B., Lessor, K. S., Eslinger, P. W., Ouderkick, S. J., and Nichols, W. E. 1994. User Instructions for the DESCARTES Environmental Accumulation Code. PNWD-2251 HEDR. Battelle Pacific Northwest Laboratories. Richland, WA.

Paas, R. J. and Soldat, J. K. 1951. Summary of Measurements for the Activity Density from I-131 for the Period September 1950 to July 1951. HW-21891. Development Division, Radiological Sciences Department. Hanford Works. Richland, WA.

Warren, J. H. 1961. Control of I-131 Releases to Atmosphere. HW-68392. Hanford Atomic Products Operation. Richland, WA.

Radiation Studies Branch | NCEH | Contact Us

CDC Home | Search | Health Topics A-Z

This page last reviewed January 6, 2000

Centers for Disease Control and Prevention
National Center for Environmental Health
Radiation Studies Branch