Conferences, Videos & Testimony

Report of September 8, 1997 Workshop

by Kevin L. O'Neill

September 8, 1997

report author: Kevin L. O'Neill with review and concurrence of David Albright, Chair of the Source Term Subcommittee Final draft submitted: November 18, 1997


On September 8, 1997 the Health Advisory Panel's (HAP) Source Term Subcommittee conducted a workshop for concerned stakeholders to review Radiological Assessment Corporation's (RAC's) estimates of plutonium releases from the 1957 fire in Building 71 at the Rocky Flats Plant. Attending the meeting were HAP members David Albright (Subcommittee Chair), Eugenia Abbott, Neils Schoenbeck and Hank Stovall. Several stakeholders attended the workshop. Kevin O'Neill of the Institute for Science and International Security, Shannon Clark of MGA, and Dr. Norma Morin and Pat O'Brien of the Colorado Department of Public Health and the Environment (CDPHE) also attended the workshop. This report summarizes the presentation and comments from the workshop and recommends further issues for investigation. The workshop's primary objective was to provide an opportunity for the public to review and give detailed comments on RAC's assessment of the 1957 fire plutonium releases. The workshop was designed to provide detailed information to stakeholders about RAC's methodology and conclusions, and to allow the public to question and challenge the principal RAC investigator, Paul Voillequé, about his findings. The workshop sought to provide more detailed and thorough comments than is possible at HAP meetings or other public meetings sponsored by HAP or CDPHE. The workshop also proved to be a useful public education tool, notably by giving the public an opportunity to thoroughly understand RAC's assessments and to constructively challenge these assessments. RAC's methodological approach is to estimate plutonium releases over time during the fire, resulting in a number of plausible scenarios that give rise to a range of plutonium releases. The principal presentation during the workshop was given by Paul Voillequé, who described his methodology and offered a preliminary estimate of plutonium released from the fire. Much of the discussion centered on developing alternative scenarios for Paul to consider as he continues his investigation and prepares his report.

Reanalysis of Plutonium Releases During the 1957 Fire

Paul Voillequé introduced his presentation by citing the reasons for reanalyzing the plutonium releases from the 1957 fire. He noted that his analysis had been broadened in response to technical comments received from stakeholders following his initial report. Many of these comments were received during and after the March and May source term subcommittee workshops. Voillequé noted that the outcome of these comments was to develop a release chronology to better describe the fire and estimate the quantity of released plutonium. Voillequé said that the revised analysis considered additional release mechanisms, including a review of the assumed location of the main filter plenum explosion, the significance of plutonium hold up in the booster system vents, and the importance of the natural draft in the stack during and after the plenum filter fire. To account for the greatest sources of uncertainty, the reanalysis considers a range of reasonable scenarios that primarily involve different oxidation conditions of the plutonium and different filter degradation rates and sequences. Voillequé then described the layout of Building 71, the ventilation systems leading from Room 180 to the main filter plenum and the known sequence of events during the fire. He noted that a key assumption in his analysis is that the fire started at about 9:45 PM, or about 20 to 30 minutes before it was discovered at 10:10 PM. He stressed that the actual time that the fire started was unknown. Shortly after the fire was discovered, the building's fans were turned to high speed. Initial efforts to extinguish the fire with carbon dioxide failed, and the fire in room 180 was not extinguished until 10:38 PM. According to the chronology, an explosion occurred in the ventilation system at 10:39 PM and a fire in the main filter plenum was discovered shortly thereafter. The explosion knocked out the filter plenum's exhaust fans, but the supply fans likely remained on until 11:10 PM, when the building lost power completely. Voillequé noted that the fire in the main filter plenum was declared "knocked down" at 2:00 AM, approximately 4 hours after it was discovered. He interpreted this to mean that the there were no flames in the plenum, although filters were probably still smoldering. The fire was not declared "out" until 11:20 AM on the following day. Voillequé then turned to the main sources of plutonium that were released from the fire. This included the plutonium involved in the fire in room 180 (an estimated 9 to 18 kg in various forms), plutonium previously deposited on the glovebox filters and in the booster system vents, and plutonium retained on the booster system and main filter plenum filters. Voillequé noted that a continuing source of uncertainty was the quantity of plutonium retained by the main filter plenum filters prior to the fire. He noted that there were four large releases of plutonium to the filters prior to the fire. These releases constituted the vast bulk of material that he estimated to be held up on these filters. He estimated the quantity from these four events to be approximately 130 grams to the main filter bank. Voillequé was questioned about the completeness of his data. Stakeholders expressed a concern that other, unrecorded events may have contributed additional plutonium to the filter bank. If the sampling data was incomplete, it may not capture possibly large releases caused by these events. Voillequé responded by stating that the daily sampling data from the building's ventilation system were largely complete. He said that he had been unable to identify any large gaps in the sampling data where additional significant loadings to the filters may have occurred. Another source of uncertainty was the quantity of plutonium in the ducts at the time of the fire. Voillequé noted that he had discovered that measurements in 1974 from research and development areas of the building indicated that 130 grams of plutonium were contained in the ducts. This material had accumulated over 20 years. A linear extrapolation would indicate that 6 grams of plutonium was in the ducts at the time of the fire, but the actual amount would depend upon the production methods used during the early years of the plant, prodution rates (assumed to be less than in later years), and whether or not the ducts had been replaced prior to the fire. Regardless, Voillequé said, the amount of plutonium in the ducts that could have been resuspended in the fire was not a significant contributor to the overall release. Voillequé then turned to his estimates of plutonium releases. He remarked that his reanalysis was based on the same experimental airborne release fractions (ARFs) and quantities of material involved in the fire in Room 180. Voillequé said that the principal difference from his prior estimate was that the revised analysis considered release rates over time, given known events. This revision also assumed that release rates increased over time, although these rates remained within the total estimated ranges. Voillequé noted that the net effect of this methodology would be to estimate higher releases. He said that other scenarios could be developed that would lead to lower estimates. Voillequé then turned to the question of filter failure scenarios. If the filters failed early in the course of the fire, Voillequé said, then larger quantities of plutonium released from Room 180 would be released directly up the stack. The revised analysis included three filter performance scenarios. Under the first scenario, the analysis assumes that the booster system fails within forty minutes of the fire's ignition, but that the main filter plenum remains largely intact until the explosion at 10:39 PM. Scenarios B and C assume a more rapid penetration and failure of both the booster system and main filter plenum filters. No scenario assumes that the main filter plenum fails completely prior to the explosion. Voillequé further remarked that the revised analysis took into account plutonium released to the atmosphere after the fans were shut off once the building lost power. This plutonium came from smoldering filters and escaped to the environment as a result of the natural draft in the stack. He noted that the fire produced temperatures high enough to deform the lead molding around the top of the stack, and that the fire itself in the main filter plenum had produced temperatures that were in ranges similar to incinerator off-gases. He assumed that a draft of approximately 1 inch of water pressure differential continued for one to three hours after the explosion in the main filter plenum.

Comments on the Reanalysis

Voillequé's presentation elicited many comments and discussion among workshop attendees. The discussion touched on many aspects of presentation, although the timing and nature of the explosion, and the need to further develop worst-case scenarios, constituted the bulk of the comments and reaction from the audience. One participant asked about Voillequé's use of airborne release fractions (ARFs). ARFs are used to estimate the amount of plutonium that is made airborne in a fire, taking into account different forms of material being combusted and different characteristics of the combustion. This participant argued that the quantities of released material, as calculated by these fractions, appeared small. This participant noted that the ARF for plutonium contained in organic cutting oil was especially small, and that a larger release was likely. Voillequé remarked that the release fractions that were used in the analysis were derived from experimental data. He acknowledged that not all of these experiments conformed exactly to the conditions experienced in the fire. He said that the uncertainty analysis (yet to be completed) would consider ranges of ARFs in the final calculations. Another participant argued that the analysis did not take into consideration that operations in Building 71 were resumed after the fire was out and before the main filter plenum and booster system were repaired. These releases might be very high, given the lack of filtering. Voillequé agreed that he had not taken these releases into account. He and another participant suggested, however, that actual production in the building following the fire may have been extremely limited and may not have involved operations normally associated with large releases. Voillequé said that he did not think that these releases would significantly contribute to the total release. Voillequé said that he thought most of the operations in the building after the fire were related to cleanup activities and not production. He noted that these activities may have resulted in significant releases that were not well documented. Voillequé noted that there was no air sampling data for the week following the fire. He noted that the first air sample taken from the main filter plenum after the fans were restored, one week after the fire, was relatively high. Ironically, he noted that the sample likely fell on the low end of the distribution of the actual release. This would indicate relatively high releases during the week before the sample was taken. As discussed above, one participant commented that there may be gaps in the duct sampling data collected prior to the fire that coincided with operations normally associated with large releases. Such releases would increase the amount of plutonium on the main filter plenum or booster system filters prior to the fire, leading to a higher overall release. Voillequé said that he could not identify any large gaps in the data. Daily operating records before the fire took place, Voillequé said, are largely complete. The discussion about production following the fire and the completeness of sampling data lead to a discussion about throughput rates in Building 71 prior to the fire. It was noted that this information remained classified. Participants agreed, however, that the trend at the plant was for throughput to increase over time. One participant asked about the disposition of the water used to put out the fire in Room 180 and the main filter plenum. A member of the HAP remarked that the quantity of plutonium released in the water was not considered in the report. The panel member noted that the issue under investigation was the quantity of plutonium released off site. As a result, the analysis focussed on airborne releases. However, participants agreed that the disposition of this water was not well documented and that it likely contained an unknown quantity of plutonium that could potentially migrate off-site over time. Much of the discussion focussed on the nature and location of the explosion. Voillequé remarked that the analysis concluded that the explosion (he termed it a "rapid deflagration") occurred on the upstream side of the main filter bank. The analysis concluded that the explosion was caused by the either the ignition of unburned gases from the plexiglass gloveboxes and other material in Room 180 and the booster system, or the ignition of hydrogen generated by a metal water reaction, or both. One participant questioned this analysis, arguing instead that the explosion was the result of a criticality. He argued that a chemical explosion was unlikely to have occurred since the airflow rate was too high to allow for sufficient quantities of unburned gasses or hydrogen to accumulate on the upstream side of the filter bank. He argued strongly that the filters failed early in the fire, causing a rapid airfolw through holes in the filter plenum that would prevent combustive gasses to collect in sufficient quantity to cause an explosion. Therefore, according to this participant, only a criticality could have caused the explosion. Voillequé, HAP members and other workshop participants disputed this conclusion. There was no evidence of fission products or of high numbers of neutrons that would indicate that a criticality took place. While the exact nature of the explosion remained uncertain, Voillequé and these participants argued, a criticality was highly unlikely. A related issue that elicited much discussion concerned the timing of the explosion relative to the failure of the main filter plenum. Several participants argued that the analysis should include a scenario where a significant portion of the filter plenum fails prior to the explosion. While this scenario may be unlikely, it would provide a worst case to bound other more likely scenarios.

Further Work

The comments and discussion led to several recommendations for Voillequé to consider as he completes he revised analysis of the 1957 fire:
  1. More scenarios need to be considered. In particular, a worst case scenario where the main filter plenum fails early in the fire (and before the explosion) needs to be developed.
  2. The nature of the explosion remains controversial. Voillequé and the HAP should identify and consult with an explosions expert for an independent analysis.
  3. The Airborne Release Fractions for cellulose and for organic liquids in the analysis need to be further justified. The experimental conditions of the ARFs may not be relevant to the assumed conditions of the 1957 fire. Are there other ARFs that may be relevant to Voillequé's analysis?
  4. The analysis needs to consider post-fire production and clean-up activities. The nature of these activities may have been inconsequential. However, these activities may also have led to significant releases. Sampling data for the week following the fire is unavailable.
  5. The argument against a criticality in the main filter plenum needs to be further defended. Voillequé should include as an attachment to his report an analysis that explains why a criticality was a highly unlikely if not impossible event. This attachment should consider the data from any neutron detectors were in Building 371 at the time of the fire, whether or not waste from the fire (ash, unburned filters, filiter frames, etc.) was tested for fission products, and other relevant data that would or would not indicate a criticality.
  6. There remains a need to seek the declassification of the plutonium production quantaties in bulding 71 during the 1950s and early 1960s. Such information would contribute to an understading of how much plutonium was held up in the ducts prior to the fire and how much plutonium may have been released during the week following the fire.
  7. The workshop also demonstrated that, by itself, recreating the events of 1957 fire may be insufficient to answer every outstanding question. Information remains unaccounted for, with the data either uncollected, classified or lost. As a result, environmental sampling, employing techniques to distinguish plutonium from Rocky Flats from plutonium from testing fallout, may need to be pursued.
The Source Term Subcommittee also agreed to hold another workshop on December 8, 1997 prior to the December quarterly HAP meeting. This purpose of this meeting will be to discuss outstanding issues related to the 1957 fire, the 1969 fire, the 903 pad and from routine releases.

email us twitter share on facebook