Click for next page ( 2

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 1
Executive Summary There are 2,253 tons of chemical agents in the 661,559 individual munitions and 108 ton containers stored at Anniston Army Depot in Anniston, Alabama. This is almost 10 percent of the current 23,416 tons of mustard and nerve agents in the U.S. chemical weap- ons stockpile. The Anniston Chemical Agent Disposal Facility (ANCDF) has been constructed and is being readied for operation in 2003 with the mission of de- stroying the aging munitions and bulk containers of agent in this stockpile safely and expeditiously. The quantitative risk assessment (QRA) performed for Anniston indicates that the risk to the public is domi- nated by accidents that could arise from the storage of M55 rockets filled with GB nerve agent. The QRA sug- gests that as the M55 rockets are safely destroyed dur- ing the first disposal campaign, the risk to the public decreases. During disposal of GB M55 rockets at Tooele, Utah, about 5,000 rockets could not be drained be- cause the GB contents had gelled. These gelled rock- ets were processed at a much slower rate that required modification and extension of the disposal process. About 20 percent (at least 8,706) of the Anniston GB M55 rockets are now estimated to be gelled. The Army has developed modified plans for their safe and expeditious disposal. This report reviews those plans. The discussions in this report focus on technical con- siderations and related issues in going from a gelled GB M55 rocket processing rate of 1.0 or 1.6 rockets per hour to 9.2 rockets per hour, which would work out to 6.4 rockets per hour on the basis of an expected 70 percent availability for the deactivation furnace 1 system (DFS). Other rates that are reported are given to reflect the variability of operational experience to date in the processing of both gelled and ungelled (drained) GB M55 rockets. Drained rockets are de- fined as rockets from which at least 95 percent of the agent has been removed. The M55 Committee's formal findings and recom- mendations can been found in Chapter 5. Major points from the findings and recommendations have been in- corporated into the narrative text of this Executive Sum- mary, along with abbreviated background material. THE BASELINEINCINERATION SYSTEM As of July 2002, about 26 percent (8,082 tons) of the original 31,495-ton stockpile had already been de- stroyed in baseline incineration system facilities at Johnston Island in the Pacific Ocean and at Tooele, Utah. ANCDF, with minor changes, is patterned after those two facilities, the Johnston Atoll Chemical Agent Disposal System (JACADS) and the Tooele Chemical Agent Disposal Facility (TOCDF). In baseline facilities such as these, munitions are trans- ported from the depot storage area to a receiving dock at the disposal facility, unloaded, and conveyed into explosive containment rooms, where they are disas- sembled by machines (or, in the case of rockets, sheared into sections) and drained of agent. The agent is disposed of in a liquid incinerator (LIC) furnace. The metal shell casings or ton container sections are decontaminated in a metal parts furnace (MPF). The energetics burster charges, fuzes, and propellants-

OCR for page 1
2 are destroyed in a DFS, which is a rotating kiln. There was also a fourth furnace at JACADS and TOCDF, the dunnage incinerator (DUN), which was designed to destroy various nonmunition wastes such as ship- ping pallets. However, when ways were found to ship these wastes off-site or to destroy them in the MPF, the DUN was no longer required, so none has been included in the ANCDF design. Each of the three furnaces in the ANCDF baseline design has its own pollution abatement system (PAS). In a PAS, the hot flue gas, containing some acidic prod- ucts from agent combustion, is treated with caustic to form brine, filtered, and discharged through a stack. The brine can be either shipped off-site to a permitted disposal facility or treated in a brine reduction area (BRA). Again, because the Army has found that off- site disposal can be done safely and economically, the BRA at Anniston probably will not be used. A third difference from JACADS and TOCDF is that the ex- haust gas that leaves a PAS at ANCDF is then passed through a filter system (high-efficiency particulate air [HEPA] and activated carbon) before being discharged through the stack. These HEPA and carbon filters, known as the PAS filter system (PFS), provide a sec- ond line of defense to ensure that agent, metals, and other potentially harmful products are not released to the environment. M55 ROCKETS Among the munitions stored at Anniston there are 42,738 GB M55 rockets and 24 GB M56 rocket war- heads containing 10.7 lb of GB agent each. Another 35,636 rockets and 26 warheads contain 10 lb of VX nerve agent each. An explosive burster charge activated by an impact fuze disperses the agent when the rocket hits a target. M55 rockets are powered by a stabilized double-base propellant, nitrocellulose and nitroglyc- erin, which ignites when the rocket is fired. The rocket has an aerodynamically shaped, finned aluminum body that is contained and shipped ready to use in a fiber- glass firing tube. Altogether, the combustibles in each rocket agent, burster charge, fuze components, pro- pellant, and epoxy resin in the fiberglass shipping tube weigh about 40 lb. GB M55 rockets carry the highest risk potential of any chemical stockpile munitions. This is in part be- cause GB is more volatile than VX and can disperse farther in an accidental release, and in part because the ASSESSMENT OF PROCESSING GELLED GB M55 ROCKETS AT ANNISTON rockets, which are arranged in compact arrays of 15 per pallet, are stored in igloos, as are most other items in the chemical stockpile. Thus, if one rocket were to ignite, it might ignite the others. While this is also true for rockets containing VX, which is more toxic than GB. VX is not as volatile and will not disperse as widely as GB. The net effect of this in terms of risk management is the scheduling of GB M55 rockets to be processed first. This is how the processing was man- aged for JACADS and TOCDF and how it is planned for ANCDF. Storage Stability of GB M55 Rockets Concerns about the storage stability of GB-filled rockets led Congress in 1985 to legislate destruction of the stockpile. Among the concerns was the fact that the propellants degrade slowly, possibly leading to auto- ignition. A stabilizer compound, 2-nitrodiphenylamine, had been added to the propellant during manufacture to scavenge products of the degradation reaction. How- ever, once stabilizer concentrations fall below a critical level, there is some risk of a runaway reaction that could cause the propellant to autoignite. Enough stabi- lizer was therefore added to protect the rockets from autoignition over what was originally thought to be a safe storage period. The stockpile has been monitored in the interim, and rocket sampling studies show that the expected depletion of stabilizer continues. Another concern was that GB munitions were found to leak at about five times the rate of other munitions (about 0.25 percent versus 0.05 percent). Leaking mu- nitions, including rockets, generally have been over- packed in tightly sealed steel containers. Overpacking usually controls leaking but heightens concerns about autoignition because the container acts as a barrier to heat transfer from within the rocket to the surrounding air in the storage igloo. Also, if GB leaks into the pro- pellant, there is a further possibility of accelerated pro- pellant degradation. Since 1985, a number of scientific evaluations have assessed the likelihood of autoignition over time, as the stockpile ages. All concluded that completion of stockpile destruction, currently programmed to meet the 2012 extended deadline of the international Chemical Weapons Convention treaty, will occur long before the risk of autoignition is appreciable, even for overpacked, leaking rockets. The frequency estimate for an M55 autoignition event in the latest study is site specific; currently, it is approximately 3 x 10-5

OCR for page 1
EXECUTIVE SUMMARY per year for the Anniston site. This is much lower (by a factor of 60) than the estimated 2 x 10-3 per year frequency (an average of once in 500 years) of igni- tion from a lightning strike on a storage igloo. Ac- cording to the stability model used by the Army, the autoignition probability will increase gradually but will stay below the lightning strike probability mea- sure until 2020. The M55 Committee has broadly re- viewed the rationale, structure, and results of this study and concluded they are sound. A more detailed review will be presented in a forthcoming National Research Council report on the status of stockpile degradation. Gelled GB M55 Rockets GB and VX M55 rockets are normally processed by first draining at least 95 percent of the agent from a rocket, shearing it into sections in a rocket shear ma- chine (RSM) in the explosion containment room (ECR) and processing the sections in the DFS kiln system. The drained agent (less a 5 percent or smaller heel of agent that usually remains in the rocket) is fed to the LIC. The MPF is not used in processing rockets. At JACADS, all the GB and VX rockets contained ungelled agent and were drained and processed in this manner, and most of the GB rockets processed at TOCDF were also processed this way. However, three particular lots in the Tooele stockpile totaling 5,287 GB rockets (of the 28,945 total) contained gelled agent that would not drain. Gelling is apparently the result of GB degradation, which increases acidity, leading to a reaction with the aluminum tank material. It is believed that this produces an aluminum phosphonate species that can, over time, link GB derivatives and cause them to gel. Most of the originally more acidic agent lots were restabilized with diisopropylcarbodiimide (DICDI), and some of the restabilized lots are those that are gelled. Most other remaining lots contain an- other agent stabilizer, tributylamine, and no gelling ef- fects were detected when that stabilizer had been used. iThis is the median site-specific annual autoignition probability for overpacked rockets at Anniston and is equivalent to about one chance in 33,000 per year. The median site-specific annual autoignition probability for nonoverpacked (undetected) leaking rockets at Anniston is approximately 1.4 x 1 o-6 (about one chance in 700,000 per year). The lower frequency estimate for nonoverpacked leaking rockets is due to their lower peak heat gen- eration and slightly higher heat losses compared with overpacked rockets. 3 At Anniston, the current estimate is that 8,706 of the 42,738 GB M55 rockets are probably gelled, based on munition lot numbers that have been associated with gelled agent. Very few gelled rockets are expected to be encountered at the other two sites using the baseline incineration system for disposal (Pine Bluff, Arkansas, and Umatilla, Oregon). For this reason, any gelled rockets at those two sites can be processed at the lim- ited rate of 1.6 rockets per hour used at TOCDF with- out causing serious delays. The committee suggests the Army proceed at the Pine Bluff and Umatilla sites un- der existing permit applications, which (as in the case of TOCDF) provide for a processing rate in the DFS based on agent loadings from processing rockets with no more than a 5 percent agent heel. Lessons Learned at JACADS and TOCDF in Rocket Processing Operations The very first agent disposal operation undertaken in the Chemical Stockpile Disposal Program (CSDP) was the processing of drained (ungelled) GB M55 rock- ets at JACADS starting in 1989. The planned goal was 32 rockets per hour. A "best shift" rate achieved for one day was 27 rockets per hour during the first opera- tional verification test (OVT 1) campaign. The best operation over an extended period (the "full rate") was 15.3 rockets per hour. The processing at JACADS rep- resented first-time experience with an (at that time) untried, extremely complex process system. Much of the processing rate shortfall can be attributed to the learning that was necessary for the operations team, notwithstanding the months of training that had been invested before start-up. After some modifications to equipment and procedures as a result of problems en- countered, the VX rocket operation (OVT 2) went more smoothly. The full rate was 20.6 per hour, although 32 rockets were processed per hour over one full 10-hour shift. TOCDF benefited significantly from the JACADS experience, and operations there on ungelled rockets went well. However, the processing of gelled rockets caused serious delays. These rockets were processed in a substantially different way than the ungelled ones. In the ECR, gelled rockets bypassed the drain station and were sheared in the RSM into segments containing large amounts of agent as well as fiberglass and ener- getics. This mix of rocket components and agent was fed to the DFS kiln. Neither the LIC nor the MPF was used to process rockets in this instance. The total com-

OCR for page 1
4 bustible content per rocket delivered to the DFS kiln thus went from roughly 30 lb to 40.4 lb. TOCDF man- agement obtained approval from the Utah regulatory authorities for 1.6 gelled rockets to be processed through the DFS kiln each hour. This contrasts with the permit rate of 32 drained, ungelled rockets per hour. The rate for gelled rockets was reduced so that the same amount of agent would be destroyed per hour in the DFS as would have been the case when drained rockets with a 5 percent heel were processed. When downtime is taken into account, the actual average rate for pro- cessing gelled rockets over the entire campaign was approximately 0.6 rockets per hour. To speed destruction of the entire inventory of GB munitions at the Deseret Chemical Depot at Tooele, TOCDF management developed the concept of "co- processing." While gelled rockets were being pro- cessed through the DFS, GB projectiles, reconfigured to remove the energetics, were coprocessed through the LIC and MPF. Utah regulatory authorities gave per- mission to do this if the gelled GB rocket processing rate was reduced further, to 1.0 rocket per hour. The Army believes that gelled GB rockets could have been processed through the DFS system at TOCDF safely and effectively at a faster rate, but this was not demonstrated. The M55 Committee agrees with the Army's judgment and recommends that the Army pursue means to demonstrate the safety of a faster rate. Another way to process rockets and projec- tiles containing the same agent is called "complemen- tary processing." In this variation, rockets are pro- cessed by themselves for a few days, and then projectiles are processed by themselves while mainte- nance is performed on the rocket processing equipment. Complementary processing was also tried successfully at TOCDF. At both JACADS and TOCDF, mandatory trial burns were undertaken to test for agent destruction and removal efficiency (DRE) and for emission of metals or toxic substances such as dioxins and furans. The pre- scribed 99.9999 percent DRE for agent was met in all but one of the eight trial burn tests at JACADS. The temperature and residence time in the DFS afterburner were increased in the TOCDF (and ANCDF) design to ensure more complete combustion. In a few tests, mer- cury and lead emissions exceeded standards at both JACADS and TOCDF. A PFS is being employed at ANCDF to ensure compliance with the more stringent requirements that have since been instituted for control of emissions. Comprehensive measurements of ASSESSMENT OF PROCESSING GELLED GB M55 ROCKETS AT ANNISTON nonagent emissions are made only during infrequent trial burns. This is in accord with standard industrial practice and regulatory requirements. As long as the furnace functions within normal operating limits, these emissions should not change. However, the M55 Com- mittee believes that more frequent monitoring could reassure the public. The lack of similar data for regular operations makes it difficult to convince the public that emissions are always within permit limits. DISPOSAL SCHEDULES FOR ANCDF The ANCDF Original Plan The first schedule for disposal operations at ANCDF called for processing in the following order: GB M55 rockets agent changeover VX M55 rockets VX munitions agent changeover GB munitions agent changeover HD/HT (mustard agent) ton containers and munitions This schedule was developed before there was a rec- ognition that gelled rockets would have to be processed. After gelled rockets were discovered at TOCDF and it became clear that there are many of them in the Anniston stockpile as well, the Army commissioned experienced furnace consultants who had worked on the baseline furnaces at JACADS and TOCDF to esti- mate the rate at which gelled GB rockets could be pro- cessed through the DFS system. The consultants con- structed a mathematical model that was necessarily a simplification of the actual DFS kiln. Backed by some detailed modeling work, assumptions were made about how much of the GB, propellant, burster, fuze, and ep- oxy resin would burn at various points along the DFS kiln. Based on the modeling, the consultants concluded that the DFS kiln could safely process up to 34 gelled rockets per hour. They recommended, however, that this goal be approached gradually during the GB agent trial burn period. The M55 Committee believes the 34 per hour rate is optimistic, but it supports the idea of ramping up pro- duction gradually, to a higher rate than was employed at TOCDF. Some of the committee's concerns about

OCR for page 1
EXECUTIVE SUMMARY the 34 per hour rate are these: (1) the maximum rate of heat release at the inlet to the DFS may be higher than assumed in the model; (2) unless properly managed, the instantaneous rate of heat release may lead to tem- perature spikes and resultant pressure puffs that could release agent into the DFS room and/or the ECR; and (3) agent will probably melt and may then vaporize and undergo thermal decomposition and oxidation in the feed chute, since the chute would be hotter than in the TOCDF runs. This could require limiting the feed rate. Also, the current DFS instrumentation the Process Data and Recording System should be reconfigured and used during the ramp-up period of the agent trial burn to measure and record instantaneous peak gas tem- peratures, differential pressures, and feed chute metal temperatures. In May 2002, ANCDF successfully conducted trial burns in the DFS kiln on gelled rocket surrogates (i.e., nonagent materials). The DREs were better than the 99.9999 percent permit limit, and emissions of metals, carbon monoxide, and other toxic materials were within limits when the PFS was in service. Levels of cadmium, lead, and mercury exceeded regulatory limits when the PFS was not in service during some of the surrogate trial burns, but it will be placed in service during agent operations. The surrogate trial burn demonstration, which fed combustibles through the DFS equivalent to the weight of combustibles in 15 gelled rockets per hour, suggests that a larger number of rockets contain- ing gelled agent can be safely processed per hour than were processed at TOCDF. The Army has derived another target rate for pro- cessing gelled rockets based on having only one rocket in the DFS kiln at a time. At the planned rotation rate of 1.85 rpm, it takes 6.5 min for solids to traverse the kiln length. This produces a rate of 60/6.5, or 9.2 gelled rockets per hour. This rate was used in both the original plan and the modified plan discussed below. It is prob- ably achievable, but it should be approached gradually and with a fully instrumented DFS system. Using this rate, plus the normal rates for the other munitions, the Army has estimated it would take 7.2 years to process the entire Anniston stockpile (see Appendix B). As noted above, there would be three agent changeovers. During agent changeover operations, all areas exposed to agent are decontaminated by workers in demilitari- zation protective ensemble (DPE) suits. Since the monitors are agent specific, once the area is cleaned so that the previous agent is nondetectable, they are re- placed with instruments calibrated for the agent that s will be processed in the next campaign. Changeover operations are labor intensive and typically take about 4 months. The ANCDF Modified Plan Based on the successful experience at TOCDF in coprocessing GB rockets and other munitions, the Army developed a modified plan for ANCDF. In this plan, GB munitions are processed in a complementary manner with GB rockets, gelled or ungelled (see the description of complementary and coprocessing in Chapter 3~. The gelled rockets are processed at the rate of 9.2 per hour as in the original plan, and the drained ungelled rockets are processed at a rate as high as 32 per hour. The processing sequence in the modified plan is as follows: complementary and coprocessing of GB rockets and munitions agent changeover VX rockets VX munitions agent changeover HD/HT ton containers and munitions This schedule is estimated to take 6.3 years to com- plete, 10 months fewer than the original schedule. If gelled rockets are processed at the rate of 1.6 per hour, demonstrated at TOCDF, complete destruction of the ANCDF stockpile is estimated to take 7.6 years (see Appendix B). Although the modified plan would thus provide more expeditious elimination of the storage risk from the overall Anniston stockpile, local officials and members of the public have questioned the safety of processing gelled rockets at a rate higher than that used at TOCDF. Assessment of Public and Worker Risks Under the Two Plans QRAs can be used to identify all conceivable acci- dent sequences that might lead to a harmful release of agent. For each sequence, a frequency of occurrence and the potential impacts on public and worker safety are estimated. Overall risk estimates are computed by summing over all the individual sequences to give com- posite expected values of risk over the duration of the disposal program. QRAs are site specific since the population densities, terrain, weather patterns, and po-

OCR for page 1
6 tential for natural disasters (earthquakes, lightning strikes, forest fires, etc.) are different at each site. The QRAs conducted for each site have consistently shown that the risk of accidental release of agent during stor- age is larger than the risk during processing. As noted earlier, different munition-agent combinations repre- sent different levels of risk, with GB M55 rockets rep- resenting the highest risk. Assuming a maximum processing rate of 9.2 gelled GB rockets per hour, the QRA estimate of the total risk to the public over the 7.2 years necessary to destroy the Anniston stockpile according to the original plan is 0.058 expected fatalities. Over the 6.3 years necessary for the modified plan, which processes gelled GB rock- ets at the same rate and coprocesses other GB muni- tions, the total risk is 0.065 expected fatalities. This is the estimated number of fatalities expected from start to completion of disposal processing. The total public risk level in the modified plan if the TOCDF rate of 1.6 rockets per hour is employed climbs to 0.095 expected fatalities for the 7.6 years necessary to destroy the en- tire Anniston stockpile. The slightly higher level in the modified plan results from keeping the VX rockets in storage an extra 4 months while the remainder of GB munitions is destroyed. The worker risk assessment has not been revised for these options, but it seems reason- able that the elimination of an extra agent changeover operation will reduce overall worker risk. If the assumptions made in developing the risk esti- mates in the QRA are accepted and the inherent uncer- ASSESSMENT OF PROCESSING GELLED GB M55 ROCKETS AT ANNISTON tainty surrounding such estimates and the trade-offs between public and worker risk are taken into account, it is not possible to differentiate meaningfully between the processing plan options based on calculated risk alone. The committee therefore recommends that the modified plan be undertaken with precautionary ramp- up of the production rate until a safe upper production limit is established or the maximum permitted rate is achieved. The overall risk from the stockpile is increased by any programmatic delays, because the risks in storage increase with time and remain greater than the risks of disposal operations. Unresolved issues between the Army and the Chemical Stockpile Emergency Pre- paredness Program, as well as between the Army and regulatory groups, need to be addressed expeditiously. Further, it is important that the Army improve commu- nications with the local communities, both to promote a better understanding of the risk issues and to address any valid public concerns. The health risk assessment (HRA) for ANCDF has not been completed because the agent trial burns have not been done. The HRA is concerned with exposures to possible toxic emissions other than agent, for example, metals and organic emis- sions such as dioxins and furans. The Army should complete the HRA for the ANCDF as soon as feasible. The fact that emissions in the surrogate trial burns were low suggests that the agent trial burns will meet rel- evant standards.