1
The Chemical Demilitarization Challenge

For more than 50 years the United States has maintained an extensive weapons stockpile containing chemical agents, stored primarily in military depots distributed in the continental United States. Largely manufactured 40 or more years ago, the chemical agents and associated weapons in this stockpile are now obsolete. Under a congressional mandate (Public Law 99-145), in 1985 the Army instituted a sustained program to destroy elements of the chemical weapons stockpile and extended this program to destroy the entire stockpile when Congress enacted Public Law 102-484 in 1992.

Chemical weapons stored overseas were collected at Johnston Island, southwest of Hawaii, and destroyed by the Johnston Atoll Chemical Agent Disposal System (JACADS), the first operational chemical demilitarization facility. JACADS began destruction activities in 1990 and completed processing of the 2,031 tons of chemical agent and the associated 412,732 munitions and containers in the overseas stockpile in November 2000 (U.S. Army, 2001a).

The largest continental U.S. stockpile component, which initially contained 13,616 tons of agent, is stored at the Deseret Chemical Depot (DCD) near Tooele, Utah. This component of the stockpile is being processed by the Tooele Chemical Agent Disposal Facility (TOCDF), which started operation in August 1996 and destroyed 5,320 tons of agent and processed more than 880,000 munitions and containers in its first 5 years of activity. As of September 2002, the first two chemical demilitarization facilities had destroyed over 25 percent of the original chemical agent tonnage (U.S. Army, 2002a).

The Army, through its Office of the Program Manager for Chemical Demilitarization (PMCD), now has more than 15 years of cumulative operating experience in chemical weapons demilitarization. PMCD plans to open three additional facilities in the near future to meet Chemical Weapons Convention1 requirements for destruction of the U.S. stockpile. Despite the progress made to date, however, operations at JACADS and TOCDF have not been without incident. Several “chemical events” at the two plants have resulted in either unplanned discharge of significant amounts of agent within the facilities and/or the release of very small amounts of agent to the atmosphere above these plants.

The following sections in this chapter discuss the chemical demilitarization challenge: how to safely destroy the stockpile of chemical weapons within the available time constraints imposed by a dangerous and deteriorating stockpile. To put this challenge in context the Committee on Evaluation of Chemical Events at Army Chemical Agent Disposal Facilities describes technology for the chemical stockpile’s disposal, defines and describes chemical events, discusses the significance of risk assessment to the chemical weapons disposal process, and categorizes institutional issues associated with chemical demilitarization.

STOCKPILE CONTENT, DISPOSAL DEADLINE, AND DISPOSAL TECHNOLOGY

The chemical weapons stockpile contains two types of chemical agents: the cholinesterase-inhibiting nerve agents (GB and VX), and blister agents, primarily mustard (H, HD, and HT) but also a small amount of Lewisite. Both types of chemical agents, which are liquids at room temperature and normal pressures, are frequently, but erroneously, referred

1  

Formally known as the Convention on the Prohibition of the Development, Production, Stockpiling and Use of Chemical Weapons and on Their Destruction (P.L. 105-277), the CWC requires the destruction of chemical weapons in the stockpile by 2007 and any non-stockpile weapons in storage at the time of the treaty ratification (1997) within 2, 5, or 10 years of the ratification date, depending on the type of chemical weapon or on the type of chemical with which an item is filled.



The National Academies | 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 7
1 The Chemical Demilitarization Challenge Convention1 requirements for destruction of the U.S. stock- For more than 50 years the United States has main- tained an extensive weapons stockpile containing chemical pile. Despite the progress made to date, however, operations agents, stored primarily in military depots distributed in at JACADS and TOCDF have not been without incident. the continental United States. Largely manufactured 40 or Several “chemical events” at the two plants have resulted in more years ago, the chemical agents and associated weap- either unplanned discharge of significant amounts of agent ons in this stockpile are now obsolete. Under a congres- within the facilities and/or the release of very small amounts sional mandate (Public Law 99-145), in 1985 the Army of agent to the atmosphere above these plants. instituted a sustained program to destroy elements of the The following sections in this chapter discuss the chemi- chemical weapons stockpile and extended this program to cal demilitarization challenge: how to safely destroy the destroy the entire stockpile when Congress enacted Public stockpile of chemical weapons within the available time con- Law 102-484 in 1992. straints imposed by a dangerous and deteriorating stockpile. Chemical weapons stored overseas were collected at To put this challenge in context the Committee on Evalua- Johnston Island, southwest of Hawaii, and destroyed by tion of Chemical Events at Army Chemical Agent Disposal the Johnston Atoll Chemical Agent Disposal System Facilities describes technology for the chemical stockpile’s (JACADS), the first operational chemical demilitarization disposal, defines and describes chemical events, discusses facility. JACADS began destruction activities in 1990 the significance of risk assessment to the chemical weapons and completed processing of the 2,031 tons of chemical disposal process, and categorizes institutional issues associ- agent and the associated 412,732 munitions and contain- ated with chemical demilitarization. ers in the overseas stockpile in November 2000 (U.S. Army, 2001a). STOCKPILE CONTENT, DISPOSAL DEADLINE, AND The largest continental U.S. stockpile component, which DISPOSAL TECHNOLOGY initially contained 13,616 tons of agent, is stored at the Deseret Chemical Depot (DCD) near Tooele, Utah. This The chemical weapons stockpile contains two types of component of the stockpile is being processed by the Tooele chemical agents: the cholinesterase-inhibiting nerve agents Chemical Agent Disposal Facility (TOCDF), which started (GB and VX), and blister agents, primarily mustard (H, HD, operation in August 1996 and destroyed 5,320 tons of agent and HT) but also a small amount of Lewisite. Both types of and processed more than 880,000 munitions and containers chemical agents, which are liquids at room temperature and in its first 5 years of activity. As of September 2002, the first normal pressures, are frequently, but erroneously, referred two chemical demilitarization facilities had destroyed over 25 percent of the original chemical agent tonnage (U.S. Army, 2002a). 1Formally known as the Convention on the Prohibition of the Develop- ment, Production, Stockpiling and Use of Chemical Weapons and on Their The Army, through its Office of the Program Manager Destruction (P.L. 105-277), the CWC requires the destruction of chemical for Chemical Demilitarization (PMCD), now has more than weapons in the stockpile by 2007 and any non-stockpile weapons in storage 15 years of cumulative operating experience in chemical at the time of the treaty ratification (1997) within 2, 5, or 10 years of the weapons demilitarization. PMCD plans to open three addi- ratification date, depending on the type of chemical weapon or on the type tional facilities in the near future to meet Chemical Weapons of chemical with which an item is filled. 7

OCR for page 7
8 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES Newport Chemical Activity Umatilla Chemical VX -TC Depot (4.2%) HD-TC GB - P, R, B Edgewood VX - P, R, M, Chemical ST Activity (12.2%) HD - TC (5.3%) Deseret Chemical Depot Blue Grass H-P; HT - C, Chemical HD - C, TC Activity GB - C, P, R, B, TC HD - P VX - P, R, M,ST GB - P, R GA - TC VX - P, R (44.5%) (1.7%) Pueblo Chemical Anniston Depot Chemical HD - C, P Activity Pine Bluff HT - C HD - C, P, TC Chemical Activity (8.5%) HT - C HD -TC GB - C, P, R HT - TC VX - P, R, M GA, GB, VX, H, HD, HT = Chemical agent GB - R (7.4%) VX - R, M TC = Ton container B = Bombs (12.6%) R = Rockets C = Cartridges M = Mines P = Projectiles ST = Spray Tanks FIGURE 1-1 Location and size (percentage of original stockpile) of eight continental U.S. storage sites. SOURCE: NRC (1997). to as gases. The stockpile contains both bulk (“ton”) con- schedule estimates indicating that chemical demilitarization tainers of nerve and blister agent and munitions, including activities at the three disposal facilities scheduled to com- rockets, mines, bombs, projectiles, and spray tanks loaded mence operation in the near future may not be completed with either nerve or blister agents. Many munitions contain until 2008 at Pine Bluff, Arkansas, and until 2009 at both chemical agent and energetic materials (propellants and/ Anniston, Alabama, and Umatilla, Oregon (U.S. Army, or explosives), a combination that poses particular chal- 2001b). lenges for safe and efficient destruction. The disposal technology selected by the Army for stor- The disposal of stockpiled chemical weapons is a major age sites that contain a full range of chemical agents and undertaking. In 1990, the stockpile included 31,496 tons of munitions types is a multifurnace incineration process (NRC, chemical agents. The current stockpile is stored at eight chemi- 1999a). In this “baseline” technology approach munitions cal weapons depots operated by the Army in the continental and containers are drained of agent, which is burned in dual United States. The location, size, and composition of the origi- liquid incinerators (LICs). Robotic machinery disassembles nal continental U.S. stockpile is presented in Figure 1-1. munitions containing energetic charges and the separated The U.S. Chemical Stockpile Disposal Program (CSDP) energetic materials are burned in a rotary kiln-based deacti- has evolved in parallel with international initiatives to elimi- vation furnace system (DFS). Sheared bulk containers and nate chemical weapons. After many years of negotiation, metal munitions parts are fed though a large heated metal the terms of the CWC were agreed upon in 1993 to deal with parts furnace (MPF) designed to burn off any residual agent this issue. As of June 2002, the CWC had been signed by or energetic material, decontaminating metal components to 174 countries and ratified by 145. The convention went into the point that they can be recycled as normal scrap metal. effect on April 29, 1997, after ratification by 65 countries. The LIC, DFS, and MPF furnaces are all equipped with ex- The CWC requires that signatories, which include the United tensive pollution abatement systems (PASs) designed to sub- States, destroy their chemical weapons stockpiles within 10 stantially eliminate gaseous and particulate exhaust material years of its initiation, making April 29, 2007, the deadline of potential concern and exhaust remaining gases through a for destruction of the U.S. stockpile. A provision in the treaty common stack. allows a 5-year extension of the deadline under some cir- The first-generation incineration system was deployed cumstances. As of early October 2001, PMCD released new at JACADS. The second-generation system, deployed at

OCR for page 7
9 THE CHEMICAL DEMILITARIZATION CHALLENGE TOCDF, is described in detail in the National Research 2. Discovery of an actual or suspected chemical agent Council report Tooele Chemical Agent Disposal Facility— container or munition in a place where it is not sup- Update on National Research Council Recommendations posed to be that may require emergency transporta- (NRC, 1999a)2 ; this detailed description is reprinted in tion or disposal. Appendix A. Third-generation incineration systems are 3. Confirmed detection of agent above the threshold currently close to operational status at the Pine Bluff, concentration for any period outside the primary en- Anniston, and Umatilla sites. The new facilities will use gineering control. basically the same process as that used at TOCDF and 4. Actual exposure of personnel to agent above the al- JACADS. Weapons will be taken apart in the same way, lowed limits specified in various Army regulations. and there will be the same three lines of incineration: a 5. Loss of chemical agent. rotary furnace for destroying propellant and explosive 6. Any terrorist or criminal act directed toward a chemi- materials (see Appendix A, Figure A-6), a furnace with a cal agent storage, laboratory, or chemical demilitari- moving conveyor primarily for decontaminating metal zation facility or any deliberate release of chemical parts (see Figure A-7), and a furnace for burning liquid agent. agent (see Figure A-8). Improvements to the new facili- 7. Any malfunction or other significant activity at a ties have been made compared with TOCDF and JACADS, chemical demilitarization plant that could reasonably however; these are noted in Chapter 5. be expected to cause concern within the local com- In addition, the Army has selected liquid-phase hy- munity or the press, or that in the judgment of the drolysis processes, supplemented by various secondary hy- local facility or installation management or leader- drolysate treatment and/or disposal processes, to destroy the ship could cause embarrassment to the U.S. Army. chemical agents contained only in bulk “ton” containers at Newport Chemical Depot in Indiana and at the Edgewood At the eight continental U.S. storage sites, the Army’s Chemical Activity site on Aberdeen Proving Ground in local depot commander has the responsibility to decide Maryland. Significant problems in introducing these new whether an upset or incident within the storage yard or at the technologies to dispose of even the simplest case of “bulk associated chemical demilitarization facility is a chemical only” chemical agent have been recognized (NRC, 2000a). event. Examples 1 through 6 above seem to imply that, in A disposal technology has not yet been selected for the rela- most cases, chemical events are those in which chemical tively small stockpile of nerve and mustard munitions at the agent ends up where it should not be, i.e., in the ambient Blue Grass Chemical Depot in Kentucky. At the Pueblo atmosphere or under the control of an unauthorized indi- Chemical Depot in Colorado the Department of Defense has vidual. However, no such requirement is inherent in ex- decided to use neutralization, followed by bio-treatment of ample 7. The wide latitude in judgment about what might the secondary waste to dispose of the mustard munitions “cause concern within the local community or the press, or stored there. . . . could cause embarrassment to the Army” that is del- egated to the depot commander suggests that some incidents defined as chemical events by one commander may not be CHEMICAL EVENTS considered chemical events by another. During the 10 years of JACADS operation to destroy Whatever the local Army commander deems to be a the chemical weapons stockpile at Johnston Island and the chemical event is subject to strict reporting procedures de- first 5 years of operation of TOCDF, a range of operating tailed in Army Regulation 50-6 (U.S. Army, 1995). Both incidents occurred that were designated as chemical events telephone reports (within 3 hours) and initial written reports by the local Army depot commanders. Army Regulation following specified formats (within 24 hours) must be made 50-6 on Chemical Surety (U.S. Army, 1995) defines chemi- to both the Army Operations Center and Headquarters–De- cal events very broadly: “The term chemical event encom- partment of the Army. These reports are usually shared with passes all chemical accidents, incidents and politically/pub- local authorities and serve as the basis for press releases is- lic sensitive occurrences.” The regulation goes on to give sued by the local depot and/or PMCD, but there appear to be specific examples, such as: no general guidelines for the form and the timing of such notification. A tabulation of chemical events is also pro- 1. Confirmed releases of agent from munitions outside vided in the Army’s annual reports to Congress summarizing a closed containment system, such as a filtered bun- chemical agent storage and chemical demilitarization activi- ker, storage igloo, or overpack container. ties (see, for example, U.S. Army, 2000a). Chemical Events Associated with Disposal 2This update report also details TOCDF technology and management The Program Manager for Chemical Demilitarization issues identified by an NRC oversight committee during that facility’s first (PMCD) is required to prepare and provide reports of chemi- 3 years of operation.

OCR for page 7
10 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES cal events for incidents so designated that occur within its est destruction, according to the overall risk mitigation strat- facilities, as specified in Army Regulation 50-6.3 egy outlined below in this chapter. Inspection and remediation PMCD provided to the committee a chronicle of 81 in- of corroded or leaking munitions prior to their shipment to cidents that occurred over the 10 years of JACADS opera- Johnston Island probably also contributed to the fact that so tion and the initial 5 years of TOCDF operation (U.S. Army, few munitions and containers leaked in the Johnston Island 2001c). As shown in the Army’s document (see Appendix storage site. B), a significant number reported under example 7 listed The statistics for stockpile leakage at the Tooele site above did not involve chemical agent. Of those events in- (DCD) for the same period from 1990 to 2000 differ consider- volving chemical agent, only a few resulted in release of ably from those for JACADS. SBCCOM tabulated 31 inci- agent outside of engineering control and into the atmosphere. dents for 1990, 34 for 1991, 40 for 1992, 37 for 1993, 38 for The total mass of chemical agent released to the environ- 1994, 33 for 1995, 26 for 1996, 14 for 1997, 14 for 1998, 10 ment in these incidents was almost certainly less than a gram for 1999, and 11 for 2000, for an 11-year total of 288 (U.S. (U.S. Army, 2001d), which is equivalent to no more than a Army, 2001e). Again the record of events suggests that as few drops. The committee’s analysis of PMCD-reported stockpile destruction proceeds, with the most problematic chemical events at the two chemical demilitarization facili- weapons and containers scheduled for early destruction, the ties is presented in Chapter 2. stockpile leakage statistics improve. From 1990 through 1995, In addition to a list of PMCD-reported chemical events, 72 percent (159) of the 213 tabulated stockpile leakage inci- the committee also received lists of possible chemical-agent- dents involved GB, which was the agent type destroyed at related incidents from local officials and concerned citizens TOCDF through March 2002. groups (see Appendixes C and D). These are also addressed Some of the 288 storage chemical agent incidents at DCD in Chapter 2. from 1990 through 2000 involved only relatively low storage igloo vapor readings with most, possibly all, of the released agent captured in the carbon air filters present in igloos storing Chemical Events Occurring During Storage the highest-risk munitions (generally those containing GB) A listing and analysis of the chemical incidents involving (U.S. Army, 2001e). In more that 100 different incidents, leaking containers and munitions at the Johnston Island stor- however, storage personnel who entered an igloo reported age site from 1990 until the end of disposal operations in 2000 observing liquid agent leaks with volumes estimated to range and at the Tooele site (now the DCD) from 1990 through 2000 between 1 teaspoon and 2 quarts, with one leak from a GB ton were provided to the committee by the Army’s Soldier Bio- container totaling 10 gallons. In dozens of additional inci- logical and Chemical Command’s (SBCCOM’s) Stockpile dents smaller amounts of leaked liquid were observed. Since Management Team (U.S. Army, 2001e). As a result of its most of these incidents involved high-vapor-pressure GB, continuous stockpile inspection program SBCCOM has many resulted in release of small amounts of agent to the at- records of the frequency of chemical agent leaks occurring in mosphere, if only when the igloo door was opened to allow stockpiled munitions and containers. Most of the incidents entry. Many of these leaks were initially detected by monitor- listed by SBCCOM involved a single leaking munition or con- ing igloo air drawn through sampling ports. Mobile powered tainer, although incidents involving more than one leaking air filtration units were often used to minimize agent migra- munition discovered in a storage igloo were not uncommon. tion out of the igloo. One incident involved 20 leaking munitions treated over the In addition, in 11 incidents at DCD from 1990 through course of a month. The most serious incidents, including all 2000, ton containers of mustard, stored outside, were found to those known to have discharged a significant amount of agent be leaking agent directly into the environment (U.S. Army, outside of engineering control, were designated as chemical 2001e). When these incidents occurred, storage site person- events and reported as required by Army Regulation 50-6. nel attempted to quantify the amount of agent lost by estimat- According to the SBCCOM statistics on stockpile leak- ing the volume of contaminated gravel or soil underneath the age at Johnston Island, 13 incidents involving leaking muni- leaking container, or, in the case of large leaks, measuring the tions were reported from 1990 through 2000. Ten of these agent remaining in the container. For most of the outdoor occurred from 1990 through 1992, with only 3 occurring in incidents documented at DCD, relatively small amounts of 1993 or later. Some of the later falloff is likely due to reduc- agent (a few drops to a few cups) were estimated by depot tion of the stockpile as chemical demilitarization proceeded, workers to have been lost. However, in the most serious event, with the most problematic munitions scheduled for the earli- a leak of distilled mustard estimated at 78 gallons (~375 kg) was discovered on September 9, 1993. The volume of agent released in this incident alone swamped the total mass of 3Detailed guidance on the preparation and distribution of these reports known emission of agent from chemical demilitarization fa- and on associated record keeping is presented in a periodically updated cilities by at least a factor of several hundred thousand. document designated PMCD Regulation 385-3, “Accident and Chemical Large outdoor releases from storage facilities are an Event Notification, Investigation, Reporting and Record Keeping” (U.S. ongoing concern. In fact, while the committee was gather- Army, 1999a).

OCR for page 7
11 THE CHEMICAL DEMILITARIZATION CHALLENGE ing data for this report during a visit to TOCDF on July 26, based on tests and performance of other units) and mild up- 2001, a leaking plug in a ton container of HD in the DCD set conditions (assumed to lead to release of a multiple of the storage site produced a vapor plume large enough to force conservative routine release for a specific fraction of the workers at adjacent TOCDF to don respirators. That leak, year). Accidents, specific systems failures, and specific hu- according to chemical event reports and related memoranda man actions are not considered. The HRA is an upper-limit supplied by SBCCOM, was determined to be about 9 pounds risk estimate for routine operations. Because it does not pro- (~4 kg). This incident delayed the committee’s access to vide a realistic estimate (accounting for uncertainty), does TOCDF for several hours. not consider accidents, and does not address worker risk, it While chemical demilitarization operations at JACADS is not helpful in evaluating chemical events, other than pro- and TOCDF have released small amounts of chemical agent viding a baseline, of sorts, against which the consequences into the environment, these releases are negligible compared of chemical events can be evaluated. For an example HRA with releases to the environment from associated chemical see the analysis of TOCDF sponsored by the Utah Division weapons storage sites. The rate of agent leaks and releases of Solid and Hazardous Waste (Utah DEQ, 1996). does decrease significantly as the stockpile is processed. Systems Hazards Analysis TOOLS FOR ASSESSING HAZARDS IN THE A systems hazards analysis (SHA) is a systematic and OPERATION OF CHEMICAL STOCKPILE DISPOSAL comprehensive search for and evaluation of all significant FACILITIES failure modes of facility systems components that can be The Army has developed a suite of risk assessment and identified by an experienced team. The hazards assessment risk management tools to permit analysis of potential risks in often includes failure modes and effects analysis, fault tree terms of the scenarios that can contribute to risk, the likeli- analysis, event tree analysis, and hazards and operability hood of those scenarios, and the consequences associated with studies. Generally, the SHA does not include external fac- them. Those consequences are as follows (NRC, 1997, p. 16): tors (e.g., natural disasters) or an integrated assessment of systems interactions. However, the tools of SHA are valu- For humans (both workers and the public) there are three able for examining the causes and the effects of chemical potential measures of risk either from the stockpile or from events. They provide the basis for the integrated analysis stockpile destruction: acute lethality; acute and latent non- known as quantitative risk assessment. For an example SHA cancerous health effects; and latent cancer. The potential see the TOCDF Functional Analysis Workbook (U.S. Army, adverse consequences for the environment are the contami- 1993-1995). nation of land and/or water and adverse effects on native or endangered species. Quantitative Risk Assessment These tools can be used to evaluate the risk associated with specific chemical events. Real-world events can also A quantitative risk assessment (QRA) is an integrated, then be used as a check on the analyses, enabling revision of quantitative analysis (including uncertainty) of accident sce- risk analyses to include new classes of events when surprises narios, their likelihood, and possible consequences. Current occur. QRAs examine human actions as well as systems failures, The variety of analysis tools is useful because of the external events as well as internal failures, and worker risk differing needs of various program elements. To understand as well as public risk. A salient feature of a QRA is that it is how they are related, the committee first groups these tools integrated, in that it: into two large classes: prospective (or predictive) tools and retrospective (or documentation) tools.4 • considers the interactions of systems and their ef- fects on each component, considers common causes of failures, and considers all forms of system depen- Prospective Risk Analysis Tools dencies • considers the integrated impact of multiple system Health Risk Assessment and human failures on the potential for releases A health risk assessment (HRA) is a compliance-ori- • considers the impacts of weather and emergency pro- ented analysis that examines the risk to a set of stylized re- tection on public consequences ceptors (e.g., the subsistence fisherman) associated with rou- tine releases (intended to be conservative upper bounds Thus, the QRA provides an effective tool for evaluating the significance of chemical events. In fact, scenarios leading to chemical events and the frequency and consequences of these events are exactly what a QRA describes and calcu- 4The committee uses the Army’s names and acronyms for these methods. lates. Real-world events provide a check on the analysis. If Use of these names is not consistent with language in other environments.

OCR for page 7
12 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES potentially risk-significant events occur that were not previ- chemical events. In its published QRA the Army performed ously modeled, the QRA can and should be updated to ac- a detailed assessment of the risk of public fatalities and can- count for that event and any similar events that could occur. cers associated both with the stockpile storage sites and For an example QRA see the Army’s TOCDF QRA (U.S. chemical weapons processing activities at Tooele (U.S. Army, 1996a). The committee notes that the TOCDF QRA Army, 1996a), and it has performed ongoing risk assess- was the first PMCD QRA and does not include all the fea- ments for the planned third-generation incineration system tures in the current analyses being finalized for the facilities chemical demilitarization facilities and associated stockpile at Anniston, Umatilla, and Pine Bluff. At the time of this storage areas at Pine Bluff, Anniston, and Umatilla. In its writing the TOCDF QRA is the only one that has been pub- QRA for TOCDF, the Army’s analyses indicated that, over lished. Similar QRAs are being completed at the remaining the facility’s projected operating schedule, the risk associ- sites. It is possible that portions of these may be unavailable ated with accidental releases of agent due to disruption of publicly because of security concerns. the stockpile, most likely due to earthquake or leaks from Key elements of the Army’s approach to quantitative ton containers of GB, greatly outweighed the risk of release risk assessment are summarized in Appendix E for the inter- of agent due to chemical demilitarization activities (U.S. ested reader. More details are available in the NRC Stock- Army, 1996a). This risk assessment does not examine po- pile Committee’s risk report (NRC, 1997). tential terrorist activities, threats that are addressed by other federal agencies in addition to the Army. The Army’s risk assessment for TOCDF and its associ- Retrospective Analysis Tools ated storage facility was reviewed by the NRC and found to be sound (NRC, 1997). Even in the event of an earthquake Monitoring Systems or plane crash that damages the disposal plant, the risk of Monitoring systems detect releases of hazardous chemi- public fatalities due to release of agent from the disposal cals, providing warning of hazardous conditions and a record facility is calculated to be about 5 percent of the expected of their occurrence and extent. They can also measure the risk of fatality due to releases of agent from the storage yard burden of chemicals on the human body. They are not pre- (U.S. Army, 1996a; NRC, 1997). A more detailed discus- dictive, but instead provide real-time observations. For a sion of the TOCDF QRA and of advances incorporated in description of monitoring schemes, see Box 1-1 and the NRC subsequent QRAs is presented in Appendix E. Stockpile Committee’s report Occupational Health and Until the last few days of the disposal schedule, the Workplace Monitoring at Chemical Agent Disposal Facili- amount of agent in the storage yard greatly exceeds the ties (NRC, 2001a). amount in the chemical demilitarization plant; as the stock- pile is depleted, the risk posed by the storage facility drops proportionally. A key risk management strategy adopted by Chemical Event Investigations the Army is to order the stockpile destruction so that the When chemical events occur, investigations identify most volatile, highly toxic agent and associated munitions what actually happened and when, the reasons, and the con- are processed first (those containing the nerve agent GB), sequences; they usually suggest corrective actions for the while less volatile and/or deadly agents are processed later. future. An investigation (separate from possible corrective Finally, it is important to note that the original TOCDF action) is most effective when it focuses on what actually QRA focused on public risk, and little effort was devoted to happened from the viewpoint of those involved (i.e., why examining worker risk. One consequence of this limitation the actions of people involved made sense to them at the in scope was that very little modeling of human performance time, what they could see and what they knew, how they was done in the TOCDF QRA. As attention in the program viewed their alternatives) (Weick and Sutcliffe, 2001). Too shifted to include worker risk, more significant modeling of often these investigations are biased by hindsight and focus human action has been performed. None of these improved on what the operators might have done rather than why they analyses have yet been published. A variety of human reli- did what they did. An effective investigation identifies the ability analysis methods have been used (Gertman and organizational and management issues that made the actions Blackman, 1994). For ongoing work, new approaches that seem reasonable to those involved, and it can provide a basis account for details of context and human cognitive function for real improvement. Chemical event investigation and are being adapted (Hollnagel, 1998; USNRC, 2000). With analysis are the subjects of Chapter 2 of this report. more careful and complete analysis, new scenarios espe- cially important to worker risk are being developed. These methods, integrated into the risk assessments, can be used to Putting It All Together quantify the impact of human actions on situations posing From this brief introduction, it is clear that the QRA, risk. Human performance not only is a significant compo- chemical agent monitoring, and event investigation are the nent in risk assessment, but also, as the committee learned key tools for addressing the safety issues associated with in its study, is directly involved in most chemical events. In

OCR for page 7
13 THE CHEMICAL DEMILITARIZATION CHALLENGE BOX 1-1 Details on Airborne Chemical Agent Monitoring Methods and Standards at Chemical Demilitarization Facilities and specific laboratory GC/FPD system, the ACAMS alarm is not con- Two systems are currently used to monitor concentrations of firmed. If the laboratory GC/FPD system does not show a chromato- airborne chemical agents at chemical demilitarization facilities. One gram consistent with agent, a second DAAMS sample may be run on system, the automatic continuous air monitoring system (ACAMS), is a laboratory GC equipped with a mass spectrometric detector (MSD). designed for “near-real-time” monitoring (currently ~3- to 8-minute The GC/MSD analysis is designed to identify interferent compounds cycle time, dependent on agent, for a single instrument). The ACAMS that may have caused a false positive ACAMS alarm. consists of an air sampling system connected to a gas chromatograph The Army currently mandates very conservative alarm thresh- (GC) equipped with a flame photometric detector (FPD). olds for it chemical demilitarization facilities (U.S. Army, 1997a; NRC, Specific columns and detector filters are used for each agent. 2001a). Current exhaust stack alarms are set at 0.2 of the allowable The nerve agents, GB and VX, are detected by phosphorus oxide stack concentration (ASC), for GB the ASC is just three times the time chemiluminescence, due to their P content, excited in the FPD detec- weighted average (TWA), which serves as the worker population limit tor, while mustard is detected by sulfur dimer chemiluminescence, (WPL) for the demilitarization workforce. The TWA is the level of from its sulfur content. Since VX has high molecular weight (298 agent an unmasked person can breathe for an 8-hour shift without amu), it is catalytically cleaved at the entrance to the GC column to harm. Thus, GB, the most volatile agent and therefore the greatest shorten its detection time. This detection scheme relies on the char- airborne threat to surrounding communities, is monitored at stack acteristic GC column transit time of the agent or agent fragment (in the concentrations equivalent to 0.6 of the level currently deemed safe for case of VX) plus the P or S spectrally specific flame chemilumines- a worker to breathe for a full shift without protection. cence detection signal to identify the agents. The method is quite Stack exhaust monitoring levels for the less volatile and less sensitive; ACAMS are often run at threshold detection volume-mixing threatening HD and VX are monitored at levels a factor of 2 and 6 above ratios of a part per trillion (pptv) or lower. However, at these low their TWA (WPL) levels, respectively. The 0.2 ASC stack level for GB is threshold levels false positive alarms often occur because other chemi- a factor of 10,000 below the “immediately dangerous to life and health” cal species can “interfere” by producing chemiluminescent signals (IDLH) level for this agent. In-plant air levels breathed by unmasked that overlap the time gate and spectral band pass associated with the workers and the output of the scrubbing system for air exiting the de- agents. For time-critical applications, like exhaust stack monitoring, militarization plant are monitored at similarly conservative levels (gen- the GC cycle time can be mitigated by time phasing two or more erally 0.2 TWA). Since any agent in either the stack exhaust gas or ACAMS sampling the same gas stream. scrubbed plant air will be greatly diluted before reaching the facility’s ACAMS alarms must be verified to ensure that they are not a fence line, air flowing into the surrounding communities will be well false positive due to an “interferent” species or instrument malfunc- below the “general population limit” (GPL) defined as the level believed tion. This verification is done using a depot area air monitoring sys- to pose no threat to the public. The GPL for the three stockpile agents is tem (DAAMS) deployed near an ACAMS. DAAMS is a passive system set at 30 to 33 times lower than their TWA (U.S. Army, 1997a; NRC, that draws an air stream through a sorbent tube. The tubes are col- 2001a). The current ASC, TWA, and GPL levels for GB are 3 × 10−4, lected and replaced periodically if there are no ACAMS alarms or 1 × 10−4, and 3 × 10−6 mg/m3. For VX these values are 3 × 10−4, 1 × shortly after an alarm occurs. They are transported to a laboratory and 10−5, and 3 x 10−6 mg/m3, while for HD they are 3 × 10−2, 3 × 10−3, and thermally desorbed onto a sample tube and analyzed on a laboratory 1 × 10−4 mg/m3 (NRC, 2001a; U.S. Army, 1997a). scale GC/FPD system. Without confirmation by the more sensitive Chapter 2, the committee examines the more significant of that . . . personal injury statistics are indicative of a system’s the chemical events at JACADS and TOCDF to determine their vulnerability (or resistance) to organizational accidents. The number of personal injuries sustained in a given time period characteristics with respect to facility performance and human must surely be diagnostic of the “health” of the system as a performance. How these events are related to safety perfor- whole. Unfortunately, this is not so. The relationship is an mance is not a simple question. In his widely referenced book asymmetrical one. An unusually high [personal injury rate] (Reason, 1997), in a chapter devoted to the relationship be- is almost certainly the consequence of a “sick” system that tween frequent, low-consequence events and the risk of high- could indeed be imminently liable to an organizational acci- consequence events, James Reason concludes that: dent. But the reverse is not necessarily true. A low . . . rate (of the order of 2-5 per million man hours)—which is the If both individual and organizational accidents have their case in many well-run hazardous technologies—reveals very roots in common systemic processes, then it could be argued little about the likelihood of an organizational accident.

OCR for page 7
14 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES The problem of human-caused events, how to control chemical agent detection and monitoring methods spurred them, and how to discern the difference between high- and by homeland defense concerns may lead to better and more low-risk events continues to be studied in many industries robust technology. The committee urges PMCD to vigor- (Reason, 1997; Hollnagel, 1998; IOM, 2000). ously seek out and exploit any suitable developments arising from these activities. Previous NRC reports have also noted the lack of ro- Monitoring Methods bust techniques for rapidly measuring agent and agent The occurrence and the extent of a release of chemical breakdown products present in liquid waste streams and agent are tracked through PMCD’s workplace chemical associated with solid materials (NRC, 2000a; NRC, 2000b; agent monitoring system as described in NRC (2001a). NRC 2001a). These reports recommend vigorous efforts Monitoring for airborne agent is a major activity at each to develop better methods to measure agent contamination chemical agent disposal facility. Box 1-1 provides details in these media. on monitoring. Sensitivity requirements for the near-real-time auto- Event Analysis and Significance matic continuous air monitoring system (ACAMS) for air- borne agent are demanding. This is because the allowable The committee notes the importance of chemical event stack concentration and time-weighted-average levels used analysis that focuses on the viewpoint of the operators dur- for exhaust stack and in-plant action levels are quite low and ing the sequence of events. Understanding why their actions because the ACAMS alarms are currently set at 0.2 of the seemed appropriate to them, at the time, is the key to effect- relevant action level. ing real improvement in performance. Gaining an under- This demand for sensitivity results in relatively fre- standing of the factors within their work environment—train- quent false positive alarms, particularly for the ACAMS ing, equipment, and operational indications, as well as goals monitoring the individual incinerator exhaust flows and the and rewards—which led them to conclude that their actions common exhaust stack (NRC, 1999b). Previous NRC re- were appropriate is an essential element of developing a real ports have noted that the frequency of false positive safety culture at the facility. ACAMS alarms disrupts plant operations, particularly An associated effort is to ensure that the QRA includes when stack alarms trigger an automatic shutdown of agent the class of events that actually have occurred. Mapping real feed to the liquid incinerator, and can lead to an unsafe event scenarios onto scenarios modeled in the QRA allows “crying wolf” mind-set that tends to discount ACAMS one to see a particular action integrated into the larger system alarms (NRC, 1999b, NRC, 2001a). In fact, one of these for each chemical event and thus determine its effect on safety. studies found evidence that the May 8-9, 2000, agent stack release at TOCDF was exacerbated by an expectation that CHEMICAL DEMILITARIZATION INSTITUTIONAL what proved to be real exhaust system ACAMS alarms ISSUES were instead just false positives (NRC, 2001a). While pre- vious NRC reports have urged PMCD to improve both the Trust and Institutional Arrangements reliability and time response of its airborne agent monitor- ing systems (see NRC, 1999b for a summary), progress in The chemical demilitarization program necessarily de- this area has been modest. pends on a combination of trust and institutional arrange- Another weakness of the airborne monitoring system is ments to accomplish the destruction of the chemical stock- the lack of real-time (<10 seconds) agent detection. The NRC pile. Because extremely hazardous materials and complex has previously recommended that the Army develop a real- technologies are involved, those seeking destruction of time system that uses a measurement technology indepen- chemical agent and munitions must rely on agencies and dent of the gas chromatography with flame photometric de- firms expert in these processes to carry out the chemical de- tector methods used by the ACAMS and the depot area air militarization program. In essence, legislation and regula- monitoring system (DAAMS) (NRC, 1994). To date, the tory agency rule making establish institutional and contrac- Army’s attempts to develop and demonstrate such a real- tual arrangements for the chemical demilitarization program, time system have not been successful (NRC, 1999b, 2001a). stipulating what is to be accomplished and (in some cases) New interest in chemical agent detection as a key compo- how it is to be done. As in any contract, the “principal” relies on an “agent”5 to accomplish a task or service, and provides nent of antiterrorism activities has spurred government and commercial activities focused on developing better sensors for airborne agent (IOM, 1999). The NRC has also urged 5It is unfortunate that use of the term “agents” to indicate those that carry the Army to continue to monitor technological advances in out tasks for “principals” might in this report be a source of confusion in the trace gas detection and to consider implementing any that context of the chemical demilitarization program (where “agent” usually are appropriate for monitoring agent in chemical munitions refers to chemical agent). Where agent is used in this report in the institu- disposal facilities (NRC, 1999b). Renewed interest in tional sense, it is italicized to reduce the potential for confusion.

OCR for page 7
15 THE CHEMICAL DEMILITARIZATION CHALLENGE the means to ensure that the task is accomplished according Thus, at any of the six continental sites where a chemical to the principal’s needs (Wood, 1992; Scholz and Wei, demilitarization facility is either operating or under construc- 1986). The U.S. Congress and the public it represents rely tion, a concerned citizen needs to receive a consistent and on agencies of the U.S. Army, state and federal regulators, accurate message from a range of state and federal entities private contractors, and a host of other entities to carry out including PMCD, SBCCOM, FEMA, and CSEPP. In the the chemical demilitarization program. At specific chemical past a consistent message from these Army or Army-related demilitarization sites, the local public and the officials who entities has been hard to achieve (NRC, 1999a). represent them similarly depend on these agents to carry out In addition, chemical demilitarization facilities must the task safely and effectively. obtain environmental permits from state environmental When principals delegate complex tasks, they create a agencies in order to commence operations and must seek relationship in which the agents on whom they rely are more permit amendments and renewals from these same agencies knowledgeable about the task than are the principals. Agents in order to sustain operations. Permit conditions may vary design, test, construct, operate, and modify the chemical de- widely from state to state even though the state environmen- militarization facilities and have intimate knowledge of these tal agencies operate largely under authority delegated from steps, while principals often rely on agents for such knowl- and overseen by the federal Environmental Protection edge. This kind of information asymmetry may place the Agency. State-to-state discrepancies in chemical demilitari- principal at a disadvantage in overseeing the safety and ef- zation facility operating permits or amendments to existing fectiveness of the program, and necessitates monitoring and operating permits may raise public concerns. Hearings re- control mechanisms that are specified in the relevant laws lated to environmental permit applications and amendments and contracts. Monitoring mechanisms include permitting give citizens an important opportunity for input into the op- and reporting requirements, inspections, investigations, and eration of chemical demilitarization plants. rules governing whistle-blowers, while control mechanisms The PMCD receives guidance from the Army Medical include arrays of incentives (such as contract fee structures) Services (U.S. Army Center for Health Promotion and Pre- and sanctions (civil and criminal punishments, fee deduc- ventive Medicine), working in conjunction with the Depart- tions, and so on). A trade-off implicit in this relationship is ment of Health and Human Services’ Centers for Disease that of the principal’s control over agents versus the scope of Control and Prevention, about the levels of exposure to the agent’s discretion, some of which is typically necessary chemical agent that are considered safe for both workers and for complex and demanding tasks that require the agent to the public (U.S. Army, 1990, 1991). Recent reevaluations push the boundaries of known processes and technologies. have led to proposals for significantly lower recommended Greater trust reduces the need to rely on formal monitoring standards related to exposure to chemical agents. This pos- and control, and conversely loss of trust increases the need sibility has raised citizen concern about the safety of Army for monitoring and controls. stockpile storage and chemical demilitarization operations designed to meet the current exposure limits. Chemical events have raised questions about the safety The Institutional Setting of Chemical Demilitarization of the stockpile storage and the demilitarization process. The U.S. government’s approach to chemical demilita- Understanding whether an event results from flaws in de- rization involves a complex amalgam of institutional stake- sign, fundamental problems with technologies, organiza- holders. The Army’s SBCCOM is the operator of the eight tional failures, or personnel lapses is essential for determina- remaining stockpile storage facilities. PMCD is responsible tion of appropriate responses. Because the answers to these for the construction, operation, and subsequent closure of questions materially affect the circumstances of the agent, JACADS, TOCDF, the three new incinerator system facili- concerns about whether agents are sufficiently forthcoming ties at Anniston, Umatilla, and Pine Bluff, and the two-bulk and responsive are inevitable. The U.S. Congress has re- only, hydrolysis-based facilities under construction at New- sponded to such concerns with diligent oversight (including port and Edgewood (Aberdeen). By law, evaluation of alter- the request for this report), requirements that whistle-blow- native (non-incineration) technologies that may be used to ers be protected from retaliation, and requests for formal dispose of the stockpiles located at Pueblo, Colorado, and annual reports from the Army on the progress of chemical Blue Grass, Kentucky, has been delegated to an independent demilitarization and the occurrence of any chemical events Program Manager for Assembled Chemical Weapons As- associated with either chemical demilitarization or storage sessment (PMACWA) within the Army. facilities. Protection of the public from harm due to accidental releases of agent near storage depots and associated chemi- REPORT ROADMAP cal demilitarization facilities is the responsibility of the Chemical Stockpile Emergency Preparedness Program The chemical events that have occurred at JACADS (CSEPP), which is funded by the Army but administered by and TOCDF are characterized and a selected subset ana- the Federal Emergency Management Agency (FEMA). lyzed in Chapter 2. Chapter 3 discusses protocols and pro-

OCR for page 7
16 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES cesses for reporting chemical events, outlines how selected TOCDF and future operations at Anniston, Umatilla, and events were reported at both facilities, and discusses how Pine Bluff. Prudent preparations to minimize the occur- these events affected plant operator interactions with other rence and impact of future chemical events at incineration stakeholders, including environmental regulators, elected system chemical demilitarization facilities are discussed in state and local officials, and the public. Chapter 4 dis- Chapter 5. Chapter 6 contains focused findings and recom- cusses the implications of lessons learned from past chemi- mendations drawn from material presented in the first five cal events and their impact on continuing operations at chapters.