5
Preparing for Potential Future Chemical Events at Baseline Chemical Demilitarization Facilities

SUMMARY OF CHEMICAL EVENTS ANALYSES

The committee’s analyses of past chemical events at Johnston Atoll Chemical Agent Disposal System (JACADS) and Tooele Chemical Disposal Facility (TOCDF) indicate that the causal factors are similar to those associated with breakdowns of other safety-critical systems. Release of chemical agent may be triggered by equipment design flaws and failures, by procedural deficiencies, and by human actions—i.e., by both latent and active failures (see Chapter 2).

The task of dismantling and destroying chemical weapons is inherently hazardous, but the Program Manager for Chemical Demilitarization (PMCD) has incorporated extraordinary safety precautions into both plant design and personnel training. The chemical demilitarization incineration plants are virtual fortresses built to withstand the consequences of accidents, and, to date, releases of chemical agent from these facilities have been rare, isolated events involving only small amounts of agent, even under upset conditions (NRC, 1996, 1997, 1999a). State-of-the-art quantitative risk assessments have determined that the major hazard to the surrounding communities arises from potential releases of agent from stockpile storage areas, not the demilitarization facilities (U.S. Army, 1996a; NRC, 1997; see also Chapter 1 and Appendix E). Further, to date by far the largest releases of agent have occurred in the storage areas, as described in Chapter 1.

The Army has sought to build in the process of learning by experience to avoid accidents where possible, and to avoid repeating them in any case. The centerpiece of this effort, the programmatic lessons learned (PLL) database, is admirable as a personnel-training tool but requires further modification to improve its accessibility (see Chapter 4). Despite considerable effort in plant design and personnel training, mistakes have been made and problems have occurred in the chemical demilitarization process.

The Army has established extraordinarily low agent threshold concentrations to trigger site alarms and a subsequent shutdown of the plant (see Chapter 1). While laudable as an effort to protect worker and public health, these overly sensitive alarms introduce their own kinds of operating problems. Difficulty in reliably detecting agent at such low concentrations leads to recurring false positive alarms. It also means that alarms triggered by chemical events in which agent levels stay near threshold will actually pose no risk to the worker or the public.

Given the inherent complexity of the chemical demilitarization task at the assembled weapons stockpile sites, it is almost certain that new problems will continue to arise, particularly from aging and deteriorating weapons and the challenges of demilitarization plant closure and decommissioning. There will be future chemical events, and serious consequences to both plant personnel and surrounding communities cannot be ruled out. This chapter focuses on prudent ways to reduce their number and to minimize their consequences.

CHEMICAL EVENT RESPONSE AND REVIEW BY MANAGEMENT

Army Regulation 50-6 presents in detail the response to a chemical event and its reporting expected from the depot commander (U.S. Army, 1995). The objective is to:

. . . encompass those actions to save life, preserve health and safety, secure chemical agent, protect property, prevent further damage to and remediate the environment, and help maintain public confidence in the ability of the Army to respond to a military chemical accident or incident. . . . The major army commands (MACOM) commanders will establish procedures to review each chemical event and to initiate safety investigations when warranted . . . .

The extent of the review process generally varies with the seriousness of the incident. The review process for a



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 44
5 Preparing for Potential Future Chemical Events at Baseline Chemical Demilitarization Facilities SUMMARY OF CHEMICAL EVENTS ANALYSES threshold concentrations to trigger site alarms and a subse- quent shutdown of the plant (see Chapter 1). While laudable The committee’s analyses of past chemical events at as an effort to protect worker and public health, these overly Johnston Atoll Chemical Agent Disposal System (JACADS) sensitive alarms introduce their own kinds of operating prob- and Tooele Chemical Disposal Facility (TOCDF) indicate lems. Difficulty in reliably detecting agent at such low con- that the causal factors are similar to those associated with centrations leads to recurring false positive alarms. It also breakdowns of other safety-critical systems. Release of means that alarms triggered by chemical events in which chemical agent may be triggered by equipment design flaws agent levels stay near threshold will actually pose no risk to and failures, by procedural deficiencies, and by human ac- the worker or the public. tions—i.e., by both latent and active failures (see Chapter 2). Given the inherent complexity of the chemical demilita- The task of dismantling and destroying chemical weap- rization task at the assembled weapons stockpile sites, it is ons is inherently hazardous, but the Program Manager for almost certain that new problems will continue to arise, par- Chemical Demilitarization (PMCD) has incorporated ex- ticularly from aging and deteriorating weapons and the chal- traordinary safety precautions into both plant design and lenges of demilitarization plant closure and decommission- personnel training. The chemical demilitarization incinera- ing. There will be future chemical events, and serious tion plants are virtual fortresses built to withstand the conse- consequences to both plant personnel and surrounding com- quences of accidents, and, to date, releases of chemical agent munities cannot be ruled out. This chapter focuses on pru- from these facilities have been rare, isolated events involv- dent ways to reduce their number and to minimize their con- ing only small amounts of agent, even under upset condi- sequences. tions (NRC, 1996, 1997, 1999a). State-of-the-art quantita- tive risk assessments have determined that the major hazard CHEMICAL EVENT RESPONSE AND REVIEW BY to the surrounding communities arises from potential re- MANAGEMENT leases of agent from stockpile storage areas, not the demili- tarization facilities (U.S. Army, 1996a; NRC, 1997; see also Army Regulation 50-6 presents in detail the response to Chapter 1 and Appendix E). Further, to date by far the larg- a chemical event and its reporting expected from the depot est releases of agent have occurred in the storage areas, as commander (U.S. Army, 1995). The objective is to: described in Chapter 1. . . . encompass those actions to save life, preserve health and The Army has sought to build in the process of learning safety, secure chemical agent, protect property, prevent fur- by experience to avoid accidents where possible, and to avoid ther damage to and remediate the environment, and help repeating them in any case. The centerpiece of this effort, maintain public confidence in the ability of the Army to re- the programmatic lessons learned (PLL) database, is admi- spond to a military chemical accident or incident. . . . The rable as a personnel-training tool but requires further modi- major army commands (MACOM) commanders will estab- fication to improve its accessibility (see Chapter 4). Despite lish procedures to review each chemical event and to initiate considerable effort in plant design and personnel training, safety investigations when warranted . . . . mistakes have been made and problems have occurred in the chemical demilitarization process. The extent of the review process generally varies with The Army has established extraordinarily low agent the seriousness of the incident. The review process for a 44

OCR for page 44
45 PREPARING FOR POTENTIAL FUTURE CHEMICAL EVENTS AT BASELINE CHEMICAL DEMILITARIZATION FACILITIES serious incident can be quite lengthy. Every chemical event sessment implies more than knowing the summary numeri- should be investigated promptly, particularly those consid- cal results of quantitative and health risk assessments. It ered potentially or actually serious. Memories of the event also requires knowing the details, including the assumptions, will change with time. Having people identified in advance simplifications, and omissions, of the analyses. The results as potential candidates for a review team would appear must be viewed in the full context of the risk assessment, as worthwhile. well as in the context of the actual safety performance of the One of the objectives of Army Regulation 50-6, stated plant. This perspective must be accompanied by a better above, is to “help maintain public confidence.” The com- understanding of explicit and implicit uncertainties. mittee believes that building trust requires regular and reli- Understanding the results of risk assessment also means able communication between the Army and the communi- knowing the significant contributors to risk, i.e., knowing ties around the demilitarization plants. It does not appear how improved performance can reduce risk and how de- that these communities feel that such communication has graded performance could increase risk. With this knowl- been achieved. Public trust is not easily established and is edge: very difficult to rebuild once lost. The recent report of the U.S. Commission on National Security (a commission • Managers and workers can develop options for re- headed by former senators Gary Hart and Warren Rudman) ducing risk or for ensuring that risk does not in- comments on the general lack of confidence in federal em- crease. They can also consider how proposals for ployees (USCNS, 2001). This general lack of confidence, change affect risk. exacerbated by the unfortunate pattern of interactions be- • Workers, emergency response personnel, and others tween PMCD and external stakeholder groups (NRC, 1996), can better understand their personal risks and how has created a serious deficit of trust in the Army’s chemical best to protect themselves and each other. demilitarization program on the part of important segments • Emergency preparedness managers can focus their of the public. In addition to addressing the public’s lack of planning and training programs on the most impor- confidence in federal officials, at some sites PMCD must tant scenarios or sources of risk to the surrounding also deal with public distrust of state and local officials. A communities. recent NRC letter report (NRC, 2000c) points out that: • State and local officials can provide more informed oversight in their decision making. . . . open, two way communications between PMCD and • Everyone can participate knowledgeably in the risk stakeholders are necessary, but insufficient. PMCD needs to management process. encourage public trust in official representatives of the pub- lic (i.e., Citizens Advisory Commissions and local regula- Quantitative and health risk assessments are complex tory bodies) as much or more than it needs to build trust in and, of necessity, include simplifications. The plant safety the Army. professionals should review the assessments thoroughly to The memorandum of understanding between TOCDF be aware of their basic assumptions and/or limitations. Plant and Tooele County (see Appendix G) should help build con- operating requirements may change, and changes need to be fidence that public officials are fully informed and respon- viewed in the light of the risk assessments. sive to chemical events, thereby contributing to building Several lessons can be learned about risk management trust. This approach might serve as a model for other com- from thinking about possible responses to certain kinds of munities with similar concerns (Utah DEQ, 2000b). chemical events. • False positive alarms. The history of false positives BUILDING ON THE RESULTS OF RISK ASSESSMENT has contributed to a number of chemical events as Risks associated with the chemical demilitarization fa- described in Chapter 2. These result from a mind- cilities have been studied in depth, through quantitative and set that develops in operators. Faced with a series of health risk assessments and systems hazards analyses (see false positive alarms, they tend to disbelieve future Chapter 1). The quantitative risk assessment, in particular, alarms, at least to the extent that they seek confirma- is a living document, subject to change as new information tion before taking action. A question that has been arises or facilities or operations are altered. It provides ex- raised is whether a similar complacency could de- cellent guidance on where risk is the highest, and thus where velop among emergency response managers and the greatest care is needed. The Army’s “Guide to Risk even the general public? If they too are subjected to Management Policy and Activities” provides a process for false alarms, they may delay ordering or responding managing risks, particularly when changes are made, and for to orders for evacuation or sheltering. Generally, communicating information on change to the public (U.S. these people have not been subjected to the false Army, 1997b). alarms, but if it should happen, similar problems Understanding and building on the results of risk as- could arise.

OCR for page 44
46 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES • Evacuation versus sheltering. At some sites, there difficulties, but based on the committee’s site reviews, a cul- has been controversy over the question of evacua- ture of questioning processes and constantly improving op- tion versus sheltering. Countering the belief that erations does not seem to exist. To be fair, it is clear that evacuation is always the safe path are at least two plant management is aware of the importance of being proac- circumstances. First, evacuation itself can create tive on safety, rather than being reactive only. Certainly there hazards. It disrupts the economy and daily life and has been real improvement in plant layout, equipment, and can create high stress. It has led to injuries due to so on (see Chapter 4). Based on the committee’s observa- traffic accidents and improper use of safety equip- tions and discussions with operating personnel, TOCDF is ment. Second, analyses by Chemical Stockpile clearly a better designed and engineered plant than JACADS, Emergency Preparedness Program (CSEPP) plan- and the third-generation incineration plants, as exemplified ners have shown that for some release scenarios, by the Anniston Chemical Agent Disposal Facility, appear to be a significant improvement1 on TOCDF. Many of these evacuation can place the evacuees directly in the path of the hazardous plume (Blewett et al., 1996). improvements were made by seeking better ways of doing For some scenarios, sheltering in place (remaining things, and anticipating possible future problems rather than indoors with the doors and windows sealed) as the reacting after a problem has occurred. The committee en- plume passes, followed by evacuation, can greatly courages a continued vigorous questioning of plant operation reduce exposure. Continued sheltering after the and equipment by management and operating personnel. cloud has passed may lead to exposure as severe as This open-minded, questioning approach should apply to being caught in the plume. In these cases, shelter- operating practices and even equipment design. ing as the cloud passes, followed by evacuation through contaminated areas, can be the most effec- 1Although the basic processes for weapon destruction will remain the tive protective action. same three lines of incineration as at TOCDF and JACADS (a furnace for injecting and burning liquid agent, a rotary furnace for propellant and ex- plosive materials, and a furnace with a moving conveyor primarily for metal BUILDING A SAFETY CULTURE parts), improvements have been made compared with TOCDF and JACADS. For example: TOCDF has clearly made an effort to promote plant safety. Two examples are (1) the use of the Safety Training • The pollution control systems of the new plants will include acti- Observation Program (purchased from the DuPont Com- vated carbon filters for the incinerator exhaust gas. This is fairly pany) and (2) the use of the Voluntary Protection Program new technology, not in common use when JACADS and TOCDF developed by the Occupational Safety and Health Adminis- were designed. Trial burn data on those two early plants showed that carbon filters were not needed to meet environmental standards. tration (OSHA). A good safety organization on paper, how- More recently, however, some samples of mustard have shown un- ever, does not ensure a high-quality safety culture. Some of expectedly high levels of mercury that could be a problem in exhaust the past events at both JACADS and TOCDF arose from ob- emissions. Carbon filters represent the technology of choice for viously poor safety practices. The recordable injury rate handling this problem. Other changes in the pollution abatement (RIR) at TOCDF, for example, has been unimpressive (Chap- system are required to accommodate the carbon filters. The exhaust gas must be cooled and its humidity reduced to maintain the carbon ter 4). The NRC has emphasized the need to focus on safety filter’s function. with constant attention to detail, starting with a complete and • The ventilation air through the plant as well as the combustion air persistent commitment from management (NRC, 1999a). will have variable-speed motors driving the fans. This should be a great improvement in controlling airflow rates, particularly at low rates (a problem in the May 8, 2000, TOCDF incident). The tech- OPERATIONAL CHANGES nology for doing this with very large motors was just being intro- duced when TOCDF was designed and was not included. It is clear that (1) serious mistakes have been made in • Isolation valves are included in the duct between the DFS burner and chemical demilitarization plant operations in the past and after burner. (The same valve was added to TOCDF after the May 8, (2) strict standards of operating practice have not been uni- 2000, event.) They should permit improved control during start-up. formly enforced (see Chapters 2 and 3). These are failures of • The DFS tipping gate has been redesigned to prevent jamming (part of the problem in the May 8, 2000, event). management to fulfill their responsibilities. Improvement will • The large isolation valves on the individual HVAC carbon filter come only with serious management effort, significantly banks have a small “bleed” valve connecting to the exhaust flow in greater than in the past. Strong safety cultures and an adher- the new plants. The purpose is to maintain the filter bank at negative ence to defined operating procedures have been established pressure even when the filter is temporarily out of service. This in other industries. The goal for chemical demilitarization should prevent migration of agent from the filter bank into the con- necting vestibule when the filter is out of service, a problem in the plants should be to match the best achieved in industry. past. A criticism that is easily voiced but difficult to respond • The DFS cyclone is in an enclosure that is to be monitored with an to is the general acceptance of the status quo by chemical ACAMS and that has a carbon-filter on an exhaust. (This modifica- demilitarization operating people and management. Changes tion was made in response to a JACADS event where VX was de- are made in response to chemical events or obvious operating tected on the cyclone ash.)

OCR for page 44
47 PREPARING FOR POTENTIAL FUTURE CHEMICAL EVENTS AT BASELINE CHEMICAL DEMILITARIZATION FACILITIES The 2000 letter report of the NRC Stockpile Committee the systems engineers with design knowledge into the train- recommended a similar open-minded approach for a public ing program could help convey that knowledge to the opera- involvement program (NRC, 2000b), naming as one require- tors. These engineers are probably in the best position to ment for a successful program “the capability to identify know the equipment and its characteristics and limitations, (even anticipate) serious problems and the flexibility and cre- information that plant operators need when unusual or unex- ativity to address them.” The current study suggests that this pected conditions occur. Many plant operators seemed to approach is also needed in plant operations and technology. the committee to have only a superficial knowledge of the A better understanding of the limitations of plant equip- operating principles and data processing algorithms of im- ment might also be helpful. In the May 8-9, 2000, incident portant process instrumentation and controls. But such at Tooele, for example, there were some serious operating knowledge is crucial to determine how to interpret reported errors. But they were compounded by the operator’s struggle instrument console readings during upset conditions which to bring the system back under control. It is the judgment of may exceed the normal ranges over which key instruments the committee that some technical education and more are calibrated or can be expected to operate reliably. A care- hands-on testing on the system simulator would have helped ful walk-through of any new procedure should precede its (see Chapter 2). start-up. There are usually surprises when new processes are first The Army’s more recent quantitative risk assessments tried. In view of the particular sensitivity of the chemical (QRAs) include detailed human reliability analyses that iden- agent disposal program, the committee emphasizes the need tify potential human performance problems. Bringing this for a hazardous operations (HAZOP) analysis for any new information into the training program will provide operators process (see Chapter 4). A HAZOP analysis by suitably with a view of what activities are especially vulnerable and trained people, and with input from operating people, could why that is so. In addition, training simulators, which mimic be particularly useful: it might identify problems and at the the operation of the various components of the instrument same time warn the operating people about what to expect. and control systems and demonstrate the effects of various New plant start-up represents a special problem with operator actions or inactions, are now being provided in the inexperienced people. Trial burns with surrogate feeds and chemical demilitarization plants. Targeted training with with the pollution abatement system in full operation, as well simulators and knowledge-based thinking exercises on plant as disassembly trials with blank munitions, should provide operation need to be developed. substantial operating experience before any chemical agent Training on overall plant operations should cover ev- is fed to the process. It is fairly common experience in in- eryone in the plant and analytical laboratory, not just the dustry to include design people on start-up teams for new operating and control people. However, this training has to facilities. As suggested earlier, their detailed knowledge of be tailored to the specific jobs and knowledge levels of each the process equipment and its limitations could be helpful to group of workers. Workers need to understand how what the operating people. they do fits into the overall operation and how things going wrong in their operations affect the whole plant and the like- lihood of accidents and releases. The QRA and HAZOP WORKER EDUCATION, TRAINING, AND analysis are a good potential source of this information. INVOLVEMENT Some of the reports of operational mistakes coming Safe plant operation depends on an educated, well- from within the plant and circulated widely within the af- trained staff. The risk to workers in an incineration plant is fected communities have come from people who are simply greater than the risk to the public (NRC, 2000c). Training uninformed and do not know normal procedures. Box 5-1 should emphasize that processing agent demands a mind-set provides examples of such uninformed observations. that always accepts a positive analysis as “real” until proven Chemical demilitarization plants are complex. A better otherwise. knowledge of the complexity of the plant and the care and One approach to safe operation is through the use of design that have been incorporated may instill pride in being standard operating procedures (SOPs). These have been part of the important national effort of weapons disposal. used extensively at JACADS and TOCDF. The most seri- The potential costs (e.g., lost trust) of having the local public ous chemical events of the past have occurred, however, alarmed by reported misperceptions of uninformed workers when there was no SOP. There will always be combinations can be substantial. of circumstances for which no SOP has been written and the operating people must rely on knowledge-based decision DESIRED PRINCIPAL-AGENT INTERACTIONS making. Even with SOPs, there is no guarantee that mis- takes will not occur. It is vital that decisions be made on the It is imperative that officials at the chemical demilitari- basis of accurate operational knowledge. Operating people zation facilities communicate openly, frequently, and in a should know their equipment and its limitations. They need timely fashion with nearby residents and officials. The pat- to know the why of their job as well as the what. Bringing tern of communication with and responsiveness to the local

OCR for page 44
48 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES BOX 5-1 Examples of Observations That the Committee Concluded Were Uninformed “December 9, 2000—Agent break through in HVAC filter bank. A video given to the committee and referred to in Appendix C showed ACAMS readings of 3.01 [TWA].” rockets being sheared and the pieces dropping to the deactivation furnace system (DFS) below. “October/November 1997: Sources inside TOCDF (who wish to re- Approximately every 11/2 minutes a large cloud of condensing vapor, main anonymous) communicated to CWWG [Chemical Weapons Working Group] several shutdowns/incidents at TOCDF due to com- referred to by the citizen group as “agent volatization,” rose into the picture, puter malfunctions, slag build-up in the PAS, numerous agent migra- undoubtedly coinciding with opening of the gate to the DFS. In fact it was a tions within the facility, and alarm ring-offs in the common stack, cloud of condensing steam, as cooling water from the shear blade and the MDB [munitions demilitarization building] and HVAC stack (averag- sliding gate dropped into the hot furnace to be instantly vaporized. ing 2-3 per week).” “Site-masking alarm and/or stack alarm. Potential case of chemical war- fare agent release or release of other related toxic chemicals (unidentified These entries suggest that agent may have been released through to date).” [the most common incident listed by the CWWG (Appendix C)] the heating, ventilation, and air conditioning (HVAC) filter to the envi- ronment. In fact, the HVAC was operating as designed. The carbon filter bank consists of six carbon beds, with exhaust gas flowing It is almost certain that the ACAMS alarm was not due to agent, be- through all six in series. The gas spaces between beds 1 and 2, 2 and cause there was no depot area air monitoring system (DAAMS) confirma- 3, and 3 and 4 are monitored by an automatic continuous air monitor- tion. The committee concluded that the event reports as written are mislead- ing system (ACAMS) (on a timer). Eventually agent will break through ing but considers them to be from a source unfamiliar with the stringent bed 1 as that bed approaches saturation, and this is undoubtedly the laboratory procedures used to analyze DAAMS samples taken coincident “agent break through” referred to by the whistle-blower. Agent break- with each ACAMS alarm to confirm or deny the presence of agent and to through of bed 1 usually follows many weeks of operation, and with attempt to identify the cause of the alarm in the absence of agent. the gas having to traverse 5 more beds the agent breakthrough of the first bed does not call for immediate shutdown. However, it does NOTE: Observations quoted are drawn from the Chemical Weapons Work- indicate that the carbon should be replaced soon. ing Group list of events provided to the committee (Appendix C). public and local officials can have substantial effects. Be- tions under which agents can carry out the demilitarization yond addressing the immediate health and environmental task (where some discretion may be essential) requires en- concerns posed by a chemical event, frequent and open dia- gendering and maintaining a degree of trust by principals for logue can alter perceptions of risk and trust, influence de- agents. mands for policy change, and mitigate undesirable effects on Effective handling of the principal-agent relationship in local economic growth and property values. As discussed in the chemical demilitarization program setting appears to the Chapter 3, the agents in the demilitarization process (regula- committee to require (1) demonstrable and timely assess- tory agency officials, the Army, and contractors at the chemi- ments of the problems leading to chemical events and means cal demilitarization facilities) must gain and retain the trust for their correction, (2) complete and timely disclosure of of the principals (local public and the officials who represent events by the agents, and (3) overview processes that assure them) in order to effectively destroy the chemical weapons principals of effective oversight. stockpile in a safe and timely manner. In its assessment of chemical events (Chapter 2), the Absent complete trust, the mechanisms by which prin- committee found specific instances (e.g., the TOCDF May cipals gain confidence in adequate performance by agents 8-9, 2000, incident) that resulted in a damaging erosion of include effective monitoring of agent behavior and appro- the confidence of principals in the monitoring and control priate inducements and sanctions to obtain desired perfor- processes. It is essential that plant operators remain cogni- mance. The lower the level of trust, the greater the need for zant of the needs of principals for high degrees of confi- monitoring and incentives. At the same time, more stringent dence in the monitoring and control protocols (incentives monitoring and incentives can limit the discretion necessary and sanctions) and mechanisms over the entire chemical de- for agents to effectively and efficiently accomplish their militarization program. Apparent weaknesses or failures at complex task. The trade-off between effective monitoring one facility or in one phase of operations will be seen to and controls by principals over agents and optimal condi- carry over to others. Protocols for reporting and responding

OCR for page 44
49 PREPARING FOR POTENTIAL FUTURE CHEMICAL EVENTS AT BASELINE CHEMICAL DEMILITARIZATION FACILITIES to events should stress meeting the needs of the array of mate of the required frequency of slag removal, e.g., external principals for assurance of timely, accurate reports for TOCDF, after 250,000 lb of agent destroyed. of events and rapid, thorough assessment and corrections. • Mist eliminator candle replacement (plugged dur- One important step to increase confidence in the moni- ing deactivation furnace system (DFS) rocket runs, toring process will be to ensure that representatives of prin- probably due to fiberglass). Time minimized by cipals (e.g., local stakeholder groups) are included in the having a spare eliminator on hand. teams assembled to investigate any serious chemical events. • LIC rebricking, maintenance that can also probably In addition, each site should develop clear and specific pro- be planned ahead and done during “contingency tocols that reflect the need to quickly, openly, and thoroughly time” (i.e., when the plant will be shut down for inform all relevant principals of chemical events. other activities such as agent changeover). More broadly, program officials should consider ways Rebricking is needed after approximately 2,000,000 in which principals and their representatives can participate lb of agent (with decontamination fluid) have been in ongoing oversight efforts. The NRC has suggested else- processed. where (NRC, 1999b) that representatives of the local public serve on monitoring teams whose purpose is to ensure that A further complication arises from the age of the weap- chemical weapons destruction processes (and the associated ons as well as their varied history—“leakers” and “gelled organizations) are operating as they should. Such an effort— agents” require changes in “standard” operations, for ex- ranging from temporary appointment of community observ- ample. The shutdowns and restarts resulting from these feed ers on investigation teams to more permanent participation stock variations can be planned for and shutdown times can of community representatives in incident review boards— be minimized. may increase the confidence of local principals that effective The Operations Schedule Task Force 2000 recommended oversight is in place. study teams to suggest how to minimize downtime (U.S. Army, 2000c); these teams should be very helpful. The com- mittee suggests that industrial experience with carefully RAPID AND SAFE RESTART REQUIREMENTS planned shutdowns for maintenance at regular intervals might be applicable. It is not clear that “project management,” Restarts After Changeovers and Maintenance which has developed into a distinct engineering subdiscipline, The chemical weapons plants have very frequent shut- is being fully integrated into the chemical demilitarization downs and restarts—“frequent” compared with industrial program. The suggested study teams noted above would rep- plants of comparable size. These shutdowns are required by resent a step in that direction. the variable nature of the plant feed: a variety of weapon types with differing disassembly requirements, containing Restarts After a Chemical Event three different chemical agents. The times required for changeover have been estimated to be surprisingly long (U.S. Major chemical events can impose further shutdowns Army, 2000c). For example: with unpredictable shutdown times. Some of these have led to major structural changes and changes in some operating • A change in agent: procedures. These changes stem from the incident reviews, —17 weeks—the time required for decontamination, and they all require regulatory approval. Shutdown times monitor conversion and baselining, and some may be long, e.g., 4 1⁄2 months for the May 8-9, 2000, incident equipment changeovers. at TOCDF. The Operations Schedule Task Force 2000 sug- • A change in munitions type: gested that a 2-week outage every 6 months be included in —5 weeks without complete equipment removal advance planning, to accommodate unplanned major mainte- (e.g., projectile to projectile types). nance (U.S. Army, 2000c). The committee believes that this —8 weeks with equipment removal (e.g., mines or unplanned shutdown allowance is less than past experience rockets to projectiles). would indicate is necessary, but these unplanned shutdowns should decrease with time, as operating experience is gained. There are other normal maintenance items that require There also may be opportunities for reducing the required extended shutdown periods but can probably be scheduled shutdown times after such incidents. Maintaining a larger during other changeovers. For example: inventory of critical spare parts has been suggested as one strategy to reduce lost operational time. • Slag removal from the liquid incinerator (LIC) sec- Obtaining regulatory approval to restart after a chemical ondary burner. [The slag removal system at TOCDF incident may cause delays, although the committee heard no failed before the refractory failed, so that slag had to specific complaints of this. The Army Audit Agency, how- be removed manually.] Time required: 10 days. The ever, has been critical of the chemical demilitarization pro- experience at JACADS and TOCDF permits an esti- gram for its handling of funds, based in part on regulatory

OCR for page 44
50 EVALUATION OF CHEMICAL EVENTS AT ARMY CHEMICAL AGENT DISPOSAL FACILITIES delays. Funds obtained for current planned programs could event probably extended the post-event shutdown associated not be spent because of delays in regulatory approvals (U.S. with that event unnecessarily. Preagreement at each demili- Army, 2001g). It is not clear, however, that regulatory delay tarization site on the composition of a joint event investiga- has been a serious problem in connection with unpredicted tion team, representing all regulatory and operational stake- shutdowns, where there was no opportunity for advance holders and chartered to produce a single, comprehensive planning. investigation report, could save significant shutdown time Finally, as noted in Chapter 3, effort spent on the mul- and clearly focus all parties on the steps necessary to achieve tiple investigations of the May 8-9, 2001, Tooele chemical safe restart of operations after future chemical events.