Risk Assessment and Management at Deseret Chemical Depot and the Tooele Chemical Agent Disposal Facility (1997)

The principles of risk management are presented in the last section of Appendix A, where risk management is described as the process by which risks are understood and controlled. Step 2 of the risk management process at the end of Appendix A also provides some general examples of different options for risk reduction that can be used in risk management at DCD/TOCDF. All affected parties have roles to play in the risk management process. The Army is responsible for managing the chemical stockpile and its destruction. However, the Army's contractors, individual workers, local governments, and the affected public must all participate for the process to proceed efficiently and safely (NRC, 1996c). Risk management usually involves the following steps:

• understanding the risk (including identifying major contributors to risk)
• suggesting alternative ways to reduce risk
• evaluating risk reduction alternatives
• selecting preferred alternatives (including implementing decisions)

As techniques for risk management have changed rapidly over the past 15 years, approaches to risk management at DCD/TOCDF have also evolved. In the early history of the stockpile, surveillance and maintenance were internal Army responsibilities. When the TOCDF was planned and constructed, the Army focus expanded to comply with regulations imposed by outside authorities with jurisdiction over disposal operations. Stemming from a growing base of knowledge and experience in risk management, and with the encouragement of the Stockpile Committee (NRC, 1996b), the Army is now attempting to move beyond a command and control compliance culture toward a more open and comprehensive risk management process.

The types of risk-related activities that must be managed at DCD/TOCDF are described in Chapter 2 and include the storage and handling of agent and munitions and the operation of the agent destruction facility. The QRA and HRA studies provide a current indication of the sources of risk, the magnitude and distribution of risks, and the levels of uncertainty. Parties interested in DCD/TOCDF operations include the Army, Army contractors, site workers, the local community, communities near other stockpile sites, state and local governments, state emergency preparedness programs, and state citizens advisory commissions (CACs). For risk management to be effective, each group needs to understand the assessment process, the results, and the significance of the results. This requires:

• effective risk communications among the various levels of programmatic and on-site Army and contractor personnel
• effective risk communications to technical audiences

• • -	presentation of materials in language understandable to engineers, technicians, and technical managers

  • -	presentation of detailed material for QRA/HRA practitioners and risk managers, including emergency preparedness officials and local regulatory groups

• effective risk communications to nontechnical audiences, which are essential to the community involvement program

Finally, effective risk management requires a program to track the present status of risk estimates, monitor changes that might shift risk levels significantly, evaluate suggestions for risk reduction, and assess proposed operational improvements in a way that considers the concerns of interested parties.

The current CSDP risk management program is a multilevel program that defines policy, sets requirements, provides guidance on implementation, and, at the facility level, defines specific requirements the facility must meet and specific management processes that must be implemented. The CSDP risk management program is evolving and has been formalized in the last two years, based on a long history of safety and hazard analysis and regulation by the Army. An informal risk management process was developed at the TOCDF parallel with the site-specific QRA. This process was described in the Stockpile Committee's March 1996 report, Review of Systemization of the Tooele Chemical Agent Disposal Facility (NRC, 1996b), which summarized a number of changes that had been implemented as a result of accident scenarios identified in preliminary work on the QRA. As part of the risk management process, the following risk monitoring activities have been introduced:

• performance evaluation (based on feedback from activities and incidents)
• emergency response exercises (periodic exercises on site, with CSEPP personnel)
• risk tracking (as new data become available, as risk models are improved, and when changes occur in the facility, the related changes in risk related to safety, environmental protection, and emergency preparedness will be calculated and tracked)
• lessons learned programs (PMCD now invites all facilities to participate in meetings about design lessons learned and programmatic lessons learned)

The committee supported these activities and recommended that they be formalized before the end of the first year of agent operations. However, the Stock-pile Committee is concerned that the emphasis on safety at the TOCDF has been focused on agent-related issues, with a corresponding lack of emphasis on traditional industrial safety practices and procedures (NRC, 1996b). The committee believes that failure to wear required protective equipment (e.g., to protect the eyes), poor housekeeping, and unsafe conditions (e.g., obstructions blocking access to safety equipment and walkways), which were observed by the committee during recent site visits to the TOCDF, indicate the lack of an established safety culture or mind-set. Although the absence of a pervasive safety culture that emphasizes agent-related and nonagent-related safety matters equally is not likely to change the QRA public risk estimates, it may significantly increase worker risk.

The first step in the Army's attempt to formalize the risk management process was the publication of the Tooele Chemical Agent Disposal Facility Risk Management Plan in April 1995 (U.S. Army, 1995b). After several reviews, that document was replaced by a program-wide document, Chemical Agent Disposal Facility Risk Management Program Requirements (U.S. Army, 1996e), which provides a basis for the CSDP risk management program. The risk management program is a framework for understanding and controlling all elements of risk within the disposal facility and the stockpile storage area. It links risk management needs to other specific requirements of the Army and other

parties at top levels of management and identifies specific documents and references that apply to all CSDP facilities (as recommended in the committee's Systemization report [NRC, 1996b]).

Currently, two additional steps are being taken—plans for a site-specific risk management program and a programmatic policy guide are being developed. The TOCDF Risk Management Program Plan (EG&G, 1996), initiated by Edgerton, Germerhausen and Grier, Incorporated (EG&G), the site contractor at the TOCDF, includes definitions of contractor responsibilities as well as responsibilities of various risk management program elements. The plan also includes a "Compliance Matrix" that identifies site implementing documents used to meet the requirements listed in the program requirements document cited above (U.S. Army, 1996e). The site-specific plans are to be "living documents" (i.e., they will be continually updated.)

In January 1997, the Army issued its draft, A Guide to Risk Management Policy and Activities (the Guide) (U.S. Army, 1997c). This draft provides an overview of the processes for managing risks associated with PMCD activities. It defines risk terminology and describes risk management processes and decision bases. The role of each organizational element is described in terms of specific risk management activities: assessment, requirements, monitoring, management of change, and public participation. The rest of the draft Guide explains these activities in some detail, identifying the products of each organizational element, the evaluation tools, the source of authorization, and the techniques for tracking performance. This draft breaks new ground in Chapter 7 by presenting a process for managing changes that may affect the risk associated with PMCD activities. It defines issues that are matters of risk assessment and issues that are matters involving policy (value judgments) and attempts to establish an approach to integrating them and involving the public in that integration.

The draft Guide explains the PMCD's risk management policy in the context of the organizational structure shown in Figure 3-1 (p. 63 of the Guide). Although the role of public affairs programs and two-way communications between the Army and the public are acknowledged, and a specific public participation element is defined for the change management process, Figure 3-1 suggests that the public outreach program is separate and will not be integrated into the risk management process in a manner consistent with past NRC recommendations (NRC, 1996b, 1996c).

The PMCD policy indicates that risk management is integrated into the normal functioning of the organization:

• Operations are now based on the Risk Management Program Requirements document (U.S. Army, 1996c).
• The Risk Management and Quality Assurance Office has been assigned the task of integrating risk management for operations, design, and construction.
• The Environmental and Monitoring Office has been assigned the task of assessing hazards to the environment, the populace, and biota in terms of regulatory requirements.
• The CSEPP has been assigned the task of planning for potential emergencies and providing liaisons with other emergency preparedness organizations.
• The Public Affairs Office is charged with providing liaisons among the public, the CAC, state authorities, and the Army to facilitate public involvement.

Another significant element in risk management is the management of change. Although changes are usually made for good reasons, overall safety of the facility could be compromised if the effects of change on risk levels are not understood. Changes need to be documented and analyzed to see if they affect procedures, training, or other aspects of the program. In Figure 3-2 (p. 46 in the Guide), the change process is initiated from an "Established Configuration," which is the current state of the facility. This Established Configuration is based on the initial design of the facility and incorporates changes that have been approved and implemented. The Established Configuration is the basis for the plant's up-to-date HRA and QRA.

A description of how TOCDF evolved to the Established Configuration can be found in Example 1 later in this chapter. The change process continues with the definition of a change package and a coarse screening process to preclude detailed analyses of changes that do not significantly affect risk. The effects of relevant changes on risk are then determined by revising the

Figure 3-1

PMCD's organizational elements directly related to risk management (p. 63 in the Guide).

Source: U.S. Army, 1997c. Note: FEMA is the Federal Emergency Management Agency.

HRA, QRA, and other Army safety evaluations. If a proposed change meets HRA standards and its effects on the QRA, the disposal schedule, and costs are documented, then the proposed change is presented for public comment.

If a change is significant, assessing its value is acknowledged to be both a policy question and a factual question. Structured discussions focus attention on all factors that affect the decision. Information on the impact of the proposed change is made available to the public, the CAC, and state regulators, and public comments are solicited. For the most significant changes (RCRA Class 3 modifications), the Army will conduct a public workshop. The Army's decision will take into account community desires and needs as well as important facts and intangible factors, which are summarized in Table 3-1 (p. 53 in the Guide). Note that factor 6 in Table 3-1, "comparison to previous decisions," ensures either that decisions are consistent or that the reasons for inconsistencies are clearly stated. A thorough consideration of uncertainties is also required. The Army will prepare responses to all public comments and inform regulators and the CAC of their decisions and rationale.

The committee agrees with the Army that "each change proposal is likely to involve unique circumstances and factors, so it is not possible (or desirable) to prescribe a set decision process with fixed criteria."

Figure 3-2

The change process (p. 46 in the Guide). Source: U.S. Army, 1997c.

Note: ECP means engineering change proposal; HE means hazard evaluation.

If all issues are considered in an appropriate and timely manner, general consensus may be possible. But even if consensus is not reached, the Army, as decision maker, will provide a "synopsis of the considerations and a summary of the overall decision basis, listing the rationale for each factor." In this way, interested parties can see if their concerns were considered and the affect they had on the decision.

In general, the Stockpile Committee agrees with the proposed management of change process developed by the CSDP and encourages its use. As this process is applied to change proposals, the Army will learn a great deal about the utility, benefits, and difficulties of the process, which may lead to improvements. The committee is concerned that the proposed process will not work well unless the current description of public

outreach is expanded to include public involvement that is fully integrated into the management of change process (NRC, 1996b, 1996c). In some parts of the draft Guide, broader public involvement is acknowledged, but more effort will be required to develop a fully integrated approach.

Although the Stockpile Committee generally agrees with the January 1997 draft, A Guide to Risk Management Policy and Activities (U.S. Army, 1997c), further development is needed in some areas. For example, the QRA has played a significant role in defining the current Established Configuration for the TOCDF, and it is reasonable to expect that the same will be true at other sites. The TOCDF is a good example of how information from the QRA can be used to facilitate risk reduction. However, Chapter 5 of the draft Guide barely mentions this.

The draft Guide only notes that interfaces and communication among the program elements are needed. It does not describe how interfaces will be managed across the functions of systemization/operation, safety, environmental protection, emergency involvement, and public participation. Some important cross-functional risk management issues are not described in detail and could benefit from a detailed description of managerial responsibilities. The draft Guide does indicate that maintaining the Established Configuration is currently the responsibility of the PMCD; that updating the QRA and HRA is the responsibility of the Army' s Risk Management and Quality Assurance Office; and that significant changes that impact risk need to be closely coordinated with emergency preparedness personnel, environmental managers, and the public. However, the coordination and chain of management responsibilities are not clearly defined.

The draft Guide appears to be an ideal place for the PMCD to elucidate the policy on safety culture; on the importance of industrial safety practices (a responsibility of each individual that can only be realized with a corporate commitment); on relationships among Army organizations, both at the PMCD level and on site; and with contractors and other agencies concerned with overall safety.

Table 3-2 (p. 14 in the Guide) provides a matrix of functions of the risk management plan and the five necessary activities defined in the draft Guide. The table shows few discrete links between the public outreach program and other activities. The committee is concerned that public outreach has not been conceptually linked to assessment, requirements, and monitoring. The public outreach function requires further thought and development.

In Table 3-3 (p. 67 in the Guide), the need for development of more fully-integrated public outreach and public involvement is indicated by the lack of an explicit method for tracking how public involvement and the CSEPP affect risk assessment and vice versa.

The process described above for risk management at the TOCDF is practical and has already been tested. The Established Configuration referenced in the Guide

represents the current state of the TOCDF and is the basis for the current QRA and HRA; improvements in TOCDF safety are ongoing.

Some initial changes to reduce risk were introduced in the committee's March 1996 Systemization report (NRC, 1996b). First, the QRA identified potentially high risk scenarios. Then, an engineering change analysis and test options were developed. In some cases, determining the technical merits of alternatives was expected to take significant time, so the order of the processing of munitions was changed, which had no impact on the HRA but greatly reduced the QRA risk for certain scenarios based on broad uncertainties. Reordering reduced the risk, and engineering changes were made to head off potential problems. Risk management improvements to reduce disposal-related worker risk for the first two disposal campaigns were modeled in the QRA. As additional measures are implemented for succeeding campaigns, the Army expects further reductions in worker risk.

Examples of how the QRA was used to make significant changes in the facility are given in the following paragraphs, which are summarized from the TOCDF QRA. The following changes are reflected in the current QRA:

Metal Parts Furnace Airlock. Earlier risk models identified the potential of a buildup and ignition of agent vapors in the feed airlock of the metal parts furnace (MPF). The most significant risk was for bulk agent containers held in the airlock for a longer-than-normal interval, but there was also a risk during the processing of projectiles. The public risk of a potential accident was estimated to be relatively small; however, the risk to workers was estimated to be higher than for other accidents. In light of these findings, the PMCD took steps to minimize the potential risk by changing the hardware to vent the airlock and minimize agent buildup. Operational changes were also made to limit the time an item could be held in the feed airlock.

MPF Processing of Weteye Bombs. The QRA postulated that the occurrence of an energetic reaction between molten aluminum and liquid agent could not be ruled out during the processing of aluminum weteye bombs. Calculations supporting the QRA indicated that molten aluminum could be present when significant agent was still left in the weteye bomb. The potential for explosive interactions of molten aluminum with water is known, although the exact conditions for explosions have not been determined. The interaction of molten aluminum and liquid agent has not been studied, but the potential for an energetic explosive reaction could not be ruled out. The QRA results indicated that this scenario could be a significant contributor to TOCDF worker risk, but the frequency and consequences were considered to be very uncertain.

Because of these uncertainties, the PMCD determined that it would be advantageous to change the TOCDF processing campaigns to exclude the weteye bombs from the first campaign. This change was made for two reasons. First, the processing campaign cannot begin without a trial burn protocol and schedule in place, and the MPF trial burn requirements for the first campaign would require larger-than-normal heel quantities (agent remaining after the draining operation).

The QRA identified the presence of liquid agent as a key element of a potential accident, and the trial burn requirement would increase the likelihood of liquid agent availability. Second, the uncertainties in the QRA results suggested that the issue required further study; a delay in processing weteye bombs allowed time for a thorough review of the calculations and the development of a strategy to eliminate the potential for aluminum-agent interaction. This work is ongoing and is to be incorporated into a QRA update before the weteye campaign begins.

Seismic Anchorage of the Liquid Propane Gas Tank. One contributor to the risk from earthquakes identified in the risk model was the 50,000-gallon liquid propane gas (LPG) tank installed at the site to provide a temporary fuel supply in case the natural gas supply via pipeline was interrupted. The QRA assessed the risk from earthquakes that exceeded the design basis and found that the LPG tank, although appropriately designed for seismic zone three, had a lower seismic fragility than other plant equipment and structures. Scenarios involving explosions and fires after an earthquake that could dislodge the tank from the outlet pipe or cause the tank to fail completely were found to contribute about one-half of the seismic risk. After reviewing this finding, the PMCD reconsidered the need for LPG and concluded that the original criterion for the size of the tank, which involved maintaining the operational status of furnaces and incinerators, was not necessary and that the LPG would only be used to provide fuel for the boilers. Thus, the fuel inventory in the tank could be reduced to less than 10,000 gallons.

The reduction had a twofold effect. First, the lower fuel volume and resultant lower weight have increased the seismic capacity of the LPG tank to the point that only much larger accelerations can cause the anchorage to fail and cause a gas leak. Second, even if the tank fails, the lower fuel quantity reduces the likelihood of an LPG explosion that would be significant enough to cause an agent release.

Campaign Order. The initial QRA results indicated that the original order of disposal did not eliminate the munitions with the highest risk first. For example, two early campaigns to destroy mustard would have had a minimal effect on the storage risk. Sensitivity studies were performed to optimize the schedule to eliminate

the higher storage risk items first. The change has resulted in a significant reduction in the public risk of storage during disposal.

Stockpile Earthquake Preparedness (Mitigation). DCD has been provided with a prioritized list for inspecting igloos after an earthquake. The QRA models, augmented by detailed analyses of seismic vulnerability, were used to determine which munitions and igloos are most vulnerable to damage so the personnel checking the condition of the stockpile could do so in the most risk-effective manner. Detailed vulnerability analyses have also resulted in some reduction in the estimates of seismic risk for the stockpile.

MPF Restart Procedures. While investigating the potential for an MPF explosion on restart, QRA analysts identified a path through the procedure where warnings against restart might not be clear in certain situations. The restart procedure was modified based on this finding.

MPF Exit Airlock Automatic Continuous Air Monitoring System (ACAMS). The QRA indicates that the risk is sensitive to an operational step—ACAMS monitoring in the MPF exit airlock. Therefore, the risk management plan will identify ACAMS monitoring in the MPF exit airlock as a critical activity that requires special attention during operation.

While QRA models were being developed, interactions between the PMCD, the TOCDF staff, and the QRA team led to improvements in the QRA analysis as well as to refinements in facility operations. For example, the identification of potential accident-initiating events required the development of process operational diagrams, which involved a step-by-step delineation of facility operations. Interaction ensured that the QRA models were accurate; at the same time, the QRA process helped to refine operations.

The risk assessment/risk management process needs to be orderly. The first QRA results identified several events that were high contributors to risk, including the seismic failure of the LPG tank. These high contributors were addressed immediately (as described above). The changes resulted in a residual risk dominated by seismic events. However, these events involve earthquakes that exceed normal design codes, with mean accelerations of 0.2 to nearly 1.0 g (1 g equals the acceleration of gravity). The recurrence intervals for earthquakes of this severity in the Tooele area are greater than one thousand years. Since the weakness in the LPG tank was corrected, the remaining risk is reasonably low. The remaining significant contributors to seismic risk are stacks of stored munitions that could tip and fall leading to agent leakage or explosions and fire, steel-arch igloos that could collapse and crush munitions, and structural failures in the container handling building/unpack area of the processing facility.

Two risk management processes are continuing: (1) the QRA team is examining all contributors at all stockpile sites to consider mitigation measures akin to those adopted at TOCDF for any high contributors; and (2) the management of change process described earlier in this chapter is being developed to ensure that changes proposed for any reason do not inadvertently increase risk. A discussion on using carbon filters with the baseline system (Example 2 later in this chapter) is provided because a decision on using them will probably be the first full-scale application of the management of change process. The addition of carbon filters has been under consideration for some time.

The ongoing use of the QRA in risk management has been effective in optimizing facility operations and improving safety. The PMCD reviews of the QRA include examining all of the assumptions and verifying all models. The QRA allows for the continuous investigation of ways to minimize the risks of given operations, which can, and has, resulted in refinements to operations and procedures.

The changes listed in Example 1 illustrate the effectiveness of using the QRA as a tool for risk management and indicate the acceptance of the QRA as an integral part of risk-informed decision making by the PMCD and the TOCDF staff. The interaction among the PMCD, the TOCDF staff, and the QRA team, and resulting improvements, were the basis for the development of the new management of change process that includes explicit consideration of effects on the HRA and solicits public comments as input to the decision process. Although public comments were not used as input to establish the initial system configuration, they will be solicited for major changes in the future.

The first full-scale application of the new management of change process is expected to be the evaluation of the proposed change to modify the pollution abatement system (PAS) by adding carbon filters to control emissions from the incinerators. An earlier NRC Stock-pile Committee report, Recommendations for the Disposal of Chemical Agents and Munitions (NRC, 1994b), recommended, in part, the following:

The application of activated charcoal filter beds to the discharge from baseline system incinerators should be evaluated in detail, including estimations of the magnitude and consequences of upsets, and site-specific estimates of benefits and risks. If warranted, in terms of site-specific advantages, such equipment should be installed.

This recommendation was prompted by the committee' s belief that adding a carbon filter system downstream of the existing PAS might provide further protection against an accidental release of agent from the stack (impacting the QRA) and might further reduce volatile organic emissions during normal and upset operations (impacting the HRA), even though organic emissions have been shown to be at trace levels and below the level of regulatory concern. The committee also recognized that adding a PAS carbon filter system might have adverse effects, e.g., the potential for filter fires and the consequent sudden accidental release of contaminants stored in the filter, and increased worker exposure to dioxins/furans and agent during disposal of the carbon filters.

In response to the NRC recommendation quoted above, the Army developed a conceptual design of a PAS filter system (PFS) and conducted a preliminary generic assessment of the application of this system to the common stack gases from the PAS of the baseline incineration system furnaces at the TOCDF. This evaluation concluded that the PFS could potentially enhance environmental performance of the baseline system but would increase the cost and complexity of the system. At the time (early 1994), data were insufficient to determine whether an installation was warranted because specific filter performance data and site-specific QRAs and HRAs had not been completed. Since then, site-specific QRAs and HRAs are either nearing completion or have been completed for most of the remaining stockpile sites.

Consistent with CSDP policy outlined in the draft Guide (U.S. Army, 1997c), the Army has developed a draft evaluation methodology for assessing PFS risk (U.S. Army, 1996f). The methodology requires the Army to conduct site-specific evaluations of the PFS, applying the same risk assessment methods, i.e., QRA and HRA, used to evaluate the risks of the established configuration of the baseline system. The need for site-specific evaluations derives from variations in site-specific factors, such as the types of chemical agents and munitions to be processed, the proximity of population, and different meteorological conditions.

The PFS can enter the new management of change process as a proposed change to the baseline system in one of two ways:

• a way to achieve regulatory compliance if the HRA indicates that the existing system does not comply with health risk standards because of anticipated levels of pollutants in emissions
• a safety improvement of the existing baseline system configuration (i.e., by reducing risk estimated by the QRA) if the existing baseline system HRA results satisfy established health standards

The first step will be to review the site-specific HRA for the baseline system established configuration to confirm that it meets the health risk standards established for the facility. If it does, the installation of a PFS is not warranted on the basis of health risk. If the results of the site-specific HRA indicate that the facility would still be in regulatory compliance with the PFS, a sensitivity analysis of the HRA-derived risk estimates will be performed. This analysis is expected to confirm the conservatism in the current HRA methodology, thereby providing additional assurance that reductions in the calculated health risk estimates, which are already below established regulatory thresholds, will have no practical benefits. Potential benefits of the PFS will then be examined from a broader perspective, such as the impact on the QRA, cost effectiveness, or public acceptance. This proposed methodology appears to be appropriate for evaluating whether the PFS should be implemented on a site-specific basis and is consistent with the Army's new management of change process as described in the Guide.

Carbon filters are not usually used for treating stack emissions from hazardous waste incinerators. The optimal design for a chemical demilitarization facility must be determined from analyses of the exhaust gases, carbon bed filter performance, and anticipated plant upsets. Designs must include data for both the benefits (filter performance) and the risks (plant upsets and a delay in the destruction of the chemical weapons stockpile).

The performance of granulated activated carbon under the conditions of interest, initially determined from the published literature, had to be augmented by carefully planned and controlled laboratory experiments to ensure the correct modeling and accurate predictions of results. Laboratory tests have been completed, and a simulation model of the system has been developed (U.S. Army, 1997d). Members of the Stockpile Committee have been actively following the development of the carbon filter simulation model for the past two years. The model indicates that the carbon filters will effectively remove the hazardous components of principal interest (i.e., dioxins/furans) to levels below the detection limits (if they are present at all) for at least one year, prior to bed replacement. Detection limits are the minimum levels assumed in a normal HRA analysis.

Dioxin/furan emissions at JACADS were extremely small (NRC, 1994a). The use of natural gas at the TOCDF for supplemental fuel (in addition to the agent being burned), rather than the JP-5 fuel used at JACADS, should further reduce emissions of dioxins/ furans.

In addition to the filter model, the Army has developed a conceptual design of a generic PFS. This design involves the use of a gas conditioning system to reduce the water content of the gas and adjust the exit temperature and relative humidity so that trace impurities will ad-sorb on the bed without being displaced by water vapor. A subsequent vertical stack of fixed horizontal carbon beds is similar to the carbon filters in the existing building ventilation system, except that the ventilation system beds are stacked vertically and require a stronger housing because the gas pressure is higher. Additional safeguards are provided for the PFS to reduce the probability of an equipment failure, which could result in hot gases reaching the filters causing the desorption of contaminants from the filters.

The installation of a PFS introduces the following possible sources of risk:

• sudden releases of accumulated hazardous contaminants (e.g., dioxins/furans) to the atmosphere at higher concentrations than are generated during normal operations with no filter
• additional plant downtime because of gas handling equipment failures and the resulting extension of the stockpile storage risk (delays in processing automatically result in increased risk from storage)
• worker exposure to accumulated hazardous contaminants during replacement and disposal of the carbon filters

If there were a sudden release of agent that had accumulated on the filter, the concentrations could be above the lower detection limit. The HRA for the baseline system without a PFS requires the assumption of agent in the stack gases at the lower detection limit, so the accumulation of agent on the PFS during normal operations must also be accounted for. Thus, potentially, agent would be available for a sudden release. This possibility must be taken into account to ensure consistency between the QRA and HRA.

In summary, the committee supports the PFS evaluation methodology and finds that the carbon filters will effectively remove dioxins/furans to levels below the detection limits, with a useful bed life of at least one year. However, more information on the effects of the PFS on QRA risks, on costs and schedules, and on public concerns must be evaluated before the Army can make a final decision.