Executive Summary
The Committee on Review and Evaluation of the Army Chemical Stockpile Disposal Program (Stockpile Committee) of the National Research Council has endorsed incineration (with comprehensive air pollution control systems) as a safe and effective procedure for destroying chemical agents and munitions. Recognizing, however, that some public opposition to incineration (based primarily on substances of potential concern [SOPCs] that could escape into the atmosphere with the combustion gas) has always existed, the committee also recommended that the Army study the addition of a carbon filtration system to improve the existing pollution abatement system. This recommendation reflected the committee's belief that (1) reductions in emissions resulting from carbon filtration systems, however small, could increase public confidence, and (2) a carbon filter would virtually eliminate the possibility of an accidental release of a chemical agent through the stack.
When the first recommendations were made in 1991 and 1992, carbon filters were being introduced in Europe. Since then, the Army has evaluated the European experience and decided to add carbon filters to the baseline incineration systems for the disposal of chemical weapons stockpiles at Anniston, Alabama; Umatilla, Oregon; and Pine Bluff, Arkansas. Carbon filters are called for in the Resource Conservation and Recovery Act (RCRA) permits for the Anniston, Umatilla, and Pine Bluff sites, where construction of the disposal facilities is already under way.
Since these decisions were made, data from trial burns conducted at the operating Tooele Chemical Agent Disposal Facility (TOCDF) near Tooele, Utah, have become available. Although this facility does not have a carbon filtration system, the data show very low emitted concentrations of SOPCs, including dioxins and metals. The concentrations measured at the TOCDF were either the lowest or among the lowest emitted concentrations in the Environmental Protection Agency's (EPA's) Hazardous Waste Combustor Emissions Database. Chemical agent, if present at all, was below the detection limit, which is also below the levels generally believed to have deleterious environmental or health effects. Nevertheless, an Army study modeling the performance of carbon filters concluded that they would reduce many SOPCs to even lower levels. The committee concurs with this judgment.
The carbon filter system, including associated gas conditioning equipment designs, had not been finalized at the time this report was prepared. Suggested design alternatives were available, however, and the committee concluded that an effective pollution abatement system carbon filter system (PFS) design could be implemented.
The Utah Department of Environmental Quality's Division of Solid and Hazardous Waste, which conducted the health risk assessment (HRA) for the Tooele facility, determined that the health risk to the public posed by the incinerator stack gas emissions was below the level of regulatory concern. HRAs have also been conducted by Army contractors for the Anniston and Umatilla facilities in which the effects of adding carbon filters to the baseline incineration system pollution abatement systems were considered, but only in terms of changes in the exhaust gas flow rate and temperature, not reduction in emissions of SOPCs. These studies did not quantitatively evaluate the potential benefits of the PFS, but even without carbon filtration systems, emissions are expected to be below the levels of regulatory concern.
Based on quantitative risk assessments (QRAs) (estimates of the probability and consequences of accident scenarios that could lead to a release of agent) completed at Tooele and under way at Anniston and Umatilla, the increased risk to the public from an accidental release of agent associated with carbon filters was found to be negligible (i.e., orders of magnitude below the risks people face every day). This was not so for worker risk. In the Anniston QRA analysis carried out using the Phase 2 QRA from the TOCDF, modified for the presence of a PFS, the only type of upset condition that would increase the risk of agent release was blockage of the exhaust gas flow by the PFS coupled with loss of the induced draft (which maintains the pressure drop for the exhaust gas flow). The risk of an explosion of agent vapor caused by blockage of the PFS represents 3 percent of the total worker risk. Individual worker fatality risk from agent over the facility life attributable to upsets in the pollution abatement system are estimated at 3.3 × 10-5 with the PFS and 1.1 × 10-5 without the PFS. This is in contrast to total worker risk from agent over the facility life of 4.1 × 10-4 as estimated for TOCDF. These findings also can be compared with the worker accidental death rates of 3 × 10-5per year for manufacturing and 1.5 × 10-4per year for construction industries during 1996. The increased risk at the TOCDF is within the range of the uncertainty of worker risk analysis at the facility but significant enough to warrant further evaluation.
The QRAs assess the risk of accidental releases of chemical agent, but they do not address ''normal'' industrial risk to workers. Hazards to workers from operating and maintaining an industrial facility (hazards not related to agent) will be evaluated during design and prior to commissioning, as part of the health, safety, and environmental evaluations for baseline facilities. If carbon filters are used, they will be included in these evaluations and the risk management and safety programs of each facility. Two risks that are frequently mentioned in this connection are risks associated with potential fires and risks during disposal of the carbon. PFS design and monitoring plans substantially mitigate the risk of potential carbon fires. The amount of potentially contaminated carbon from the PFS that will require disposal is small in comparison to the amount of agent-contaminated carbon that will require disposal from the treatment of the ventilation air for the facility.
The QRAs for three sites (Tooele, Anniston, and Umatilla) to date all confirm the committee's previous observations: (1) the major hazard to the public is from the stored agent and munitions in the stockpile itself; and (2) the risk introduced by stockpile disposal processing is relatively small (less than 1 percent of the stockpile storage risk). Major changes in a RCRA permit may engender a considerable delay that would increase the overall risk to the public. However, the magnitude of the increased storage risk depends on the length of the delay (which is uncertain). The increased risk from prolonged stockpile storage has been estimated on a per year of storage basis. For the population 2 to 5 km from the Anniston Chemical Agent Disposal Facility, the individual public fatality risk is 1.4 × 10-5 per year, and the societal public fatality risk is 2.6 × 10-2 per year. This risk is in contrast to the disposal processing risks for the same population of 3.8 × 10-8 per year (individual public fatality risk) and 1.8 × 10-5 (societal public fatality risk). Thus, the per year risk from storage is at least three orders of magnitude higher than the risk from disposal processing. Hence, very short delays would increase public risks more than the total public risk from disposal. A delay of approximately one year would result in increased individual public risks of the same order of magnitude as the estimated increase attributable to the PFS in individual worker fatality risk over the entire period of disposal processing. Consequently, public risk will be minimized by the expeditious safe destruction of the stockpile.
Conceptually, the committee agrees with the Army's decision to proceed with the current designs at Anniston and Umatilla and not to alter the operating configurations of JACADS and the TOCDF. Removing or adding carbon filters at this point is likely to cause delays that will increase the risk to workers and the public. However, potential increases in worker risk from the carbon filters, which were initially estimated to be small, require further evaluation. To mitigate the potential adverse consequences of adding carbon filters at Anniston and Umatilla, worker risk should be evaluated quickly and managed effectively, including changing the PFS design, if necessary.
The Army's initial attempts at public outreach using its change management process (CMP) in PFS decision making did not elicit meaningful public involvement or comment during the decision process, and several shortcomings of the CMP have now become apparent. First, public involvement must be initiated much earlier in the process of evaluating change. For example, public involvement could have helped the
Army formulate the questions to be answered during the PFS risk evaluation. Second, public involvement should allow for public input prior to making decisions on major process changes, even if initial assessments indicate that no change is preferred. Third, for the CMP public involvement process to be credible and engender public trust, the Army must provide clear guidelines for initiating the CMP, which should not be circumvented by executive decision.
The Army's decisions not to change the configurations at Tooele, Anniston, and Umatilla were made in the context that the original intent of the PFS was to reduce risk and increase public confidence. These goals were to be achieved by adding another air pollution control system component to polish the effluent and curb whatever pollutants would have been emitted without the PFS. However, the results of the Army's analysis showed that changes to risk would be small, that these changes could be improvements or degradations depending on the population considered and the uncertainty analysis, and that the risks could be different for the public and workers. In addition, the Army's presentation of the risk evaluations was difficult to understand and was not issued in a self-contained document delineating (1) comparisons of each risk component with and without the PFS and (2) the Army's rationale for making no changes to the current site configurations. These crucial lapses all but precluded the public from following the process or influencing the results.
FINDINGS AND RECOMMENDATIONS
The estimated concentrations and emission rates of SOPCs from chemical agent incinerator operations developed during the permitting processes for the Anniston Chemical Agent Disposal Facility and the Umatilla Chemical Agent Disposal Facility were below the thresholds of regulatory concern, whether or not a passive carbon filtration system (like the PFS) was included in the facility design. Therefore, the committee considers PFS to be risk neutral to off-site populations.
The addition of a PFS to the PAS would probably reduce the already low emissions of some SOPCs during normal, transient, and upset operating conditions. However, a PFS would also increase worker risk by making the facility more complex and by introducing new scenarios for potential facility upsets and failures. The extent of the increase in worker risk is not clear because all of the applicable risk evaluations (e.g., Phase 2 QRAs and health, safety, and environmental evaluations) and resulting risk mitigation measures have not yet been completed. Preliminary assessments, however, indicate that the increase in worker risk would be small.
Significant changes in permitted facility designs require permit modifications, which could cause substantial delays. Because risk analyses consistently indicate that the storage risk to the public and workers is much greater than the processing risk, changing the permitted configuration at any stockpile site is likely to increase the overall risk by delaying destruction of the stockpile.
Finding 1a. The reported emitted concentrations of SOPCs measured during trial bums at the JACADS and TOCDF incinerators are among the lowest reported to the EPA. TOCDF emissions are the lowest, or at least one of the lowest, in dioxins, mercury, cadmium, lead, arsenic, beryllium, and chromium. The reported emissions of some SOPCs were based on the analytical detection limit for the constituent, which means the actual concentration could be much lower than the reported concentration. Maximum emitted concentrations from JACADS were used for the HRAs for other baseline facilities to ensure that estimates of risks would be conservative.
Finding 1b. In 1992 and 1994, the NRC recommended that the Army investigate using carbon filters for two purposes: (1) to contain transient stack emissions or accidental releases of agent and (2) to increase public confidence in incineration. Activated carbon filters in use at several large incinerators in Europe meet very stringent regulations on emissions of chlorinated dioxins/furans and are considered to be the state-of-the-art technology for this purpose. Based on preliminary design evaluations, activated carbon in the PFS of the Army's baseline incineration system is likely to have sufficient adsorption capacity to reduce emitted concentrations of dioxins, furans, HD, VX, and GB for more than a year of normal operations before the activated carbon would have to be replaced. The activated carbon would also have the capacity to adsorb a chemical agent in case of a major upset; however, a major upset would necessitate the immediate replacement of the activated carbon.
The addition of carbon filters to a baseline incineration PAS does not appear to reduce the health risk to
the surrounding population substantially because the health risk is already small (see Finding 1a). Nevertheless, reinforcing public and worker confidence is an important goal.
Recommendation 1. The Army should only consider removing the carbon filtration system from the permitted designs of the Anniston, Umatilla, or Pine Bluff facilities if, after a thorough implementation of the change management process to ensure meaningful public involvement, the public supports that decision.
Finding 2. Based on the evaluation of preliminary PFS design alternatives, an effective design for the PFS is feasible. Operating facilities in several countries now have significant experience in the design and operation of activated carbon filters.
Recommendation 2. The Army should take advantage of the experience of other users of carbon filters through appropriate consultation.
Finding 3. The Army has evaluated the implications of adding or removing passive carbon filter systems to the baseline incineration systems at the Tooele, Anniston, and Umatilla disposal facilities. Some of the impacts on risk to public health from stack emissions were evaluated by comparing the HRAs for the existing baseline facilities to estimates of the upper bound of public health risk posed by the addition of the PFS. However, the potential reductions in public health risk were not estimated, and the evaluations of impacts to off-site populations were incomplete.
An estimate of the impact on risk of accidents leading to agent-related public fatalities was made by expanding the Anniston and Umatilla Phase 1 QRAs to consider the addition of the PFS. The impact of the PFS on worker risk, which is not evaluated in the Anniston and Umatilla Phase 1 QRAs, was estimated by extrapolating the Tooele Phase 2 QRA results (which does include worker risk) to these other facilities. The Phase 1 QRAs for the Anniston and Umatilla facilities were also used to estimate increases in risk to the public from extended storage of the stockpile due to the PFS. Thus, the QRA evaluations completed to date are initial estimates of the magnitude of increased risk to the public from accidental releases of agent resulting from the addition of the PFS, but they are not complete evaluations of worker risk. Moreover, the range of potential delays to stockpile destruction caused by permit modifications and physical changes to the current site-specific baseline incineration configurations has not been defined.
Based on these estimates, the Army concluded that "[the] current plan to install and operate the PFS at the ANCDF [Anniston] and the UMCDF [Umatilla] remains the best course of action for maximizing human health and environmental protection," and that the TOCDF should continue to operate without a PFS. The decision to continue with the current configurations at permitted facilities eliminates increases in risks to the public and workers from potential delays in stockpile destruction caused by facility modifications or permit changes. Although worker risk from current PFS configurations is uncertain, based on the available risk estimates and projected schedules, the committee concurs with the Army's conclusion.
Recommendation 3. To minimize increased risks to off-site populations and on-site workers from delays in stockpile destruction, the Army should proceed with the current configurations, which include carbon filtration systems at Anniston and Umatilla, and should continue operations at Tooele, which does not have a carbon filtration system.
Finding 4. Only the Phase 1 Anniston and Umatilla QRAs have been completed. The risk of acute hazards to workers, probably the receptors at greatest risk from a mishap involving the PFS, has not been adequately characterized. Early initiation of the Phase 2 QRAs could identify these risks while facility design and construction are in progress and give the Army greater flexibility to modify facility designs and operating procedures, if necessary.
Recommendation 4a. The site-specific Phase 2 QRAs for Anniston, Umatilla, and Pine Bluff, which would identify and analyze specific failure modes, should include a complete evaluation of worker risk associated with the addition of the pollution abatement system filter system. The Phase 2 QRAs for each site should be initiated as soon as possible and should be completed and reviewed by independent technical experts before systemization of the facilities at Anniston, Umatilla, and Pine Bluff is completed.
Recommendation 4b. A risk management plan should be developed to minimize worker risk during the operation and maintenance of the pollution abatement system
filter systems. The evaluation of operating and maintenance risks should include the operational experience of similar systems. If the increased risk to on-site workers is found to be substantial, the Army should consider making modifications, as long as they do not substantially increase overall worker or public risk from prolonged storage.
Finding 5. If increased worker risks and hazards are identified, it is not clear what steps the Army would take to mitigate them. Nor does the Army have a clear decision basis for balancing reductions in public risk and increases in worker risk.
Recommendation 5. The Army should clarify to the public and facility workers the risk management actions that would be taken if increased worker risks are identified. The Army should also clarify the decision basis for balancing reductions in public risk against increases in worker risk while fulfilling its mandate to protect both workers and the public.
Finding 6. The PFS was assumed to have no effect on concentrations of SOPCs in the HRA calculations for Anniston and Umatilla. The effects of SOPCs emitted from the stacks at these facilities have been estimated to be below the thresholds of regulatory concern without the benefit of the PFS. However, changes from installing a PFS have not been determined in a way that facilitates quantitative comparisons.
Recommendation 6. Future health risk assessments should include estimates of emitted and ambient concentrations of SOPCs, with and without the PFS, for all substances that contribute significantly to the overall risk. Because PFS performance cannot be based on actual measurements, the analysis should consider the implications of reducing emissions to both the method detection limit and the levels indicated by engineering calculations, including quantitative evaluations of the uncertainties associated with each risk estimate. The results, including the acute and latent risks, should be reviewed by independent technical experts. The results should then be presented in a way that facilitates public input to decision making.
Finding 7. Because of the length of time required to complete the preliminary PFS risk assessment, the fact that this evaluation is still incomplete, and the status of construction activities at Anniston and Umatilla, meaningful public involvement in the decision to include the PFS at these sites is no longer possible. The CMP Plan and the CMP Public Involvement Outreach Plan were not effectively implemented during the Army's analysis of the PFS. The lack of public involvement in this process represents a lost opportunity for the Army to develop its CMP and to implement the CMP public outreach process.
Recommendation 7a. The health risk assessment and quantitative risk assessment for Pine Bluff should be completed as quickly as possible and communicated to the public in a timely manner so that there can be meaningful public involvement in the decision process to retain or remove the carbon filter system. The risk assessments should be subject to independent expert review and the findings incorporated into the decision-making process.
Recommendation 7b. The Army should continue to refine its change management process and the change management process public involvement plan. Public involvement should be an integral part of future evaluations of the pollution abatement system filter system, especially at Pine Bluff. The committee repeats its recommendation that the Army involve the public meaningfully in the Chemical Stockpile Disposal Program as a whole.
