2
Deseret Chemical Depot/Tooele Chemical Agent Disposal Facility Site-Specific Risk Assessments

Overview

In this chapter, a brief overview of the sources of risk at DCD/TOCDF and of the risk assessments performed by Science Applications International Corporation (SAIC) and the state of Utah is presented followed by descriptions of the objectives and scope, approach and methodology, oversight, results and analysis, and integration of results of the assessments. SAIC performed a QRA that examined the risk from agent accidents (U.S. Army, 1996c). The state of Utah performed an HRA that examined the maximum risk from normal and upset operations (Utah DSHW, 1996). (See Appendix A for background information on risk assessment). The SAIC methodology is carefully implemented and consistent with previous recommendations of the committee (NRC, 1993b; NRC, 1994b) except that SAIC was not asked by the Army to present an integrated assessment of the QRA and HRA results. Therefore, under each topic, the QRA and the HRA are discussed sequentially. The risks from both the QRA and HRA are discussed at the end of this chapter. Figure 2-1 illustrates some of the elements of risk discussed in this chapter.

Deseret Chemical Depot Stockpile

As of March 1997, the chemical weapons storage facilities at DCD contained nearly 45 percent of the remaining U.S. chemical weapons stockpile, with more than 13,000 tons of agent. The DCD stockpile contains nerve agents GB, VX, and small quantifies of GA, as well as all three types of mustard (blister) agents. In addition, every type of U.S. chemical weapon containment system (mines, rockets, projectiles, bombs, ton containers, and spray tanks) is present in the DCD stockpile, with more than one million individual items in the inventory. The DCD stockpile has the largest quantity of chemical agents and the most complex combination of agent/containment systems.

Sources of Risk

For a chemical agent and munitions storage and destruction site like DCD/TOCDF, there are two primary sources of risk: (1) risk associated with the stockpile itself (stockpile risk) and (2) risk associated with destruction of the stockpile (operational risk). The actual risk from either or both sources depends upon whether risk-initiating events occur. Such events can be either internal or external in nature. Internal risk-initiating events are events associated with the storage and routine maintenance of the stockpile and with the operation of the destruction facility. External risk-initiating events are events not associated with site operations, such as earthquakes, floods, lightning strikes, and airplane crashes. (Note that there are also external risks to the stockpile from war or sabotage, which are reportedly evaluated and managed by specific government agencies and are not considered in publicly available site-specific risk assessments. The Stockpile Committee has not been involved in or reviewed any of these evaluations.)

Stockpile Risk at DCD

The principal hazards associated with the stockpile at DCD are from the inherent toxicity of the anti-



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 14
2 Deseret Chemical Depot/Tooele Chemical Agent Disposal Facility Site-Specific Risk Assessments Overview In this chapter, a brief overview of the sources of risk at DCD/TOCDF and of the risk assessments performed by Science Applications International Corporation (SAIC) and the state of Utah is presented followed by descriptions of the objectives and scope, approach and methodology, oversight, results and analysis, and integration of results of the assessments. SAIC performed a QRA that examined the risk from agent accidents (U.S. Army, 1996c). The state of Utah performed an HRA that examined the maximum risk from normal and upset operations (Utah DSHW, 1996). (See Appendix A for background information on risk assessment). The SAIC methodology is carefully implemented and consistent with previous recommendations of the committee (NRC, 1993b; NRC, 1994b) except that SAIC was not asked by the Army to present an integrated assessment of the QRA and HRA results. Therefore, under each topic, the QRA and the HRA are discussed sequentially. The risks from both the QRA and HRA are discussed at the end of this chapter. Figure 2-1 illustrates some of the elements of risk discussed in this chapter. Deseret Chemical Depot Stockpile As of March 1997, the chemical weapons storage facilities at DCD contained nearly 45 percent of the remaining U.S. chemical weapons stockpile, with more than 13,000 tons of agent. The DCD stockpile contains nerve agents GB, VX, and small quantifies of GA, as well as all three types of mustard (blister) agents. In addition, every type of U.S. chemical weapon containment system (mines, rockets, projectiles, bombs, ton containers, and spray tanks) is present in the DCD stockpile, with more than one million individual items in the inventory. The DCD stockpile has the largest quantity of chemical agents and the most complex combination of agent/containment systems. Sources of Risk For a chemical agent and munitions storage and destruction site like DCD/TOCDF, there are two primary sources of risk: (1) risk associated with the stockpile itself (stockpile risk) and (2) risk associated with destruction of the stockpile (operational risk). The actual risk from either or both sources depends upon whether risk-initiating events occur. Such events can be either internal or external in nature. Internal risk-initiating events are events associated with the storage and routine maintenance of the stockpile and with the operation of the destruction facility. External risk-initiating events are events not associated with site operations, such as earthquakes, floods, lightning strikes, and airplane crashes. (Note that there are also external risks to the stockpile from war or sabotage, which are reportedly evaluated and managed by specific government agencies and are not considered in publicly available site-specific risk assessments. The Stockpile Committee has not been involved in or reviewed any of these evaluations.) Stockpile Risk at DCD The principal hazards associated with the stockpile at DCD are from the inherent toxicity of the anti-

OCR for page 14
Figure 2-1 Schematic illustration of risk elements at the TOCDF. cholinesterase agents, GB and VX, and mustard agents, H, HD, and HT. Risks associated with the stockpile are almost all related to agent releases from either internal or external events. Agent GB presents the greatest hazard because of its toxicity and volatility; GB also represents the greatest potential risk because it constitutes about half of the total amount of agent on site and is contained in more than 75 percent of the inventory items at DCD. Agent releases initiated by internal events could result from handling accidents during stockpile manipulation and maintenance; the deterioration of containment systems; the spontaneous detonation of munitions; or the spontaneous ignition of propellant. External events that could cause releases include earthquakes, floods, lightning strikes, and airplane crashes. Operational Risk at the TOCDF Agent destruction imposes risks above and beyond the inherent risks associated with the existence and maintenance of the chemical agent and munitions stockpile. The transportation of agents from storage to the destruction facility, the unpacking and disassembly of munitions and containment systems, and the actual agent destruction processes provide additional opportunities for agent releases caused by internal or external events. Like the stockpile risk, the predominant operational risks are associated with agent toxicity. However, the quantities of agent being processed at any given time are small compared to the original inventories in the stockpile. The maximum quantity of agent present in the disposal facility at any given time would be the equivalent of about three ton containers (i.e., approximately 5,000 pounds of agent).

OCR for page 14
In addition to agent, other risks from the agent destruction process must be considered, such as products of incomplete combustion from agent destruction and toxic materials used in the disposal process. However, because the quantity of these toxic products is substantially smaller than the original quantity of agent, they generally represent a smaller hazard. Risks from toxic products are primarily initiated by internal events, such as process upsets. External risks are virtually nonexistent because they would usually result in a shutdown of the process. However, extreme external events could cause the release of hazardous materials, such as propane or sodium hydroxide, from on-site storage tanks. Risk Receptors There are three potential risk receptors: workers, the public, and the environment. Because of their proximity to the stockpile and agent processing operations, workers are at risk from the acute lethal (and nonlethal) hazards associated with agent releases, regardless of the initiating event (an at risk situation). They are also potentially at risk from long-term exposure to very low concentrations (i.e., below the eight-hour time weighted average) of agent and the products and by-products of agent destruction. Workers are also susceptible to injury from ordinary industrial accidents (e.g., falls, burns, eye injuries, overheating in protective clothing), but these risks are not included in the risk assessments performed for the TOCDF. They can be better understood through safety inspections and analyses of injury rates and can be managed by following proven safety practices. Risks to the public stem primarily from agent releases caused by external (catastrophic) events. The public could also be at risk from long-term exposure to the products and by-products of agent destruction, if they were released into the environment as a result of destruction processes. Environmental risk is associated almost exclusively with the release of agent and the products and by-products of agent destruction to the environment. Risk Measures For humans (both workers and the public) there are three potential measures of risk either from the stockpile or from stockpile destruction: acute lethality; acute and latent noncancerous health effects; and latent cancer. The potential adverse consequences for the environment are the contamination of land and/or water and adverse effects on native or endangered species. Risk Mitigation The most effective mitigation of risk takes place before a hazardous material is released and is often called prevention rather than mitigation. However, after a hazardous material has been released, but before it reaches a receptor, risk mitigation is also possible, i.e., the consequences of the release can be reduced. Risk mitigation can include taking measures at the spill site (e.g., containing the spill), measures at the receptor site (e.g., using protective masks), and emergency response measures (e.g., shelters, evacuation, etc.). The QRA takes into account some of these measures. However, the primary intent of the QRA is to calculate a realistic estimate of risk to the public. The analysis is not structured to measure the effectiveness of the local Chemical Stockpile Emergency Preparedness Program (CSEPP). The QRA uses simple models and average data for mitigation (e.g., radial evacuation, evacuation time estimates for broadly defined conditions [time of day and weather, for example], using a representative evacuation speed of 8 m/sec (18 mph) and an assumption that 95 percent of the populace will take protective action). Representatives of the QRA team, the Army, and the CSEPP concluded that the QRA is the best estimate of risk. However, it may be more pessimistic than CSEPP calculations for some scenarios because the CSEPP uses more sophisticated models. Objectives and Scope of the DCD/TOCDF Risk Assessments Two separate risk assessments were performed for DCD/TOCDF. The first, a QRA, evaluated internal and external event-initiated risks to workers and the public. The second, an HRA, evaluated human health (public) and environmental risks associated with normal operation of the destruction facility. The committee's Systemization report (NRC, 1996b) observed that "the multiplicity of assessments can cause misunderstanding among reviewers, government agencies, and the

OCR for page 14
public. The Army should adopt a standard language that recognizes the ensemble of risk-related projects as 'the risk assessment,' and individual studies should always be referred to as components of the wider 'risk assessment.'" The report advocated the preparation of a single TOCDF risk assessment summary report to present integrated QRA/HRA results. The current report does not dwell on the risk assessment methodologies but deals explicitly with the DCD/TOCDF risk assessment and risk management processes. A more detailed description of the QRA methodology can be found in the Systemization report (NRC, 1996b). A very brief overview can be found in the following sections of this report. The HRA was performed following methodology described in public documents (EPA, 1994). Quantitative Risk Assessment The DCD/TOCDF QRA (U.S. Army, 1996c) had several objectives: to evaluate quantitatively the health risks from accidental releases of chemical agents to the public and to workers at the site to rank the plant and operational features at the facility that govern risk (Insights are to be used as a basis for risk management at the facility.) to compare the risks associated with the disposal process with the risks of continued storage (Insights are useful to the Army in making decisions regarding stockpile disposal, especially the specific order of items scheduled for disposal.) to provide a "living model" QRA that can be updated as changes are made to the facility or as additional insights into accident behavior become available (The living model will be one of the analytical tools supporting decision making within the TOCDF risk management program throughout the life cycle of the plant.) The TOCDF QRA estimates the risk to the public and to workers from accidental releases of chemical agent associated with all activities during storage at DCD and throughout the disposal process at the TOCDF. Activities associated with the disposal process include: munitions storage at DCD prior to disposal munitions handling at DCD in preparation for transport to the disposal facility transport of munitions to the disposal facility the disposal processes The study includes all identified potential causes of release, except for intentional acts, such as sabotage. Releases resulting from both internal initiating events (events that originate inside the facility or that directly result from activities during the disposal process) and external events (such as earthquakes, aircraft crashes, and tornadoes) are included. Results of the TOCDF QRA are presented in terms of both public and worker risks. For public risks, both the risk of acute fatalities and the risk of exposure-induced cancer from accidents (mustard agents are potential carcinogens) are estimated. The risk of fatalities is presented in three ways: a risk profile showing the probability of exceeding a given number of deaths during the disposal period risk profiles as a function of distance from the site an average measure, e.g., the expected number of deaths during the disposal period Appendix A of this report develops the bases for the presentations of risk (risk profiles and expected fatalities), explains how to interpret results, and discusses various measures for comparing risks. In estimating worker risk, the TOCDF QRA addresses only acute fatalities from accidents involving agent release caused by processing. Latent risks to workers from exposure were calculated but are not included in the results for several reasons: Workers directly involved in an accident are assumed to be killed, either from agent or from an explosion. Reporting the health effects for workers who are not directly involved, but who work in adjacent areas, would be deceptive for several reasons: - The model may not properly capture the close-in dose. - A convincing argument can also be made that projected latent effects from everyday activities (e.g., maintenance) are much greater than the latent effects from an agent accident. No

OCR for page 14
study of routine exposures has been done because no problem is apparent. - The calculated latent risk to workers is very, very small compared to the acute risk. The committee agrees that latent risks to workers from exposure to accidents is very small and including them in the QRA is not warranted. Worker risk from continued storage was not assessed in the QRA because processing has already begun and activities related to disposal are of most interest to the Army. Worker risk from continued storage would require assessing limited worker populations and restricted activity schedules that no longer exist at DCD. Worker risks associated with industrial-type accidents also were not included in the QRA. Uncertainty analyses showing the possible range of results, which were presented only for the public risks, incorporate the types of uncertainty discussed in Appendix A. All other risks were expressed as expected risk levels. The upper uncertainty bound shown for the QRA estimates is a measure of the analysts' confidence in the results. There is a 95 percent chance that the risk is less than the upper bound. Health Risk Assessment To complement the QRA and to meet Resource Conservation and Recovery Act (RCRA) permitting requirements, a screening-level HRA to estimate possible human health risks associated with exposure to airborne emissions from the TOCDF has been completed by the Utah Division of Solid and Hazardous Waste (DSHW) (Utah DSHW, 1996). The HRA also evaluated risks to wildlife and the environment. The scope of the HRA was limited to anticipated normal operating conditions with a fairly large allowance for emissions associated with operational process upsets. The HRA was a screening estimate in the sense that the results represent extreme upper bounds for normal and upset releases, well beyond the 95 percent upper bound described in the QRA. Approach and Methodology Quantitative Risk Assessment The Army completed and published the final report of the TOCDF QRA in 1996 (U.S. Army, 1996c). The TOCDF QRA, which was conducted following guidelines recommended by the NRC (NRC 1993b; NRC 1994b), quantitatively analyzes the probability and consequences of accidental releases of agent at the TOCDF facilities and the DCD storage area. QRA Team The DCD/TOCDF QRA was performed for the Army by SAIC. The major part of the QRA was performed by the SAIC analysts themselves (SAIC has strong in-house technical capability and extensive experience conducting QRAs for large-scale engineered systems). In areas where special expertise was required, external subcontractors or independent consultants were used. These areas included seismic hazards, structural mechanics, munitions fragility, and the latent health effects of agent. Operation of the QRA team was independent of the TOCDF site staff. However, the QRA team frequently interacted with the TOCDF staff to ensure the validity and completeness of the analysis. Approach The DCD/TOCDF QRA used the state-of-the-art approach to probabilistic-based risk assessment methodology that was first introduced to the nuclear industry in the 1970s in the WASH-1400 report (U.S. NRC, 1975). Since then, QRA methodology has gradually evolved into a sophisticated decision-support tool and is now well accepted and widely used to analyze complex engineering systems in the nuclear and chemical process industries (U.S. NRC, 1990; CCPS, 1989). Other approaches to risk assessment have been evaluated by the NRC and are more commonly used to assess health risks where assessing exposure depends heavily on dose-response characteristics (NRC, 1983, 1994c). Because of the complexity of DCD and TOCDF operations, and because conservative assumptions were made about agent lethality, the committee recommended the U.S. Nuclear Regulatory Commission method be used for the QRA (NRC, 1993b). The QRA is based on a comprehensive set of logic models developed from the engineering design and operation of the disposal system and from various scenarios of potential system accidents. The risk at the site is then represented by the likelihood of these accident

OCR for page 14
Typical Examples Initiators Progression Release Consequences Composite Risk Earthquake of 0.35-0.4g (recurrence interval 7,300 years) 5 pallets of GB M55 rockets topple, 12 igloos set on fire 20 tons of GB See note 1.3 expected fatalities GB ton container leak (recurrence interval 21 years) localized spill several pounds of GB .003 expected fatalities Earthquake of .25g-35g (recurrence interval 2.000 years) CHB/UPA falls and fails; 5 GB ton containers rupture @ 4 tons of GB .15 expected fatalities Operational emissions over 7.1 years (HRA) atmospheric dispersion trace amounts less than 1 in 100.000 chance of excess cancer to "maximally exposed individual" over 70-year lifetime Note: A simple tabular display for the consequences column is not effective. For a given initiator and one progression sequence from that initiator leading to a particular release, many possible consequence sequences can evolve, each with its own conditional probability (conditional on the previous events). Factors affecting the range of consequences Include wind speed and direction, other weather conditions, and emergency response. The weighted average of the consequences is used to calculate the composite risk shown in the last column. Figure 2-2 Overview of QRA process. Source: Adapted from U.S. Army, 1996d. Note: CHB/UPA means container handling building/unpack area. scenarios and the severity of their consequences. The DCD/TOCDF QRA process is shown in Figure 2-2 and is summarized in the following paragraphs. Figure 2-2 also contains a table showing how a few elements of the QRA and HRA process are analyzed. Identifying and Modeling Risk Initiators The QRA starts with a systematic identification of deviations from normal process operations. These deviations are called "initiators." As suggested by the 1993 NRC letter report (NRC, 1993b), initiators considered in the DCD/TOCDF QRA include "internal" initiators, such as equipment failures and human errors, and "external'' initiators, such as earthquakes, plant fires, floods, tornadoes, and aircraft crashes. Fault tree analyses, a well-accepted QRA logic modeling technique (Roberts et al., 1981), were used to identify the causes of internal initiators, a combination of equipment failures and human errors, for example.

OCR for page 14
Figure 2-3 Rocket handling system fault trees for agent spilled during shear operation. Source: U.S. Army, 1996c. To illustrate the method, the top logic from the DCD/TOCDF QRA fault tree for "Agent Spill during Shear Operation" is displayed in Figure 2-3, which has been extracted from an appendix of the QRA (U.S. Army, 1996c). This is one of many fault trees used to analyze operations at the facility. The rocket shear operation takes place inside a contained operating area (an enclosed room with a sealed environment to contain agent spills) and presents a hazard to workers in protective suits who would have to clean up any spills. Here the authors show that an agent spill during shearing operation can occur in one of two ways: The rocket is not drained (but automatic processing continues), OR The rocket is stopped before it is completely drained (and later is sent to the shear process). The symbol with the pointed top and labeled RHSSHSP is called an "OR gate," which means that the event above it happens if either or both of the two events below it occur. The figure shows that the logic for the ways to "Rocket stopped prior to full drain" appears later in the fault tree, where one would learn that the draining is stopped if both of the following events occur: Processing stops before the rocket is fully drained, AND A human error is made (an operator sends an undrained rocket to shear). An example of an "AND gate" in the figure can be found below the event "Rocket not drained; auto mode maintained." The AND gate has a rounded top and indicates that the event above it occurs only if all the events below it occur. In this case, "Rocket not drained, auto mode maintained" occurs if both "Rocket agent cavity is not punched'' AND ''Drain failure fails to stop rocket process" occur. The rest of the AND/OR logic for failure is easy to follow. On the bottom line of this

OCR for page 14
figure is one new symbol, the diamonds below two of the event blocks, which indicate that the event will not be further decomposed. For example, "A misoriented rocket is processed" is an event that will be quantified directly, using available data. Modeling Accident Progression Sequences To specify potential accident sequences following an initiator, the TOCDF QRA uses the accident progression event tree (APET). Based on the engineering and operational information collected from the facility, the analysts identify and model (using the event trees to track different failure pathways) sequences of events following an initiator that could lead to an agent release. Process accident flowchart models, called process operational diagrams (PODs) in the QRA, have been devised to encode process information. Upsets are identified in the POD. A POD is described in the Tooele Chemical Agent Disposal Facility Quantitative Risk Assessment Methodology Manual (U.S. Army, 1994): A POD is a step-by-step search for events and upsets . . . By asking a set of what-if questions after each successive operational step, a thorough assessment of potential upsets can be generated. During this process, existing analyses are referenced to ensure that previously suggested events are covered . . . [Start by] listing the major steps of the normal operations . . . Given each normal step, it is necessary to consider all deviations that could occur during that step or if that step did not happen properly . . . The PODs are used to document the steps in the process and allow efficient review by operational staff. Quantifying Model Parameters Data on equipment failures and human errors are collected from both industrial and CSDP experience and used as a basis for evaluating the likelihood of initiators as well as the likelihood of subsequent events leading to accidents and potential agent releases. The probability of accident sequences resulting in agent releases are then estimated based on the accident sequence model and the basic event data. Determining the Magnitude and Conditions of an Agent Release Following the identification in the APET of accident sequences that could lead to a release, the size of the release is estimated based on the event sequences. The amount of agent released and the conditions associated with the release are modeled for each accident sequence. Estimating Health Consequences to the Public and Workers Health effects to the public and workers are identified as the consequences of the accidental releases and have to be estimated. Mathematical models are used to estimate the dispersion of agent releases for site-specific weather conditions and to evaluate the exposure and resultant consequences to the public and to workers at the site. The Army's air dispersion code, D2PC (Whitacre et al., 1987; IEM, 1993), includes extensive chemical agent-specific data and models. However, it does not include statistical weather sampling, health effects models for agent exposure levels, population distribution modeling, or evacuation and sheltering models. The CHEMMACCS code (Haskin et al., 1995), developed at Sandia National Laboratory for use in QRA consequence analyses, uses the same agent-specific data and dispersion model as D2PC. The underlying Gaussian plume dispersion model is similar to the codes used by the Environmental Protection Agency (EPA). Input includes site-specific data for the TOCDF and the surrounding area. Assembling and Calculating Public and Worker Risk The risk of each accidental release is represented by the probability of the accident event sequence and the consequence of the release. The total risk from the disposal facility and the storage area is represented by combining risks from individual releases. Presentation of Results The DCD/TOCDF QRA provides detailed analyses of risk levels from several perspectives. Risk Profile. A risk profile is a plot of the likelihood of "x or more fatalities" plotted as a function of "x." The uncertainty bounds (at least the 5th and 95th percentile) are usually included for calculated mean and median values. The mean value represents an average

OCR for page 14
of the range of estimates. The median value is the point at which half the estimates are higher and half lower. Expected Fatalities or Expected Number of Induced Cancers. The expected number of fatalities or the expected number of induced cancers are statistical summations over all impact scenarios of the individual expected values (the product of the probability of an individual accident release and its consequence in terms of expected fatalities). This number is most often referred to as the risk value. Dominant Contributors to risk. Dominant contributors are sequences of events that rank high in risk value. Effective site-specific risk management would identify dominant contributors in seeking the most effective ways to reduce risk within the CSDP. Health Risk Assessment The HRA performed by the Utah DSHW followed guidelines and methods that have been established and prescribed by the EPA and used for assessing the acceptability of a broad range of health and environmental risks (EPA, 1994). Screening-level risk assessments are conservative by design in that they are based upon worst-case assumptions when operational data are not available. Because TOCDF was a new facility with no operational history at the time the HRA was prepared, a great many default assumptions were used. Because site-specific values were not available for wind speed profiling exponents, terrain adjustment factors, surface roughness, and scavenging coefficients, default values were used. Site-specific inputs were confined to local geographical, hydrological, meteorological, and agricultural information. Site emissions data were approximated based on data from JACADS operational experience because the TOCDF had not yet begun operations. For the HRA, six point sources of emissions are identified, including the TOCDF incinerators and two other areas where the products of the agent destruction process are handled. A description of the facilities is given in the Recommendations report (NRC, 1994b). The six point sources included in the HRA are: liquid incinerators metal parts furnace deactivation furnace system dunnage incinerator brine reduction area stack heating, ventilation, and air conditioning filter stack The potential impact from each point source was evaluated. Impacts from the TOCDF combined stack (the metal parts furnace, liquid incinerators, and deactivation furnace system in simultaneous operation), from all sources at maximum TOCDF operations, as well as from combined TOCDF and CAMDS operations, were considered. The HRA identified four categories of constituents of potential TOCDF emissions: chemical agents, metals, and volatile and nonvolatile agent decomposition products (i.e., products of incomplete combustion). Sixty individual constituents were identified. Human exposures were considered to occur both directly and indirectly. The inhalation of emissions (direct) as well as the ingestion of contaminated soil and food (indirect) were exposure mechanisms deemed appropriate for purposes of the HRA. Consistent with EPA guidelines for screening-level risk assessments, an adult resident, a child resident, a subsistence fisher, and three different subsistence farmers were identified as likely receptors. The adult and child residents were considered to reside at the off-site point of maximum emissions impact. The subsistence fisher was located in an area where subsistence fishing was thought to be practiced, and the three subsistence farmers were located based upon a survey of farming in the area. The HRA considered potential human health risks based on scenarios of 10, 15, and 30 years of continuous TOCDF/CAMDS operation, although the TOCDF is scheduled to operate for only 7.1 years. Stockpile Committee Oversight As a standing committee of the NRC, the Stockpile Committee reviewed the technical developments that led to the design of JACADS and the first risk assessment in the FPEIS (U.S. Army, 1988). That review and concern about the need to understand the risk at each site led to the committee's letter report on risk assessment (NRC, 1993b), which essentially laid out a

OCR for page 14
specification for site-specific risk assessments. Later the committee's Recommendations report (NRC, 1994b) further defined that specification and reiterated the need for site-specific risk assessments. As the Army's site-specific risk assessment for DCD/TOCDF took shape, the committee's role was defined as oversight of the risk assessment/risk management process and oversight of the Expert Panel review process. Actual detailed technical review of the QRA was the charge of the independent panel of experts in risk assessment and chemical engineering (information on the members of the panel is given in Appendix B). However, the committee took advantage of many opportunities to examine the technical details of the risk assessment work. The committee's Systemization report (NRC, 1996b) reviewed the methodology of the QRA and the efforts of the Expert Panel. The committee found that: The QRA methods met the recommendations of the committee's earlier reports. The SAIC QRA team was being responsive to committee questions and Expert Panel comments; they were developing new analysis tools for first-of-a-kind QRA calculations, retaining outside expert groups in difficult technical areas to advise them in areas where the literature was incomplete, conducting tests and new mechanistic analyses to answer new technical questions, and revising the QRA analyses based on new information. Thus the QRA was being modified and extended as work progressed to respond to advice from the Expert Panel and prior recommendations of the committee. Furthermore, answers to questions sometimes required revised approaches to specific aspects of the QRA analysis. Quantitative Risk Assessment The committee closely followed the risk assessment activities in three ways. The Army and its contractors made presentations on the technical progress of the QRA at all regular quarterly meetings of the committee and at some special meetings. During these sessions, the QRA team responded to detailed questions from the committee. In addition, two members of the committee attended the meetings of the Expert Panel, observing the process and also having the opportunity to question the analysts actually involved in all aspects of the QRA. Finally, all members of the committee received the draft of the Main Report of the QRA to review, and three members received the entire report including the extensive appendices. Questions generated in this process were raised at subsequent Expert Panel meetings. Health Risk Assessment The committee received the protocols under which the HRA was to be performed. It also received a briefing by the state of Utah and its consultants on the HRA results, assumptions, and models during a regular quarterly committee meeting. Three members of the committee received the HRA document. The HRA was performed in accordance with EPA and Army protocols because the HRA is required to meet legal requirements and must be done in accordance with standard methods. The Army and the state of Utah agreed to perform an EPA-style HRA using conservative worst-case analysis rather than best-estimate and uncertainty analysis. The HRA showed that the risk is low and meets the permitting requirements so no special risk management efforts are required for normal and mild upset conditions. It is also clear that the risk of accidents is much higher than the risks examined in the HRA. Therefore, the committee can find no compelling reason for the Army to extend the HRA for the purpose of directly combining and comparing the results of the two studies. Additional Review of The Risk Assessments In addition to the Stockpile Committee, other organizations have been involved in the review and guidance of the risk assessments. Quantitative Risk Assessment Three principal reviews were used throughout the development of the QRA: intraproject reviews, PMCD and TOCDF reviews, and independent external reviews.

OCR for page 14
Intraproject Reviews Intraproject reviews were conducted according to the quality assurance requirements established by the PMCD. Analysis and integration models and results at the subtask level, the task level, the integration level, and the assembly level were reviewed by SAIC analysts and engineers with experience performing QRAs on large-scale integrated engineering systems. PMCD and TOCDF Reviews The PMCD and TOCDF management provided most of the engineering and operational data used in the DCD/TOCDF QRA. They also reviewed the QRA models and results to ensure that the facility was correctly modeled in the QRA. The operational diagrams and models developed by the QRA team to analyze potential accidents at the TOCDF were reviewed by field engineers familiar with the TOCDF processes. As failure sequences were modeled during TOCDF systemization, PMCD and site personnel were asked if the results were consistent with their general knowledge and operating experience at JACADS. They were also asked to brainstorm on types of failures that might have been omitted. Integration of the PMCD and TOCDF reviews into the QRA process at an early stage led to the establishment of an effective communication network. This not only kept the QRA team analysts well informed of the status of the facilities and of ongoing activities, but also helped the PMCD and TOCDF staff understand the risks associated with the disposal processes and the significance of the risk assessments being done as early as possible in the project. Independent External Reviews The PMCD also established the Expert Panel (see Appendix B) through a separate contractor, MITRETEK Systems, to oversee the conduct of the QRA. This independent review group consisted of five experts, each of whom was either a specialist in the QRA field, a professional in the chemical industry, or an expert who specialized in chemical process safety. The Expert Panel, which met for the first time in November 1994 and regularly thereafter, followed the progress of the QRA through regular, interactive meetings with the project team. The Expert Panel had full access to all analytical activities and maintained an ongoing dialogue with the QRA team. Representatives of the Stockpile Committee attended the second panel meeting in February 1995 (as observers) and have attended all panel meetings since then. The Expert Panel reviewed and evaluated the QRA methodology, data, procedures, and assumptions. On March 28, 1996, the panel briefed the Stockpile Committee and indicated that the panel members were, in general, satisfied with the QRA methodology. The panel indicated that the SAIC study extended the state of the art in several areas (MITRETEK Systems, 1996). The panel also pointed out that the QRA analysts responded positively to comments. Appendix S of the QRA provides documentation of comments from the Expert Panel and responses of the QRA analysts (U.S. Army, 1996c). The final report of the Expert Panel is now available (MITRETEK Systems, 1996). The committee concurs with the Expert Panel's findings (MITRETEK Systems, 1996): The methodology was sound and has extended the state of the art in several areas. The methodology was well implemented. The panel had some reservations about a few technical aspects of the QRA but was reasonably satisfied that these did not affect the overall conclusions. The committee notes that the panel had a significant impact on several key areas of the QRA: The treatment of uncertainty is now more clearly addressed. The seismic vulnerability analysis for the liquid propane gas tank has been improved. The model for workers donning masks after a strong earthquake is more realistic. The mechanistic modeling of munitions handling accidents is much improved. The interactive independent review process was effective. Significant improvements in the QRA methodology have been made.

OCR for page 14
Figure 2-7 Mean public acute fatality complementary cumulative distribution function for munition storage during the 7.1 years of disposal processing, by distance from DCD. Source: Adapted from U.S. Army, 1996c. For disposal processing at the TOCDF, the QRA results show that public fatality risk is dominated by earthquakes (97.4 percent) as the most dangerous risk-initiating event (Figure 2-10). The consequences of an earthquake at the TOCDF are further dominated by the potential for a structural failure in the unpack area of the container handling building area caused by an earthquake stronger than the building is designed to withstand. The severe consequences would result partly because munitions are unpacked in this area and are not protected by transport containers. The QRA results shown in Figure 2-10 also indicate that internal events associated with processing account for less than 1 percent (i.e., 0.8 percent) of the TOCDF risk and that nearly all of this risk is associated with handling rather than with actual agent destruction. The study credits the low risk of processing to the safety and mitigation features of the baseline system and the limited quantities of agent available for release during processing. Operational Risk to Workers Workers at the TOCDF, including all support and administrative staff located at the facility or in nearby buildings and munition handlers responsible for removing munitions from the stockpile and transporting them to the disposal facility, were included in the risk assessment. The study includes only worker risks associated with accidents involving agent releases. Processing and handling workers can be directly affected by the blast of an explosion, for example, or by agent dispersion from an accident, and both effects are included. However, industrial-type risks, e.g., being crushed by a lift-truck, were not considered. The QRA results indicate a 1 in 7 probability of a worker fatality in the total disposal-related worker population in the 7.1 years of disposal processing. Figure 2-11 shows the contributions of various causes to worker risk. Maintenance activities account for 44 percent of the risk; seismic events, 36 percent; metal parts furnace explosions, 6 percent; handling, 6 percent; and other causes, 8 percent.

OCR for page 14
The QRA indicates that risks to disposal-workers from agent-related accidents are substantially higher than the public risks, as would be expected because of the proximity of the workers to the agent. Small releases that would not have an impact at a significant distance could still be lethal to workers in the immediate area. According to the QRA, there are about 500 workers at the TOCDF. If the 0.13 expected fatalities per 7.1 years of operation are dominated by single fatality accidents, then the individual disposal-worker risk at the TOCDF is about 4 × 10-5 per year. By comparison, this risk is about equal to the total occupational risk for all occupations (based on 1995 Occupational Safety and Health Administration [OSHA] data for fatalities by occupation) (U.S. Department of Labor, 1995). Higher risk levels are encountered by workers in the construction industry (about 12 × 10-5 per year). The individual risk levels for the general population of operators, fabricators, and laborers is about 11 × 10-5 per year. Presumably, most of the TOCDF workers would bear this sort of job-related risk plus the agent-related risk, which would increase the level of risk by about one-third. The OSHA data are averages across a wide spectrum of work environments and do not represent the lower individual risk levels that can be achieved by companies that emphasize safety. Because the disposal-worker risk levels from agent exposure are added to the normal occupational risk level, the committee believes that emphasizing job safety, both for agent and non-agent activities, is very important. The risk for other on-site workers (outside the TOCDF and DCD storage area) is evaluated in the same manner as public risk. The probability of one or more fatalities for other on-site workers during the 7.1 years of disposal processing is 5 × 10-4 (1 in 2,000). With about 100 workers in this category, and assuming that most accidents cause a single fatality, the Figure 2-8 Comparison of public risks during processing at DCD and TOCDF. Source: Adapted from U.S. Army, 1996d.

OCR for page 14
Figure 2-9 Comparison of public risks during processing at DCD and TOCDF (logarithmic scale). Source: Adapted from U.S. Army, 1996d. Figure 2-10 Contributors to the average public fatality risk from processing at DCD and TOCDF. Source: Adapted from U.S. Army, 1996d individual annual risk is less than 1 × 10-6 (1 in 1 million per year) for other on-site workers. This risk is small in comparison to risk levels in standard occupations (on the order of 1 × 10-4 per year). Thus other on-site workers are not significantly affected by the movement and disposal operations at DCD/TOCDF. Overall Risk Acute Fatality Risk. The public risk of an acute fatal poisoning from agent release is shown for DCD and the TOCDF in Figure 2-12 as a risk profile. The vertical axis shows the probability of a release at the site in which the number of fatalities would equal or exceed the number on the horizontal axis. The risks associated with the three situations of concern are summarized in three curves: the risk of disposal processing at the TOCDF; the risk of storage at DCD during the disposal process (with allowance for depletion of the stockpile during disposal); and the risk of continued munitions

OCR for page 14
Figure 2-11 Contributors to the average risk of fatality to disposal-related workers at DCD and TOCDF. Source: Adapted from U.S. Army, 1996d. storage (without disposal) at DCD. The analysis assumes that the disposal process will last approximately 7.1 years. The risk for 20 years of continued storage considers the case of a 20-year delay in initiating disposal operations. According to Figure 2-12, the average probability of incurring one or more public fatalities is about 1 × 10-5 (approximately 1 in 100,000) for the 7.1 years of disposal processing at the TOCDF; about 1.4 × 10-4 or (1 in 7,000) for stockpile storage at DCD during the disposal period; and about 5 × 10-3 (1 in 200) for continued stockpile storage at DCD for the next 20 years, with no processing. Figure 2-13 shows the same mean risk profile for disposal processing for 7.1 years of TOCDF operations with uncertainty bounds. The expected number of public fatalities over the time period is the probability-weighted sum of each possible number of fatalities (see Appendix A). The results of the QRA indicate that the expected number of fatalities is approximately 0.00016 for the 7.1 year disposal processing period, 0.002 for the stockpile storage at DCD during disposal processing, and 0.03 for continued stockpile storage at DCD for 20 years. The total expected public acute fatalities during the disposal operations is the summation of both the processing risk Figure 2-12 Summary of mean public risk from storage and processing at DCD and TOCDF. Source: Adapted from U.S. Army, 1996d.

OCR for page 14
Figure 2-13 Public societal acute fatalities for all campaigns (TOCDF disposal processing). Source: Adapted from U.S. Army, 1996c. and the storage risk during processing. This probability is only slightly higher (0.0021 versus 0.0020) than the storage risk alone because the processing risk is relatively small in comparison with the storage risk. As anticipated, the risk decreases with distance from the site. Figure 2-14 shows how disposal processing risk profiles vary with distance from the site. The distances reported are measured from the main processing building. Since the TOCDF is not very large (approximately 250 yards × 300 yards), these distances are essentially the same as the distance from the TOCDF fence. Risk associated with the disposal process also varies with time, as the various agents and items are sequentially destroyed. Figures 2-8 and 2-9 show these variations as the disposal processing moves from one agent and munition configuration to another. The disposal sequence was adjusted after an early QRA draft showed it would be desirable to eliminate the most hazardous agents (GB and VX) earlier in the program. More than 90 percent of the accidental risk at the Tooele site (storage and processing) is associated with agent GB, largely because of its higher evaporation rate in comparison to the more acutely toxic, but low volatility, agent VX. Accident sequences involving VX contribute about 10 percent of the overall public risk. The risk from a release of mustard is very small in comparison. An aircraft crash into the mustard ton container storage area and an ensuing fire pose the greatest threat from mustard. The public risk is dominated by seismic initiating events. In the absence of an earthquake-initiated event, the mean fatality risk would be 40-fold less for the public, 10-fold less for other (nondisposal) on-site workers, and 16-fold less for disposal-related workers. Worker Risk. Figure 2-15 illustrates the risk profile for other on-site workers at DCD/TOCDF. Risk profiles for disposal-related workers are not provided because current methods are not capable of combining the remote (dispersed agent-related) and direct (explosion, close-in agent, etc.) effects. The remote agent effects include variations in weather conditions and other external factors. The direct effects are calculated on a scenario-by-scenario basis. The QRA team could have produced risk profiles for remote effects alone but believes that they would have been incomplete. The committee agrees.

OCR for page 14
Figure 2-14 Mean public acute fatality risk by distance from TOCDF during disposal processing. Source: Adapted from U.S. Army, 1996c. Public Cancer Risk. There is essentially no risk of cancer from accidental releases during processing or during 20 years of continued storage. The probability of an individual developing cancer is vanishingly small. The mean cancer risk to the public (the expected number of excess cancers) from storage for 7.1 years is only 0.000002; the mean cancer risk from processing is only 1 percent of the risk from storage. These are small risks in comparison with the acute fatality risk to the public associated with the facility. Health Risk Assessment To evaluate human health risks, both carcinogenic and noncarcinogenic health effects from chemical agents, metals, volatile and semivolatile products of incomplete combustion, and other combustion products were considered in the HRA screening risk assessment. Because carcinogenic effects dominate public concerns, the following section focuses on this aspect of the HRA. According to EPA guidelines (EPA, 1994), to protect human health and the environment, emissions of carcinogens should not exceed a cancer risk level of 1 × 10-5 to a maximally exposed individual over a 70-year lifetime. This corresponds to a 1 in 100,000 chance of developing cancer from exposure under a particular scenario. The selection of this level acknowledges that background exposures from drinking water, food, and air, that is, sources other than the emission source being evaluated in the screening assessment, also contribute to the risk in the study area. Setting the cancer risk level at 1 × 10-5 rather than at a less protective level (e.g., 1 × 10-4) is intended to protect the public from an unacceptable total exposure to carcinogens. Conservative assumptions were used to derive an upper limit estimate of risk to the various populations considered in this assessment. The calculated concentrations of heavy metals in the air and soil were compared to specific EPA criteria. Potential health risks from the inhalation of particulates were characterized by comparing modeled annual air concentrations of particulates (at the maximum point of impact) to national ambient air quality standards. Potential impacts to environmental receptors were evaluated by comparing modeled surface water concentrations to ambient water quality criteria.

OCR for page 14
Figure 2-15 Acute fatalities for other on-site workers at TOCDF from accidents during disposal processing. Source: Adapted from U.S. Army, 1996c. In addition to the conservative assumptions for selecting parameters for the HRA, a number of scale-up adjustments were made to the data used to model air and deposition concentrations. The HRA was conducted using data from JACADS to represent TOCDF emissions because actual data on emissions from the TOCDF incinerators had not yet been collected. The JACADS emissions data were scaled up to reflect the maximum anticipated feed rate at the TOCDF and modified to reflect the maximum metals composition in munitions to be treated at the TOCDF. The emissions data were also modified to reflect potential upset conditions when the incinerator units might emit higher than usual concentrations of constituents. As suggested in the EPA screening assessment guidance (EPA, 1994), it was assumed that 5 percent of the time organic emissions from the incinerators would be 10 times higher than normal and that 20 percent of the time metals emissions would be 10 times higher than normal. These assumptions were chosen to account for abnormal combustion conditions that might occur during startup, shutdown, or production upsets, and are very conservatively based. To account for unknown organic constituents (i.e., constituents that were not specifically analyzed by the laboratory analytical methods), the estimated emissions were weighted according to a recommended EPA method (EPA, 1994). This adjustment was made only for data for which associated total organic carbon information was available for the emissions (the method requires total organic carbon data). This approach assumes that the unidentified organic compounds are similar in toxicity and chemical properties to the identified organic compounds taken as a whole. Computations are made by increasing the emission rate of each identified organic compound by the ratio of the concentration of total organic compounds to the total concentration of all identified organic compounds. The risk assessment was then conducted using the adjusted (i.e., increased) emission rates for each identified organic compound. Total organic carbon data were available for the metal parts furnace, the dunnage incinerator,

OCR for page 14
and the liquid incinerator units (Utah DSHW, 1996). The adjustments to the JACADS emissions data increased the overall risks associated with operation of the TOCDF and provided a conservative screening risk assessment (EPA, 1994). HRAs were conducted for an adult resident, a child resident, a subsistence fisher, and three different farmers, an approach that is consistent with EPA guidelines. The adult resident was assumed to reside for 30 years at the maximum off-site point of impact along the northern boundary of the TOCDF facility. The child resident was assumed to reside for six years at the same maximum point of impact. The subsistence fisher was assumed to reside for 30 years 40 kilometers (25 miles) north-northwest of the TOCDF, where subsistence fishing was thought to be practiced. The three farmers were assumed to reside for 40 years in the vicinity of the TOCDF. Pathways for human exposure to incinerator vapor and particulate emissions were specified in the HRA for each of the six individuals analyzed. Exposure pathways included: consumption of fish, meat, and homegrown vegetables; incidental ingestion of soil; and inhalation. For example, in the case of the subsistence fisher, the fish consumption rates used in the analysis were considered to be representative of a subsistence fisher rather than the general population; all fish consumed were assumed to be caught in water impacted by incinerator emissions; and 25 percent of aboveground and below-ground vegetables consumed was considered to be homegrown. In the case of Farmers A and C, 100 percent of the beef was assumed to be from home-raised stock that grazed in various locations near the TOCDF. For Farmer B, 12.5 percent of aboveground and 100 percent of below-ground vegetables consumed were assumed to be homegrown and contaminated by emissions from the TOCDF. The study assumed that all individuals were exposed 350 days per year, except Farmer A, who was assumed to be exposed 175 days per year. For all individuals, the TOCDF was considered to be operating continuously (i.e., 24 hours per day, 365 days per year) for a period of 10, 15, and 30 years. Incinerator emissions evaluated in the risk assessment included: each incinerator or heating, ventilation, and air conditioning filter stack operating individually the combined stack (liquid incinerators, metal TABLE 2-1 Summary of the Human Health Risk—Overall Risk of Cancer for Combined TOCDF and CAMDS Disposal Operations Period of Operation Receptor 10 years 15 years 30 years Adult Resident <1 × 10-6 <2 × 10-6 <4 ×10-6 Child Resident <3 × 10-6 <3 × 10-6 <3 × 10-6 Fisher <5 × 10-8 <5 × 10-8 <7 × 10-8 Farmer A <8 × 10-6 <8 × 10-6 <8 × 10-6 Farmer B <1 × 10-7 <1 × 10-7 <2 × 10-7 Farmer C <9 × 10-6 <1 × 10-5 <1 × 10-5 Source: Adapted from Utah DSHW, 1996. parts furnace, and deactivation furnace system operating simultaneously) maximum TOCDF operations (all TOCDF units operating simultaneously) maximum TOCDF plus CAMDS operations (all TOCDF units plus CAMDS deactivation furnace and heating, ventilation, and air conditioning filter stack operating simultaneously). The latter scenario, involving total TOCDF operations plus CAMDS operations, resulted in the largest overall cancer risk. The summary of the human health risk calculations are shown in Table 2-1. In no case did the carcinogenic risks exceed the 1 × 10-5 carcinogenic risk level established in the EPA screening risk assessment guidelines (EPA, 1994). Keeping Assessments Current Quantitative Risk Assessment The results presented in this report reflect the results of the QRA as of December 1996. The committee saw several earlier drafts of the QRA, and the risk estimates were different, sometimes significantly, in each draft. For example, in the first draft, it became obvious that several risk reduction options were available, which were highlighted in the Systemization report (NRC, 1996b). They included deferring the processing of weteye bombs until some safety concerns had been addressed, which resulted in a risk-based reordering of the disposal schedule. Operation of the liquid propane tank was also modified to reduce flammable fuel

OCR for page 14
inventory in the event of seismic damage. As changes were implemented, the QRA was revised accordingly. Because the QRA process was interactive and included input and comments from the Expert Panel and data from accumulating experience captured in the Army's ''Lessons Learned'' program and other site-specific risk management and reviews, more changes were incorporated. As the QRA proceeded, some major sources of risk were analyzed more thoroughly than they had been in the first draft. For example, outside experts were brought in to refine the estimates of the seismic response of structures and stacked munitions, which were subjected to a much more detailed analysis. Some of the updated analyses have led to significant reductions in the estimates of stockpile risk and to some reductions in uncertainty. As this discussion shows, QRAs represent a state-of-knowledge and a state-of-the-system at a given point in time. Because both of these factors change with time, the Army is treating the QRA as a "living model" that will continue to be updated as changes affect risk or new information becomes available. This "living model" QRA will be the basis for ongoing risk management at the TOCDF. For example, an analysis of stockpile failure modes under seismic conditions might lead to ideas for further reducing interim stock-pile risks. Health Risk Assessment The HRA reflects pre-operational assumptions about emissions from the incinerators and the frequency of upset conditions. As the TOCDF continues operation and actual operational and trial burn data become available, these levels will be substituted for the initial assumptions. Analyzing and Integrating Results Different types of risk factors are evaluated on different bases, making the integration of results difficult. However, some comparisons can be helpful. The QRA for Tooele estimates the risk to an individual living within a ring 2 to 5 kilometers (1.2 to 3.1 miles) from the TOCDF. This mean risk level can be interpreted as typical of the individual public fatality risk from an agent release because about half of the people in the zone will have a somewhat higher risk and half will have a somewhat lower risk. Note that government property around the TOCDF provides at least a 2-km buffer zone. Individual risk levels decrease with distance from the facility. These results are summarized in Box 2-1. BOX 2-1 Individual Risk at DCD and the TOCDF in Perspective According to the QRA, the public mean individual fatality risk levels in the 2 to 5 kilometer zone are as follows: during one year of exposure to the storage area, 6.4 × 10-6 (1 in 160,000) during one year of exposure to processing operations, 1.7 × 10-7 (1 in 6 million) The disposal-worker mean individual fatality risk during one year of exposure to processing operations is: 4 × 10-5 (1 in 25,000). To put these numbers in perspective, consider that the risk for a typical American of dying from a fall is about 7 × 10-5 per year (1 in 14,000), and the risk of dying in a fire is about 3 × 10-5 per year (1 in 33,000). The risk of dying from a lightning strike is about 5 × 10-7 per year (1 in 2 million). An individual living in the 2 to 5 kilometer ring around the TOCDF is subject to the four general categories of risk shown in Table 2-2. The numbers shown are the probabilities that that individual will die or will contract cancer as a result of either the 7.1 year TOCDF processing schedule or of the same 7.1 year period of exposure to storage of the full stockpile. In fact, once disposal has begun, an individual would be exposed to both the processing risk and a diminishing stockpile risk as the stockpile was depleted (Figure 2-8). For an individual who lives more than 10 kilometers away, the individual risk of fatality associated with disposal accidents is insignificant compared to his or her exposure to ordinary risks, and the residual risk of fatality of 4 × 10-8 per year is from stockpile storage accidents. The HRA only estimates the risk to a "maximally exposed individual" and does not address health risks to more distant individuals because atmospheric dispersion dilutes source streams with distance. This is

OCR for page 14
TABLE 2-2 Risks for an Individual Living 2 to 5 Kilometers from the TOCDF Consequence 7.1 Years of Exposure to DCD Storage during Disposal Processing 7.1 Years of Exposure to Disposal Processing at the TOCDF Disposal on Schedule Acute Fatality from Agent Release Accidents 4.5 × 10-5 probability of fatality 1.2 × 10-6 probability of fatality Latent Fatality from Agent Release Accidents 8.5 × 10-6 probability of delayed cancer (only HD is a carcinogen) 1.7 × 10-11 probability of delayed cancer (only HD is a carcinogen) Eventual Chance of Cancer from Exposure to Stack Emissions Not applicable Less than 1 × 10-5 probability of delayed cancer to a "maximally exposed individual" Delayed Disposal Total storage risk increases linearly with time Disposal risk shifts to the future Source: Adapted from U.S. Army, 1996c; Utah DSHW, 1996. the same reason that minor stockpile accidents have no impact beyond 10 kilometers. There is a 1 in 22,000 (4.6 × 10-5) chance of a fatality to a person located 2 to 5 kilometers from the TOCDF from stockpile or storage accidents over a 7.1 year period (Table 2-2). The health risks from normal operations are computed in the HRA as a conservative upper limit for a maximally exposed individual outside the site. The criterion established by the EPA of 1 × 10-5 probability of delayed cancer is based on a very conservative estimate of the risk. With proper treatment, not all of these cancers will cause death. All that can be concluded is that the cancer risk from normal operations is lower, probably much lower, than the fatality risk from accidents. The EPA, which has set the basic standards and approach for the HRA, suggests an acceptable risk of a lifetime probability of getting cancer of less than 1 in 100,000 for a 70-year lifetime after exposure to the health risk associated with the facility. For an average one year period within that person's lifetime, the risk of latent cancer would be roughly on the order of 1 in 7 million. To put these risk criteria in perspective, consider the familiar cancer risks, some of which are voluntary (lung cancer is linked to smoking), others of which are not. The general incidence of lung cancer in the U.S. population means that the annual chance of an individual dying are about 1 in 2,000, an annual risk level of 5 × 10-4. For all cancers, the chance is about 1 in 600 per year. (Note that there are about 500,000 deaths from all cancers and 150,000 deaths from lung cancer each year in the United States.) Thus, the change in cancer risk as the result of exposure at the maximum level considered acceptable by the EPA would be much less than 1 percent of the normal incidence of all cancer. Table 2-3 presents measures of public risk (expressed as the number of fatalities expected as the TABLE 2-3 Expected Number of Fatalities (Societal Risk) Processing Period 7.1 Years of Exposure to DCD 7.1 Years of Exposure to Disposal Consequence Storage during Disposal Processing Processing at the TOCDF Fatality from Agent Release Accidents Public risk, 1.6 × 10-3 expected fatalities Public risk, 1.6 × 10-4 expected fatalities 2.4 × 10-6 risk of latent cancers 2 × 10-8 expected cancers Worker risk not estimated in QRA Disposal workers, 1.3 × 10-1 expected fatalities Other workers, 6.6 × 10-4 expected fatalities Eventual Chance of Cancer as a Result of Exposure to Stack Emissions Not applicable Not included in HRA Source: Adapted from U.S. Army, 1996c.

OCR for page 14
result of storage and disposal operations). Worker risks are also shown. The HRA does not address public risk; thus the health risk from normal operations cannot be put into the same framework. The HRA sets a criterion for a "maximally-exposed person" but does not evaluate the number of individuals who might be exposed at that or lower levels. The criterion is set low enough to compensate for the potential of multiple exposures. The EPA has established acceptable risk criteria for the potential for latent cancers associated with plant operating emissions. However, criteria have not been established by a regulatory group for risks associated with accidents. The foregoing discussion of DCD/TOCDF risk assumes the timely disposal of agent over a 7.1 year period. Delays in processing would have a minimal effect on the disposal risk, but the risk of continued storage would increase.