Search Options
Index | Site Map | FAQ | Facility Info | Reading Rm | New | Help | Glossary | Contact Us

5. Environmental Impacts of Postulated Accidents

[ Prev | Next | Table of Contents ]

5.1 Introduction

[ Prev | Next | Table of Contents ]

This section discusses each aspect of postulated accidents that is normally treated in the final environmental statements (FESs) for the operation of nuclear power plants. Methodologies that estimate future risks at each existing nuclear power plant site in the United States are developed in Section 5.3.3, considering an additional 20-year period of operation beyond the current license term.

The characteristics of nuclear power plant accidents are discussed (Section 5.2.1) to acquaint the reader with (1) the sources of radiation from postulated accidents, (2) the potential pathways of radiation to the environment, and (3) the possible health effects of exposure to such accidental releases. Historical experience and observed impacts of nuclear power plant accidents are discussed next (Section 5.2.2), followed by a description of the various measures taken in the design and operation of a power plant to reduce the likelihood or consequences of an accident (Section 5.2.3).

The impacts of accident risks during a license renewal period are discussed in Section 5.3. A brief discussion of the primary concern arising from extending the operational life of nuclear power plants is provided (Section 5.3.1). This concern centers on the effects that plant aging and increasing population can have on the probability and consequences of accidents. Calculation of the estimated environmental impacts and risks due to postulated accidents during the license extension period is discussed in Sections 5.3.2 and 5.3.3. Consequences of design-basis and severe accidents are reviewed. The potential pathways for radiation release examined are (1) direct release to the atmosphere, (2) fallout on open bodies of water, and (3) groundwater. Existing severe accident analyses were reviewed and used to predict consequences at all sites. The potential economic impacts of accidents during the renewal period were also reviewed (Section 5.3.4). To maintain a perspective on the results of this analysis, a discussion of the uncertainties associated with the types of consequence analyses used in this evaluation is provided (Section 5.3.5). Finally, a discussion is given on the role of severe accident mitigation design alternatives (SAMDAs) in the license renewal process (Section 5.4).

5.2 Plant Accidents

[ Prev | Next | Table of Contents ]

5.2.1 General Characteristics of Accidents

The term "accident" refers to any unintentional event outside the normal plant operational envelope that results in a release or the potential for release of radioactive materials into the environment. Generally, the U.S. Nuclear Regulatory Commission (NRC) categorizes accidents as "design basis" (i.e., the plant is designed specifically to accommodate these) or "severe" (i.e., those involving multiple failures of equipment or function and, therefore, whose likelihood is generally lower than design-basis accidents but where consequences may be higher), for which plants are analyzed to determine their response. The predominant focus in environmental assessments is on events that can lead to releases substantially in excess of permissible limits for normal operation. Normal release limits are specified in the NRC's regulations (10 CFR Part 20 and 10 CFR Part 50, Appendix I).

Many features combine to minimize the risk of accidents at nuclear power plants. These features include high-quality reactivity control and reactor cooling systems and containment and backup safety systems to respond to equipment failure. The incorporation of safety into design, construction, and operation is to a very large extent devoted to minimizing the possibility of the release of radioactive materials from their normal places of confinement within the plant. Descriptions of these safety features are provided in each licensee's final safety analysis report (FSAR) and in the NRC's safety evaluation report.

The plant design, including the types and number of safety systems, takes into consideration the specific locations of radioactive materials within the plant; their amounts; their nuclear, physical, and chemical properties; and their potential for transport into the environment and for creating health hazards.

5.2.1.1 Fission Product Characteristics

By far the largest inventory of radioactive material in a nuclear power plant is produced as a by-product of the fission process and is located in the uranium oxide fuel pellets in the reactor core in the form of fission products. During periodic refueling shutdowns, the assemblies containing these fuel pellets are transferred to a spent-fuel storage pool; the second largest inventory of radioactive material is located in this storage area. Much smaller inventories of radioactive materials are also normally present in the water that circulates in the reactor coolant system and in the systems used to process gaseous and liquid radioactive wastes in the plant.

Radioactive materials in power plants exist in a variety of physical and chemical forms. Their potential for dispersal into the environment depends not only on mechanical forces that might physically transport them, but also on their inherent properties, particularly their volatility. The majority of these materials exist as nonvolatile solids over a wide range of temperatures. Some, however, are relatively volatile solids, and a few are gaseous at normal temperatures and pressures. These characteristics have a significant bearing on the assessment of the environmental radiological impacts of accidents.

The gaseous materials include radioactive forms of the chemically inert noble gases krypton and xenon. These two gases have the highest potential for release into the atmosphere. If a reactor accident involving degradation of the fuel cladding were to occur, the release of substantial quantities of these radioactive gases from the fuel into the reactor cooling system would be virtually certain. Such accidents are low-frequency, but credible, events. For this reason, the safety analysis of each nuclear power plant incorporates a hypothetical design-basis accident that postulates the release of the entire contained inventory of radioactive noble gases from the fuel in the reactor into the containment structure. If the noble gases were further released to the environment as a result of failure to maintain the containment boundary, the hazard to individuals from these noble gases would arise predominantly through the external gamma radiation from the airborne plume. The reactor containment structure and containment support systems are designed to minimize the possibility of this type of release.

Radioactive forms of iodine are produced in substantial quantities in the fuel by the fission process, and in some chemical forms they may be quite volatile. For these reasons, iodine has traditionally been regarded as having a relatively high potential for release from the fuel into the containment during certain design-basis accidents. Because iodine concentrates in the thyroid gland, the release of radioiodines to the atmosphere is controlled by containment and by the use of special systems (i.e., filters) designed to retain the iodine.

The chemical forms in which the fission product radioiodines are found are generally solid materials at room temperatures; hence, they have a strong tendency to condense (or "plate out") on cooler surfaces. In addition, most of the iodine compounds are quite soluble in, or are chemically reactive with, water. Although these properties do not prevent the release of radioiodines from degraded fuel, they would act to inhibit release from the containment structure that has large internal surface areas and may contain large quantities of water as a result of an accident. The same properties affect the behavior of radioiodines that may "escape" into the atmosphere. Thus, if it rains during a release, or if there is moisture on exposed surfaces (for example, dew), the radioiodines will show a strong tendency to be absorbed by the moisture.

Other radioactive materials formed during the operation of a nuclear power plant are less volatile and have a much smaller tendency to escape from degraded fuel unless the temperature of the fuel becomes very high. Such materials tend to condense quite rapidly when they are transported to a lower temperature region or to dissolve in water when it is present. This mechanism can result in production of very small particles that can be carried some distance by a moving stream of gas or air. If such particulate materials are dispersed into the atmosphere as a result of containment failure, they tend to be carried downwind and deposited on surfaces by gravitational settling (fallout) or by precipitation (washout or rainout), where they will become "contamination" hazards in the environment.

All of these radioactive materials exhibit the property of radioactive decay with half-life periods ranging from fractions of a second to many days or years. Many of the radioactive materials decay through a sequence of decay processes, and all eventually become stable (nonradioactive). The radiation emitted during these decay processes renders the radioactive materials hazardous.

5.2.1.2 Meteorological Considerations

Two separate analyses of accident sequences are performed during the licensing process for a nuclear power plant. The first analysis is the determination of the consequences for design-basis accidents and is performed for the NRC's safety evaluation report. This analysis is performed to ensure that the doses to any individual at the exclusion area boundary over a period of 2 hours, or at the outer boundary of the low population zone (LPZ) during the entire period of plume passage, will not exceed the siting dose guidelines of 25 rem to the whole body or 300 rem to the thyroid, pursuant to 10 CFR Part 100. This analysis is used to examine site suitability (10 CFR Part 100) and the mitigative capability of certain plant safety features (10 CFR Part 50). The atmospheric dispersion model for this evaluation, as described in the NRC Regulatory Guide 1.145, uses on-site meteorological data (typically, a multiyear record) considered representative of the site and vicinity to calculate relative dilutions that will be exceeded no more than 0.5 percent of the time in any one sector (22.5° ) and no more than 5 percent of the time for all sectors (360° ) at the exclusion area boundary and LPZ. These dilution factors, because they provide little plume spread, ensure site-specific calculated doses that could be exceeded only 5 percent of the time.

The second analysis of accident consequences is generally found in the environmental documentation for the most recently licensed nuclear plants and considers a spectrum of releases, including those for severe accidents. Actual meteorological conditions from a representative 1-year period of record of on-site data are used in this environmental analysis. A detailed description of the atmospheric dispersion model used to estimate the environmental impacts of such accidents is contained in NUREG-75/014 (formerly WASH-1400), Appendix VI.

5.2.1.3 Exposure Pathways

The radiation exposure (hazard) to individuals is determined by the individual's proximity to the radioactive materials; the duration, intensity, and type (external versus internal) of exposure; and factors that act to shield the individual from the radiation. Many of the pathways for radiation and the transport of radioactive materials that lead to radiation exposure hazards to humans are the same for accidental as for "normal" releases. These pathways are depicted in Figure 5.1. Two additional possible pathways that could be significant for accident releases are not shown in Figure 5.1. One of these pathways is the fallout of radioactivity onto open water or onto land with runoff into open water bodies. The second pathway would be unique to an accident involving temperatures high enough to cause melting of the reactor core and subsequent penetration of the reactor vessel and underlying base mat by the molten core debris. Such an occurrence would create the potential for the release of radioactive material into the hydrosphere via groundwater beneath the plant. These pathways may lead to external exposure to radiation and also to internal exposure if radioactive material is contacted, inhaled, or ingested from contaminated food or water.

It is characteristic of all these pathways that during the transport of radioactive material by wind or water, the material tends to spread and disperse--like a plume of smoke from a smokestack--becoming less concentrated in larger volumes of air or water. The result of these natural processes is to lessen the intensity of exposure to individuals downwind or downstream of the point of release, but the processes also tend to increase the number who may be exposed. For a release into the atmosphere, the degree to which dispersion reduces the concentration in the plume at any downwind point is governed by the turbulence characteristics of the atmosphere, which vary considerably with time and from place to place. This fact, taken in conjunction with the variability of wind direction and the presence or absence of precipitation, means that accident consequences depend largely upon the

Figure 5.1 Potential exposure pathways to individuals.

weather conditions existing at the time of the accident.

5.2.1.4 Adverse Health Effects

The cause-and-effect relationships between radiation exposure and adverse health effects are quite complex. Whole-body radiation exposure resulting in a dose greater than about 25 rem over a short period of time (hours) is necessary before any physiological effects to an individual are clinically detectable¹ shortly thereafter. Doses about 10 to 20 times larger, also received over a relatively short period of time (hours to a few days), can be expected to cause some fatalities. At the severe (but extremely low probability) end of the accident spectrum, very high exposures of these magnitudes are theoretically possible for persons in proximity to the plant if measures are not or cannot be taken to provide protection, such as by sheltering or evacuation.

Lower levels of exposures may constitute a longer-term health risk. The effects of such exposures may include randomly occurring cancer in the exposed population and genetic changes in future generations after exposure of a prospective parent. Relating a given health effect to a known exposure to radiation is most often not possible because of the many other possible causes for such effects. For this reason, it is necessary to assess radiation-induced cancer effects on a statistical basis.

Occurrences of cancer in the exposed population may begin to develop only after a lapse of 2 to 15 years (latent period) from the time of exposure and continue over a period of over 40 years (plateau period). However, in the case of exposure of fetuses (in utero), occurrences of cancer may begin to develop at birth (no latent period) and end at age 10 (that is, the plateau period is 10 years). The occurrence of cancer itself is not necessarily indicative of a fatality because the ratio of mortality to incidence of cancer depends upon the cancer type and advances in medical treatment.

The estimates of health consequences used for latent fatalities in this document are the same estimators used in the development of the revised 10 CFR 20 regulations. A discussion is provided in Sections 3.8 and 4.6, and a more detailed discussion is provided in Section E.4, which details the recent developments in radiation risk estimation that lead to the health-consequence risk estimates in this section. The discussion in Section E.4 includes background information about epidemiology as well as health-effects information compiled by the United Nations Scientific Committee on the Effects of Atomic Radiation, by the National Academy of Sciences (reports of Advisory Committees on the Biological Effects of Ionizing Radiation--BEIR-I, BEIR-III, BEIR-V), and by the International Commission on Radiological Protection. The risk estimates for fatal cancers are considered to be nominally 500 per million person-rem, consistent with the risk factors described earlier (Section 3.8.1.3 and Appendix E.4). In addition, approximately 100 genetic disorders per million person-rem are projected for the succeeding generations.

5.2.1.5 Avoiding Adverse Health Effects

Radiation hazards in the environment disappear by the natural process of radioactive decay. Where the decay process is a slow one, however, and where the material becomes relatively fixed in its location as an environmental contaminant (such as in soil), the hazard can continue to exist for a long period of time--months, years, or even decades. Thus, a possible environmental societal reaction to severe accidents is avoidance of the potential health hazards by restrictions on the use of the contaminated property or contaminated foodstuffs, milk, and drinking water.

5.2.2 Accident Experience and Observed Impacts

A limited number of accidents have been recorded in the experience data of the world's nuclear programs. The United States, Great Britain, and the Soviet Union have all experienced accidents of varying magnitudes and consequences. The following paragraphs will discuss first the United States experience, followed by the British and then the Soviet accident experience.

As of September 1990, 112 commercial nuclear power reactor units were licensed for operation in the United States (Table 2.1) with power-generating capacities ranging from 72 to 1270 MW(e). The combined experience with these operating units represents approximately 1300 reactor-years (RYs) of operation over an elapsed time of about 28 years. [An additional 6 plants, with individual generating capacities of up to 1314 MW(e), are expected to receive an operating license within the next 10 years.] Several of these facilities have experienced accidents (ORNL 1980; NUREG-0651; Thompson and Beckerley 1964), some of which have resulted in small releases of radioactive material to the environment. None is known to have caused any radiation injury or fatality to any member of the public, nor any significant contamination of the environment. Although the experience base with light-water reactors (LWRs) having containments such as those licensed in the United States is not large enough to permit reliable statistical prediction of accident probabilities, it does, however, suggest that significant environmental impacts caused by accidents are not at all likely to occur over time periods of a few decades.

Melting or severe degradation of reactor fuel has occurred in only one U.S.-licensed commercial LWR--during the accident at Three Mile Island Unit 2 (TMI-2) on March 28, 1979. It has been estimated that about 2.5 million Ci of noble gases (about 0.9 percent of the core inventory) and 15 Ci of radioiodine (about 0.00003 percent of the core inventory) were released to the environment at TMI-2 (NUREG/CR-1250).² No other radioactive fission products were released in measurable quantities. It has been estimated that the maximum cumulative off-site radiation dose to an individual was less than 100 mrem (NUREG/CR-1250; President's Commission 1979). The total population exposure has been estimated to be in the range from about 1000 to 5000 person-rem. (This range is discussed on page 2 of NUREG-0558.) This exposure could statistically produce between zero and one additional fatal cancer over the lifetime of the exposed population of approximately 2 million in the site area. The same population receives about 240,000 person-rem each year from natural background radiation, and approximately a half million cancers are expected to develop in this group over the lifetime of the population (NUREG/CR-1250; President's Commission 1979), primarily from causes other than radiation. Trace quantities (barely above the limit of detectability, below allowable limits, and less than that from fallout due to nuclear tests) of radioiodine were found in a few samples of milk produced in the area. No other food or water supplies were affected.

Accidents at U.S. nuclear power plants have also caused occupational injuries and a few occupational fatalities, but these were not due to radiation exposure. Individual worker exposures have ranged up to about 4 rem as a direct consequence of reactor accidents (although there have been higher exposures to individual workers as a result of other unusual occurrences). However, the collective worker exposure levels (person-rem) from accidents are a small fraction of the exposures experienced during routine operation; during the 1982-1986 time period, routine exposures ranged from 23 to 2880 person-rem/year in pressurized-water reactors (PWR) and 84 to 4080 person-rem/year in boiling-water reactors (BWR) per RY (NUREG-0713).

Accidents have also occurred at other nuclear facilities in the United States and in other countries (ORNL 1980; Bertini 1980). Reactor fuel melted in at least seven of these accidents: Fermi 1 (Lagoona Beach, Michigan), St. Laurent (France), NRX Reactor (Chalk River, Canada), Experimental Breeder Reactor 1 (Idaho), Heat Transfer Reactor Experiment Facility (Idaho), Westinghouse Reactor (Waltz Mills, Pennsylvania), and Oak Ridge Research Reactor (Tennessee). Because of inherent differences in design, construction, operation, and purpose of these other facilities, their accident record has only indirect relevance to current nuclear power plants. The most relevant accident was the one in 1966 at Enrico Fermi Atomic Power Plant Unit 1. Fermi Unit 1 was a sodium-cooled fast breeder demonstration reactor designed to generate 61 MW(e). The damages were repaired and the reactor reached full power 4 years after the accident. It operated successfully and completed its mission in 1973. The Fermi accident did not release any measurable radioactivity to the environment.

A reactor accident in 1957 at Windscale, England (renamed Seascale), released a significant quantity of radioiodine, approximately 20,000 Ci, to the environment and minor quantities of ¹³⁷Cs, ⁸⁹Sr, and ⁹⁰Sr (Eisenbud 1987) and ²⁴⁰Po (Crick and Linsley 1983). This reactor, which was not operated to generate electricity, used a graphite core design and circulated air rather than water to cool the uranium fuel. During a special operation to heat the large amount of graphite in this reactor (an operation normal for this graphite-moderated reactor), the fuel overheated and radioiodine and noble gases were released directly to the atmosphere from a 123-m (405-ft) stack. Milk produced in a 518-km² (200-mile²) area around the facility was impounded for up to 44 days. The United Kingdom National Radiological Protection Board (1957) estimates that the releases may have caused as many as 260 cases of thyroid cancer, about 13 of them fatal, and as many as 7 deaths from other cancers or hereditary diseases. This kind of accident cannot occur in a water-moderated and -cooled reactor like those in the U.S. nuclear power program.

On April 26, 1986, a major accident occurred at reactor 4 of the Chernobyl Nuclear Power Station in the Soviet Union. This reactor differs substantially from LWRs licensed to operate in the United States. The initiating event, a reactivity insertion, was recognized as a potential problem early in U.S. power reactor design; consequently, licensed U.S. power reactors are designed to prevent or accommodate occurrences of reactivity insertions. A major difference in safety between the U.S. designs and Chernobyl is that the Chernobyl reactor did not have a containment similar to those found on U.S. reactors. Also, the Chernobyl plant, which used graphite as a neutron moderator rather than water as with U.S. designs, had a positive power coefficient for the off-normal plant conditions that were present at the time of the accident. Thus, the accident has only indirect relevance to LWRs. The release of radioactive material from the accident was initially reported by the Soviets to be about 100 million Ci of fission products, but (except for the noble gases) that estimate included only material deposited within the European part of the Soviet Union. As a result of the accident, radionuclides were deposited throughout the Northern Hemisphere.

Of the almost 3 billion people in the Northern Hemisphere receiving Chernobyl radiation, about 800 million people account for 97 percent of the total risk increment. The remaining 3 percent of the dose commitment in Asia and North America represents a minuscule risk increment. Outside of the 30-km (19-mile) zone surrounding Chernobyl, the incremental increase in fatal cancer risk is a fraction of a percent and is not likely ever to be detected epidemiologically (DOE/ER-0332; Goldman 1987).

5.2.3 Mitigation of Accident Consequences

5.2.3.1 Design Features

All U.S. power reactors contain system features designed to prevent accidental release of radioactive fission products from the fuel and to lessen the consequences should such a release occur. Many of the design and operating specifications of these features are derived from the analysis of postulated events known as design-basis accidents. These accident-preventing and -mitigating system features are collectively referred to as "engineered safety features." Safety injection systems are incorporated to provide cooling water to the reactor core during a loss-of- coolant accident to prevent or minimize fuel damage. Heat-removal capability is provided inside the containment to prevent containment failure from overpressure. Long-term decay heat removal systems are also provided to remove decay heat from the core. All the mechanical systems mentioned above are supplied with emergency power from on-site diesel generators in the event that normal off-site station power is interrupted.

Containment structures are used as a mitigating system to provide a nearly leaktight barrier to minimize the escape of fission products to the environment in the event of a fission product release inside containment. These structures are designed to withstand the internal pressure and temperature associated with design-basis accidents.

The fuel-handling structures also have accident-mitigating systems. Spent fuel is handled and stored under water, which would tend to greatly reduce the amount of radioactive material released to the building environment in the event of fuel failure. A safety-grade exhaust air ventilation subsystem contains both charcoal and high-efficiency particulate filters. The ventilation systems are also designed to keep the area around the spent-fuel pool below the prevailing barometric pressure during fuel-handling operations to minimize the outleakage through building openings. Upon detection of high radiation, exhaust air is routed through the filter units, and radioactive iodine and particulate fission products which escaped from the spent fuel pool would be removed from the flow stream before exhausting to the atmosphere.

Much more extensive discussions of the safety features and characteristics of a particular plant may be found in the FSAR for that plant. In addition, the implementation of the lessons learned from the TMI-2 accident--in the form of improvements in design, procedures, and operator training--has significantly reduced the likelihood of a degraded-core accident that could result in large releases of fission products to the containment. These TMI-2-related requirements are specified in NUREG-0737.

5.2.3.2 Site Features

The NRC's site criteria, found in 10 CFR Part 100, require that every power reactor site have certain characteristics that tend to reduce the risk and potential impact of accidents. First, the site must have an exclusion area around the reactor. A typical exclusion area radius is about 0.8 km (0.5 mile). No residents are allowed in the exclusion area. Public transportation routes and other public activities are allowed within the exclusion area, but these routes and activities must be demonstrated to be controllable in the event of an emergency. Second, beyond and surrounding the exclusion area is an LPZ. A typical LPZ radius is about 5 km (3 miles). Within this zone, the licensee must ensure that there is a reasonable probability that appropriate protective measures could be taken on behalf of the residents and other members of the public in the event of a serious accident. Third, 10 CFR Part 100 requires that the distance from the reactor to the nearest boundary of a densely populated area containing more than 25,000 residents be at least one and one-third times the distance from the reactor to the outer bound of the LPZ.

5.2.3.3 Emergency Preparedness

Each licensee is required to establish emergency preparedness plans to be implemented in the event of an accident, including protective action measures for the public. The NRC, as well as other federal and state regulatory agencies, review the subject plans to ensure that the condition of on- and off-site emergency preparedness provides reasonable assurance that adequate protective measures can and will be taken in the event of a radiological emergency. Among the standards that must be met by these plans are provisions for two emergency planning zones (EPZs). A plume exposure pathway EPZ (requiring preplanned evacuation procedures) of about 16 km (10 miles) in radius and an ingestion exposure pathway EPZ (where interdiction of foodstuffs is planned) of about 80 km (50 miles) in radius are required. Other standards include appropriate ranges of protective actions for each of these zones; provisions for dissemination to the public of basic emergency planning information; provisions for rapid notification of the public during a serious reactor emergency; and methods, systems, and equipment for assessing and monitoring actual or potential off-site consequences in the event of a radiological emergency condition.

5.3 Accident Risk and Impact Assessment for License Renewal Period

[ Prev | Next | Table of Contents ]

5.3.1 Regulatory Interface Between License Renewal and Accident Impacts

In general, the safety and environmental issues associated with license renewal fall into two general categories: (1) effects of aging on the physical plant itself and the associated impact of these effects on accident frequency and radiological releases and (2) effects on accident consequences due to the changing environment in which the plant exists.

The potential effects of age on the physical plant are addressed through engineering and research programs. Potential deterioration of plant components and structures due to physical processes such as corrosion, erosion, mechanical wear, and embrittlement could result in the increased likelihood of component or structure failure. These increased failure rates, in turn, could lead to a higher frequency of accidents or more severe consequences. Therefore, control of these effects is necessary if the plant is to be assured of continuing to operate in a safe manner. As a result, NRC has developed the license renewal process within the context of the aging issue. The process will provide assurance that these age-related impacts are controlled and adequate protection of the health and safety of the public is maintained during the 20-year license renewal period. To supplement the control that the normal regulatory process has over the aging effects on the plant, the NRC requires that the renewal applicant specifically address the issue of age-related degradation by identifying, in an integrated plant assessment process, those structures and components which are susceptible to age-related degradation and whose functions are necessary to ensure that the facility's licensing basis is maintained. The licensee will further be required to demonstrate that the effects of aging will be adequately managed so that the intended functions of these structures and components will be maintained for the period of extended operation. The combined impact of these actions should be to provide high confidence that significant incremental increases in public risk will not result from aging effects related to the plant. A comprehensive discussion of the NRC rule, programs, and activities to provide this assurance is found in 55 FR 29043, dated July 17, 1990.

In assessing the impact on the environment from postulated accidents during the license renewal period, the assumption has been made that the license renewal process will ensure that aging effects on the plant are controlled and that the probability of any radioactive releases from accidents will not increase over the license renewal period.

The effects due to the changing environment around the plant during the license renewal period are less predictable, are generally not subject to regulatory controls, and could cause an increase in public risk as the plant continues to operate. These effects manifest themselves primarily by increasing the consequences of a given accident. For example, as the general population in the vicinity of a nuclear plant increases, the number of persons that could be affected by an accidental release also increases. Because these impacts are "noncontrollable," their potential for increasing risk as well as the magnitude of any such increase in risk must be specifically examined. Such an examination is presented in the following sections, which will discuss and assess the potential adverse impacts to the environment from postulated accidents during the license renewal period.

5.3.2 Design-Basis Accidents

Two classes of accidents are evaluated. The first class of accidents, design-basis accidents, is discussed in this section. The second, severe accidents, is discussed in Section 5.3.3. As noted previously, design-basis accidents are those that both the licensee and the NRC staff evaluate to ensure that the plant meets acceptable design and performance criteria. The environmental impacts of design-basis accidents are evaluated during the initial license process, and the ability of the plant to accommodate these accidents is demonstrated to be acceptable before issuance of the license. The results of these evaluations are found in license documentation such as FESs and FSARs. The licensee is required to maintain these acceptable design and performance criteria throughout the life of the plant, including any extended-life operation. The consequences for these events are evaluated for the hypothetical maximum exposed individual; as such, changes in the plant environment will not affect these evaluations. Because of the requirements that continuous acceptability of the consequences and aging management programs be in effect for license renewal, the environmental impacts as calculated for design-basis accidents should not differ significantly from initial licensing assessments over the life of the plant, including the license renewal period. In addition, any refurbishment necessary to prepare for license renewal would be done in a fashion consistent with the limits set for design-basis accidents and would not alter their consequences. Accordingly, the design of the plant relative to design-basis accidents during the extended license period is considered to remain acceptable and the environmental impacts of those accidents will not be examined further in this section.

5.3.3 Probabilistic Assessment of Severe Accidents

This section presents the staff's assessment of impacts of severe accidents during the license renewal period. Methodologies were developed to evaluate each of the dose pathways by which a severe accident may result in adverse environmental impacts and to estimate off-site costs of severe accidents.

Three pathways for release of radioactive material to the environment from severe accidents are evaluated in this section for each plant site for the license renewal period. These pathways are (1) air, (2) air to surface water, and (3) groundwater to surface water. For most plants, the air pathway represents the most likely pathway for significant dose to the public. The air to surface water pathway is significant for only a few sites that are close to large but confined bodies of water. The third pathway represents a less significant potential for dose because of reduction in radioactivity due to retention in the ground and greater flexibility and time to implement interdiction measures. Separate methodologies were developed for quantifying the potential consequences resulting from each pathway for each site. Economic impacts from severe accidents during the license renewal period are also described in this chapter.

Section 5.3.3.1 reviews the existing analyses available for use in this Generic Environmental Impact Statement (GEIS) study; Section 5.3.3.2 examines the effects of atmospheric releases, including vegetation pathways; Section 5.3.3.3 examines the effects from direct fallout onto open bodies of water; Section 5.3.3.4 reviews effects from releases to groundwater; and Section 5.3.3.5 examines the economic impacts of severe accidents. All analyses will adhere to a process that uses the results of existing analyses and site-specific information to conservatively predict the environmental impacts of severe accidents for each plant during the license renewal period.

5.3.3.1 Review of Existing Impact Assessments

The public risk due to nuclear power accidents has a range of values. The staff believes that current regulatory practices ensure that the basic statutory requirement, adequate protection of the public, is met (51 FR 28044). These risk estimates are representative of the magnitude of risk associated with current regulatory practices. Since the early 1970s, there have been increasing efforts to determine severe accident risks more precisely and on a plant-specific basis. The first comprehensive plant-specific examination of risk was the Reactor Safety Study (RSS), published in 1975 (NUREG 75/014, formerly WASH-1400). The risk values calculated in RSS were later updated in NUREG-0773 and used in NRC FESs published after 1980. Later, more complex and more intensive plant-specific risk studies were developed, both by NRC and the industry. The most recent NRC studies of severe accident consequences are found in the NUREG-1150 analyses. To date, about 40 percent of the 118 operating plants and plants under construction have had some level of plant-specific risk analysis reviewed by NRC. This body of knowledge was used in the prediction of environmental impacts of severe accidents for all plants. Further details of these studies are provided in the following paragraphs.

FES reports since 1980 (Table 5.1) have provided assessments of impacts resulting from postulated severe accidents. Both the frequency and magnitude of the source terms ("source term" is a descriptive name for the releases of radioactive material to the environment under various accident conditions) for such assessments were usually taken from the rebaselined RSS (NUREG-0773). [These values are the result of updating the original RSS (NUREG-75/014, formerly WASH-1400) results using improved methods relative to the original WASH-1400 methodology.] Table 5.2 provides more information on the source-term data used in the FES analyses. These rebaselined source terms were then used with site-specific meteorological and demographic data to calculate off-site risk. A separate rebaselined set of source terms was provided for each of the two types of reactor designs, BWRs and PWRs. In most FES assessments, these same sets of data, without change, were used to evaluate off-site risks. Accordingly, the risk values provided in these FESs are based upon the plant designs analyzed in WASH-1400. As such, they do not represent plant-specific analyses for the FES plants but are sufficient to illustrate the general magnitude and types of risks that may occur from reactor accidents. There were some exceptions in that several studies included some further modification of the rebaselined RSS frequency estimates to better account for plant-specific design differences from the RSS plants. When available, other studies used plant-specific information on severe accident risks [such as probabilistic risk assessments (PRAs)]. Once the source-term data were established, all plants used the Calculation of Reactor Accident Consequences (CRAC) code to determine environmental consequences. Site-specific information regarding meteorology, population, and evacuation was used. Assumptions regarding exposure pathway, exposure limits, and plume behavior remained largely unchanged for all analyses.

The NUREG-1150 study is an NRC-sponsored risk examination of five U.S. nuclear power plants.³ These analyses used state-of-the-art technology in evaluation of source-term release frequency, source-term characteristics, and consequence evaluation. Efforts were made to explore uncertainties in accident frequency, containment behavior, and radioactive material release and transport so that from this distribution of results, mean values of risk could be determined. Source terms and frequencies specific to the plant were determined. Advanced computer codes were used. For example, the MELCOR Accident Consequence Code System (MACCS) computer code for consequence evaluation was used instead of CRAC.

^aNSSS = nuclear steam supply system, W = Westinghouse, GE = General Electric, CE = Combustion Engineering.
^bIndian Point 2 and 3 consequence information was obtained from Atomic Safety and Licensing Board testimony.
^cWNP-2 = Washington Nuclear Project 2.

^aWashington Nuclear Project 2.

The industry-sponsored risk assessments (e.g., Oconee 3, Seabrook, and Millstone 3) are similar in that efforts are made to reduce the degree of conservatism and to use the best information available. For these studies, source-term levels and frequencies specific to the plant are calculated.

Finally, studies exist that provide a detailed assessment of the risk due to specific types of accidents. For example, two such studies are NUREG-0440, which is a generic study of the radiological risks that could result from a severe accident that releases significant contamination into the groundwater, and NUREG-0769 (Addendum 1), which estimates the risks from direct contamination of the Great Lakes due to fallout from a severe accident at the Enrico Fermi 2 power plant. These two as well as other specific risk studies are used in this GEIS to provide a projection of risk during the license renewal period.

Severe accidents initiated by external phenomena such as tornadoes, floods, earthquakes, fires, and sabotage have not traditionally been discussed in quantitative terms in FESs. With the exception of sabotage, the NRC staff has, however, reviewed or performed detailed probabilistic assessments of external events for Zion Units 1 and 2, Indian Point Units 2 and 3, Limerick Units 1 and 2, Surry Unit 1, Peach Bottom Unit 2, and Millstone Unit 3. In most cases, the external event risks were determined to be comparable to internal event risks. However, for Zion and Limerick, the licensee's PRAs indicated that external events could be significant contributors to risk. For the Indian Point units, NRC staff evaluations also indicated that external events could significantly contribute to severe accident risk. The most recent NRC analysis of external events has been the NUREG-1150 external events assessment for Surry Unit 1 and Peach Bottom Unit 1. This analysis examined a broad range of external events and found that they could range from negligible to significant contributors to risk when compared with internal initiators. It should be noted, however, that in cases where external event risk was shown to be a significant contributor to the overall risk, the majority of the estimated risk arose from large beyond design basis earthquakes; but in all cases, the total risk (internal and external) is still small.

NRC's earthquake design standards have been conservatively developed to ensure protection of the public health and safety from earthquakes whose magnitudes are well above the most likely earthquake magnitude when considering the collective earthquakes history for specific plant sites in the United States. Therefore, earthquakes exceeding NRC seismic design standards are extremely unlikely. However, in the unlikely event of such an earthquake, there would be substantial damage to older residential structures, commercial structures, and high-hazard facilities such as dams whose seismic design standards are below nuclear seismic design standards. The societal impact due to the non-nuclear losses alone from an earthquake larger than the design basis of a nuclear plant, including property damage, injuries, and fatalities, would be major. The technology for assessing losses from such large earthquakes is a developing one, and there are several ongoing studies of this technology, including work at the United States Geological Survey. Presently there is no agreed-upon method for performing such assessments, although a recent report of the National Academy of Sciences suggests some broad guidelines (NAS 1989). The NRC has not developed a method for assessing the societal losses from large earthquakes such that the reactor contribution to accident consequences can be quantitatively compared with the non-nuclear losses. However, as supported by at least one study (Lee et al. 1979), the commission expects that the reactor accident contribution to the losses from large beyond design basis earthquakes would be small relative to the non-nuclear losses. While this in itself does not mean the reactor consequences from such an earthquake would be small, the commission concludes that even with potentially high consequences from a beyond design basis earthquake, the extremely low probability of such earthquake yields a small risk from beyond design basis earthquakes at existing nuclear power plants.

With regard to sabotage, quantitative estimates of risk from sabotage are not made in external event analyses because such estimates are beyond the current state of the art for performing risk assessments. The commission has long used deterministic criteria to establish a set of regulatory requirements for the physical protection of nuclear power plants from the threat of sabotage, 10 CFR Part 73, "Physical Protection of Plants and Materials", delineates these regulatory requirements. In addition, as a result of the World Trade Center bombing, the Commission amended 10 CFR Part 73 to provide protection against malevolent use of vehicles, including land vehicle bombs. This amendment requires licenses to establish vehicle control measures, including vehicle barrier systems to protect against vehicular sabotage. The regulatory requirements under 10 CFR part 73 provide reasonable assurance that the risk from sabotage is small. Although the threat of sabotage events cannot be accurately quantified, the commission believes that acts of sabotage are not reasonably expected. Nonetheless, if such events were to occur, the commission would expect that resultant core damage and radiological releases would be no worse than those expected from internally initiated events.

Based on the above, the commission concludes that the risk from sabotage and beyond design basis earthquakes at existing nuclear power plants is small and additionally, that the risks form other external events, are adequately addressed by a generic consideration of internally initiated severe accidents.

Although external events are not discussed in further detail in this chapter, it should be noted that the NRC is continuing to evaluate ways to reduce the risk from nuclear power plants from external events. For example, each licensee is performing an individual plant examination to look for plant vulnerabilities to internally and externally initiated events and considering potential improvements to reduce the frequency or consequences of such events. Additionally, as discussed in Section 5.4.1.2, as part of the review of individual license renewal applications, a site-specific consideration of alternatives to mitigate severe accidents will be performed in order to determine if improvements to further reduce severe accident risk or consequences are warranted.

5.3.3.2 Dose and Adverse Health Effects from Atmospheric Releases

5.3.3.2.1 Methodology for Predicting Future Risk Summary of methodology

The assessment of environmental impacts due to the atmospheric release pathway are described in this section. This pathway includes the exposure of individuals directly from the passage of the cloud of radioactive material released from an accident and from material deposited on the ground, as well as the longer-term effects from other terrestrial pathways such as the ingestion of crops. Doses and the resulting health effects (early and latent fatalities) will be estimated for the middle year of relicense (MYR) population. The MYR is the estimated midpoint of the renewal period for a given plant rounded upward to the next year of available population data. Predictions of MYR risk were generated by taking the results of existing risk calculations (i.e., plant-specific estimates of early fatalities, latent fatalities, and dose) and regressing those values against a composite site-specific variable called the exposure index (EI). EI is a function of population surrounding the plant weighted by the site-specific wind direction frequency and, thus, is a site-specific parameter. Because meteorological patterns, including wind direction frequency, tend to remain constant over time, EI changes as populations change or become redistributed.

A straight-line regression of the total risk values (taken from FES analyses) for each plant listed in Table 5.1 versus the EI for that plant (at the date assumed in the FES analyses) was calculated. Average and 95 percent upper confidence bound values of total risk were estimated. Risks for individual plants for their license renewal period were then estimated from the upper confidence bound values based on MYR population data converted to MYR EI. In the prediction of risk using EI (discussed in the preceding paragraph), the assumption was made that future plant risk is primarily a function of population and wind direction. Secondary factors--such as terrain, rainfall, and wind stability--also have some effect on risk, but their impact was judged to be much smaller than the effects of population and wind direction.

Selection of appropriate existing analyses for use in regression

Currently, 118 nuclear plants are in operation or under active construction in the United States. These 118 plants represent 72 sites for the evaluation of air pathway consequences bsp;sites are used for the other two pathway evaluations).⁴ As noted previously, only a portion of these nuclear plants have severe accident analyses available for review.

The data selected for use in this GEIS analysis were taken from the FESs published since 1981. As discussed previously, these FES analyses are based upon source terms resulting from the RSS (NUREG-75/014, formerly WASH-1400) rebaselined in NUREG-0773. As such, these source terms (and the resulting risk and environmental impacts calculated using them) reflect the plant designs used in WASH-1400. However, this approach is considered conservative because the source terms developed in WASH-1400 generally reflect an "as found" (late 1970s) and, as such, do not reflect the improvements that have been made in nuclear industry plant design and operations since the early 1980s. Accordingly, the use of WASH-1400 source terms in the FESs may, in many cases, tend to overestimate the actual environmental consequences and risks.

Since the RSS study was completed, the NRC has implemented several industry-wide risk-reduction programs. These programs, such as station blackout (10 CFR 50.63), anticipated transient without scram (10 CFR 50.62), resolution of other generic safety issues, improvements resulting from the extensive reviews of the accident at Three Mile Island (NUREG-0737), and the individual plant examination and containment performance improvement programs, have served to lower the overall average values of nuclear plant risk relative to their values prior to the changes. Because the programs are implemented on an industry-wide basis, risk values should be smaller at all plants. No quantification of the overall risk reduction has been performed, but it is believed that the risk reduction is significant. As a result of the changes, the staff believes that the spectrum of risk for the entire nuclear industry shifted downward to lower overall risk values, and the average total risk for all nuclear plants is smaller than the risk estimated in the original RSS analyses. Thus, RSS risk estimates are more representative of the upper end of the total nuclear plant risk spectrum than the actual current risks.

The preceding discussion shows that the use of the FES risk values provides reasonable estimates of the actual average risk of the general nuclear plant population and that the use of the FES values in this analysis results in appropriate risk values in the GEIS. Where there were choices of methodology and the best method was not obvious, the staff chose the method that would lead to higher predicted values. The use of the 95 percent upper prediction confidence bounds from the regression in this analysis (discussed later) provides even greater assurance that the GEIS does not underestimate potential future environmental impacts.

As for use of detailed PRA analyses in the GEIS, particularly the NUREG-1150 studies, the plants represented in these detailed PRAs have had the benefit of considerable risk reduction feedback and improvement; consequently, their predicted risk values are not considered to be representative of the absolute values of the general plant population risk. However, these studies do provide significant risk information on the relative risks to the public as a function of distance from the plant. Because of the much more advanced computational tools available during the NUREG-1150 studies (which could better model secondary effects such as rainfall pattern), as well as more than 10 years of additional knowledge about severe accidents, the information on the distribution of risk at a specific plant, as estimated by the NUREG-1150 reports, is considered more realistic and representative of the actual environmental impacts due to the air pathway for severe accidents. The GEIS uses this relative risk information in its analysis process.

Enveloping of all plants with FES analyses

Many factors could potentially increase the consequences to the general public resulting from a severe-accident release. A comprehensive listing and description of factors that influence consequences are provided in the PRA Procedures Guide (NUREG/CR-2300). The purpose of this section is to use, to the extent possible, the available severe accident results (Table 5.3), in conjunction with those factors that are important to risk and that change with time to estimate the consequences of nuclear plant accidents for all plants for a time period that exceeds the time frame of existing analyses. This estimation process was completed by predicting increases or decreases in consequences as the plant lifetime is extended past the normal license period by considering the projected changes in the risk factors. The primary assumption in this analysis is that regulatory controls will ensure that the physical plant condition (i.e., the predicted probability of and radioactive releases from an accident) will be maintained at a constant level during the renewal period; therefore, the frequency and magnitude of a release will remain relatively constant. In other words, significant changes in consequences will result only from changes in the plant's external environment. The most logical approach, then, would be to incorporate the most significant environmental factors into calculations of consequences for subsequent correlation with existing analyses (which use the consequence computer codes). The two parameters selected for this analysis are population and wind direction, as discussed in the following paragraphs.

Table 5.3 Comparison of general site characteristics. Italics indicate that the final environmental statement contains severe accident evaluations

^aMYR = Middle year of license renewal period rounded up to the next year for which population forecasts were available.
^b50 miles = 80 km.
^cAnnual average in inches.
^dTerrain categories: 1 = coastal plain, 2 = central lowlands, 3 = plateaus, 4 = parallel valleys and ridges, 5 = rolling hills to high mountains, 6 = steep mountains.
^eSevere accident information obtained from Atomic Safety and Licensing Board testimony and not from the final environmental statement.
^fWNP-2 = Washington Nuclear Project 2.

Many factors can affect the amount of radiation to which the public is exposed. The magnitude of impact varies for any individual factor and generally is very specific to a particular plant or site. If the FES risk results are to be used to predict future risk values for all plants, it should be demonstrated that the FES plants provide a reasonable envelope of the more significant risk factors for all plants. Such factors include population density, meteorology, evacuation, and interdiction. Studies have shown that some factors have a greater degree of influence than others; for example, population has a very strong influence over risk (NUREG/CR-2239, NUREG-1150). Evacuation can have a significant influence on early fatality risk but a much more limited impact on latent fatality risk. Interdiction primarily reduces latent fatality risk. While particular aspects of meteorology, such as rainfall, can have a significant impact on peak risk values, mean health effect values are relatively insensitive to meteorology. When the basic reasons for the risk influence of each factor are examined, these factors can generally be reduced to three issues: (1) the number of people exposed to the severe accident release, (2) the likelihood that any given individual receives an exposure, and (3) the amount of radiation the individual receives. Consequently, site population (which reflects the number of people potentially at risk to severe accident exposure) and wind direction frequency (which reflects the likelihood of exposure) have been chosen as the primary factors affecting risks.

Although there are other secondary factors (e.g., source term and dose response relationship) that can influence risk and were not specifically analyzed on a plant-specific basis in this GEIS, these factors were not ignored as their impact is included in the FES analyses whose results are the basis for the GEIS analyses. Consequently, their effects are indirectly considered in the prediction of future risks and are reflected within the uncertainty bounds generated by the regression of the FES risk values. To ensure that the existing FES analyses cover a range of secondary factors representative of the total population of plants, the more significant secondary factors were examined as discussed below. The secondary factors examined are as follows:

• average annual precipitation,
• residential population within a 50-mile (80-km( radius of the plant,
• population [50 miles (90 km)] in highest frequency wind direction,
• general terrain, and
• emergency planning.

Average annual precipitation. After an atmospheric release caused by a severe accident, the fallout rate of the released radionuclides is generally the result of gravitational settling and, consequently, is not a rapid process. This slow fallout allows a given release to be suspended for sufficient time to allow for some radioactive decay of the shorter-lived radionuclides, resulting in lower individual doses to the public. In addition, releases are distributed over a wide area, resulting in relatively low individual doses (although the overall total population dose is not greatly affected). However, precipitation counteracts both of these effects by washing the radionuclides out of the atmosphere and not allowing time for extensive dispersion or decay. Thus, plant sites with higher levels of annual precipitation may indicate higher levels of risk for those measures that are based on individual doses.

Residential population within a 50-mile radius of the plant. This factor is a rather understandable selection in that plant sites with larger resident populations will have a larger number of persons at risk for a given severe accident release. Population projections were made based on the 1980 census data and projected growth (decline) factors derived from the U.S. Bureau of Census evaluations. A radius of 50 miles was selected for comparison purposes because existing analyses indicate a large majority (although not all) of early health effects from a severe accident release occur within 50 miles of the plant site.

Population (50 miles) in highest-frequency wind direction. This factor highlights a "higher risk" sector of the overall population around a specific plant site. The sector is 22.5° and the population is 0 to 50 miles from the plant in that sector. Higher populations combined with higher frequency of wind in that direction may indicate higher risks in that sector.

General terrain. This factor is chosen because the dispersion behavior of the plume may be influenced by the general terrain surrounding the plant (e.g., plains versus mountains). Six terrain classifications were selected as described in footnote c to Table 5.3.

Table 5.3 shows the values for these four factors for all nuclear plant sites. As can be seen, the existing severe accident analysis as provided in those FESs containing a severe accident evaluation provides a reasonable envelope for precipitation (rainfall and snowfall), 50-mile population, and 50-mile population in the direction of highest wind frequency. All six terrain classifications are also covered by referenced FES analyses. From review of these data, it is concluded that the FES plants sufficiently envelop these factors. Likewise, any plant risk projections that are developed from the FES severe accident results will reasonably account for secondary effects from these factors if the upper confidence bounding values from the projections are used to estimate the risk from atmospheric releases for plants during their license renewal period.

Emergency planning. Even in the event of a release of radioactive material from a plant, protective actions can be taken to move or shelter members of the public in the projected path of the radioactive cloud. The success of these actions in preventing exposure of members of the public to the radioactive material is dependent upon the warning time available prior to the release and the time it takes to carry out the protective actions. In general, this latter item (the time to carry out the protective action) is mostly influenced by the size of the population around the plant. Each FES that addresses severe accidents considers the effects of site-specific emergency planning in calculating exposures and risks to the public. Since the FES plants include sites with populations that reasonably cover the range of populations at all 74 sites, a range of emergency planning is considered in the data used for the predictions of early and latent fatalities during the license renewal period. Thus, this GEIS analysis should reasonably account for the effects of emergency planning.

Projections of estimates of risk

Detailed severe accident consequence (early and latent fatalities and total dose) evaluations are not available for all plants. Therefore, a predictor for these consequences was developed using correlations based upon the calculated results from the existing FES severe accident analyses. This predictor was then used to infer the future consequence level of all individual nuclear plants. Correlations were developed using two environmental parameters that are available for all plants. This correlation process is described below.

Discussion of exposure index

Population, which changes over time, defines the number of people within a given distance from the plant. Wind direction, which is assumed not to change from year to year, helps determine what proportion of the population is at risk in a given direction, because radionuclides are carried by the wind. Therefore, an EI relationship was developed by multiplying the wind direction frequency (fraction of the time per year) for each of 16 (22.5° ) compass sectors times the population in that sector for a given distance from the plant and summing all products. An example calculation for an EI value for 1990 at 10 miles (16 km) is shown in Table 5.4. The EI value, as calculated in Table 5.4, can be considered to be the expected population at risk for the year 1990 out to a distance of 10 miles from the nuclear power plant. Population varies with population growth and movement, and with the distance from any given plant. As the population changes for that plant, the EI also changes (the larger the EI, the larger the number of people at risk). Thus, EI is proportional to risk and an EI for a site for a future year can be used to predict the risk to the population around that site in that future year.

Regression of FES values

Several relationships of EI versus risk were developed by regressing total early fatality, normalized total latent fatality, and normalized total dose values on various EI values for the FES plants (see Appendix G). The EI values at 10 miles were found to best correlate with early fatalities, which is to be expected because, in the FES analyses, early fatalities tend to be clustered close to the plant. The EI values at 150 miles (241 km) were found to best correlate with latent fatalities and total dose. This finding is to be expected because the magnitudes of these risk values are largely influenced by the exposure of large populations around the plant.

^a10 miles = 16 km.

Note: To calculate EI value: A x B = C; EI = sum of C.

Because the magnitude of the source term is generally proportional to plant power for a given accident sequence, the FES estimates for total latent fatalities used in the latent fatality regression were first normalized to 1000 MW(t) to minimize the regression variance due to the differing plant sizes. A linear dose response function is used in the FES analyses, and because of the assumptions of downwind and crosswind spread, radioactive material is predicted to be widely dispersed. Thus, the larger the amount of radioactive material released, the larger the predicted latent fatality level (slightly reduced from strict linearity by the interdiction assumptions). Similar logic is applicable to normalization of total dose. Normalization was not used for early fatalities because of the highly nonlinear dose response function used in the FES analyses and the use of a threshold effect (that is, there is a dose level below which no early fatality is predicted). Nonetheless, early fatalities are also highly influenced by the amount of radioactive material released (i.e., plant size), and to help ensure that early fatality data from smaller plants do not distort the regression results for the larger plants, only the early fatality data for plants greater than 3025 MW(t) were used in the regression of early fatalities (Table 5.5, footnote f). The inability to correct fully for the effects of plant size and the dose-early fatality relationship leads to a higher dispersion in the regression estimates, which influences the upper confidence bound (UCB) as will be seen in subsequent sections.

Also, in several of the FES documents, two sets of early fatality values were provided, one set which assumed minimal medical support was available to aid the exposed population and a second set which assumed normal and expected levels of medical support were available. The regression used those early fatality values associated with expected medical support levels. The assumption there would be only minimal or no medical support after an accident was considered to be unrealistic. A detailed discussion of the regression analyses and UCB is provided in Appendix G.

5.3.3.2.2 Results

The data in Table 5.5 summarize the information for 28 nuclear plant sites that were used to develop the relationship between EI and consequences of severe accidents analysis for both PWRs and BWRs. Because of fundamental design differences between PWRs and BWRs, separate regression analyses were performed for each to better account for the BWR-PWR differences in plant failure modes and source terms. Accordingly, the PWR regression was used to determine the best fit relationship for PWR risk values and the BWR regression was used to determine the best fit relationship for FES BWR risk values. As can be seen in Figures 5.2-5.7, two lines (representing average and UCB values) result from the regression analyses for total early fatalities, total latent fatalities, and total dose. The 95 percent UCB (dashed line) was developed based on the scatter in the data. Two points need to be made about the UCB. First, two UCB values were calculated: one value assuming that the data points (i.e., early and latent fatalities and population dose) had a normal distribution about some mean and the second value assuming that the data points did not have a normal distribution about the mean. The larger of the two UCB values was then used in making plant risk projections. The second point to be noted is that because of the small number of data used in the regressions (18 PWR data points and 10 BWR data points), the scatter in the data (expressed as residuals) for all 28 data points was used in determining the UCB for both the PWR and BWR regressions.

Table 5.5 Information used for regression analyses for expected early, latent, and total dose at 28 nuclear plant sites for the license renewal period

Normalized values are obtained by converting nonnormalized values to the equivalent of a 1000-MW(t) plant