-------
(less than 30 pCi/L), cesium-137 (less than or equal to 22 pCi/L), and Co-60 (less than or
equal to 20 pCi/L) (JAC88). As with tritium, these radionuclides are also moving east and
southeast following the movement of the ground-water toward the Columbia River. Except
for technetium and iodine, the ground-water plumes associated with these radionuclides are
not as extensively dispersed as the one due to tritium. The reported concentrations for
technetium, iodine, and strontium in several wells exceed the EPA DWS.
The presence of alpha-emitting radionuclides were detected in several wells located in
the 100, 200, and 300-Areas. The total gross-alpha radioactivity is thought to be due to
uranium since plutonium concentrations were noted to be below the limit of detection (about
0.1 pCi/L) (JAC88). The highest concentrations were noted to occur in the 200-Areas
(West) while much lower concentrations were detected in the eastern sector of the 200-Areas.
The peak concentrations in the 200-Areas (West) were reported to range from 100 to 10,500
pCi/L. Uranium has been also been noted in the vicinity and downgradient of the fuel
fabrication facilities (300-Areas) and near inactive waste disposal sites. The average uranium
concentrations were reported to range from 2 to 310 pCi/L, with peak concentrations ranging
from 100 to nearly 12,000 pCi/L. Other locations on the Hanford Site are characterized by
uranium concentrations ranging from non-detectable levels (0.5 pCi/L) to less than 100
pCi/L. The reported gross-alpha concentrations in several wells exceed the EPA DWS.
The Hanford radiological environmental surveillance program also routinely monitors
other areas, at both on and off-site locations. These locations include three on-site ponds and
one lake, soils at 38 different on and off-site locations, and at upstream and downstream
points on the Columbia River.
Radionuclide concentrations in the three ponds and West Lake have been noted to
vary (DOESTb). The 1987 survey results indicate that tritium is the dominant radionuclide
with peak concentrations ranging from 160 to 9,500 pCi/L. The next predominant
radionuclide is cesium-137 which was reported to range from 1.1 to 50 pCi/L. Strontium-90
was also detected in pond and lake water samples with peak concentrations ranging from 0.4
to 2.8 pCi/L. Total gross beta and alpha water sample activity revealed peak water
concentrations of 490 and 267 pCi/L, respectively, for West Lake. The gross beta and alpha
water activity in the three ponds were typically one to two orders of magnitude lower than
those noted for West Lake.
Soil sample analyses at 15 on-site locations revealed four radionuclides have been
routinely detected in measurable levels. Strontium-90 is known to be present in concentration
varying from 0.02 to 0.38 pCi/g with an average of 0.31 pCi/g. Cesium-137 has been
measured at concentrations varying from 0.01 to 16 pCi/g with an average of 2.0 pCi/g.
Plutonium-239 and 240 have also been measured at concentrations varying from 0.001 to
0.17 pCi/g with an average of 0.027 pCi/g. Finally, uranium was reported at concentrations
ranging from 0.19 to 3.8 pCi/g with an average of 0.58 pCi/g. Typically, the average on-
site measurements are higher than those noted off-site by factors ranging from about 2 to 5.
Analyses of water samples taken downstream in the Columbia River indicate that
radionuclides identified with Hanford Site operations were noted at very low concentrations,
typically well below the applicable drinking water standards (DOESTb). The water samples
4-8
-------
were taken at two different locations, one at the 300-Areas Water Intake and the other at the
Richland Pumphouse located about 3 km downstream from the site boundary. The Richland
Pumphouse is the first downstream point on the river where water is withdrawn for public
use. Water sample analyses revealed that tritium is the most predominant radionuclide with a
reported peak concentration of 200 pCi/L. Other radionuclides were also reported, including
strontium-89 and -90 (0.2 and 0.15 pCi/L, respectively), total uranium (0.61 pCi/L), gross
beta (2.8 pCi/L), and gross alpha (0.79 pCi/L). Other radionuclides, including plutonium-
239 and -240 as well as other fission products, were reported at lower concentrations,
typically ranging from l.OxlO'6 to 4.5xlO'2 pCi/L.
Savannah River Plant — At the Savannah River Plant, high-level wastes, in the form
of alkaline liquids, alkaline sludges, and salt cake, are stored underground in high integrity,
double walled, stainless-steel tanks. By 1993, hot operation of a waste processing facility to
vitrify these wastes into borosilicate glass is scheduled to begin.
Idaho National Engineering Laboratory — In Idaho, high-level wastes, in the form of
acidic liquids, are first stored in underground tanks and later converted to calcine. Stainless
steel tanks housed in concrete vaults are used to store liquid wastes, and stainless steel bins
in concrete vaults are used for the calcine wastes. According to DOE plans, a facility for
inimobilizing newly generated wastes will start operations early in the next century. It will
also process the stored calcine wastes. Evaluations of waste forms and immobilization
processes are being pursued.
4.3.2 Transuranic Wastes and Defense Waste Programs
The research and development (R&D) efforts for defense wastes are divided into three
major categories: 1) the immobilization of high-level wastes, 2) the preparation of transuranic
wastes for shipment to the WIPP facility, and 3) investigations to demonstrate the
performance of the WIPP site. Also these R&D efforts include the development of
technology for in-place immobilization of wastes stored in tanks and evaluation of methods
for immobilizing wastes stored at the Idaho National Engineering Laboratory.
• Waste Isolation Pilot Plant (WIPP)
The DOE is developing the Waste Isolation Pilot Plant (WIPP) in a bedded-salt
formation near Carlsbad, New Mexico to demonstrate the disposal of defense TRU wastes.
The WIPP project was authorized in 1980 by Public Law 96-164 to provide a research and
development facility for demonstrating the safe disposal of transuranic wastes produced by
national defense activities. If testing proves satisfactory, the DOE is expected to open the
site for the permanent disposal of TRU waste.
The WIPP site is in a sparsely populated area on land owned by the Federal
government. The WIPP plant consists of surface facilities (mainly a waste-handling
building), four access shafts, and underground facilities designed to emplace approximately
6.5 million cubic feet of TRU waste in a 100-acre repository. About 12 acres have also been
set aside as an underground test area to conduct experiments and study the behavior and
4-9
-------
performance of the repository. The repository has been excavated in a bedded-salt formation
(the Salado Formation) 2,150 feet beneath the surface.
By mid-1989, the initial major construction activities at the WIPP had been nearly
completed (DOE89a). The surface facilities were essentially complete, and most of the
underground rooms for experimentation and initial waste emplacement had been excavated.
A five-year test phase is planned to develop data for incorporation into the performance
assessment. All of the wastes will be retrievably emplaced should the site be declared
unsuitable at the end of the test period. During this phase, the DOE will monitor the site
and facility as part of the environmental monitoring programs it has been conducting since
1980.
For shipment to the WIPP site, TRU wastes will be contained in Type B shipping
containers (TRUPACT-II) certified by the NRC and carried by truck. The DOE's purpose in
using track transportation for moving waste to the WIPP is to have greater accessibility to
the site and greater control of the transportation system, routes, and speed. The proposed
routes from the waste storage locations use the interstate highway system to the maximum
extent possible (DOE89a). To ensure safe and efficient transport, the DOE will use a
transportation tracking and communication system that will combine navigation, satellite
communication, and computer network technologies to monitor the movements of TRU waste
shipments to the WIPP.
All of the wastes received by the WIPP will have to meet acceptance criteria covering
factors such as waste forms and characteristics, gas generation, immobilization, presence of
toxic and corrosive substances, and thermal power. All incoming packages will be checked
for surface contamination and external radiation exposure rates, and repackaged or repaired if
necessary.
• Greater Confinement Disposal Facility
In 1981, the National Low-Level Waste Management Program and the DOE's Nevada
Operations Office began a project to demonstrate the feasibility of "greater depth" burial in
the alluvial sediments of the Nevada Test Site (REY83, EPA87). The purpose of the project,
named Greater Confineinent Disposal Test (GCDT), was to evaluate the feasibility of
disposing of classified TRU wastes and high specific-activity low-level wastes at intermediate
depths in large-diameter augered holes. These wastes originate from weapon facilities across
the nation. The basic concept involves sinking a shaft 3 meters in diameter and nearly 40
meters deep. The shaft has a capacity of about 1,100 m3. Wastes are then lowered into the
hole and stacked up to depth of about 20 meters from the surface. At this point, the hole is
backfilled with soil all the way up to the surface. The goal of the GCDT program is collect
and analyze data on radionuclide migration and to develop waste handling procedures and
equipment. Plans are also being developed to retrieve these wastes after emplacement, if
necessary.
4.4 POTENTIAL HOST ROCKS FOR GEOLOGIC REPOSITORIES
4-10
-------
Many types of rocks are potentially suitable as host rocks for a repository, depending
on the natural attributes of the rock and the geohydrologic setting. Ideally, the host rock
should be suitable for the construction of the repository and for waste containment, and the
surrounding rock formations should provide adequate isolation (DOESOb). Important natural
attributes include thermal, mechanical, hydraulic, and chemical characteristics that affect the
response of the host rock to heat, the movement and chemistry of ground water, and the
ability to retard the migration of radionuclides. The desirable geohydrologic properties
include low rates of ground-water flow, long path lengths to the accessible environment, and
evidence of long-term stability (DOESOb).
In the United States, early plans for geologic disposal were based on bedded salt and
salt domes. Salt was the rock investigated most extensively as part of a site screening
program. Later, when the DOE began to study Federal lands dedicated to nuclear activities,
several other host rocks came under investigation. They included argillaceous rocks and tuff
in Nevada and basalt in the State of Washington. For the second repository, DOE began to
study crystalline rock formations (DOE86a-g). Other rocks that have been considered are
limestone, sandstone, anhydrite, chalk, and argillaceous rocks like shale (GON85). The
sections that follow briefly review the properties of host rock media most studied in the
United States.
4.4.1 Basalt
Basaltic rock masses are among the strongest of common rock types. In addition,
basalt has moderate thermal conductivity and a high melting temperature, which enable it to
withstand high thermal loads. The basaltic formation that had been investigated in the first
repository program was a thick section, about 950 meters below the surface, near the middle
of the extensive basalt flows of the Columbia Plateau. The basaltic rock in this section
contains openings filled with alteration products (mainly clay minerals), and as a result the
rock mass is of low permeability. On the other hand, the basalts of the Columbia Plateau
commonly have columnar joints or rubbles that are potential channels for water flow.
Water-bearing sedimentary interbeds within the basalt section are also common.
A potential site in basalt is located in the State of Washington. Thick basaltic
formations also occur in the States of Idaho and Oregon.
4.4.2 Bedded Salt and Salt Domes
Of the nine sites identified as potentially acceptable for the first repository, seven
were in salt: four sites in bedded-salt formations and three in salt domes (DOESSa).
Salt is suitable as a host rock because of its structural strength, radiation shielding
capability, high plasticity (which enables fractures to heal or seal themselves at repository
depths), low moisture content, and low permeability. In addition, salt deposits are abundant
in the United States and are relatively easy to mine. Desirable features of many salt basins
are their relatively simple structure and predictable stratigraphy over large areas.
4-11
-------
Although salt deposits are widespread, the salt itself and the other deposits with which
it is often associated (e.g., hydrocarbons or potash) could increase the probability of human
intrusion into a repository. Furthermore, the solubility of salt is greater than that of any
other potential host rock. The potential for this failure mode must be carefully assessed in
analyzing the long-term performance of a repository sited in salt.
4.4.3 Granite and Related Crystalline Rocks
Granite and related crystalline igneous and metamorphic rocks, such as gneiss, are the
most abundant rocks in the upper 10 kilometers of the Earth's continental crust. These rocks
underlie virtually all of the United States; they occur at the surface in stable areas, in the
cores of many mountain ranges, and beneath all of the younger sedimentary rocks. Their
strength, structural and chemical stability, and low porosity make them attractive for geologic
repositories. The water content of these rocks is low and is held mainly in fractures and in
hydrous silicate minerals. The permeability of these rocks is largely dependent on the
presence of fractures, and it is reduced considerably by the closure of fractures, which
occurs at depths in excess of several hundred meters. The depth for a repository is likely to
vary from region to region, depending on how the permeability is affected by the tectonic
history of the region.
Granite as a potential host rock is being investigated in some European countries. In
the United States, the DOE had conducted preliminary investigations of near-surface and
exposed crystalline rock formations in 17 States in a search for sites for the second
repository. However, the Amendments Act directed DOE to terminate site-specific activities
for a second repository and limited such activities only to tuff.
4.4.4 Tuff
Tuff is the dominant component of the voluminous and widespread volcanic strata in
the Basin and Range province of the western United States. The tuff formation at the Yucca
Mountain site, located in southern Nevada, currently being characterized for the first
repository, consists of a sequence of welded and non-welded tuffs.
The site selected as the potential host rock is moderately to densely welded and
devitrified, with a minor number of cavities. This section of the rock formation has high
density, low porosity and water content, good compressive strength, and the ability to
withstand the heat generated by radioactive waste. However, the characteristics that affect the
thermal and mechanical properties of tuff, such as porosity, degree of saturation, and stress
state, are known to vary both laterally and vertically. Consequently, the thermal and
mechanical properties are also likely to vary spatially.
Lying beneath the welded tuff are non-welded tuffs containing zeolite, a hydrous
silicate. These tuffs are characterized by low density, moderate compressive strength,
moderate thermal conductivity, and excellent capability for sorption. The latter is important
to the waste isolation performance of a repository because it would allow these rocks to
significantly retard the migration of radionuclides into the accessible environment.
4-12
-------
4.5 INTERNATIONAL ACTIVITIES
Countries which are committed to use nuclear power or in which nuclear power
already makes up a significant fraction of the total electrical generating capacity are
establishing long-term programs for the safe management and disposal of spent fuel and high-
level radioactive wastes. Such programs include adopting a national strategy, assigning the
technical responsibility for research and development activities to designated agencies,
selection of disposal technologies and geological media, and setting the appropriate
regulatory standards to protect the public and environment.
Typically, the objective of a geological disposal program is to immobilize and isolate
radioactive wastes from the environment for a sufficient period of time under conditions such
that any radionuclide releases from the repository will not result in unacceptable radiological
risks. For illustrative purposes, the disposal programs of eight countries are summarized
below (NEA86, NEA88, SCH88, SCH91, IEAL87). These countries are Canada, the United
Kingdom, France, the Federal Republic of Germany, Belgium, Switzerland, Sweden, and
Japan. A summary of these countries' institutional and regulatory programs is also provided
in Chapter 2, Section 2.3.
4.6.1 Canada
Atomic Energy of Canada Limited (AECL), a Crown corporation reporting to the
Federal Minister for Energy, Mines and Resources, has been assigned the responsibility for
the permanent disposal and isolation of radioactive wastes in Canada. Currently, the
program considers only direct disposal of spent fuel without reprocessing, although the
reprocessing option has not been completely ruled out. Until a repository is available, spent
fuel will initially be stored at each reactor site and, later, possibly at a central facility.
Under a joint agreement, Ontario Hydro (a provincially owned utility) has been mandated to
develop the technologies needed for the interim storage and transportation of spent fuel.
The Canadian disposal concept considers siting a repository in a granitic formation
located in the Canadian Shield. The repository will be located at depths of 500 to 1,000
meters. The spent fuel canisters will be inserted in floor cavities located in excavated
disposal rooms. Once filled, the floor cavities and room excavations will be backfilled and
sealed using engineered barriers. The AECL facility design is already well defined (Concept
Assessment Documentation) and the concept was submitted for public and regulatory review
in 1988. AECL is now preparing a final Environmental Impact Statement which it will
submit to a government-appointed Review Panel by mid-1993, after which public hearings
will be held. The Panel is expected to present findings and recommendations to the
government in early 1995; subsequently, the government will reach a finding on the
acceptability of the concept. AECL estimates that siting, licensing and construction of a
disposal facility will take 25 to 30 years and that the facility could therefore be in operation
by 2025.
4-13
-------
In 1986, AECL established an underground research laboratory (URL) in undisturbed
granitic rock at a depth of 240 meters at Lac du Bonnet, in the Province of Manitoba.
AECL has since deepened the facility to 440 meters. The purpose of the URL is to conduct
large-scale, in-situ experiments in the type of rock envisioned under the Canadian disposal
concept, demonstrating some of the components of the disposal concept (the facility is not a
candidate repository site). AECL is developing methodologies and analytical techniques to
evaluate the geomechanical and geohydrological properties of granitic rock. Construction of
the URL was completed in 1988.
4.6.2 United Kingdom
In the United Kingdom, the responsibility to develop a national strategy for
radioactive waste management lies with the Department of the Environment. The
organizations which produce the wastes have the direct responsibility for their safe
management and funding. An industry consortium, the UK NIREX Ltd, has been established
to develop and operate new low- and intermediate-level radioactive waste disposal facilities in
England.
The United Kingdom's radioactive waste disposal program strategy has postponed the
development of a disposal facility in deep geological media. Rather, the current plans call
for reprocessing of spent fuel, solidification, and surface storage for about 50 years. The
United Kingdom has also adopted a policy of monitoring the results of research activities
being conducted by other countries. Depending on the outcome of research being conducted
abroad, Britain would then identify a high-level waste disposal strategy and repository
program development activities using concepts that best fit British needs.
However, some in-situ research has been conducted by the UK Atomic Energy
Authority and UK NIREX Ltd in heat transfer properties of Cornish granite, statistical
analysis of fracture occurrence, orientation, and aperture in granite, and fractured flow in
Cornwall shale. Other research activities have included geohydrological and geophysical
measurements, geochemistry, radionuclide migration and transport, integrated site
characterization and model validation, and characterization of model parameters and
measurement methods.
4.6.3 France
The French radioactive waste disposal program is based on a closed fuel cycle
involving spent fuel reprocessing, interim storage, and recovery and re-use of plutonium in
breeder and light-water reactors. The nuclear waste program has been entrusted to the
National Radioactive Waste Management Agency (ANDRA), an arm of the French Atomic
Energy Commission (CEA). Since 1969, short-lived radioactive wastes have been emplaced
in engineered near-surface disposal facilities at Centre de la Manche, near Cherbourg on the
English Channel. This facility will reach its design capacity in 1994; a new facility, Centre
de 1'Aube, began operation January 1992 about 100 miles southeast of Paris.
ANDRA was previously investigating four geological media for HLW disposal —
clay, salt, granite, and schist - and had begun investigative work at a site in each medium.
4-14
-------
An underground research laboratory was to be established at one or more of the candidate
sites; if found suitable, one of these was to have been converted to an operating repository to
receive TRU wastes by 2000 and HLW by 2010. However, in light of the serious public
protests at three of the sites under investigation, former Prime Minister Michel Rocard
declared a one-year moratorium on siting activities in February 1990 to allow a reassessment
of the overall French waste management strategy. The Parliamentary Office for the
Assessment of Tehnological Options published a report in January 1991 recommending major
changes to the program, and the Parliament enacted a new Law on Radioactive Wastes on
December 30, 1991.
The 1991 law allows the government to resume site selection efforts for underground
laboratories. A waste "negotiator" will be appointed to discuss proposed investigations with
local and regional officials, and the government is expected to select two sites to host
laboratories. Only research quantities of waste may be emplaced in these laboratories. The
law calls on the government to submit a report to Parliament within 15 years assessing the
results of studies on partitioning and transmutation of actinides, use of test facilities for
retrievable and permanent storage of HLW, and technologies for waste conditioning and
surface storage. In addition, the government report to the Parliament must propose a bill to
authorize an underground waste repository. The law does not establish a schedule for
developing a HLW repository; Parliament will reassess the program based on the results of
the 15-year research phase.
In preparation for the underground laboratory phase, the Institute for Nuclear
Protection and Safety (EPSN) within CEA is independently preparing facilities to evaluate the
long-term safety requirements of a HLW repository, on behalf of the French regulatory
authority. IPSN operates two Methodological and Instrumental Laboratories in a granite
formation near Limoges; it is preparing two similar facilities in clay and schist formations.
4.6.4 Germany
Germany sends spent fuel to foreign reprocessors and will receive vitrified HLW in
return, which it intends to dispose in deep geological formations. The Federal government's
Institute for Radiation Protection (BfS) is responsible for the design, construction and
operation of waste disposal facilities. BfS intends to dispose of non-heat generating low- and
intermediate-level wastes in the Konrad repository, an abandoned iron ore mine in Lower
Saxony in the north central part of unified Germany. Vitrified HLW returned from foreign
reprocessors and other heat-producing wastes will eventually be disposed at the Gorleben
facility, a salt dome also located in Lower Saxony, if the site proves acceptable. Vitrified
waste will be stored at Gorleben and another facility, Ahaus, until the repository is ready for
operation.
A former salt mine at Asse, which served until 1978 as a repository for the disposal
of 125,000 containers of low-level and smaller quantities (1,300 drums) of intermediate-level
radioactive wastes, now serves as an underground research laboratory.
4-15
-------
The Gorleben facility will be situated at depths ranging from 250 to 3,000 meters.
The geology of the site has been widely investigated by exploratory drilling and by
geophysical measurements. In 1986, the construction of an underground research laboratory
was initiated. In 1987, all work was stopped for over one year because of a construction
fatality. Pending completion of the site characterization studies in the mid- 1990s, the
construction of the repository could start at the turn of the century. Once opened, the facility
is planned to be remain operational for as long as 60 years.
4.6.5 Belgium
The Belgian research and development program to establish a radioactive waste
repository was initiated in 1974. A national agency, ONDRAF, was established to take the
responsibility for implementing and managing a multi-year national program. The Belgian
waste management program has included domestic spent fuel reprocessing in the past, but
spent fuel is now either sent to France for reprocessing or stored in reactor pools. Long-
term storage of solidified wastes is planned, followed by construction of a repository located
in a deep clay formation at the Mol-Dessel site.
Investigation of the Mol-Dessel site as a candidate for the Belgian radioactive waste
repository began in 1975. The site is situated in a deep clay formation and is the only
suitable geological medium identified in Belgium. By 1980, a repository conceptual design
was developed for a clay site, and by 1985 an underground research laboratory at Mol-
Dessel (Project HADES) was declared operational. The underground laboratory extends to a
depth of 224 meters, and since 1987 a new experimental gallery has been added to the
original facility. The purpose of the additional gallery is to conduct high specific-activity
disposal experiments and pilot studies. The studies include experiments in corrosion
properties of containers and engineered barriers, geochemistry and radionuclide migration,
backfilling and sealing technology, and near-field effects of heat and radiation on clays.
Based on the outcome of these studies, a larger underground facility will be constructed for a
full-scale demonstration project.
Assuming that the results of investigations at Mol-Dessel are favorable, repository
construction could begin around 2025 and operation around 2030.
4.6.6 Switzerland
The responsibility for establishing radioactive waste disposal facilities lies with the
National Cooperative for the Storage of Radioactive Waste (NAGRA). NAGRA plans to
begin construction of an intermediate-depth repository for low- and intermediate-level wastes
by no later than 2000 and field studies have been conducted at four candidate sites (Bois de
la Glaivaz, Oberbauenstock, Piz Pian Grand and Wellenberg). NAGRA plans to select the
preferred site by mid-to-late 1993 for full characterization.
With regard to developing a deep geologic repository for HLW and TRU waste,
NAGRA has performed extensive field work in crystalline rock formations and will
synthesize all project work based on crystalline rock during 1992-1993. To support the
crystalline rock studies, investigative techniques and equipment have been tested in an
4-16
-------
underground laboratory at the Grimsel Test Site (which is not a candidate repository site). In
addition, a broad survey of sedimentary formations under way since 1988 has resulted in the
selection of clay and a variant of freshwater molasse for further study. NAGRA plans to
choose between clay and molasse by the end of 1993 (NAGRA. gives higher priority to clay)
as the sedimentary medium for further study, and conduct field work in the selected medium
by 1997. NAGRA must submit a program — the Siting Feasibility Project - for government
approval by 2000 demonstrating the feasibility of siting a repository in one or more of the
crystalline or sedimentary media under consideration, and intends to choose by 1997 which
medium or media to present in that program. Commissioning of a repository will not occur
before 2020 to allow a 40-year spent fuel/HLW cooling period. Participation in any
international repository projects that may develop is also under consideration.
4.6.7 Sweden
Following a 1980 national referendum, the Swedish Parliament decided to phase out
nuclear power plants by the year 2010. Consequently, Swedish utilities sold their contract
rights to foreign reprocessing services. The Swedish Nuclear Fuel and Waste Management
Company (8KB) began operating a centralized spent fuel storage facility (CLAB) in 1985 that
will eventually hold all Swedish spent fuel (about 8,000 metric tons) for about 40 years. The
facility is situated in an underground granite cavern at a depth of 30 meters, near an existing
nuclear power plant (Oskarshamn). A repository for short-lived low- and intermediate-level
wastes, SFR, began operating in 1988 near the Forsmark nuclear power plant.
SKB's reference disposal concept for spent fuel is to encapsulate it in high-integrity
copper canisters and emplace the canisters in a repository built in crystalline rock at a depth
of about 500 m, backfilling the deposition holes with highly-compacted bentonite and the
tunnels and shafts with a mixture of sand and bentonite. SKB is evaluating alternative
concepts such as deep boreholes and tunnel emplacement, as well as alternative canister
designs. Three candidate repository sites are to be identified in 1993, followed by
preliminary characterization of the sites, to be completed around 1996. Subsequently, two
sites would be characterized in detail, beginning in 1997 and lasting about six years. SKB
would file a license application for one site in 2003. Construction is anticipated to begin
around 2010 and operation around 2020.
The international OECD/Nuclear Energy Agency conducted an international research
project in an underground research laboratory at Sweden's Stripa mine from 1980 to 1991.
SKB has decided to build a second laboratory under the island of Aspo, 2 km north of
Oskarshamn, as a means of preparing for site selection, site characterization and licensing the
spent fuel repository. Construction of the Aspo Hard Rock Laboratory began in October
1990; the facility is scheduled to begin operation by the end of 1994 at a depth of 500 m.
4.6.8 Japan
Under 1985 plans for waste management published by the Atomic Energy
Commission and an R&D plan announced by the Science and Technology Agency in 1986,
the Power Reactor and Nuclear Fuel Development Corporation (PNC) has the lead
responsibility for HLW disposal R&D, while the Japan Atomic Energy Research Institute
4-17
-------
(JAERI) and others share in the R&D work. The current waste management strategy
includes spent fuel reprocessing using domestic and foreign facilities, on-site spent fuel
storage, waste solidification followed by long term storage (30-50 years), and eventual
disposal in a suitable deep geological formation.
The site selection process for an HLW repository consists of four phases: 1) selection
of effective formations (completed in 1984); 2) selection of a candidate disposal site (now
underway); 3) demonstration of the disposal technology at the candidate site; and 4)
construction, operation and closure of the disposal facility. The conclusion of the first phase
was that HLW disposal should be possible in any geologic formation excluding
unconsolidated media (e.g. soil and sand). The site selection phase currently in progress
emphasizes generic R&D at sites that are not candidates to host the repository. The
demonstration phase is expected to begin at a candidate repository site by 1995. Because of
geological heterogeneities in Japan, geological characterization is expected to be difficult,
causing uncertainties in predicting the performance of natural barriers. Thus, Japan is
assigning a major role to the engineered barrier system, while defining a small number of
critical natural characteristics for the site which are expected to be achievable in various
geological settings.
PNC operates an underground test facility in the Tono Uranium Mine in central
Japan, in both sedimentary and crystalline rock environments. Major experiments in the
Tono Mine include a groundwater flow investigation, studies on the effects of excavation on
the mechanical and hydraulic behavior of the repository, natural analogue studies and
evaluations of the chemical durability of simulated waste glasses and the corrosion rates of
candidate overpack materials. In addition, PNC is conducting tests in the Kamaishi iron ore
mine in northern Honshu. Major investigations at Kamaishi have included detailed fracture
mapping, cross-hole hydraulic and geophysical testing, drift excavation-effect studies and in-
situ stress measurements, single-fracture flow tests and observations of seismic activity.
Furthermore, PNC is conducting analogue studies on the stability of gkss, iron, concrete and
bentonite in natural settings.
4-18
-------
Chapter 4 References
AH92 P.-E. Ahlstrom, "Swedish High-Level Radioactive Waste Management Issues,
Third International Conference on High Level Radioactive Waste Management,
Las Vegas, Nevada, April 12-16, 1992.
AL92 C.J. Allan et al, "Canadian High-Level Radioactive Waste Management
System Issues," Third International Conference on High Level Radioactive
Waste Management, Las Vegas, Nevada, April 12-16, 1992.
DOESOa U.S. Department of Energy, Final Environmental Impact Statement,
Management of Commercially Generated Radioactive Waste, DOE/EIS-0046F,
Washington, D.C., October 1980.
DOESOb U.S. Department of Energy, Statement of Position of the United States
Department of Energy in the Matter of Proposed Rulemaking on the Storage
and Disposal of Nuclear Waste (Waste Confidence Rulemaking), DOE/NE-
0007, Washington, D.C., April 1980.
DOE85a U.S. Department of Energy, Mission Plan for the Civilian Radioactive Waste
Management Program, DOE-RW-0005, Washington, D.C., June 1985.
DOE85b U.S. Department of Energy, An Evaluation of Commercial Repository
Capacity for the Disposal of Defense High-Level Waste, DOE/DP-0020,
Washington, D.C., January 1985.
DOE86a U.S. Department of Energy, Environmental Assessment, Deaf Smith County
Site, Texas, DOE/RW-0069, Washington, D.C., May 1986.
DOE86b U.S. Department of Energy, Environmental Assessment, Reference Repository
Location, Hanford Site, Washington, DOE/RW-0073, Washington, D.C., May
1986.
DOE86c U.S. Department of Energy, Environmental Assessment, Davis Canyon Site,
Utah, DOE/RW-0071, Washington, D.C., May 1986.
DOE86d U.S. Department of Energy, Environmental Assessment, Richton Dome Site,
Mississippi, DOE/RW-0072, Washington, D.C., May 1986.
DOE86e U.S. Department of Energy, Environmental Assessment, Yucca Mountain Site,
Nevada Research and Development Area, DOE/RW-0073, Washington, D.C.,
May 1986.
DOE86f U.S. Department of Energy, Recommendation by the Secretary of Energy of
Candidate Sites for Characterization for the First Radioactive-Waste
Repository, DOE/S-0048, Washington, D.C., May 1986.
4-19
-------
DOE86g U.S. Department of Energy, Draft Area Recommendation Report for the
Crystalline Rock Repository Project, DOE/CH-15, Chicago, 111., 1986.
DOE87a U.S. Department of Energy, Monitored Retrievable Storage Submission to
Congress, DOE/RW-0035, three volumes, Washington, D.C., March 1987.
DOESTb U.S. Department of Energy, Final Environmental Impact Statement - Disposal
of Hanfbrd Defense High-Level, Transuranic and Tank Wastes, DOE/EIS-
0113, five volumes, Washington, D.C., December 1987.
DOE88a U.S. Department of Energy, Site Characterization Plan, Yucca Mountain Site,
Nevada Research and Development Area, DOE/RW-0199, Washington, D.C.,
December 1988.
DOE89a U.S. Department of Energy, Draft Supplement, Environmental Impact
Statement, Waste Isolation Pilot Plant, DOE/EIS-0026-DS, two volumes,
Washington, D.C., April 1989.
DOE89b U.S. Department of Energy, MRS System Study Summary Report,
Washington, D.C., 1989 (in preparation).
DOE89c U.S. Department of Energy, Impacts of Proposed Revision of 40 CFR 191,
MJ. Furman, Richland Operations Office, March 22, 1989.
DOE89d U.S. Department of Energy, Integrated Data Base for 1988: Spent Fuel and
Radioactive Waste Inventories, Projections, and Characteristics, DOE/RW-
0006, Rev. 4, Washington, D.C., September 1988.
EPA87 U.S. Environmental Protection Agency, Mixed Energy Waste Study (MEWS),
Office of Solid Waste and Emergency Response, Washington, D.C., March
1987.
GON85 Gonzales, S., and K. S. Johnson, Shales and Other Argillaceous Strata in the
United States, ORNL/SUB/84-64794/1, Oak Ridge National Laboratory, Oak
Ridge, TN, March 1985.
DEAL87 International Energy Associates Limited, Regulatory Strategies for High-Level
Radioactive Waste Management in Nine Countries - Final Report, IEAL-R/87-
93, Prepared for U.S.DOE - Pacific Northwest Laboratory, December 1987.
JAC88 Jacquish, R.E., Mitchell, P.J., Environmental Monitoring at Hanford for
1987, PNL-6464, U.S. Department of Energy - Pacific Northwest Laboratory,
May 1988.
MC92 C. McCombie, "Swiss High-Level Radioactive Waste Management System
Issues," Third International Conference on High Level Radioactive Waste
Management, Las Vegas, Nevada, April 12-16, 1992.
4-20
-------
NEA86 Nuclear Energy Agency, Nuclear Spent Fuel Management - Experience and
Options, Organization for Economic Co-Operation and Development, Paris,
France, 1986.
NEA88 Nuclear Energy Agency, Geological Disposal of Radioactive Wastes - In Situ
Research and Investigations in OECD Countries, Organization for Economic
Co-Operation and Development, Paris, France, 1988.
NRC84 U.S. Nuclear Regulatory Commission, Waste Confidence Rulemaking
Decision, Federal Register, 49FR34658, August 31, 1984.
NWPA82 Nuclear Waste Policy Act of 1982, Public Law 97-425, January 7, 1983.
NWPA87 Nuclear Waste Policy Amendments Act of 1987, Public Law 100-203,
December 22, 1987.
REY83 Greater Confinement Disposal Test at the Nevada Test Site, prepared by
Reynolds Electrical & Engineering Co.,Inc. for the U.S. Department of
Energy - Nevada Operations Office, DOE/NV/00410-79, Las Vegas, NV,
June 1983.
SCH88 Schneider, K.J., Lakey, L.T., Silviera, D.J., National Briefing Summaries:
Nuclear Fuel Cycle and Waste Management, PNL-6241, Rev. 1, U.S. DOE -
Pacific Northwest Laboratory, December 1988.
SCH91 Schneider, KJ. et al, National Briefing Summaries: Nuclear Fuel Cycle and
Waste Management, PNL-6241, Rev. 2, U.S. DOE- Pacific Northwest
Laboratory, April 1991.
YA92 A. Yamato et al, "The High Level Radioactive Waste Management Program in
Japan," Third International Conference on High Level Radioactive Waste
Management, Las Vegas, Nevada, April 12-16, 1992.
4-21
-------
Chapter 5: RADIATION DOSIMETRY
5.1 INTRODUCTION
The setting of standards for radionuclides requires an assessment of the doses received
by individuals who are exposed by coming into contact with radiation sources. Two forms of
potential radiation exposures can occur from these sources —internal and external. Internal
exposures can result from the inhalation of contaminated air or the ingestion of contaminated
food or water. External exposures can occur when individuals are immersed in contaminated
air or water or are standing on contaminated ground surfaces. Internal or external doses can
result from radionuclides at the site area or from radionuclides that have been transported
from these sites to other locations in the environment. The quantification of the doses
received by individuals from these radiation exposures is called radiation dosimetry. This
chapter highlights the internal and external dosimetric models used by EPA to assess the dose
to individuals exposed to radionuclides.
The models for internal dosimetry consider the quantity of radionuclides entering the
body, the factors affecting their movement or transport through the body, and the energy
deposited in organs and tissues from the radiation that is emitted during spontaneous decay
processes. The models for external dosimetry consider only the photon doses to organs of
individuals who are immersed in air or are exposed to a contaminated ground surface. In
addition, the uncertainties associated with each model will be discussed.
5.2 BASIC CONCEPTS
Radioactive materials produce radiation as their constituent radioactive nuclides
undergo spontaneous radioactive decay. The mechanisms of emitting this energy are
characteristic of the decay process and include energetic charged particles (alpha and beta
particles) and photons (gamma rays and x-rays). Alpha particles are nuclei of helium atoms
and carry a positive charge two times that of an electron. These particles can produce dense
ionization tracks in the biological material that they traverse. Beta particles are electrons or
positrons emitted in radioactive decay. Their penetration power in material is greater than
that of alpha particles. Gamma and x-rays are electromagnetic radiation and are
distinguishable from alpha and beta particles by their greater penetrating power in material.
This section introduces some terminology used in Chapters 5 and 6 to describe internal
and external dosimetry. For a more detailed explanation, the reader is referred to reports
published in this area by the International Commission on Radiation Units and Measurements
(ICRU80), International Commission on Radiological Protection (ICRP84), and National
Council on Radiation Protection and Measurements (NCRP71).
5.2.1 Activity
The activity of a sample of any radionuclide of species, i, is the rate at which the
unstable nuclei spontaneously decay. If N is the number of unstable nuclei present at a
certain time, t, its activity, A^t), is given by
5-1
-------
Ai(t) = -dN/dt = XXN, (5-1)
where X, is the radioactive decay constant. The customary unit of activity is the curie (Ci);
its subraultiples, the millicurie (mCi), the microcurie (uCi), and the picocurie (pCi), are also
often used. The curie, which is defined as 3.7xl010 disintegrations per second, is the
approximate activity of 1 gm of radium-226.
The time variation of the activity can be expressed in the form:
= Aoi exp(- M). (5-2)
Aoi is the activity of nuclide i at time t=0. For a sample of radioactive material containing
more than one radionuclide, the total activity is determined by summing the activities for each
radionuclide:
A(t) = S, ^(t) (5-3)
5.2.2 Radioactive Half-Life
From the above equations, it is apparent that the activity exponentially decays with
time. The time when the activity of a sample of radioactive material containing species i
becomes one-half its original value (i.e., the time t that A^t) = A0/2) is called its
radioactive half-life, T], and is defined as:
T* = (In 2)/ % (5-4)
The unit for the radioactive half-life is any suitable unit of time such as seconds, days, or
years. The specific activity of a radionuclide (the activity per unit mass) is inversely
proportional to the half-life.
5.2.3 Radionuclide Chains
Radionuclides decay either to stable atoms or to other radioactive species called
daughters. For some species, a decay chain of daughter products may be produced until
stable atoms are formed. For example, strontium-90 decays by emitting a beta-particle,
producing the daughter yttrium-90, which also decays by beta emission to form the stable
atom zirconium-90:
yr) jj 90Y(64.0 h) P "Zrfctable) (5-5)
5.2.4 Biological Half-Life
5-2
-------
The biological half-life of radionuclides is the time required for biological tissues to
eliminate one-half of the activity by elimination processes. This time is the same for both
stable and radioactive isotopes of any given element.
5.2.5 Internal and External Exposures to Radionuclides
The term "exposure," in the context of this report, denotes physical interaction of the
radiation emitted from the radioactive material with cells and tissues of the human body. An
exposure can be "acute" or "chronic" depending on how long an individual or organ is
exposed to the radiation. Internal exposures occur when radionuclides, which have entered
the body through the inhalation or ingestion pathway, deposit energy in organ tissues from the
emitted gamma, beta, and alpha radiation. External exposures occur when radiation enters the
body directly from sources located outside the body, such as radiation from material on
ground surfaces, dissolved in water, or dispersed in the air.
In general, for sources of concern in this report, external exposures are from material
emitting gamma radiation. Gamma rays are the most penetrating of the emitted radiations,
and external gamma ray exposure may contribute heavily to radiation doses to the internal
organs. Beta and alpha particles are far less penetrating and deposit their energy primarily in
the skin's outer layer. Consequently, their contribution to the absorbed dose to the total body,
compared to that deposited by gamma rays, is negligible and will not be considered in this
report.
5.2.6 Absorbed Dose and Absorbed Dose Rate
The radiological quantity absorbed dose, D, denotes the mean energy imparted Ae, by
ionizing radiation to a small finite mass of organ tissue with a mass, Am, and is expressed as
D = dWm = lim (Ae/Am). (rad) (5-6)
Am—»0
Internal and external exposures from radiation sources are not usually instantaneous but are
distributed over extended periods of time. The resulting time rate of change of the absorbed
dose to a small volume of mass is referred to as the absorbed dose rate, D
'D = dD/dt = lim (AD/At). (mrad/y) (5-7)
At—»0
The customary unit of absorbed dose rate is any quotient of the rad (or its multiple or
submultiple) and a suitable unit of time. In this report, absorbed dose rates are generally
given in mrad/yr.
5.2.7 Linear Energy Transfer (LET)
5-3
-------
Linear energy transfer, L^, is the loss of kinetic energy, by collision, by charged
particles per unit length of an absorbing medium. The increment of the mean energy lost,
AE, to tissue by a charged particle of specified energy in traversing a distance, AX:
LTO = dE/dX = Mm (AE/AX) (keV urn'1) (5-8)
Ax—»0
For photons, L^ represents the energy imparted by the secondary electrons (electrons
that are knocked out of their orbitals by primary radiation) resulting from secondary
interactions between the photons and tissue material. High-LET radiation (alpha particles)
imparts more energy per unit length of organ tissue than does low-LET radiation (x-rays,
gamma rays, and beta particles). Consequently, the former are more effective per unit dose in
causing biological damage.
5.2.8 Dose Equivalent and Dose Equivalent Rate
Dose equivalent is a special radiation protection quantity that is used to express the
absorbed dose in a manner that considers the difference in biological effectiveness of various
kinds of ionizing radiation. The ICRU has defined the dose equivalent, H, as the product of
the absorbed dose, D, the quality factor, Q, and all other modifying factors, N, at the point of
interest in biological tissue (ICRU80). This relationship can be expressed in the following
manner:
H = D Q N. (rem) (5-9)
The quality factor is a dimensionless quantity that depends on the collision stopping power
for charged particles. It accounts for the differences in biological effectiveness found among
varying types of radiation. By definition, it is independent of tissue and biological endpoint.
The generally accepted values for quality factors for high- and low-LET radiation, which are
used by EPA, are given in Table 5-1. The product of all other modifying factors, N, such as
dose rate, fractionation, etc., is taken as 1.
Table 5-1. Quality factors for various types of radiation (ICRP77).
Radiation Type Quality Factors (Q)
x-rays, gamma rays, and electrons 1
alpha particles 20
The dose equivalent rate, IJ is the time rate of change of the dose equivalent to organs and
tissues and is expressed as:
H = dH/dt = lim (AH/At). (mrem/yr) (5-10)
At—»0
5-4
-------
5.2.9 Effective Dose Equivalent and Effective Dose Equivalent Rate
The ICRP has defined the effective dose equivalent, HE, as:
Hk = LrWTHr, (rem)(5-ll)
where HT is the dose equivalent in tissue and wx is the weighting factor, which represents the
estimated proportion of the stochastic risk resulting from tissue, T, to the stochastic risk when
the whole body is uniformly irradiated (ICRF77). The weighting factors recommended by the
ICRP are listed in Table 5-2.
Table 5-2. Weighting factors recommended by the ICRP for stochastic risks (ICRP77).
Organ or Tissue WT
Gonads 0.25
Breast 0.15
Red Bone Marrow 0.12
Lung 0.12
Thyroid 0.03
Bone Surfaces 0.03
Remainder 0.30
The effective dose equivalent rate is the time rate of the delivery of the dose equivalent and is
expressed as HE, where:
HE = Zr WT HT . (mrem/yr) (5-12)
5.2.10 Relationship of the Dose Equivalent and the Effective Dose Equivalent to Risk
The dose equivalent was introduced by the ICRP to allow one to combine and
compare - on the basis of biological effects - absorbed doses of different types of radiation.
Subsequently, the effective dose equivalent was introduced to provide a single-valued
indicator of risk for dose equivalents distributed nonuniformly in the body. By convention,
these concepts, in combination with the ICRP-recommended quality factors and organ-
weighting factors, are widely used in radiation protection. These recommended factors,
however, are based on dose response models that differ significantly from those used by EPA
to estimate risk (see Chapter 6).
To calculate risk, EPA first calculates age-specific, high- and low-LET absorbed dose
rates, by organ, for a uniform intake or external exposure rate. The risk from each year's
dose is then calculated using the life table procedure in conjunction with age- and organ-
specific risk models adapted from the BEIR HI report (NAS80).
5-5
-------
These models (see Chapter 6) assume a linear dose-response relationship and a
lifetime relative risk projection for cancers other than bone cancer and leukemia, for which
absolute risk projection is employed. Finally, the risks from each year's dose are summed to
arrive at the risk from lifetime exposure.
In calculating dose equivalents and effective dose equivalents, the ICKP Publication 30
convention was employed, including the same quality factors and organ-weighting factors.
Nevertheless, in calculating the risk from a given absorbed dose of alpha particle irradiation,
RBEs of 8 and 2.7 were used for the induction of cancers and genetic effects, respectively
(see Chapter 6). Since these RBEs are lower than the assumed alpha quality factor (Q=20),
EPA's estimates of the risk per unit dose equivalent (mrem) will be lower for alpha particles
than for x-rays or gamma rays. Likewise, the ICRP organ-weighting factors shown in Table
5-2 do not stand in the same proportion as the organ risks calculated using the EPA models
for cancer induction or genetic mutations. Furthermore, EPA considers somatic and genetic
risks separately. Thus, even if attention was restricted to low-LET radiation, the estimated
risk from a given effective dose equivalent will vary, depending on how the absorbed dose is
distributed within the body.
To summarize, because EPA risk models differ from those underlying the ICRP
recommendations, the risks calculated directly by EPA are not strictly proportional to the
effective dose equivalents derived using ICRP quality factors and organ weighting factors.
5.2.11 Working Levels and Working Level Months
The working level is a unit that has been used as a measure of the radon decay-
product activity in air. It is defined as any combination of short-lived radon daughters
(though polonium-214) per liter of air that will result in the ultimate emission of 1.3 x 10s
MeV of alpha energy. An activity concentration of 100 pCi/L of radon-222 in equilibrium
with its short lived daughters gives rise to a potential alpha-energy concentration of
approximately 1 WL. The WL unit could also be used for thoron daughters. The potential
alpha energy exposure is commonly expressed in units of working level month (WLM). One
WLM corresponds to an exposure to a concentration of 1 WL for the commonly used
reference period of 170 hours.
5.2.12 Customary and SI Units
The relationship between the customary units used in this text and the international system of
units (SI) for radiological quantities is shown in Table 5-3. While the SI radiological units
are almost universally used in other countries for radiation protection regulation, the United
States has not yet officially adopted their use for such purposes.
5-6
-------
Table 5-3. Comparison of customary and SI special units for radiation quantities.
Quantity
Customary Unit
Name
Definition
Special SI Unit
SI Unit
Definition
Activity (A)
Absorbed
dose (D)
Dose
equivalent (H)
Linear energy
transfer
Curie (Ci) 3.7xl010 sl
rad 10'2 J kg1
rem
keV
(kiloelectron
volts per
micrometer)
becquerel (Bq) 1.0 s'1
gray (Gy) 1.0 J kg1
lO'2 J kg1
1.602X10'10 J m'1
sievert (Sv)
1.0 J kg
.-i
5.3 EPA DOSIMETRIC MODELS
The EPA dosimetric models, to be discussed in the following sections, have been
described in detail in previous publications (Du80, Su81). Information on the elements
treated in these sections was taken directly from those documents or reports. In most cases,
the EPA models are similar or identical to those recommended by the ICRP (ICRP79,
ICRP80, ICRP81). However, differences in model parameters do exist for some radionuclides
(Su81). The basic physiological and metabolic data used by EPA in calculating radiation
doses are taken from ICRP reports (ICRP75, ICRP79).
5.3.1
Internal Dose Models
EPA implements contemporary models to estimate absorbed dose rates as a function of
time to specified organs in the body. Estimates of the doses resulting from the deposition and
retention of inhaled particulates in the lung and their subsequent absorption into the blood and
clearance into the gastrointestinal (GI) tract are made using the ICRP Task Group Lung
Model (ICRP66).
5-7
-------
5.3.1.1 Generalized Scheme for Estimating Organ Absorbed Dose Rates
5.3.1.1.1 Distribution of Activity of Radionuclides in the Body
The complex behavior of radionuclides is simplified conceptually by considering the
body as a set of compartments. A compartment may be any anatomical, physiological, or
physical subdivision of the body throughout which the concentration of a radionuclide is
assumed to be uniform at any given time. The terms "compartment" and "organ" are often
used interchangeably, although some of the compartments considered in this report may
represent only portions of a structure usually considered to be an organ, while some
compartments may represent portions of the body usually not associated with organs.
Examples of compartments used in this report are the stomach, the pulmonary region of the
lung, the blood, or the bone. Within a compartment, there may be more than one "pool" of
activity. A pool is defined to be any fraction of the activity within a compartment that has a
biological half-life which is distinguishable from the half-time(s) of the remainder of activity
within the compartment
Activity entering the body by ingestion is assumed to originate in the stomach
compartment; activity entering through inhalation is assumed to originate in a compartment
within the lung (the tracheo-bronchial, pulmonary, or naso-pharyngeal region). From the
stomach, the activity is viewed as passing in series through the small intestine, the upper large
intestine, and the lower large intestine, from which it may be excreted. Also, activity
reaching the small intestine may be absorbed through the wall into the bloodstream, from
which it may be taken in parallel into any of several compartments within the skeleton, liver,
kidney, thyroid, and other organs and tissues.
The list of organs or regions for which dose rates are calculated is found in Table 5-4.
Activity in the lung may reach the bloodstream either directly or indirectly through the
stomach or lymphatic system. The respiratory system and gastrointestinal tract models are
discussed further in later sections. Figure 5-1 illustrates the EPA model used to represent the
movement of radioactivity in the body.
EPA models separately consider the intake and subsequent behavior of each
radionuclide in the body. The models also allow for the formation of radioactive decay
products within the body, and it is assumed that the movement of internally produced
radioactive daughters is governed by their own metabolic properties rather than those of the
parent. This contrasts the ICRP assumption that daughters behave exactly as the parent
5-8
-------
Table 5-4. Target organs and tissues used for calculating the ICRP effective dose equivalent
and the EPA cancer risk.
ICRP effective
dose equivalent
EPA cancer risk
Ovaries
Testes
Breast*
Red marrow
Lungsb
Thyroid
Bone surface
Stomach wall
Small intestine wall
Upper large intestine wall
Lower large intestine wall
Kidneys
Liver
Pancreas
Brain
Spleen
Thymus
Uterus
Adrenals
Bladder wall
Breast
Red marrow
Pulmonary lung0
Thyroid
Bone surface (endosteum)
Stomach wall
Intestined
Kidneys
Liver
Pancreas6
a)
b)
c)
d)
e)
Dose to breast is assumed to equal dose to muscle.
The ICRP considers the lungs to be a composite of the trachiobronchial region,
pulmonary region, and the pulmonary lymph nodes with a combined mass of 1,000 g
(ICRP79).
The EPA calculates lung cancer risk on the basis of the dose to the pulmonary lung.
The mass of this region, which does not include venous or arterial blood, is considered
to be 570 g.
The EPA averages the values for the small, upper large, and lower large intestine using
weights of 0.2, 0.4, and 0.4 respectively for calculating the risk of bowel cancer.
The pancreas is also used as a surrogate organ for calculating the cancer risk for all
other organs and tissues.
5-9
-------
Figure 5-1. A schematic representation of radioactivity movement among respiratory tract,
gastrointestinal tract, and blood.
INGESTION
RESPIRATORY
TRACT
I
B
L
O
O
D
— >
f'
S
l?Jfr |As-24day-i
S..::;*?' B
ASI
*^jPm»
Auu
A.
SI
|V«
ULI
.-1
• A. - 1.85 day -1
1 UU
LLI
IA, -
Jlu
S = stomach
SI = small intestine
ULI = upper large intestine
LLI = lower large intestine
X. = elimination rate constant
5-10
-------
If Afcft) denotes the activity of the ith species of the chain in organ k and if that activity
is divided among several "pools" or "compartments" indexed by subscript 1, then the time rate
of change of activity can be modeled by a system of differential equations of the following
form:
B, E Ajr + Pik)
1 = 1 ..... L* (5-14)
where compartment 1 is assumed to have L^ separate pools of activity, and where:
Ajfc = the activity of species i in compartment 1 of organ k;
= (hi 2) / if where if = radioactive half of species i;
= rate coefficient (time"1) for biological removal of species i from compartment 1 of
organ k;
Ly, = number of exponential terms in the retention function for species i in organ k;
By = branching ratio of nuclide j to species i;
p^ = inflow rate of the 1th species onto the organ k; and
% = the fractional coefficient for nuclide i in the 1th compartment of organ k.
The subsystem described by these L& equations can be interpreted as a biological
compartment in which the fractional retention of radioactive species is governed by
exponential decay. Radioactivity that enters an organ may be lost by both radioactive decay
and biological removal processes. For each source organ, the fraction of the initial activity
remaining at any time after uptake at time t = 0 is described by a retention function
consisting of one or more exponentially decaying terms:
R^t) = S^ c^ exp[-Uf + A4 )t] (5-15)
The subscript 1 in the above equation represents the 1th term of the retention function, and the
coefficients c^ can be considered as "pathway fractions."
5.3.1.1.2 Dose Rates to Target Oraans
The activity of a radionuclide in a compartment is a measure of the rate of energy
being emitted in that compartment, at any time, t, and can be related to the dose rate to a
specific organ at that time. This requires estimating the fraction of the energy emitted by the
decay of the radionuclide in each compartment that is absorbed by the specific organ.
The absorbed dose rate, IJ(X;t) to target organ X at time t due to radionuclide species i in
source organs Y^Yi,...., YM is estimated by the following equation:
5-11
-------
Di(X;t) = S£t D/X^-Y^t) (5-16)
where: IJ(X<—Yk;t) = Si(X«-Yk) A^t); and A^t) is the activity, at time t of species i in source
organ Yk; Si(X«-Yk), called the S-factor, represents the average dose rate to target organ X
from one unit of activity of the radionuclide uniformly distributed in source organ or
compartment Yk. It is expressed in the following manner:
= c Sm fm Em (j>m(X<-Yk) (5-17)
where:
c = a constant that depends on the units of dose,
energy, and time being used;
fm = intensity of decay event (number per
disintegration);
Em = average energy of decay event (Mev); and
= specific absorbed fraction, i.e., the fraction
emitted energy from source organ Yk absorbed by
target organ X per gram of X,
where the summation is taken over all events of type m. The units for S-factors depend on
the units used for activity and time; thus, the S-factor units may be rad/Ci-day. The S-factor
is similar in concept to the SEE factor (specific effective energy) used by the ICRP
Committee 2 in Publication 30. However, the SEE factor includes a quality factor for the
type of radiation emitted during the transformation.
The above equations are combined to produce the following expressions for the
absorbed dose rates to target organs at any time due to one unit of activity of radionuclide
species, i, uniformly distributed in source organs Yt ... Yk:
D(X;t) = St Sm Afc(t) S^X^-Yt) (5-18)
5-12
-------
The corresponding dose equivalent rate, Hj(X;t), can be estimated by inclusion of the quality
factor, Qm, and the modifying factor, Nm(Yk):
Hi(X;t) = SfeSm A^t) Q.NJYi) S^CX^-Y,) (5-19)
Implicit in the above equations is the assumption that the absorbed dose rate to an organ is
determined by averaging absorbed dose distributions over its entire mass.
Alpha and beta particles are usually not sufficiently energetic to contribute a
significant cross-irradiation dose to targets separate from the source organ. Thus, the
absorbed fraction for these radiations is generally assumed to be just the inverse of the mass
of organ X, or if the source and target are separated, then (j>m(X<— Y) = 0. Exceptions occur
when the source and target are in very close proximity, as is the case with various skeletal
tissues. Absorbed fractions for cross-irradiations by beta particles among skeletal tissues were
taken from ICRP Publication 3 (ICRP80). The energy of alpha particles and their associated
recoil nuclei is generally assumed to be absorbed in the source organ. Therefore, (j)m(X<— X)
is taken to be the inverse of the organ mass, and (j)m(X«-Y) = 0 if X and Y are separated.
Special calculations are performed for active marrow and endosteal cells in bone, based on
the method of Thorne (Th77).
5.3.1.1.3 Monte Carlo Methodology to Estimate Photon Doses to Organs
The Monte Carlo method uses a computerized approach to estimate the probability of
photons interacting within target organ X after emission from source organ Y. The method is
carried out for all combinations of source and target organs and for several photon energies.
The body is represented by an idealized phantom in which the internal organs are assigned
masses, shapes, positions, and attenuation coefficients based on their chemical composition.
A mass attenuation coefficient, ^ is chosen, where u0 is greater than or equal to the mass
attenuation coefficients for any region of the body. Photon courses are simulated in randomly
chosen directions, and potential sites of interactions are selected by taking distances traversed
by them as -hi r/j^, where r is a random number distributed between 0 and 1. The process is
terminated when either the total energy of photons has been deposited or the photon escapes
from the body. The energy deposition for an interaction is determined according to standard
equations (ORNL74).
5.3.1.1.4 Effects of Decay Products
In calculating doses from internal and external exposures, the in-growth of radioactive
decay products (or daughters) must be considered for some radionuclides. When an atom
undergoes radioactive decay, the new atom created in the process, which may also be
radioactive, can contribute to the radiation dose to organs or tissues in the body. Although
5-13
-------
these decay products may be treated as independent radionuclides in external exposure, the
decay products of each parent must be followed through the body in internal exposure
situations. The decay product contributions to the absorbed dose rates, which are included in
EPA calculations, are based on the metabolic properties of the individual daughters and the
organ in which they occur.
5.3.1.2 Inhalation Dosimetry - ICRP Respiratory Tract Model
As stated earlier, individuals immersed in contaminated air will breathe radioactive
aerosols or particulates, which can lead to doses to the lung and other organs in the body.
The total internal dose caused by inhalation of these aerosols can depend on a variety of
factors, such as breathing rates, particle sizes, and physical activity. Estimating the total dose
to individuals over a specific time period requires specifying the distribution of particle
depositions in the respiratory tract and the mathematical characteristics of the clearance
parameters. The EPA currently uses assumptions established by the ICRP Task Group on
Lung Dynamics (TGLM)(ICRP66). This section will summarize the essential features of that
model. For a more comprehensive treatment, the reader is referred to the actual report.
The basic features of the ICRP lung compartmental model are shown in Figure 5-2.
According to this model, the respiratory tract is divided into four regions: naso-pharyngeal
(N-P), tracheo-bronchial (T-B), pulmonary (P), and lymphatic tissues.
In the model, the regions N-P, T-B, and P are assumed to receive fractions D3, D4, and
D5 of the inhaled particulates, where the sum of these is less than 1 (some particles are
removed by prompt exhalation). The values D3, D4, and D5 depend on the activity median
aerodynamic diameter (AMAD) of the inspired particles. For purposes of risk calculations,
EPA uses AMADs of 1 micron. The lung model employs three clearance classes, D, W, and
Y, corresponding to rapid, intermediate, and low clearance, respectively, of material deposited
in the respiratory passages. The clearance class depends on chemical properties of the
inhaled particles.
Like the ICRP, EPA assumes that the absorbed dose rate to the N-P region can be
neglected. Unlike the ICRP, however, EPA averages the dose over the pulmonary region of
the lung (compartments e through h), to which is assigned a mass of 570 g, including
5-14
-------
Figure 5-2. The ICRP Task Group lung model for particulates
COMPARTMENT
N-P
(D3 - 0.30)
T-B
(D4 - 0.08)
P
(D5 - 0.25)
L
a
b
c
d
e
f
g
h
i
CLASS
D
T F
0.01 0.5
0.01 0.5
0.01 0.95
0.2 0.05
0.5 0.8
n.a. n.a.
n.a. n.a.
0.5 0.2
0.5 1.0
W
T F
0.01 0.1
0.4 0.9
0.01 0.5
0.2 0.5
50 0.15
1.0 0.4
50 0.4
50 0.05
50 1.0
Y
T F
0.01 0.01
0.4 0.99
0.01 0.01
0.2 0.99
500 0.05
1.0 0.4
500 0.4
500 0.15
1000 0.9
The columns labeled D, W, and Y correspond, respectively, to rapid, intermediate, and slow clearance of the inspired material (in days, weeks,
or years). The symbols T and F denote the biological half-time (days) and coefficient, respectively, of a term in the appropriate retension
function. The values shown for D3, D4, and D5 correspond to activity median aerodynamic diameter, AMAD = urn, and represent the friction
of the inspired material depositing in the lung regions.
5-15
-------
capillary blood (ICRP75). In addition, it is assumed that the total volume of air breathed in
one day by a typical member of the general population is 22,000 liters. This value was
determined by averaging the ICRP-23 adult male and female values based on 8 hours of
working "light activity," 8 hours of nonoccupational activity, and 8 hours of resting.
5.3.1.3 Ingestion Dosimetry - ICRP GI Tract Model
According to the ICRP-30 GI tract model, the gastrointestinal tract consists of four
compartments: the stomach (S), small intestine (SI), upper large intestine (ULI), and lower
large intestine (LLI). The fundamental features of the model are shown in Figure 5-1. It is
assumed that absorption into the blood occurs only from the small intestine (SI).
This model postulates that radioactive material entering the compartments of the GI
tract is exponentially removed by both radioactive decay and biological removal processes,
and that there is no feedback. Absorption of a particular nuclide from the GI tract is
characterized by flt which represents that fraction of the nuclide ingested which is absorbed
into body fluids if no radiological decay occurs:
f _ \°*> it \a° .1 \ rs-9.0')
lj — "5/ 'x SI V/ ' ^ £\J)
where
^ = the absorption coefficient (s"1)
ASI = the transfer coefficient from the small intestine to the large intestine (s"1)
Figure 5-1 graphically presents the role of these coefficients in the gastrointestinal model.
The kinetic model, as formulated by the ICRP, does not permit total absorption of a nuclide
(^ = 1).
5.3.1.4 Dose Rate Conversion Factors
EPA uses the computer code RADRISK (Du80) for calculating radiation doses and
risks to individuals resulting from a unit intake of a radionuclide, at a constant rate, for a
lifetime exposure (50-yr dose commitment). These calculations are done for the inhalation
and ingestion pathways to individuals who are exposed by immersion in contaminated air or
by contaminated ground surfaces.
RADRISK computes doses for both chronic and acute exposures. Following an acute
intake, it is assumed the activity in the body decreases monotonically, particularly for
radionuclides with rapid radiological decay rates or rapid biological clearance. In the case of
chronic exposure, the activity in each organ of the body increases monotonically until a
steady state is achieved, at which time the activity remains constant. The instantaneous dose
rates at various times after the start of chronic exposure provide a reasonably accurate (and
conservative) estimate of the total annual dose for chronic exposure conditions. However, the
instantaneous dose rates may err substantially in the estimation of annual dose from an acute
exposure, particularly if the activity levels decrease rapidly.
5-16
-------
Since the rate of change in activity levels in various organs is more rapid at early
times after exposure, doses are computed annually for the first several years and for
progressively longer periods thereafter, dividing by the length of the interval to estimate the
average annual dose. This method produces estimates of risk that are similar to those
computed by the original RADRISK methodology for chronic exposures and provides a more
accurate estimate of the risks from acute intakes.
5.3.1.5 Special Radionuclides
The following paragraphs briefly summarize some of the special considerations for particular
elements and radionuclides.
5.3.1.5.1 Tritium and Carbon-14
Most radionuclides are nuclides of elements found only in trace quantities in the body.
Others like tritium (hydrogen-3) or carbon-14 must be treated differently since they are long-
lived nuclides of elements that are ubiquitous in tissue. An intake of tritium is assumed to be
completely absorbed and to be rapidly mixed with the water content of the body (Ki78a).
The estimates for inhalation include consideration of absorption through the skin.
Organ dose estimates are based on the steady-state specific-activity model described by
Killough et al. (Ki78a).
Carbon-14 is assumed to be inhaled as CO2 or ingested in a biologically bound form.
Inhaled carbon-14 is assumed to be diluted by stable carbon from ingestion (Ki78b). This
approach allows separate consideration of the ingestion and inhalation pathways. The
specific-activity model used for organ dose estimates is also that of Killough et al. (Ki78a).
Short-lived carbon radionuclides (e.g., carbon-11 or carbon-15) are treated as trace elements,
and the organ doses are calculated accordingly.
5.3.1.5.2 Noble Gases
Retention of noble gases in the lungs is treated according to the approach described by
Dunning et al. (Du79). The inhaled gas is assumed to remain in the lungs until lost by
radiological decay or respiratory exchange. Translocation of the noble gas to systemic organs
is not considered, but doses due to translocated decay products produced in the lungs are
calculated. The inhalation of the short-lived decay products of radon is assessed using a
potential alpha energy exposure model (see Chapter 6) rather than by calculating the doses to
lung tissues from these radionuclides.
5.3.1.5.3 Uranium and Transuranics
The metabolic models for transuranics elements (polonium, neptunium, plutonium,
americium, and curium) are consistent with those used for the EPA transuranic guidance
(EPA77). A Gl-tract-to-blood absorption factor of 10"3 is used for the short-lived nuclides of
plutonium (plutonium-239,-240, and -242), while a value of 10"* is used for other transuranics.
For soluble forms of uranium, a GI tract to blood absorption factor of 0.2 is used in
5-17
-------
accordance with the high levels of absorption observed for low-level environmental exposures
(Hu73, Sp73).
5.3.1.6 Uncertainties in Internal Dose Estimates
Estimates of radiation dose in risk assessment studies have traditionally been based on
dosimetric models derived in the context of radiation protection for adult workers. Despite
the obvious differences between risk assessment and radiation protection, the dosimetric
formulations of the latter have been generally adopted, often with no modifications, in risk
assessment activities. This approach permits use of a substantial body of information
assembled by international experts from the occupational setting and provides models that
avoid the complex problems encountered in biokinetic modeling of radionuclides for the
general public in an age-dependent sense. However, the continued use in risk assessment of
dosimetric data derived for workers, which neglects organ-specific biokinetics and age
dependence, is becoming increasingly difficult to justify. One major limitation of the current
ad hoc dosimetric formulations is the great difficulty in making informed estimates of the
uncertainties in the estimated dose.
All dosimetry models are inherently uncertain. At best, these models can only
approximate real situations in organs and tissues in humans. Consequently, without extensive
human data, the uncertainties associated with their use for risk assessment purposes is
extremely difficult, and in some cases impossible, to quantify. However, consideration of
their limitations in estimating doses to an average member of the general population is
essential.
In applying the dosimetric models in current use, as discussed in the previous sections,
the primary sources of uncertainty are attributed to ICRP model formulation and parameter
variability produced by measurement error or natural variation. The purpose of this section is
to provide a general but limited discussion of these sources and to introduce an uncertainty
scheme for classifying radionuclides. The authors gratefully acknowledge Dr. Keith
Eckerman of Oak Ridge Laboratory for discussions with respect to implementation of ICRP
models and for guidance regarding the magnitude of uncertainties. However, the conclusions
presented here are those of the Agency.
5.3.1.6.1 Uncertainties Due to ICRP Model Formulation
Uncertainty in calculations based on ICRP models arises primarily from five sources:
(1) the uncertainty in the Reference Man data; (2) the uncertainty in the lung and Gl-tract
model describing the translocation and absorption of inhaled or ingested activity into the
blood; (3) the uncertainty associated with the formulation of the ICRP Publication 30
biokinetic models describing the distribution and retention of the activity among the various
organs in the body; (4) the uncertainty in the dose models to calculate the absorbed dose to
organs from that activity; and (5) the uncertainty in the model parameters.
5-18
-------
5.3.1.6.2 Reference Man Concept
To establish a degree of consistency in occupational dosimetry calculations, the ICRP
developed the concept of Reference Man (ICRP75). Reference Man is a conceptual
individual who has the anatomical and physiological characteristics of a healthy 20 to 30 year
old male with a total body mass of 70-kg. The anatomical and physiological data of
Reference Man have been embedded in many computational models for estimating organ
doses and applied in radiation protection and in some calculations for medicine.
Although these data have been extensively applied in calculating doses, the approach
in which Reference Man data is used to represent average individuals in a specific population
introduces bias from the outset. The uncertainties in this approach are primarily due to age-
and sex- specific differences in the anatomical and physiologic parameters. Biological and
ethnic variability also contribute. Li addition, the Reference Man data do not always
represent data for a 70-kg man. Many of the data found in ICRP Publication 23 were from
adults who had anatomical or physiological characteristics significantly different from those of
a 70-kg man.
Due to the many parameters involved and the quality of the data available to define
the numerical values, it is very difficult to establish the level of uncertainty in using
Reference Man data to estimate doses to the average individual in the U.S. population.
Furthermore, the Reference Man concept was not formulated so as to facilitate a quantitative
analysis of the uncertainty in the dose estimates. Finally, Reference Man is not intended to
be representative of the U.S. population.
5.3.1.6.3 ICRP Respiratory Tract Model
When individuals inhale radioactive aerosols, the dose to the lungs and other organs in
the body depends primarily on how the aerosols are deposited in and cleared from the
airways of the respiratory tract. Mechanisms involved in the deposition of inhaled aerosols
and gases are affected by physical and chemical properties, including aerosol size distribution,
density, shape, surface area, electrostatic charge, chemical composition and gas diffusivity and
solubility. Deposition is also affected by respiratory physiology, morphometrics and
pathology.
The ICRP modeling system assumes that deposition rates for aerosols in the
respiratory tract are controlled primarily by three mechanisms: sedimentation, impaction and
Brownian diffusion. The major uncertainties associated with the ICRP deposition models for
the lungs are: (1) the uncertainty in the anatomical model of the respiratory tract, (2) the
uncertainty in the effective aerodynamic diameter of the inhaled particles, (3) the uncertainty
in the breathing patterns and rates, and (4) the questionable validity of the fluid dynamic
models used for all exposure situations.
The number of particles deposited in the lung essentially depends on physiologic,
morphometric and anatomical properties, such as airway dimensions and numbers, branching
and gravitational angles of airways, and distances to the alveolar walls. The ICRP respiratory
tract model (ICRP66) uses the anatomical model devised by Findeisen (Fi35) in its dosimetric
5-19
-------
calculations. This model assumes that lung airways are rigid tubes with symmetric
dichotomous branching patterns and that their morphometric properties are those of an adult
male. In reality, however, the airways have circular ridges or longitudinal grooves (FRC67),
and many airways, like the trachea, are irregular in shape (Br52). In addition, airways change
in diameter and length during inspiration and expiration (Ho75, Hu72, Th78), which affects
gravitational and branching angles (Ph85). Since many of these properties depend on age and
sex, using the anatomic and morphometric lung properties of an adult male for estimating
doses to other members of the population is likely to introduce considerable bias.
Clearance of particles from the respiratory tract depends on many factors, such as site
of deposition, chemical composition, physical properties of the deposited material, and
mucociliary transport rates. The uncertainties associated with using the values provided by
the ICRP are due primarily to the sparseness of data on lung clearance mechanisms, in
general, and secondarily to age, activity levels and general health status of the individual at
the time of exposure. Furthermore, as stated earlier, most of the lung deposition data and
models are derived from studies of healthy adults. Studies have shown, however, that
children's lungs differ from adults' with respect to anatomical, physiological, and
morphological properties. As a consequence, particle deposition in the respiratory tract is
expected to be higher in children than in adults.
5.3.1.6.4 ICRP GI-Tract Model
The ICRP Gl-tract model assumes that ingested material (radionuclides) moves in
sequence through the stomach, small intestine, upper large intestine, and lower large intestine.
The model depicts an exponential removal from each compartment, characterized by a single
removal rate that depends only on the compartment. The model has no provision for
addressing endogenous secretion. In addition, it is assumed that radionuclides are absorbed
into the blood from the small intestine (SI).
Uncertainties arise when applying these assumptions to the estimation of doses to
average individuals. Although radionuclides transported through the GI tract are primarily
absorbed into the blood stream from the SI, fractions can be absorbed from the other
compartments. Furthermore, the removal rates, which are model parameters, vary among
different individuals in the population. Considerable differences can exist depending on the
type of radionuclide ingested, its chemical form, the amount and composition of food in the
stomach at the time of intake and other factors which vary because of nutritional status, age,
and the sex of the individual. The fj factor, which represents the fraction of material
absorbed from the SI, generally contributes the largest uncertainty in the GI tract model. This
parameter will be discussed in a later section.
5.3.1.6.5 ICRP 30 Biokinetic Models
The ICRP biokinetic models were chosen to represent adult male members of the
population. Uncertainties are associated with the approach because they do not account for
differences in the metabolic behavior of radionuclides, which vary depending on age, sex, and
dietary intakes of an individual at the time of exposure. In addition, many of the models
5-20
-------
chosen for dosimetry calculations are based on very limited observational data that cannot be
reliably applied across the population.
Below is a list of additional uncertainties associated with the ICRP biokinetic models:
a) The models have been constructed largely from animal data in such a way that
extrapolation to humans has no strong logical or scientific support.
b) Doses to heterogeneously distributed radiosensitive tissues of an organ (e.g.,
skeletal and lung tissues) cannot be estimated accurately, since the actual
movement of radionuclides in the body is not accurately tracked.
c) Some radionuclides are assigned the model of an apparently related nuclide (e.g.,
americium, curium, neptunium are assigned the plutonium model) although
differences in metabolism are known.
d) The growth of radioactive daughters is often not handled realistically, and the
format of the models makes it difficult to supply alternative assumptions.
e) The models often yield inaccurate estimates of excretion even for the average
adult.
5.3.1.6.6 ICRP Dose Models
ICRP models estimate doses to organs of the body by considering the distribution of
the radioactivity and the interaction of radiation with cells and tissues in these organs.
Estimates of the absorbed dose in a region (referred to as the target region) depend upon the
spatial relationships of that region to the regions containing the radionuclide (referred to as
source regions) and how the activity is distributed in the source region. For organs other than
bone, it is assumed that the radionuclides are uniformly distributed in the source regions and
that the radiosensitive cells of interest are uniformly distributed in the target region.
However, this assumption may bias the dose estimates because of the nonuniformity of the
activity that is normally found in human organs.
5.3.1.6.7 Uncertainties Due to Parameter Value Variability
Most discussions concerning the uncertainties in dose estimates focus on the
uncertainty associated with model parameter values. These discussions assume that the ICRP
metabolic and dose models are correct. The most important parameters of concern for dose
assessment calculations are: radionuclide intake rates, organ masses, blood transfer factors,
organ uptake rates, and biological half-times of radionuclides. Although parameter value
variability can be attributed to measurement and sampling errors and natural biological
variation, hi many cases, age is the largest source of variability.
Depending on the type of radionuclide ingested, the age and element dependency in
the metabolic and physiological processes determines how the dose to target organs varies
with age. For example, strontium tends to follow the calcium pathways in the body and
5-21
-------
deposits to a large extent in the skeleton. In fact, the fraction of ingested strontium
eventually reaching the skeleton at a given age depends largely on the skeletal needs for
calcium at that age, even though the body is able to discriminate
somewhat against strontium in favor of calcium after the first few weeks of life.
Given the importance of age as a contributor to parameter variability in dose
estimates, the possible age dependence in thyroid dose for chronic ingestion of a fixed
iodine- 131 concentration in milk is examined in more detail below. Some other examples of
parameter variability will also be noted.
A simple model that can be used to relate the absorbed dose rate to a target organ due
to radioactivity located in that organ can be expressed as follows :
D(t) = c I fj f2 E [l-exp(-to)]/mX (5-21)
where:
D(t) = absorbed dose rate (rad/day);
I = radionuclide intake rate (Ci/day);
fj = fraction of ingested activity transferred to the blood;
£ = fraction of blood activity transferred to the organ;
m = target organ mass (g);
X, = elimination constant (day"1) = 0.693/T1/2 where T1/2 is
the effective half-time, including the effects of
both biological removal and radioactive decay.
E = energy absorbed by the target organ for each
radioactive transformation.
c = proportionality constant
(51.2 x 106g rad Cf1 MeVM'1).
For simplicity, we will consider the case where t is very large compared to the
biological half -life of the incorporated radionuclide, so that the term in the bracket is
approximately 1:
(5-22)
In addition, it is assumed that the parameters remain constant throughout the period of
investigation and are independent of each other.
5-22
-------
Equation 5-22 is a simplified form of the model used by EPA to estimate the absorbed
dose rates to target organs resulting from the ingestion of radioactive material. It represents
the absorbed dose rate to a target organ from particulate radiation due to radioactivity that is
uniformly distributed in that organ.
For this illustration, the chronic intake of iodine-131 is considered assuming that it
behaves metabolically the same as stable iodine. It is further assumed that iodine is rapidly
and almost completely absorbed into the bloodstream following inhalation or ingestion. From
the blood, iodine enters the extracellular fluid and quickly becomes concentrated in the
salivary, gastric, and thyroid glands. It is rapidly secreted from the salivary and gastric
glands but is retained in the thyroid for relatively long periods.
The intake and metabolism of iodine have been reviewed extensively in the literature.
Two papers have used published data to model the absorbed dose from radioiodine. In the
first (Du81), the authors compiled and evaluated the variability in three of the principal
biological parameters contained in Equation 5-22: m, X, and f2. In the second (Br69), the
author provided age-specific values for most of the same model parameters. Differences in
these data illustrate how parameter variability, when used in the same model, can affect
absorbed dose rate estimates for members of the general population.
Intake Rate, I
The amount of radioactive material taken into the body over a specified period of time
by ingestion or inhalation is expected to be proportional to the rate of intake of food, water,
or air containing such material, which, in turn, would depend on such factors as age, sex, diet,
and geographical location. Therefore, understanding the patterns of food intake for
individuals in the population is important in assessing the possible range of intake rates for
radionuclides.
Recent EPA analyses were done to assess the daily intake rates of food and water for
individuals in the general population. These studies showed that age and sex played an
important role (Ne84). Age significantly affects food intake rates for all of the major food
classes and, with one exception, subclasses. The relationships between food intake and age
are, in most cases, similar to growth curves; there is a rapid increase in intake at an early
stage of physical development, then a plateau is reached in adulthood, followed by an
occasional decrease after age 60.
When sex differences were significant, males, without exception, consumed more than
females. The study also showed that relative consumption rates for children and adults
depend on the type of food consumed. The amount of radioactivity taken into the body per
unit intake of food, air, and water depends on its relative density (amount of radioactivity
contained in the material per unit volume). The most likely pathway to organs in the body
for the ingestion of radioactive iodine comes from drinking milk. According to the above
analysis, the daily intake rate of milk by children (under 1 yr) was twice that for an adult (25
to 29 yr) male. The intake rates for milk used in the models are 0.7 L/day and 0.5 L/day for
the child and adult, respectively.
5-23
-------
Transfer Fraction, /y
While uncertainty in fx is not an important consideration for iodine, it can be very
significant for other elements. Experimental studies suggest that the fx value for some
radionuclides may be orders of magnitude higher in newborns than in adult mammals, with
the largest relative changes with age occurring for those nuclides with small adult f: values
(Cr83). For some radionuclides, the fj value appears to decrease rapidly in the first year of
life. This can be related to the change in diet during this time period, which could affect both
the removal rate from the small intestine to the upper large intestine and the absorption rate
from the small intestine to the bloodstream. Studies have indicated that the wall of the small
intestine is a selective tissue and that absorption of nutrients is to a large extent controlled by
the body's needs (Cr83). In particular, the fraction of calcium or iron absorbed depends on
the body's needs for these elements, so the ft value for these elements and for related
elements such as strontium, radium, and barium (in the case of calcium) and plutonium (in
the case of iron) may change as the need for calcium or iron changes during various stages of
life.
For some essential elements, such as potassium and chemically similar radioelements,
such as rubidium and cesium, absorption into the bloodstream is nearly complete at all ages,
so that changes with age and possible homeostatic adaptations in absorption are not
discernible. The fraction of a radioelement that is transferred to the blood depends on its
chemical form, and wide ranges of values are found in the literature for individuals who
ingest the material under different conditions. For example, fj values for uranium were found
to range from 0.005 to 0.05 for industrial workers, but a higher average value of 0.2 (0.12 to
0.31) is indicated by dietary data from persons not occupationally exposed (ICRP79). EPA
has used the 0.2 value for uranium ingestion by the general population.
It appears that all iodine entering the small intestine is absorbed into the blood; hence
the ^ value is taken as 1 for all ages, which is the value used in this analysis.
Organ Masses, m
To a large extent, the variability in organ masses among individuals in the general
population is related to age. For most of the target organs listed in Table 5-2, the mass
increases during childhood and continues to increase until adulthood, at which time the net
growth of the organ ceases; there may be a gradual decrease in mass (for some organs) in
later years.
Based on data reviewed by Dunning and Schwarz (Du81), the mass of an adult thyroid
ranges from 2 to 62 g. It is expected that this parameter variability would be reflected in
large dosimetric variability among adults. Children in the age group from .5 to 2 yr were
found to have a mean thyroid mass of 2.1 g, while the adult group had a mean mass of 18.3
g. For this illustration, the same values are used as employed by the ICRP (20 g for the adult
thyroid mass and 1.8 g for that of a 6-month-old child), which are also consistent with the
recommendation of Bryant (Br69).
5-24
-------
Organ Uptake Fraction, f2
The fraction of a radionuclide taken up from the blood in an organ is strongly
correlated with the size of the organ, its metabolic activity, and the amount of material
ingested. Iodine introduced into the bloodstream is rapidly deposited in the thyroid, usually
reaching a peak slightly after 24 hours. The uptake of iodine-131 by the thyroid is similar to
that of stable iodine in the diet and can be influenced by sex and dietary differences. There
can be considerable variation among populations.
Dunning and Schwarz (Du81) found a mean f2 value of 0.47 for newborns, 0.39 for
infants, 0.47 for adolescents, and 0.19 for adults. This analysis uses f2 values of .35 and .15
for a child and adult, respectively.
Effective Half-Life, T]/2
Some data suggest a strong correlation between biological half-lives of radionuclides
in organs in the body and the age of the individual. Children are expected to exhibit faster
elimination rates and greater uptakes (Ro58). For iodine, a range of biological half-lives of
21 to 200 days for adults has been observed, and a similarly wide range would be expected
for other age groups (Du81). Rosenberg (Ro58) found a significant correlation between the
biological half-life and the age of the individual and an inverse relationship between uptake
and age in subjects from 22 to 50 yr of age. Dunning and Schwarz (Du81) concluded that for
adults the observed range was from 21 to 372 days; for children in the age group from .5 to 2
yr, the range was 4 to 39 days.
In light of the possible inverse relation between the biological half-life and the f2
value, this analysis uses biological half-lives of 24 and 129 days, respectively, for children
and adults, based on the paper by Bryant (Br69). Including the effect of radioactive decay,
these values imply an effective half-life of 6 days in adults and 8 days in children.
Effective Energy per Disintegration, E
The effective energy per disintegration (MeV/dis) of a radionuclide within an organ
depends on the decay energy of the radionuclide and the effective radius of the organ
containing the radionuclide (ICRP59). It is expected, therefore, that E is an age-dependent
parameter which could vary as the size of the organ changes. While very little work has been
done in determining E for most radionuclides, some information has been published for
iodine-131 and cesium-137. Considering the differences between the child and the adult
thyroid, Bryant (Br69) estimates E to be 0.18 MeV/dis for the child and 0.19 MeV/dis for the
adult. The above values correspond to a 6-month-old child with a mass of 1.8 g and an f2
value of 0.35. The corresponding E value for the adult was calculated for a 20-g thyroid with
an f2 value of 0.3.
5-25
-------
Taking into account all the age-dependent factors discussed above, this analysis
indicates that, for a given concentration of 1-131 in milk, the estimated dose rate to the
thyroid of a 6-month-old child would be approximately 13 times that to an adult thyroid. In
other words, use of adult parameters would underestimate the thyroid dose to the child by
about a factor of 13.
5.3.1.6.8 Significance of Parameter Variability to EPA Dose and Risk Assessments
In its radiological risk assessments, EPA is generally interested in estimating the risk
to an average individual due to chronic lifetime exposures. Variation in dosimetric parameter
values among people and age groups is of reduced importance in this context because such
variation gets averaged over a population and/or over a lifetime. Nevertheless, it should be
kept in mind that some individuals in a population are going to be at higher risk from a given
exposure. Furthermore, despite such averaging, parameter value variability can contribute
substantially to the uncertainty in the dose and risk estimates.
Parameter value variation among individuals contributes uncertainty to the models by
causing random errors in any measured human data upon which the dosimetric models are
based. To the extent that the subjects from whom such data are collected are atypical of the
U.S. population (e.g., with respect to health status), parameter variation may also be a source
of bias. In this respect, since the parameters contained in the dosimetric models were
estimated for adult males, primarily, they may not provide an adequate basis for calculating
the average dose or risk in cases where age- and sex-related variations in these parameters are
large. This problem becomes more significant in light of the generally higher risks associated
with a given dose for childhood exposures (see Chapter 6); if doses are also higher in
childhood, the enhanced effect on risk will be compounded.
5.3.1.6.9 Past Approaches Used in Estimating Uncertainties in Calculated Organ Dose
As in any predictive exercise, it is useful to question the reliability of the predictions.
Variations in environmental levels, dietary and life style preferences, and the variability of
controlling physiological and metabolic processes contribute to the distribution of dose among
members of the exposed population. Superimposed on this variability is a component of
uncertainty arising from limitations in the predictive ability of the dosimetric models
themselves. Various approaches have been taken to understand and quantify these
uncertainties.
It has recently become popular to estimate the uncertainty by computing the
distribution of dose among exposed individuals.
This approach consists of repeated solution of the dosimetric model using parameter values
selected at random from a frequency distribution of potential values suggested in the
literature. It is assumed that the dosimetric model has been properly formulated, although
these models were developed to yield point estimates. Despite these and other difficulties,
propagation of parameter uncertainty through the dosimetric equation can provide a measure
of the model uncertainty. Application of these methods to the estimation of dose from
iodine-131 and cesium-137 ingestion can be found in the literature (Du81, Sc82).
5-26
-------
An alternative approach to assessing the potential variability is to consider that the
observed frequency distribution of a measurable quantity is closely related to dose. Cuddihy
and co-workers (Cu79) have investigated the variability of selected target organ deposition
among test animals and some individuals exposed. However, they did not address differences
in age, gender, magnitude or duration of exposure.
5.3.1.6.10 Uncertainty Classification of Radionuclides
In this section, radionuclides of interest are classified in terms of the uncertainties in
estimated dose per unit intake. Nuclides are placed in broad groups, largely reflecting the
general status of information on their bioMnetic behavior in the body. It is assumed that the
uncertainty associated with the calculation of the energy deposition in the target tissues is a
minor contributor to the overall uncertainty.
Classification of Uncertainty in Radionuclide Dose
Establishing numerical values of uncertainty for model dose estimates of each of the
many radionuclides, for each route of exposure, is a formidable task. Even if there is
agreement on the definition of uncertainty, any quantification will be arbitrary to a degree.
No model has been verified in man for any long-term exposure scenario; some of the models
may be fundamentally wrong in their formulation. In addition, the data selected to establish
the parameters used in the model may not be representative of the population being evaluated.
Most risk assessors use some informed scientific judgment in estimating the level of
uncertainty in a dose model.
A broad categorization of radionuclides reflecting the estimated magnitude of the
dosimetric uncertainties is presented. Because of the problems cited above with respect to the
development of models and model parameters, it is quite possible that the error in model
estimates may be larger than indicated in some cases. Nevertheless, this exercise is useful
since it provides some perspective on the magnitude of the uncertainties in light of current
evidence and focuses attention on the largest gaps in knowledge. Ultimately, however, better
quantification of dose estimates and their associated uncertainties can be obtained only
through the development and verification of improved dosimetric models.
Radioisotopes behave biologically like their stable elements. The elements, in rum,
can be broadly grouped as: (1) essential elements and their analogs, (2) inert gases, (3) well-
studied toxic metals and (4) others. Uncertainties for each of these categories will be
expressed as multiplicative factors, which roughly estimate the 95% upper and lower
confidence interval limits. [Since the interval is based on judgment, a preferable term would
be "credibility interval" (NIH85).]
Group I - Essential Elements and Their Analogs
Essential elements are controlled by homeostatic mechanisms to within narrow
tolerances. Usually, analogs of essential elements have distribution and deposition patterns
similar to those of the essential element The uncertainty expected in calculated dose for
essential elements is a factor of two or less in major critical organs, perhaps 3 or less in other
5-27
-------
significant tissues and organs. The expected dose uncertainty for analogs of essential
elements is perhaps a little greater, a factor of 3 or less in major organs and up to 5 or more
in less significant tissues. Important radionuclides of essential elements include hydrogen-3,
carbon-14, phosphorus-32, potassium-40, calsium-45, cobalt-60, iodine-129, and iodine-131;
important analogs include strontium-89, strontium-90, cesium-134, cesium-137, radium-226,
and radium-228.
Group II - Inert Gases
Uptake and retention of inhaled inert gases has been fairly well studied. The
uncertainty in dose, particularly average whole body dose, is not expected to be large.
However, the gases do not distribute uniformly in body tissues, and the effect of distribution
on organ dose estimates has not been carefully addressed. The uncertainty in the calculated
dose is expected to be about a factor of 2. This group includes, but is not limited to
argon-41, krypton-85, xenon-133, and radon-222.
Group III - Well-Studied Toxic Metals
A number of elements have been extensively studied in animals with limited
information available for man. Examples here include toxic elements encountered in
industrial activities, e.g., mercury, cadmium, lead, and uranium, for which studies were
carried out to help establish safe working conditions. Often the available information is not
sufficiently complete to identify the dominant processes governing the biokinetic behavior or
is simply fragmentary. For example, while much information exists on the biokinetics of
uranium, considerable uncertainty remains associated with the absorption to blood from the
small intestine. Uncertainties for dose estimates in this group of elements would be variable,
ranging from 2 or less for lead up to about 5 or more for polonium, thorium, uranium, and
the transuranics. Nuclides in this group include, but are not limited to lead-210,
polonium-210, uranium-235, uranium-238, thorium-230, thorium-232, plutonium-239,
plutonium-241, and americium-241.
Group IV - Other Elements
For a number of radionuclides information is largely limited to data from animal
studies. While animal studies often are the major source of detailed information on the
processes governing the biokinetics, the lack of a general framework for extrapolations to
man and the limited information upon which to judge the reasonableness of the extrapolations
suggest that the estimates must be considered to be potentially in error by at least an order of
magnitude. Nuclides in this group include, but are not limited to cerium-144 and other rare
earth elements, technetium-99, curium-244, californium-252, etc.
The groupings listed above represent the Agency's best judgment on the uncertainty of
internal radionuclide dose estimates. The primary source of uncertainty is in the biokinetic
modeling with little uncertainty in the physics. The magnitudes of the uncertainties posited
for each group of radionuclides should be regarded as only rough estimates; however, the
qualitative breakdown between groups is fairly reliable.
5-28
-------
Specific Problems
Certain radioisotopes and aspects of dosimetry pose unique problems. While the effect
of these problems may be to increase the uncertainty in dose estimates, the extent of such an
increase has yet to be evaluated.
Long-Lived Bone Seekers
Radioisotopes with effective half-lives that are short compared to the average life span
are expected to be in dynamic equilibrium. However, some bone seekers have long effective
half-lives; therefore, they do not reach dynamic equilibrium during a life span. Since the
relevant human biokinetic data are quite limited, dose estimates for such radionuclides are
more uncertain.
Nonuniformity of Distribution
The distribution of an element within an organ may not be uniform; in particular, the
distribution may be nonuniform with respect to biological targets of interest. This can be a
serious problem with respect to the estimation of relevant doses from internally deposited
alpha emitters, given the short range of alpha particles in matter. For example, where an
alpha emitter is distributed nonuniformly in bone, the calculation of doses to sensitive cells in
the bone and the bone marrow will be difficult. Another example is the uncertainty in
estimating doses to cells lining the GI tract from ingested alpha emitters passing through the
tract. In some cases, the mucus lining may effectively shield the target cells from irradiation.
5.3.2 External Dose Models
This section is concerned with the calculation of dose rates for external exposure to
photons from radionuclides dispersed in the environment Two exposure models are
discussed: (1) immersion in contaminated air and (2) irradiation from material deposited on
the ground surface. The immersion source is considered to be a uniform semi-infinite
radionuclide concentration hi air, while the ground surface irradiation source is viewed as a
uniform radionuclide concentration on an infinite plane. In both exposure modes, the dose
rates to organs are calculated from the dose rate in air.
Dose rates are calculated as the product of a dose rate factor, which is specific for
each radionuclide, tissue, and exposure mode, and the corresponding air or surface
concentration. The dose rate factors used were calculated with the DOSFACTOR code
(Ko81a,b). Note that the dose rate factors for each radionuclide do not include any
contribution for decay products. For example, the ground surface dose factors for cesium-137
are all zero, since no photons are emitted in its decay. To assess surface deposition of
cesium-137, the ingrowth of its decay product, metastable barium-137, which is a photon
emitter, must first be calculated.
5-29
-------
5.3.2.1 Immersion
For immersion exposure to the photons from radionuclides in air, EPA assumes that an
individual is standing at the base of a semi-infinite cloud of uniform radionuclide
concentration. First, the dose rate factor (the dose rate for a unit
concentration) in air is calculated for a source of photons with energy Er At all points in an
infinite uniform source, conservation of energy considerations require that the rates of
absorbed and emitted energy per unit mass be equal. The absorbed energy rate per unit mass
at the boundary of a semi-infinite cloud is just half that value. Hence
DRF° (Er) = l/2k £r/p (5-23)
where:
DRF* = the immersion dose rate per unit air
concentration (rad m3/Ci s);
Ey = emitted photon energy (MeV);
k = units conversion factor
= 1.62E-13 (J/MeV) x 3.7E+10 (dis/s-Ci) x
l.OE+3 (g/kg) x 100 (rad kg/J)
= 5.93E+2 (g rad/MeV Ci s); and
p = density of air (g/m3).
The above equation presumes that for each nuclide transformation, one photon with
energy Er is emitted. The dose rate factor for a nuclide is obtained by adding together the
contributions from each photon associated with the transformation process for that
radionuclide.
5.3.2.2 Ground Surface Irradiation
In the case of air immersion, the radiation field was the same throughout the source
region. This allows the dose rate factor to be calculated on the basis of energy conservation
without having to consider explicitly the scattering processes taking place. For ground
surface irradiation, the radiation field depends on the height of the receptor above the surface,
and the dose rate factor calculation is more complicated. The radiation flux per unit solid
angle is strongly dependent on the angle of incidence. It increases from the value for photons
incident from immediately below the receptor to a maximum close to the horizon.
Attenuation and buildup due to scattering must be considered to calculate the dose rate factor.
Secondary scattering provides a distribution of photon energies at the receptor, which
increases the radiation flux above that calculated on the basis of attenuation. Trubey (Tr66)
has provided a useful and reasonably accurate expression to approximate this buildup:
5-30
-------
B*n (wr) = 1 + Ca jj, r exp(DaMar) (5-24)
where B*n is the buildup factor (i.e., the quotient of the total energy flux and that calculated
for attenuation) only for energy in air; ju,, is the attenuation coefficient at the energy of the
released photon (m"1); r is the distance between the photon source and the receptor; and the
Berger buildup coefficients Ca and Da are dependent on energy and the scattering medium.
The buildup factor is dimensionless and always has a value greater than unity. The resulting
expression for the dose rate factor at a height z (m) above a uniform plane is
D*FYa(z,Er) = l/2k(E/p)(pen/p)a{E1(Maz) + (5-25)
Ca/(l-Da)exP[-(l-Da)Maz]}
where (uen/p)a is the mass energy-absorption coefficient (m2/g) for air at photon energy EY
(MeV); E! is the first order exponential integral function, i.e.,
Ex(x) = j exp (-u) du
u (5-26)
Ca and Da are the buildup coefficients in air at energy E^ and
k=5.93x!02 (g rad/MeV Ci s) as for the immersion calculation.
As for immersion, the dose rate factor for a nuclide combines the contribution from
each photon energy released in the transformation process.
5.3.2.3 Organ Doses
The dose rate factors in the preceding two sections are for the absorbed dose in air.
For a radiological assessment, the absorbed doses in specific tissues and organs are needed.
For this purpose, Kerr and Eckerman (Ke80, KeSOa) have calculated organ dose factors for
immersion in contaminated air. Their calculations are based on Monte Carlo simulations of
the absorbed dose in each tissue or organ for the spectrum of scattered photons in air
resulting from a uniform concentration of monoenergetic photon sources. Kocher (Ko81) has
used these data to calculate values of the ratio of the organ dose factor to the air dose factor,
Gk(EY), for 24 organs and tissues at 15 values of EY ranging from 0.01 to 10.0 MeV.
The resulting organ-specific dose rate factor for immersion is
= Gk(EY) ZM2FYa(EY) (5-27)
For a specific nuclide, the dose rate factor is obtained by taking the sum of the contributions
from each photon energy associated with the radionuclide decay.
5-31
-------
Ideally, a separate set of Gk(EY) values would be used for the angular and spectral
distributions of incident photons from a uniform plane source. Since these data are not
available, Kocher has used the same set of Gk(Ey) values for calculating organ dose rate
factors for both types of exposure (Ko81).
5.3.2.4 Uncertainty Considerations in External Dose Rate Factors
In computing the immersion dose rate factor in air, the factor of 1/2 in Equation 5-27,
which accounts for the semi-infinite geometry of the source region, does not provide a
rigorously correct representation of the air/ground interface. However, Dillman (Di74) has
concluded that this result is within the accuracy of available calculations. The radiation field
between the feet and the head of a person standing on contaminated ground is not uniform,
but for source photon energies greater than about 10 keV, the variation about the value at 1
meter becomes minimal. A more significant source of error is the assumption of a uniform
concentration. Kocher (Ko81) has shown that sources would have to be approximately
uniform over distances of as much as a few hundred meters from the receptor for the dose
rate factors to be accurate for either ground surface or immersion exposures. Penetration of
deposited materials into the ground surface, surface roughness, and terrain irregularities, as
well as the shielding provided by buildings to their inhabitants, all serve to reduce doses.
The effect of using the same factors to relate organ doses to the dose in air for ground
surface as for immersion photon sources has not been studied. The assumptions that the
radiation field for the ground surface source is isotropic and has the same energy distribution
as for immersion clearly do not hold true, but more precise estimates of these distributions are
not likely to change the organ dose rate factors substantially.
Kocher (Ko81) has noted that the idealized photon dose rate factors are "likely to be
used quite extensively even for exposure conditions for which they are not strictly
applicable... because more realistic estimates are considerably more difficult and expensive [to
make]."
5-32
-------
Chapter 5 References
Be70 Bernard, J.D., McDonald, R.A., and Nesmith, J.A., "New Normal Ranges for
the Radioiodine Uptake Study", J. Nucl. Med., ii:(7):449-451, 1970.
Br52 Bruckner, H. Die Anatomic der Lufttrohre beim lebenden Menchen, A.
AnaL, Entwicklungsgeschichte, 116:276, 1952 [cited in Li69].
Br69 Bryant, P.M., "Data for Assessments Concerning Controlled and Accidental
Releases of 131I and 137Cs to Atmosphere", Health Phvs., 17(l):51-57, 1969.
Cr83 Crawford, D.J., An Age-Dependent Model for the Kinetics of Uptake and
Removal from the G.I. Tract, Health Phys. 44- 609-622, 1983.
Cu79 Cuddihy, R.G., McCleUan, R.O., and Griffith, W.C., Variability in Target
Deposition Among Individuals Exposed to Toxic Substances, Toxicol. Appl.
Pharmacol. 49: 179-187, 1979.
Di74 Dillman, L.T., "Absorbed Gamma Dose Rate for Immersion in a Semi-
Infinite Radioactive Cloud", Health Phvs., 27(6):571, 1974.
Du79 Dunning, D.E. Jr., Bernard, S.R., Walsh, P.J., Killough, G.G. and Pleasant,
J.C., Estimates of Internal Dose Equivalent to 22 Target Organs for
Radionuclides Occurring in Routine Releases from Nuclear Fuel-Cycle
Facilities, Vol. H, Report No. ORNL/NUREG/TM-190/V2, NUREG/CR-
0150 Vol. 2, Oak Ridge National Laboratory, Oak Ridge, Tennessee, 1979.
Du80 Dunning, D.E. Jr., Leggett, R.W., and Yalcintas, M.G., "A Combined
Methodology for Estimating Dose Rates and Health Effects from Exposure
to Radioactive Pollutants," ORNL/TM-7105, 1980.
Du81 Dunning, D.E. and Schwartz, G., "Variability of Human Thyroid
Characteristics and Estimates of Dose from Ingested 131I", Health Phys.,
40(5):661-675, 1981.
EPA77 U.S. Environmental Protection Agency, Proposed Guidance in Dose Limits
for Persons Exposed to Transuranium Elements in the General Environment,
EPA 520/4-77-016, 1977.
Fi35 Findeisen, W., Uber das Absetzen Kleiner in der Luft Suspendierten
Teilchen in der Menschlichen Lunge bei der Atmung, Pflugers Arch, f d
ges. Phvsiol., 236, 367, 1935.
FRC67 Federal Radiation Council, Guidance for the Control of Radiation Hazards in
Uranium Mining, FRC Report No. 8, Revised, U.S. Government Printing
Office, Washington, D.C., 1967.
5-33
-------
Ho75
Hu72
Hu73
ICRP59
ICRP66
ICRP75
ICRP77
ICRP79
ICRP80
ICRP81
ICRP84
ICRU80
Holden, W.S., and Marshal, R., "Variations in
Clin. Radiol., 26:439-454, 1975.
Bronchial Movement",
Hughes, J.M.B., Hoppin, F.G., Jr. and Mead, J., "Effect of Lung Inflation on
Bronchial Length and Diameter in Excised Lungs", J. Appl. Physiol, 32:25-
35, 1972.
Hursh, J.B., and Spoor, N.L., "Data on Man", Chapter 4 in Uranium,
Plutonium and the Transplutonic Elements, Springer, New York, 1973.
International Commission on Radiological Protection, Report of Committee
II on Permissible Dose for Internal Radiation, ICRP Publication 2, Pergamon
Press, Oxford, 1959.
ICRP Task Group on Lung Dynamics, "Depositions and Retention Models
for Internal Dosimetry of the Human Respiratory Tract", Health Phys.,
12(2): 173-207, 1966.
International Commission on Radiological Protection, Report on the Task
Group on Reference Man, ICRP Publication No. 23, Pergamon Press,
Oxford, 1975.
International Commission on Radiological Protection, "Recommendations of
the International Commission on Radiological Protection", ICRP Publication
26, Annals of the ICRP, Vol. 1, No. 3, Pergamon Press, Oxford, 1977.
International Commission on Radiological Protection, Limits for Intakes of
Radionuclides by Workers, ICRP Publication No. 30, Pergamon Press,
Oxford, 1979.
International Commission on Radiological Protection, "Limits for Intakes of
Radionuclides by Workers", ICRP Publication 30, Part 2, Annals of the
ICRP, Vol. 4 (3/4), Pergamon Press, Oxford, 1980.
International Commission on Radiological Protection, "Limits for Intakes of
Radionuclides by Workers", ICRP Publication 30, Part 3, Annals of the
ICRP, Vol. 6 (2/3), Pergamon Press, Oxford, 1981.
International Commission on Radiological Protection, "A Compilation of the
Major Concepts and Quantities in Use by ICRP", ICRP Publication No. 42,
Pergamon Press, Oxford, 1984.
International Commission on Radiation Units and Measurements, ICRU
Report No 33, Washington, D.C., 1980.
5-34
-------
Ke80
KeSOa
Ki78a
Ki78b
KoSla
KoSlb
NAS80
NCRP71
Ne84
NIH85
ORNL81
Kerr, G.D., and Eckerman, K.F., Oak Ridge National Laboratory, private
communication; see also Abstract P/192 presented at the Annual Meeting of
the Health Physics Society, Seattle, Washington, July 20-25, 1980.
Kerr., G.D., "A Review of Organ Doses from Isotropic Fields of X-Rays",
Health Phvs., 39(1):3, 1980.
Killough, G.C., Dunning, D.E Jr., Bernard, S.R. and Pleasant, J.C., Estimates
of Internal Dose Equivalent to 22 Target Organs for Radionuclides
Occurring in Routine Releases from Nuclear Fuel Cycle Facilities, Vol. 1,
Report No. ORNL/NUREG/TM-190, Oak Ridge National Laboratory,
Tennessee, June 1978.
KUlough, G.C., and Rohwer, P.S., "A New Look at the Dosimetry of 14C
Released to the Atmosphere as Carbon Dioxide", Health Phys., 34(2): 141,
1978.
Kocher, D.C., and Eckerman, K.F., "Electron Dose-Rate Conversion Factors
for External Exposure of the Skin", Health Phvs.. 40(1):67, 1981.
Kocher, D.C., "Dose-Rate Conversion Factors for External Exposure to
Photon and Electron Radiation from Radionuclides Occurring in Routine
Releases from Nuclear Fuel-Cycle Facilities", Health Phvs.. 38(4):543-621,
1981.
National Academy of Sciences - National Research Council, The Effects on
Populations of Exposure to Low Levels of Ionizing Radiation, Report of the
Committee on the Biological Effects of Ionizing Radiation (BEIR HI),
Washington, D.C., 1980.
National Council on Radiation Protection and Measurements, Basic
Radiation Protection Criteria, NCRP Report No. 39, Washington, D.C.,
1971.
Nelson, C.B., and Yang, Y., An Estimation of the Daily Average Food
Intake by Age and Sex for Use in Assessing the Radionuclide Intake of
Individuals in the General Population, EPA 520/1-84-015, 1984.
National Institutes of Health, Report of the National Institutes of Health Ad
Hoc Working Group to Develop Radioepidemiological Tables, NIH
Publication No. 85-2748, U.S. Government Printing Office, Washington, DC
20402, p 92, 1985.
Oak Ridge National Laboratory, Estimates of Health Risk from Exposure to
Radioactive Pollutants, ORNL/RM-7745, Oak Ridge, Tenn., 1981.
5-35
-------
ORNL85 Oak Ridge National Laboratory, "Report of Current Work of the
Metabolism and Dosimetry Research Group", ORNL/TM-9690, Oak Ridge,
Tennessee, 1985.
Ph85 Phalen, R.F., Oldham, M.J., Beaucage, C.B., Crocker, T.T., and Mortensen,
J.D., Postnatal Enlargement of Human Tracheobronchial Airways and
Implications for Particle Deposition, AnaL Rec. 212: 368, 1985.
Ro58 Rosenberg, G., "Biologic Half-life of 131I in the Thyroid of Healthy Males",
J. Clin. Endocrinol. Metab., 18, 516-521, 1958.
Sc82 Schwarz, G., and Dunning, Jr., D.E., Imprecision hi Estimates of Dose from
Ingested Cs-137 due to Variability in Human Biological Characteristics,
Health Phys. 43, 631-645, 1982.
Sn74 Snyder W.S., Ford, M.R., Warner, G.G., and Watson, S.B., A Tabulation of
Dose Equivalent per Microcurie-Day for Source and Target Organs of an
Adult for Various Radionuclides, Oak Ridge National Laboratory,
ORNL-5000, 1974.
Sp73 Spoor, N.L., and Hursh, J.B., "Protection Criteria", Chapter 5 in Uranium,
Plutonium and the Transplutonic Elements, Springer, New York, 1973.
Su81 Sullivan, R.E., Nelson, N.S., Ellett, W.H., Dunning, D.E. Jr., Leggett, R.W.,
Yalcintas, M.G. and Eckerman, K.F., Estimates of Health Risk from
Exposure to Radioactive Pollutants, Report No. ORNL/TM-7745, Oak Ridge
National Laboratory, Oak Ridge, Tennessee, 1981.
Th77 Thorne, M.D., "Aspects of the Dosimetry of Alpha-Emitting Radionuclides
in Bone with Particular Emphasis on 226Ra and 239Pu", Phvs. Med. Biol..
22:36-46, 1977.
Th78 Thurlbeck, W.M. "Miscellany", 287-315 in The Lung: Structure Function
and Disease, Thurlbeck, W.M. and Abell, M.R., editors, The Williams and
Wilkins Co., Baltimore, Maryland, 1978.
Tr66 Trubey, D.K., A Survey of Empirical Functions Used to Fit Gamma-Ray
Buildup Factors, Oak Ridge National Laboratory Rep., ORNL-RSIC-10,
1966.
5-36
-------
Chapter 6: ESTIMATING THE RISK OF HEALTH EFFECTS RESULTING FROM
EXPOSURE TO LOW LEVELS OF IONIZING RADIATION
6.1 INTRODUCTION
This chapter describes how EPA estimates the risk of fatal cancer, serious genetic
effects, and other detrimental health effects caused by exposure to low levels of ionizing
radiation.
Ionizing radiation refers to radiation that strips electrons from atoms in a medium
through which it passes. The highly reactive electrons and ions created by this process in a
living cell can produce, through a series of chemical reactions, permanent changes (mutations)
in the cell's genetic material, the DNA. These may result in cell death or in an abnormally
functioning cell. A mutation in a germ cell (sperm or ovum) may be transmitted to an
offspring and be expressed as a genetic defect in that offspring or in an individual of a
subsequent generation; such a defect is commonly referred to as a genetic effect. There is
also strong evidence that the induction of a mutation by ionizing radiation in a non-germ
(somatic) cell can serve as a step in the development of a cancer. Finally, mutational or other
events, including possible cell killing, produced by ionizing radiation in rapidly growing and
differentiating tissues of an embryo or fetus can give rise to birth defects; these are referred
to as teratological effects. At acute doses above about 25 rads, radiation induces other
deleterious effects in man; however, for the low doses and dose rates of interest in this
document, only those three kinds of effects referred to above are thought to be significant.
Most important from the standpoint of the total societal risk from exposures to low-
level ionizing radiation are the risks of cancer and genetic mutations. Consistent with our
current understanding of their origins in terms of DNA damage, these are believed to be
stochastic effects; i.e., the probability (risk) of these effects increases with the absorbed dose
of radiation, but the severity of the effects is independent of dose. For neither induction of
cancer nor genetic effects, moreover, is there any convincing evidence for a "threshold," i.e.,
some dose level below which the risk is zero. Hence, so far as is known, any dose of
ionizing radiation, no matter how small, might give rise to a cancer or to a genetic effect in
future generations. Conversely, there is no way to be certain that a given dose of radiation,
no matter how large, has caused an observed cancer in an individual or will cause one in the
future.
Beginning nearly with the discovery of x-rays in 1895 but especially since World War
n, an enormous amount of research has been conducted into the biological effects of ionizing
radiation. This research continues at the level of the molecule, the cell, the tissue, the whole
laboratory animal, and man. There are two fundamental aspects to most of this work:
1. Estimating the radiation dose to a target (cell, tissue, etc.). This aspect
(dosimetry), which may involve consideration of physiological, metabolic, and
other factors, is discussed more fully in Chapter 5.
2. Measuring the number of effects of a given type associated with a certain dose
(or exposure).
6-1
-------
For the purpose of assessing the risk to man from exposures to ionizing radiation, the
most important information comes from human epidemiological studies in which the number
of health effects observed in an irradiated population is compared to that in an unirradiated
control population. The human epidemiological data regarding radiation-induced cancer are
extensive. As a result, the risk can be estimated to within an order of magnitude with a high
degree of confidence. Perhaps for only one other carcinogen - tobacco smoke - is it possible
to estimate risks more reliably.
Nevertheless, there are gaps in the human data on radiation risks. No clear-cut
evidence of excess genetic effects has been found in irradiated human populations, for
example, probably due to the limited numbers in the exposed cohort providing inadequate
power to detect a dose-response. Likewise, no statistically significant excess of cancers has
been demonstrated below about 5 rads, the dose range of interest from the standpoint of
environmental exposures. Since the epidemiological data are incomplete in many respects,
risk assessors must rely on mathematical models to estimate the risk from exposures to low-
level ionizing radiation. The choice of models, of necessity, involves subjective judgments
but should be based on all relevant sources of data collected by both laboratory scientists and
epidemiologists. Thus, radiation risk assessment is a process that continues to evolve as new
scientific information becomes available.
The EPA estimates of cancer and genetic risks used here are based largely on the
results of a National Academy of Sciences (NAS) study as given in the BEIR El report
(NAS80). The study assessed radiation risks at low exposure levels. As phrased by the
President of the Academy, "We believe that the report will be helpful to the EPA and other
agencies as they reassess radiation protection standards. It provides the scientific bases upon
which standards may be decided after nonscientific social values have been taken into
account."
In this discussion, the various assumptions made in calculating radiation risks based on
the 1980 NAS report are outlined, and these risk estimates are compared with those prepared
by other scientific groups, such as the 1972 NAS BEIR Committee (NAS72), the United
Nations Scientific Committee on the Effects of Atomic Radiation (UNSC77, 82, 86, 88), and
the National Radiological Protection Board of the United Kingdom (St88). Because
information on radiation risks is incomplete, estimates of risk based on the various models
may not be highly accurate. This discussion identifies some of the deficiencies in the
available data base and points out possible sources of bias in current risk estimates.
Nevertheless, the risk estimates made by EPA are believed to be reasonable in light of current
evidence.
Sections 6.2 to 6.2.6 consider the cancer risk resulting from whole-body exposure to
low-LET (see Chapter 5) radiation, i.e., sparsely ionizing radiation like the energetic electrons
produced by x-rays or gamma rays. Environmental contamination by radioactive materials
also leads to the ingestion or inhalation of the material and subsequent concentration of the
radioactivity in selected body organs. Therefore, the cancer risk resulting from low-LET
irradiation of specific organs is examined in Sections 6.2.7 to 6.2.9. Sections 6.2.10 to
6.2.12 summarize recent developments in radiation risk estimation and discuss the
uncertainties in the estimates.
6-2
-------
Organ doses can also result from high-LET radiation, such as that associated with
alpha particles. The cancer risks when high-LET radiation is distributed more or less
uniformly within a body organ is the third situation considered (Section 6.3). Because
densely ionizing alpha particles have a very short range in tissue, there are exposure situations
where the dose distribution to particular organs is extremely nonuniform. An example is the
case of inhaled radon progeny, Po-218, Pb-214, and Po-214. For these radionuclides, cancer
risk estimates are based on the amount of radon progeny inhaled rather than the estimated
dose, which is highly nonuniform and cannot be well quantified. Therefore, risk estimates of
radon exposure are examined separately (Section 6.4).
Section 6,5 reviews and quantifies the risk of deleterious genetic effects from radiation
and the effects of exposure in utero on the developing fetus. Finally, in Section 6.6, cancer
and genetic risks from background radiation are calculated using the models described in this
chapter.
6.2 CANCER RISK ESTIMATES FOR LOW-LET RADIATION
6.2.1 Basis for Risk Estimates
There are extensive human epidemiological data upon which to base risk estimates for
radiation-induced cancers. Most of the observations of radiation-induced carcinogenesis in
humans are of groups exposed to low-LET radiations. These groups include the Japanese A-
bomb survivors and medical patients treated with diagnostic or therapeutic radiation, most
notably for ankylosing spondylitis in England from 1935 to 1954 (Sm78). Comprehensive
reviews of these and other data on the carcinogenic effects of human exposures are available
(UNSC77, NAS80).
The most important source of epidemiological data on radiogenic cancer is the
population of Japanese A-bomb survivors. The A-bomb survivors have been studied for more
than 38 years, and most of them (the Life Span Study Sample) have been followed since 1950
in a carefully planned and monitored epidemiological survey (Ka82, Wa83). They are the
largest group that has been studied, and they provide the most detailed information on the
response pattern for organs, by age and sex, over a wide range of doses of low-LET radiation.
Unfortunately, the 1980 BEIR Committee's analysis of the A-bomb survivor data collected up
to 1974 was prepared before bias in the dose estimates for the survivors (the tentative 1965
dose estimates, T65) became widely recognized (Lo81). It is now clear that the T65 dose
equivalents to organs tended, on average, to be overestimated (Bo82, RERF83,84) so that the
BEIR Committee's estimates of the risk per unit dose are likely to be too low. A new
dosimetry system, termed the Dosimetry System 1986 (DS86), is now nearly complete, and
preliminary analyses of the risk based on DS86 have been published (Pr87,88; Sh87).
At present, the "BEIR V Committee" of the National Academy of Sciences is
preparing a report on radiation risks in light of DS86 and other new information. A detailed
reevaluation of EPA's current risk estimates is indicated when this report is issued. A brief
discussion of the new dosimetry and its likely effect on risk estimates is included.
6-3
-------
To derive risk estimates for environmental exposures of the general U.S. population
from epidemiological studies of irradiated populations requires some extrapolation. First,
much of the useful epidemiological data pertain to acute doses of 50 rad or higher, whereas
we are concerned with small chronic doses incremental to the natural background level of
about 100 mrad/year. Second, epidemiological follow-up of the irradiated study cohorts is
incomplete; hence, obtaining lifetime risk estimates involves some projection of risk beyond
the period of follow-up. Third, an extrapolation must be made from a study population to the
U.S. population. In general, these populations will differ in various respects, for example,
with respect to organ-specific, base-line cancer rates.
Data pertaining to each of these three extrapolations exist, but in no case are they
definitive. Hence, uncertainty in our risk estimates is associated with each of them. These
uncertainties are in addition to statistical uncertainties in the epidemiological data (sampling
variations) and errors in dose determinations. Generally speaking, it is the former, modeling
uncertainties, which are more important.
6.2.2 Dose Response Functions
Radiogenic cancers in humans have been observed, for the most part, only following
doses of ionizing radiation that are relatively high compared to those likely to result from a
combination of background radiation and environmental contamination from controllable
sources of radiation. Therefore, a dose response model must be chosen to allow extrapolation
from the number of radiogenic cancers observed at high doses to the number of cancers at
low doses resulting from all causes including background radiation.
The range of extrapolation is not the same for all kinds of cancer because it depends
upon the radiosensitivity of a particular tissue. For example, the most probable radiogenic
cancer for women is breast cancer. The incidence of radiogenic breast cancer does not seem
to diminish when the dose is protracted over a long period of time. For example, the number
of excess cancers per unit dose among Japanese women, who received acute doses, is about
the same per unit dose as women exposed to small periodic doses of x-rays over many years.
If this is actually the case, background radiation is as carcinogenic per unit dose for breast
tissue as the acute exposures from A-bomb gamma radiation.
Moreover, the female A-bomb survivors show an excess of breast cancer at doses
below 20 rads which is linearly proportional to that observed at several hundred rads (To84).
(Evidence of a nonlinear dose response relationship for induction of breast cancer has been
obtained in a study of Canadian fluoroscopy patients, but only at doses above about 500 rads
(Ho84). Women in their 40s, the youngest age group in which breast cancer is common,
have received about 4 rads of whole-body low-LET background radiation and usually some
additional dose incurred for diagnostic medical purposes. Therefore, for this cancer, the
difference between the lowest dose at which radiogenic cancers are observed, less than 20
rads, and the dose resulting from background radiation is less than a factor of 5, not several
orders of magnitude as is sometimes claimed. Based on data from irradiated tinea capitis
patients, induction of thyroid cancer also seems to be linear with doses down to 10 rads or
lower (NCRP85). However, for most other cancers, a statistically significant excess has not
6-4
-------
been observed at doses below 50 rads of low-LET radiation. Therefore, the range of dose
and dose rate extrapolation is often large.
The 1980 NAS report (NAS80) examined three dose response functions in detail: (1)
linear, in which the number of effects (risk) is directly proportional to dose at all doses; (2)
linear-quadratic, in which risk is very nearly proportional to dose at very low doses and
proportional to the square of the dose at high doses; and (3) quadratic, where the risk varies
as the square of the dose at all dose levels.
The 1980 NAS BEIR Committee considered only the Japanese mortality data in its
analysis of possible dose response functions (NAS80). Based on the T65 dose estimates, this
Committee concluded that the excess mortality from solid cancers and leukemia among the A-
bomb survivors is compatible with either a linear or linear-quadratic dose response to the low-
LET radiation component and a linear response to the high-LET neutron component (NAS80).
Although the 1980 BEIR report indicated risk estimates for low-LET radiation based on a
linear-quadratic response were "preferred" by most of the scientists who prepared that report,
opinion was not unanimous, and the subsequent reassessment of the A-bomb dose weakens
the Committee's conclusion. The Committee's analysis of dose response functions was based
on the assumption that most of the observed excess leukemia and solid cancers among
survivors in Hiroshima resulted from neutrons (see Tables V-13, A-7, Equations V-10, V-ll
in NAS 80). Current evidence, however, is conclusive that neutrons were only a minor
component of the dose among all but a few survivors in both Hiroshima and Nagasaki (Bo82;
RERF83, 84; Pr87; Sh87). Therefore, it is likely that most of the response attributed to
neutrons was caused by the gamma dose, not the dose from neutrons.
Under the revised DS86 dosimetry, the A-bomb survivor data is more consistent with
a linear dose response than under T65. Indeed, the linear coefficient obtained by fitting a
linear-quadratic function to the data for either leukemia or solid tumors differs only slightly
from the respective proportionality constant obtained by fitting a simple linear function
(Sh88). Thus, the linear and linear-quadratic functions derived from statistical fits to the
Japanese DS86 data yield very similar predictions at low doses. Other human data,
particularly that relating to induction of breast cancer (NAS80, NJH85), also lend support to a
linear dose response for radiogenic human cancers.
On the other hand, there is extensive laboratory evidence on irradiated animals and
cellular preparations which indicates that the effectiveness of low-LET radiation is
substantially reduced at low doses and low dose rates. Guided by those observations, as well
as by the Japanese data interpreted according to the T65 dosimetry system, the BEIR HI
committee expressed preference for a linear-quadratic dose response model for low-LET
radiations.
For low-LET radiations, the BEIR JJI Committee preferred the linear-quadratic dose
response model. In this model, the risk from an acute dose, D, of low-LET radiation is
assumed to be of the form aD + pD2. The BEIR UJ Committee assumed that the linear and
quadratic terms were equal at 116 rads, leading to a linear coefficient a, which was about a
factor of 2.5 times lower than the coefficient obtained from the linear model (NAS80). At
low doses, the quadratic term becomes negligible; at chronic low-dose rates it is ignored, for
6-5
-------
reasons discussed below. For environmental exposures, therefore, risk estimates based on the
BEIRIH linear-quadratic dose response model are only about 40 percent of those based on
the BEIR HI linear model.
Building on earlier work by Lea (Lea62), a theoretical basis for the linear-quadratic
dose response model has been put forth by Kellerer and Rossi (Ke72). In this theory of "dual
radiation action," events leading to "lesions" (i.e., permanent changes) in cellular DNA require
the formation of interacting pairs of "sublesions." The interacting pairs can be produced by a
single traversing particle, or track, or by two separate tracks, giving rise, respectively, to a
linear and quadratic term in the dose response relationship. According to the theory, a
sublesion may be repaired before it can interact to form a lesion, the probability of such
repair increasing with time. Consequently, as dose rate is reduced, the formation of lesions
from sublesions caused by separate tracks becomes less important, and the magnitude of the
D2 term diminishes. Hence, the theory predicts that at sufficiently low doses or dose rates,
the response should be a linear function of dose. Moreover, the constant of proportionality is
the same in both cases: i.e., a.
Results of many animal and cellular experiments are qualitatively consistent with the
theory: low-LET radiation often seems to have a reduced effectiveness per unit dose at low
dose rates (NCRP80). However, it is usually not possible from the data to verify that the
dose response curve has the linear-quadratic form. Another success of the dual action theory
has been in explaining observed differences between the effects of low-LET and high-LET
radiations. In this view, the densely ionizing nature of the latter results in a much greater
production of interacting pairs of sublesions by single tracks, leading in turn to higher relative
biological effectiveness at low doses and a linear dose response relationship for high-LET
radiation (except for possible cell-killing effects).
The dual action theory has nevertheless been challenged on experimental grounds, and
observed variations in response with dose, dose rate (see below), and LET can also be
explained in terms of a theory involving only single lesions and a "saturable" repair
mechanism that decreases in effectiveness at high dose rates on the microscopic scale (To65,
Go82). One property of such a theory is that the effectiveness of repair, and therefore the
shape of the dose response curve, can in principle vary substantially with cell type and
species. Hence, results obtained on laboratory animals would not necessarily be entirely
applicable to people.
The quadratic model was put forward in the BEIR m Report, in large part, to account
for observed differences in solid tumor induction between Hiroshima and Nagasaki. In
Hiroshima, the dose-response appeared linear, but in Nagasaki it appeared quadratic. Rossi
suggested that the cancers in Hiroshima were mostly due to neutron doses, while in Nagasaki
neutrons were largely absent, so the observed quadratic dose-response there reflected the
"true" response to gamma rays (NAS80). With the revisions in A-bomb dosimetry, this
rationale is lost Preliminary analyses based on DS86 dosimetry indicate that the quadratic
model generally provides a poorer fit to the data than do the other two models (Sh88). Some
laboratory evidence also suggests that the risk in humans may increase linearly with dose at
low doses (Gr85). Thus, though a quadratic dose-response at low doses (or even a threshold)
6-6
-------
cannot now be definitively ruled out, EPA does not consider such models suitable for
radiation risk assessment.
Finally, "supralinear models," in which the risk coefficient decreases with increasing
dose (downward bending, or convex, dose response curve) should be mentioned. Such
models imply that the risk at low doses would actually be greater than predicted by linear
interpolation from higher doses. The evidence from radiation biology investigations, at the
cellular as well as the whole animal level, indicates that the dose response curve for induction
of mutations or cancer by low-LET radiation is either linear or concave upward for doses to
mammalian systems below about 250 rads (NCRP80). Somewhere above this point, the dose
response curve often begins to bend over: this is commonly attributed to "cell-killing." The
A-bomb survivor data, upon which most of these risk estimates depend, is dominated by
individuals receiving about 250 rads or less. Consequently, the cell-killing phenomenon
should not produce a substantial underestimate of the risk at low doses.
Noting that human beings, in contrast to pure strains of laboratory animals, may be
highly heterogeneous with respect to radiation sensitivity, Baum (Ba73) proposed an
alternative mechanism by which a convex dose response relationship could arise. He pointed
out that sensitive subgroups may exist in the population who are at very high risk from
radiation. The result could be a steep upward slope in the response at low doses,
predominantly reflecting the elevated risk to members of these subgroups, but a decreasing
slope at higher doses as the risk to these highly sensitive individuals approaches unity.
Based on current evidence, however, it seems unlikely that the effect postulated by
Baum would lead to substantial overestimation of the risk at low doses. While there may
indeed be small subgroups at very high risk, it is difficult to reconcile the A-bomb survivor
data with a strongly convex dose response relationship. For example, if most of the
leukemias found among the cohort receiving about 200 rads or more hi fact arose from
subgroups whose risk saturated below 200 rads, then many more leukemias ought to have
occurred in lower dose cohorts than were actually observed. The U.S. population, it could be
argued, may be more heterogeneous with respect to radiation sensitivity than the Japanese.
The risk of radiation-induced breast cancer appears, however, to be similar in the two
populations, so it is difficult to see how the size of the hypothetical sensitive group could be
large enough in the former to alter the conclusion reached above. The linear dose-response
relationship seen for radiogenic breast cancer in several populations (NIH85) further argues
against Baum's hypothesis.
6.2.3 The Possible Effects of Dose Rate on Radiocarcinogenesis
The BEIR El Committee limited its risk estimates to a minimum dose rate of 1 rad
per year and stated that it "does not know if dose rates of gamma rays and x-rays of about
100 mrad/yr are detrimental to man." At dose rates comparable to the background everyone
receives from naturally occurring radioactive materials, a considerable body of scientific
opinion holds that the effects of radiation are reduced compared to high dose rates. NCRP
Committee 40 has suggested that carcinogenic effects of low-LET radiations may be a factor
of from 2 to 10 times less per unit dose for small doses and dose rates than have been
observed at high doses and dose rates (NCRP80).
6-7
-------
The low dose and low dose rate effectiveness factors estimated by NCRP Committee
40 are based on its analysis of a large body of plant and animal data that showed reduced
effects at low doses for a number of biological endpoints, including radiogenic cancer in
animals, chiefly rodents. However, no data for cancer in humans confirm these findings;
indeed, human studies where there are sufficient data to develop a dose-response function for
organ exposure seem to contradict them. Highly fractionated small doses to human breast
tissue are apparently as carcinogenic as large acute doses (NAS80, La80). Small acute doses
(less than 10 rads) to the thyroid have been found to be as effective per rad as much larger
doses in initiating thyroid cancer (UNSC77, NAS80). Also relevant in this connection,
perhaps, is the finding that a radiation-induced mutation increased linearly with dose, and
independently of dose rate, in human cells but not in rodent cells (Gr85).
While none of these examples is persuasive by itself, collectively they indicate that it
may not be prudent to assume that all kinds of cancers are reduced at low dose rates and/or
low doses. However, it may be overly conservative1 to estimate the risk of all cancers on the
basis of the linearity observed for breast and thyroid cancer. The ICRP and UNSCEAR have
used a dose rate effectiveness factor (DREF) of about 2.5 to estimate the risks from
occupational (ICRP77) and environmental exposures (UNSC77). That choice of a DREF is
fully consistent with and equivalent to the reduction of risk at low doses obtained by
substituting the BEER in linear-quadratic response model for their linear model (see above).
Therefore, use of both a DREF and a linear-quadratic model for risk estimation in the low-
dose region is inappropriate (NCRP80).
6.2.4 Risk Projection Models
None of the exposed populations have been observed long enough to assess the full
effects of their exposures if, as currently thought, most radiogenic cancers occur throughout
an exposed person's lifetime (NAS80). Therefore, another major choice that must be made in
assessing the lifetime cancer risk due to radiation is to select a risk projection model to
estimate the risk for a longer period of time than currently available observational data will
allow.
To estimate the risk of radiation exposure that is beyond the years of observation,
either a relative risk or an absolute risk projection model (or suitable variations) may be used.
These models are described at length in Chapter 4 of the 1980 NAS report (NAS80). The
relative risk projection model projects the currently observed percentage increase in annual
cancer risk per unit dose into future years, i.e., the increase is proportional to the underlying
(baseline) risk. An absolute risk model projects the average annual number of excess cancers
per unit dose into future years at risk, independent of the baseline risk.
1 Risk assessments require choosing among alternative assumptions, none of which can be
definitively shown to be more accurate than the others. A conservative choice, in this
connection, is one leading to higher estimates of risk.
6-8
-------
Because the underlying risk of most types of cancer increases rapidly with age, the
relative risk model predicts a larger probability of excess cancer toward the end of a person's
lifetime. In contrast, the absolute risk model predicts a constant incidence of excess cancer
across time. Therefore, given the incomplete data and less than lifetime follow-up, a relative
risk model projects a somewhat greater total lifetime cancer risk than that estimated using an
absolute risk model.
Neither the NAS BEIR Committee nor other scientific groups (e.g., UNSCEAR) have
concluded which projection model is the more appropriate choice for most radiogenic cancers.
However, recent evidence favors the relative risk projection model for most solid cancers. As
pointed out by the 1980 NAS BEER. Committee:
If the relative-risk model applies, then the age of the exposed groups, both at
the time of exposure and as they move through life, becomes very important
There is now considerable evidence in nearly all the adult human populations
studied that persons irradiated at higher ages have, in general, a greater excess
risk of cancer than those irradiated at lower ages, or at least they develop
cancer sooner. Furthermore, if they are irradiated at a particular age, the
excess risk tends to rise pari passu [at equal pace] with the risk of the
population at large. In other words, the relative-risk model with respect to
cancer susceptibility at least as a function of age, evidently applies to some
kinds of cancer that have been observed to result from radiation exposure.
(NAS80, p.33)
This observation is confirmed by the Ninth A-bomb Survivor Life Span Study,
published two years after the 1980 Academy report. This latest report indicates that, for solid
cancers, relative risks have continued to remain constant in recent years, while absolute risks
have increased substantially (Ka82). Smith and Doll (Sm78) have reached similar conclusions
on the trend in excess cancer with time among the irradiated spondylitic patients. More
recent analysis of the spondylitic data does show evidence of a fall-off in relative risk after 25
years post-exposure, but the decrease is not yet statistically significant (Da86).
Although considerable weight should be given to the relative risk model for most solid
cancers (see below), the model does not necessarily give an accurate projection of lifetime
risk. The mix of tumor types varies with age so that the relative frequency of some common
radiogenic tumors, such as thyroid cancer, decreases for older ages. Land has pointed out that
this may result in overestimates of the lifetime risks when they are based on a projection
model using relative risks (La83). While this may turn out to be true for estimates of cancer
incidence that include cancers less likely to be fatal, e.g., thyroid, it may not be very
important in estimating the lifetime risk of fatal cancers, since the incidence of most of the
common fatal cancers, e.g., breast and lung cancers, increases with age.
Leukemia and bone cancer are exceptions to the general validity of a lifetime
expression period for radiogenic cancers. Most of the leukemia risk has apparently already
been expressed in both the A-bomb survivors and the spondylitics (Ka82, Sm78). Similarly,
bone sarcoma from acute exposure appears to have a limited expression period (NAS80,
6-9
-------
Ma83). For these diseases, the BEIR DI Committee believed that an absolute risk projection
model with a limited expression period is adequate for estimating lifetime risk (NAS80).
Note that, unlike the NAS BEIR I report (NAS72), the BEIR m Committee's relative
and absolute risk models are age-dependent; that is, the risk coefficient changes, depending on
the age of the exposed persons. Observational data on how cancer risk resulting from
radiation changes with age are sparse, particularly so in the case of childhood exposures.
Nevertheless, the explicit consideration of the variation in radiosensitivity with age at
exposure is a significant improvement in methodology. It is important to differentiate
between age sensitivity at exposure and the age dependence of cancer expression. In general,
people seem to be most sensitive to radiation when they are young. In contrast, most
radiogenic cancers seem to occur late in life, much like cancers resulting from other causes.
In this chapter, lifetime cancer risk estimates for a lifetime exposure of equal annual doses are
presented. However, it is important to note that the calculated lifetime risk of developing a
fatal cancer from a single year of exposure varies with the age of the recipient at the time of
exposure.
6.2.5 EPA Assumptions about Cancer Risks Resulting from Low-LET Radiation
The EPA estimates of radiation risks, presented in Section
6.2.6, are based on a presumed linear dose response function. Except for leukemia and bone
cancer, where a 25-year expression period for radiogenic cancer is used, a lifetime expression
period is used, as in the NAS report (NAS80). Because the most recent Life Span Study
Report (Ka82) indicates that absolute risks for solid cancers are continuing to increase 33
years after exposure, the 1980 NAS Committee choice of a lifetime expression period appears
to be well founded.
To project the number of fatalities resulting from leukemia and bone cancer, EPA uses
an absolute risk model, a minimum induction period of 2 years, and a 25-year expression
period. To estimate the number of fatalities resulting from other cancers, EPA has used a
relative risk projection model (EPA84), a 10-year minimum induction period, and the
remaining balance of an exposed person's lifetime as the expression period.
6.2.6 Methodology for Assessing the Risk of Radiogenic Cancer
EPA uses a life table analysis to estimate the number of fatal radiogenic cancers in an
exposed population of 100,000 persons. This analysis considers not only death due to
radiogenic cancer, but also the probabilities of other competing causes of death which are, of
course, much larger and vary considerably with age (Bu81, Co78). Basically, it calculates for
ages 0 to 110 the risk of death due to all causes by applying the 1970 mortality data from the
National Center for Health Statistics (NCHS75) to a cohort of 100,000 persons. Additional
details of the life table analysis are provided in Appendix B. It should be noted that a life
table analysis is required to use the age-dependent risk coefficients in the BEIR ffl report.
For relative risk estimates, EPA has used age-specific cancer mortality data also provided by
NCHS (NCHS73). The EPA computer program used for the life table analysis was furnished
to the NAS BEIR IE Committee by EPA and used by the Committee to prepare its risk
6-10
-------
estimates. Therefore, the population base and calculations should be essentially the same in
both the NAS and EPA analyses.
Both absolute and relative risk models have been considered to project the observed
risks of most solid radiogenic cancers beyond the period of current observation. The range of
estimated fatal cancers resulting from the choice of a particular projection model and its
internal assumptions is about a factor of 3. Although the relative risk model has been tested
in some detail only for lung and breast cancer (La78), based on current evidence, it appears to
be the better projection model for solid cancers. Therefore, it has been adopted for risk
estimates in this report. Previously, EPA used an average of the risks calculated by the
absolute and relative risk projection models (EPA84).
To estimate the cancer risk from low-LET, whole-body, lifetime exposure, the analysis
uses relative risk projections (the BEIR HI L-L model) for solid cancers and the absolute risk
projection for leukemia and bone cancer (the BEIR HI L-L model). Since the expression
period for leukemia and bone cancer is less than the follow-up period, the same risk values
would be calculated for these cancers using either projection method. For a dose to the whole
body, this procedure yields about 400 fatalities per million person-rad (for the BEIR in
linear-quadratic model, a low-LET whole-body dose would yield an estimated lifetime risk of
about 160 fatalities per million person-rad).
BEIR EH also presented estimates of excess soft tissue cancer incidence risk
coefficients for specific sites, as a function of age at exposure, in its Table V-14. By
summing the site-specific risks, it then arrived at an estimate for the whole-body risk of
cancer incidence (other than leukemia and bone cancer) as given in Table V-30. Finally, by
using the weighted incidence/mortality ratios given in Table V-15 of the same report
(NAS80), the results in Table V-30 can be expressed in terms of mortality to yield (for
lifetime exposure) a risk estimate of about 242 and 776 cancer fatalities per 106 person-rad,
depending on whether an absolute or a relative risk projection model, respectively, is used to
estimate lifetime risk. These values are about 1.7 and 2.1 times their counterparts for the
BEIR ffl L-Lmodel and 4.2 and 5.2 times the LQ-Lvalues. These models all presume a
uniform dose to all tissues at risk in the body. In practice, such uniform whole-body
exposures seldom occur, particularly for ingested or inhaled radioactivity. The next section
describes how this risk estimate is apportioned for whole-body exposure when considering the
risks following the exposure of specific organs.
6.2.7 Organ Risks
For most sources of environmental contamination, inhalation and ingestion of
radioactivity are more common than external exposure. In many cases, depending on the
chemical and physical characteristics of the radioactive material, inhalation and ingestion
result in a nonuniform distribution of radioactive materials within the body so that some
organ systems receive much higher doses than others. For example, since iodine isotopes
concentrate preferentially in the thyroid gland, the dose to this organ can be orders of
magnitude larger than the average dose to the body.
6-11
-------
To determine the probability that fatal cancer occurs at a particular site, EPA has
performed life table analyses for each cancer type using the information on cancer incidence
and mortality in NAS80. NAS80 published incidence risk coefficients (NAS80 Table V-14)
and mortality to incidence ratios (NAS80 Table V-15). The data in Tables 6-1 and 6-2 are
from these tables with the exception of the mortality to incidence ratios for thyroid and lung
cancer. Since not all forms of thyroid cancer can be induced by radiation and since, for those
that are, a more reasonable mortality to incidence ratio would be 0.1 (NRC85), EPA has used
that value in its calculations. Lung cancer incidence and mortality have both shown an
increasing trend between 1970 and 1980. Since incidence leads mortality, an uncorrected
mortality to incidence ratio gives a low estimate of the fraction of those persons who, having
been diagnosed with lung cancer, will die of that disease. Therefore, a mortality to incidence
ratio of 0.94, based on long-term survival studies by the National Cancer Institute for lung
cancer (J. Horn, private communication), has been used.
Risk coefficients for a site-specific relative risk model were calculated as follows:
1. Mortality risk coefficients for an absolute risk model were calculated using the
data in Tables 6-1 and 6-2.
2. Following the procedure used in NAS80, absolute risks at an absorbed dose
rate of 1 mrad/y were calculated for each site for males and females in each
age group. A 10-year minimum latency and a 20-year plateau - i.e., a 30-year
follow up - was used for these calculations.
3. The relative risk coefficients (1/rad) for each age group providing the same 30-
year projected risk were then calculated. Following the NAS80 convention, the
values calculated for ages 10-19 were used for ages 0-9. For consistency, this
report uses this convention for all cancers including lung and breast, for which
the NAS80 absolute risk coefficients are zero in the first decade. For
calculating thyroid risks, the relevant age-specific mortality rate was considered
to be one-tenth of the corresponding incidence rate.
4. Male and female risks for lifetime expression of risk at 1 mrad/y were then
calculated and combined to obtain estimates for the general population.
EPA used the NCHS 1970 life table and mortality data for all these calculations. Male
and female cohort results were combined presuming a male:female sex ratio at birth of
1.0511, consistent with the expected lifetimes at birth for the 1970 male, female, and general
cohort life tables.
6-12
-------
Table 6-1.1 Site-specific incidence risk coefficients (10~6
per rad-y).
Age at Exposure
Site 0-9 10-19 20-34 35-50 50+
Males
Thyroid
Breast
Lung
Esophagus
Stomach
Intestine
Liver
Pancreas
Urinary
Lyraphoma
Other
All Sites
Females
Thyroid
Breast
Lung
Esophagus
Stomach
Intestine
Liver
Pancreas
Urinary
Lymphoma
Other
All Sites
2.20
0.00
0.00
0.07
0.40
0.26
0.70
0.24
0.04
0.27
0.62
4.80
5.80
0.00
0.00
0.07
0.40
0.26
0.70
0.24
0.04
0.27
0.62
8.40
2.20
0.00
0.54
0.07
0.40
0.26
0.70
0.24
0.23
0.27
0.38
5.29
5.80
7.30
0.54
0.07
0.40
0.26
0.70
0.24
0.23
0.27
0.38
16.19
2.20
0.00
2.45
0.13
0.77
0.52
0.70
0.45
0.50
0.27
1.12
9.11
5.80
6.60
2.45
0.13
0.77
0.52
0.70
0.45
0.50
0.27
1.12
19.31
2.20
0.00
5.10
0.21
1.27
0.84
0.70
0.75
0.92
0.27
1.40
13.66
5.80
6.60
5.10
0.21
1.27
0.84
0.70
0.75
0.92
0.27
1.40
23.86
2.20
0.00
6.79
0.56
3.35
2.23
0.70
1.97
1.62
0.27
2.90
22.59
5.80
6.60
6.79
0.56
3.35
2.23
0.70
1.97
1.62
0.27
2.90
32.79
Source: NAS80, Table V-14
6-13
-------
Table 6-2. Site-specific mortality to incidence risk ratios.
Site Male Female
Thyroid
Breast
Lung
Esophagus
Stomach
Intestine
Liver
Pancreas
Urinary
Lymphoma
Other
0.10
—
0.94
1.00
0.75
0.52
1.00
0.91
0.37
0.73
0.65
0.10
0.39
0.94
1.00
0.78
0.55
1.00
0.90
0.46
0.75
0.50
Source: NAS80, Table V-15, except thyroid and lung (see text).
The average risk for a uniform dose to all tissues was calculated to be 542 x 10"6, 806 x
10"6, and 678 x 10"6 per rad for males, females, and the general population, respectively.
It is generally accepted that the risk estimates for the individual sites are less certain than
are the risk estimates for all sites combined. Table 6-3 summarizes the relative risk
calculations for the BEIR HI L-Lmodel. The calculational procedure was the same as that
outlined above.
The risks tabulated in Table 6-3 are slightly different from those in NAS80. These
differences reflect a correction in the exposure interval data for each age group and the use of
final rather than preliminary 1970 mortality data. NAS80 also combined male and female
risk estimates presuming a sex ratio at birth of 1:1, which is not consistent with natality data.
Since the total risk for all sites is considered more certain than the risk for each site
individually, the lifetime risks calculated for the L-Lmodel have been used as a constraint for
the sum of the individual site estimates. The relative risk coefficient for each site shown in
Table 6-4 has been calculated by multiplying the coefficient for the unconstrained model for
each sex by the quotient of the average risk for all age groups for the L-Lunconstrained site-
specific model. The constrained risk coefficients are about one-half of the unconstrained
values.
The L-L absolute risk model coefficients for leukemia and bone cancer are shown in
Table 6-5. The risk coefficient for bone was obtained by dividing the value for alpha
particles (high-LET) in NAS80 Table A-27 by an RBE of 8 to obtain a low-LET value of
1.25 x 10"7 per rad-year. The risk coefficients for leukemia were obtained by subtracting the
risk coefficients for bone from the risk coefficients for leukemia and bone from NAS80 Table
V-17. EPA has followed the BEIR HI Committee's practice of using the absolute risk model.
6-14
-------
Table 6-3. BEIR
bone
Group 0-9
Risk Coefficients (10'6 per
Male 1.92
Female 2.567
Risk Coefficients (10~3 per
Male 4.458
Female 4.748
General 4.586
Cohort Deaths at 10'3 rad/y
Male 0 .612
Female 0 .689
General 0 .649
HI L-Lmodel for excess fatal
cancer.
10-19
Age at Exposure
20-34
cancers other than leukemia and
35-49
50+
All
rad-y) for Absolute Risk Model*
1.457
1.955
rad) for
4.458
4.748
4.586
4.327
5.807
Relative Risk Model
2.793
3.875
3.322
5.921
7.102
1.007
1.902
1.447
8.808
11.823
0.861
1.586
1.257
for Relative Risk Model
0.609
0.686
0.647
0.563
0.824
0.690
0.181
0.357
0.267
0.112
0.268
0.188
2.076
2.823
2.440
Risk per Unit Dose (10'6 per rad) for Relative Risk Model
Male 627
Female 702
General 664
629
703
665
397
568
481
134
252
193
56
101
81
310
378
345
* Source: NAS80, Table V-20
6-15
-------
Table 6-4. Mortality risk coefficients (10"3 per rad) for the
constrained relative risk model.
Age at Exposure
Site 0-9
Male
Thyroid 52.74
Breast 0.00
Lung 2.99
Esophagus 6.15
Stomach 11.71
Intestine 3.35
Liver 120.37
Pancreas 67.81
Urinary 4. 14
Lymphoma 4.41
Other 1.12
Female
Thyroid 35.30
Breast 10.52
Lung 6.36
Esophagus 13.30
Stomach 14.15
Intestine 2.63
Liver 142.77
Pancreas 11.81
Urinary 8.10
Lymphoma 6.28
Other 0.53
General
Thyroid 40.01
Breast 10.57
Lung 3.61
Esophagus 8.01
Stomach 12.63
Intestine 2.95
Liver 126.87
Pancreas 9.66
Urinary 5.48
Lymphoma 5.28
Other 0.76
10-19
52.74
0.00
2.99
6.15
11.71
3.35
120.37
7.81
4.14
4.41
1.12
35.30
10.52
6.36
13.30
14.15
2.63
142.77
11.81
8.10
6.28
0.50
40.18
10.57
3.61
8.01
12.63
2.95
126.84
9.66
5.48
5.28
0.76
20-34
38.00
0.00
2.15
1.44
4.20
1.28
25.19
2.49
1.38
1.28
1.02
35.96
2.80
6.27
3.90
7.08
1.06
46.62
3.61
3.41
1.60
0.47
6.67
2.82
2.91
2.08
5.37
1.16
32.42
3.00
2.08
1.43
0.69
35-50
28.63
0.00
1.34
0.71
1.76
0.48
7.23
1.12
0.59
0.42
0.44
34.81
1.52
6.10
2.31
3.19
0.45
16.29
1.50
1.63
0.50
0.24
33.15
1.54
2.19
1.14
2.34
0.47
10.37
1.30
0.95
0.45
0.32
50+
22.43
0.00
1.18
1.15
1.70
0.46
4.24
1.37
0.39
0.21
0.47
29.53
1.02
6.12
3.17
2.60
0.42
7.80
1.59
0.96
0.25
0.27
28.10
1.07
2.15
1.77
2.10
0.44
5.70
1.48
0.61
0.23
0.34
6-16
-------
Table 6-5. BEIR in L-L model for excess incidence of (and mortality from) leukemia and
bone cancer (absolute risk model).
Age at Exposure
Site
0-9
Risk Coeffcient (10"
Male
Leukemia
Bone
Female
Leukemia
Bone
General
Leukemia
Bone
3.852
0.125
2.417
0.125
3.147
0.125
10-19
6 /rad-v)*
1.724
0.125
1.067
0.125
1.399
0.125
20-34
2.471
0.125
1.541
0.125
1.005
0.125
35-50
1.796
0.125
1.112
0.125
1.439
0.125
50+
4.194
0.125
2.631
0.125
3.277
0.125
All
Cohort Deaths at 10'3 rad/v
Male
Leukemia
Bone
Total
Female
Leukemia
Bone
Total
General
Leukemia
Bone
Total
.0923
.0030
.0953
.0588
.0030
.0618
.0760
.0030
.0790
Risk per Unit Dose
Male
Leukemia
Bone
Total
Female
Leukemia
Bone
Total
94.7
3.1
97.8
59.9
3.1
63.0
.0405
.0029
.0435
.0257
.0030
.0287
.0333
.0030
.0363
(1Q-6 per rad)
41.9
3.0
44.9
26.3
3.1
29.4
.0829
.0042
.0871
.0543
.0044
.0587
.0689
.0043
.0732
58.5
3.0
61.4
37.4
3.0
40.4
.0508
.0035
.0543
.0357
.0040
.0398
.0435
.0038
.0472
37.5
2.6
40.1
25.3
2.8
28.1
.0968
.0029
.0997
.0932
.0044
.0976
.0950
.0036
.0987
48.6
1.4
50.1
35.3
1.7
36.9
.3634
.0165
.3799
.2677
.0189
.2866
.3167
.0177
.3344
54.2
2.5
56.7
35.9
2.5
38.4
6-17
-------
Table 6-5. (continued) BEIR HI L-L model for excess incidence of
(and mortality from) leukemia and bone cancer (absolute risk model).
Age at Exposure
Site 0-9 10-19 20-34 35-50 50+ All
Risk per Unit Dose (10"
General
Leukemia
Bone
Total
77.7
3.1
80.8
6 per rad)
34.3
3.1
37.4
48.1
3.0
51.1
31.4
2.7
34.1
41.2
1.6
42.8
44.8
2.5
47.3
* Source: NAS80, Table V-17
projections for leukemia and bone cancer with the relative risk projection for all other
cancers. Since the expression period for leukemia and bone cancer is 27 years, there is no
difference between the number of cancers projected for a 30-year and a lifetime follow-up
period.
Table 6-6 shows the average mortality risks per unit absorbed dose for the combined
leukemia/bone and constrained relative risk models. The risk, in general, decreases with
increasing age at exposure. For a constant, uniform absorbed dose rate to all organs and
tissues, about 60 percent of the risk is conferred by the exposures in the first 20 years of life.
The mortality to incidence ratios in Table 6-2 were used to convert the mortality risk
estimates in Table 6-6 to incidence risk estimates. For leukemia and bone cancer, the
incidence risks are considered to be equal as in NAS80. The resultant incidence risks are
shown in Table 6-7.
6.2.8 Thyroid Cancer from Iodine-131 and Iodine-129
Iodine-131 has been reported to be only one-tenth as effective as x-rays or gamma
rays in inducing thyroid cancer (NAS72, NCRP77, NCRP85). BEIR m reported estimates of
factors of 10-80 times reduction for iodine-131 compared to x-rays and noted the estimates
were derived primarily from animal experiments (NAS80). However, one study in rats
reported that iodine-131 was just as effective as x-rays in inducing thyroid cancer, leading an
NRC review group to select one-third as the minimum ratio of iodine-131 to x-ray effects that
is compatible with both old and new data (NRC85).
6-18
-------
Table 6-6. Site-specific mortality risk per unit dose (l.OE-6 per rad) for combined leukemia-
bone and constrained relative risk model.
Age at Exposure
Site 0-9
Male
Leukemia 94.68
Bone 3.07
Thyroid 8.25
Breast 0.00
Lung 145.90
Esophagus 25.57
Stomach 110.95
Intestine 53.49
Liver 168.01
Pancreas 74.36
Urinary 40.73
Lymphoma33.43
Other 37.48
Total 796.43
Female
Leukemia 59.93
Bone 3.10
Thyroid 15.85
Breast 309.33
Lung 78.57
Esophagus 21.47
Stomach 102.64
Intestine 57.15
Liver 115.94
Pancreas 103.00
Urinary 46.40
Lymphoma45.71
Other 27.69
Total 986.78
10-19
41.86
3.04
8.25
0.00
146.95
25.76
111.72
53.83
168.24
74.90
40.99
33.28
37.23
746.05
26.35
3.09
14.54
310.52
78.89
21.57
103.05
57.38
115.25
103.48
46.54
45.66
27.65
953.96
20-34
58.46
2.96
5.08
0.00
107.22
6.13
40.63
20.89
35.40
24.21
13.85
9.62
33.72
358.15
37.39
3.03
11.46
81.01
77.09
6.32
51.49
23.07
36.97
31.71
19.64
11.54
24.48
415.21
35-50
37.52
2.61
2.69
0.00
61.40
2.82
16.4
7.60
9.48
10.34
5.79
2.88
13.09
172.65
25.27
2.84
7.46
36.93
64.70
3.46
22.38
9.57
11.95
12.70
9.08
3.35
11.27
220.95
50+
48.64
1.45
0.80
0.00
22.55
2.03
9.36
4.30
2.50
10.34
2.22
0.71
6.93
108.06
35.27
1.67
2.24
10.30
24.96
2.26
10.73
5.01
2.80
7.11
3.06
0.79
5.80
112.01
All
54.19
2.47
4.32
0.00
84.21
9.91
46.95
22.78
58.87
30.78
16.60
12.49
22.66
366.25
35.86
2.53
8.42
107.63
56.72
8.33
45.00
23.08
40.74
38.15
18.80
15.13
16.20
416.59
6-19
-------
Table 6-6. Site-specific mortality risk per unit dose (l.OE-6 per rad) for combined leukemia-
bone and constrained relative risk model.
Age at Exposure
Site 0-9
General
Leukemia 77.69
Bone 3.09
Thyroid 12.22
Breast 151.21
Lung 112.98
Esophagus 23.56
Stomach 106.89
Intestine 55.28
Liver 142.55
Pancreas 88.36
Urinary 43.50
Lymphoma39.44
Other 32.69
Total 889.49
10-19
34.26
3.06
11.33
152.03
113.63
23.71
107.48
55.57
142.30
88.89
43.71
39.34
32.54
847.84
20-34
48.06
2.99
8.23
39.95
92.34
6.22
45.98
21.96
36.17
27.90
16.70
10.56
29.16
386.21
35-50
31.39
2.72
5.07
18.40
63.00
3.14
19.37
8.58
10.71
11.51
7.43
3.11
12.18
196.60
50+
41.20
1.58
1.61
5.75
23.91
2.16
10.13
4.70
2.67
6.87
2.69
0.76
6.30
110.32
All
44.76
2.50
6.43
55.36
70.07
9.09
45.95
22.94
49.55
34.57
17.73
13.85
19.34
392.14
6-20
-------
Table 6-7. Site-specific incidence risk per unit dose (l.OE-6 per rad) for combined leukemia-
bone and constrained relative risk model.
Age at Exposure
Site 0-9
Male
Leukemia 94.68
Bone 3.07
Thyroid 87.59
Breast 0.00
Lung 155.21
Esophagus 25.57
Stomach 147.94
Intestine 102.87
Liver 168.01
Pancreas 81.71
Urinary 110.08
Lymphoma45.80
Other 57.66
Total 1080.20
Female
Leukemia 59.93
Bone 3.10
Thyroid 158.45
Breast 793.16
Lung 83.59
Esophagus 21.47
Stomach 131.59
Intestine 103.90
Liver 115.94
Pancreas 114.44
Urinary 100.88
Lymphoma60.95
Other 55.38
Total 1802.80
10-19
41.86
3.04
82.5
0.00
156.33
25.76
148.97
103.52
168.24
82.31
110.79
45.58
57.27
1026.20
26.35
3.09
145.42
796.20
83.93
21.57
132.11
104.34
115.25
114.98
101.16
60.88
55.30
1760.60
20-34
58.46
2.96
50.84
0.00
114.07
6.13
54.18
40.16
35.40
26.60
37.44
13.17
51.88
491.27
37.39
3.03
114.59
207.73
82.01
6.32
66.01
41.94
36.97
35.23
42.70
15.38
48.97
738.28
35-50
37.52
2.61
26.92
0.00
65.31
2.82
21.87
14.63
9.48
11.37
15.65
3.94
20.15
232.28
25.27
2.84
74.60
94.69
68.83
3.46
28.69
17.40
11.95
14.11
19.74
4.47
22.54
388.58
50+
48.64
1.45
8.04
0.00
23.99
2.03
12.48
8.28
2.50
7.20
6.01
0.98
10.65
132.25
35.27
1.67
22.38
26.40
26.56
2.26
13.75
9.11
2.80
7.91
6.66
1.06
11.61
167.42
All
54.19
2.47
43.23
0.00
89.58
9.91
62.61
43.81
58.87
33.83
44.87
17.12
34.86
495.35
35.86
2.53
84.16
275.97
60.34
8.33
57.70
41.96
40.74
42.39
40.88
20.18
32.40
743.44
6-21
-------
Table 6-7. Site-specific incidence risk per unit dose (l.OE-6 per rad) for combined leukemia-
bone and constrained relative risk model.
Age at Exposure
Site 0-9
General
Leukemia 77.69
Bone 3.09
Thyroidl 22.24
Breast 387.78
Lung 120.19
Esophagus 23.56
Stomach 139.95
Intestine 103.38
Liver 142.55
Pancreas 97.71
Urinary 105.58
Lymphoma53.21
Other 56.55
Total 1433.50
10-19
34.26
3.06
113.32
389.82
120.88
23.71
140.71
103.92
142.30
98.30
106.08
53.07
56.31
1385.70
20-34
48.06
2.99
82.26
102.42
98.24
6.22
60.00
41.03
36.17
30.85
40.02
14.26
50.43
612.96
35-50
31.39
2.72
50.66
47.18
67.02
3.14
25.25
16.00
10.71
12.73
17.68
4.20
21.33
310.01
50+
41.20
1.58
16.05
14.74
25.43
2.16
13.20
8.74
2.67
7.60
6.37
1.02
11.19
151.96
All
44.76
2.50
64.28
141.95
74.54
9.09
60.08
42.86
49.55
38.23
42.28
18.69
33.60
622.96
6-22
-------
It would be prudent to use this factor until further information from animal studies or
some human data are developed. In this document, EPA has employed a thyroid cancer risk
coefficient for internal exposures to iodine-131 and 1-129 which is one-third that used for
gamma rays or beta radiations from other radionuclides.
6.2.9 Cancer Risks for a Constant Intake Rate
The fatal cancer risks shown in the tables of this chapter presume a lifetime exposure
at a constant dose rate. Even for a dosimetric model with age invariant parameters, dose rates
vary with time for a constant intake rate. This variation reflects the time-dependent activity
levels associated with the retention of the radionuclide in the organs and tissues. The
ingrowth of radioactive decay products can also contribute further to the time-dependence of
dose rates.
Traditionally, risk estimates for chronic intake of a radionuclide have been determined
using a dose commitment model to calculate dose rates following a fixed period (e.g., a 70-
year average lifespan). For the purpose of estimating risk, these dose rates are considered to
be invariant over the individual's lifetime. This approach is overly conservative for
estimating risk for many long-lived radionuclides. Therefore, EPA estimates risks for
constant radionuclide intakes by first determining dose rates to each radiosensitive organ or
tissue as a function of time. Then these dose rates and the risk models of this chapter are
used to calculate lifetime risk based on 1970 life table data. The resulting risks are consistent
with both the dosimetric and risk models, and the arbitrary choice of a dose commitment
period is avoided.
6.2.10 Effect on Risk Estimates of Recent Information Regarding A-Bomb Survivors
Since publication of the BEIR ffl report, there has been further epidemiological
follow-up of the Japanese A-bomb survivors. As discussed above, the results have lent
support to the relative risk projection model for solid tumors, which has been utilized here.
The additional data provided by the follow-up reduces statistical uncertainties in the risk
coefficients and fills in important gaps pertaining to some organ-specific risks, particularly
with respect to childhood irradiation (Pr88).
Subsequent to BEIR IE, there has also been a major reassessment of doses assigned to
the A-bomb survivors, the effect of which, in general, will be to increase the risk of low-LET
radiation calculated according to a particular model.
Investigators from Oak Ridge National Laboratory carried out careful state-of-the art
evaluation of the dose to A-bomb survivors in the early 1960s (Au67, Au77). The results of
these studies resulted in a "T65" dose being assigned to the dose (kerma) in free air at the
location of each survivor for both gamma rays and neutrons. A major conclusion of the
ORNL study was that the mix of gamma ray and neutron radiations was quite different in the
two cities where A-bombing occurred. These results indicated that at Hiroshima the neutron
dose was more important than the gamma dose when the greater biological efficiency of the
high-LET radiations produced by neutrons was taken into account. Conversely, the neutron
dose at Nagasaki was shown to be negligible compared to the gamma dose for that range of
6-23
-------
doses where there were significant numbers of survivors. Therefore, the 1980 BEIR
Committee evaluated the cancer risks to the survivors at Hiroshima on the assumption that the
combined effects of gamma rays and particularly neutrons caused the observed cancer
response.
Serious inadequacies in the T65 dosimetry system were discovered in the late 1970s.
A comprehensive revaluation of the doses to survivors was carried our under the auspices of
the U.S.-Japan Joint Committee for Reassessment of Atomic Bomb Dosimetry in Hiroshima
and Nagasaki. In 1986, this committee provided results to the Radiation Effects Research
Foundation (RERF) from which a revised dosimetry system, termed "DS86," was developed.
Although work in the DS86 is largely complete, small adjustments in dose estimates are
anticipated over the next few years (Pr87). In addition, about 1,000 survivors from Nagasaki,
who were shielded by terrain or were hi factories, have so far been excluded from the
analysis because of difficulties in estimating their doses. It is anticipated that dose estimates
for some of these survivors will be forthcoming in the near future (Pr87).
The major differences between TS65 and DS86 are: (1) the neutron dose in DS86 is
decreased to 10 percent of its former value in Hiroshima and 30 percent in Nagasaki (as a
result, neutrons now contribute relatively little to the estimated excess of cancers in the two
cities); (2) the DS86 free-in-air gamma dose increases somewhat in Hiroshima but decreases
in Nagasaki relative to T65; (3) transmission of gamma rays through wooden structures is
decreased by about a factor of 2 in DS86; and (4) transmission of gamma rays through the
body to internal organs is generally increased, partially nullifying the change associated with
the decreased transmission through structures (Pr87, Sh87).
Analysis of the A-bomb survivor data using the DS86 dosimetry is continuing.
Preliminary indications are that risk estimates corresponding to a given dose-response model
(linear or linear-quadratic) will be increased by more than a factor of 2 as compared to BEIR
HI estimates. This increase arises not only from changes in dosimetry, but also from further
epidemiological follow-up and new statistical procedures employed (Pr87, Pr88). A
preliminary estimate of low-LET radiation risk to the Japanese population based on DS86
dosimetry and the linear, relative risk model is 1.2 X 10"3 fatal cancers per rad (Pr88) -
approximately 3 times the corresponding BEIR HI estimate. Recent publications by
UNSCEAR (UNSC88) and the British NRPB (St88) obtained similar estimates for the
Japanese and U.K. populations, respectively.
It appears that either a linear or linear-quadratic dose response is consistent with the
survivor data, analyzed according to DS86 (Pr87). However, as noted above, linear and
linear-quadratic best fits to the data differ only slightly in their predictions at low doses. It
would also appear that the residual difference in risk per unit dose between Hiroshima and
Nagasaki is no longer statistically significant under DS86 dosimetry (Sh87).
6.2.11 Comparison of Risk Estimates for Low-LET Radiation
Table 6-8 summarizes various estimates of risk from low level, low-LET exposures of
the general population. As discussed above, the highest risk estimates are obtained by
assuming a linear dose response (for purposes here, equivalent to a DREF=1.0) and a relative
6-24
-------
risk projection model. EPA's current risk estimate of 392 x 10~6/rad corresponds to that
obtained by the BEIR in committee (NAS80) using these "conservative" assumptions.
However, this estimate was not derived from the most recent Japanese data; recent
calculations based on similar assumptions but revised data yield about three times higher risk
(see Pr88 in Table 6-8). Thus, as illustrated by a comparison with the UNSC88 and St88
entries in Table 6-8, the EPA89 estimate is in good agreement with the new data if one
assumes that the risks projected from a linear fit to the epidemiological data should be
reduced by a factor of about three when extrapolating to chronic low dose conditions. Such
an assumption is reasonable in view of supportive laboratory data and the apparent decreased
effectiveness of iodine-131 in causing thyroid cancer in humans relative to X-rays (NCRF77).
However, it should be noted that while the current estimate 392 x 10"6/rad is reasonable, and
well within the range of uncertainty, it can no longer be regarded as conservative, in the sense
of providing an extra margin of public health protection. The EPA plans to reevaluate its risk
models, including the choice of DREF, in light of the UNSC88 and NAS BEIR V reports.
It is expected that this review will also lead to revisions in the distribution of fatal
cancer risk among organs. To assign organ risks, evidence on the Japanese A-bomb survivors
has to be integrated with that from other epidemiological studies. As an indicator of the
possible impact that the new Japanese data may have on EPA's organ-specific risk estimates,
Table 6-9 compares EPA's current organ risk estimates with those recently published by the
NRPB for the general U.K. population (St88), which take into account recent changes in the
Japanese data. Two model estimates are presented from the NRPB publication: (a) one based
on a linear extrapolation of high dose epidemiological data and (b) one based on an assumed
DREF of two for breast cancer induction and three for all other sites. Both sets of model
estimates assume a relative risk protection for cancers other than bone cancer and leukemia.
Thus the model assumptions underlying the first NRPB set of organ risk estimates closely
parallel those employed by EPA. The difference in the risk estimates largely reflect changes
in the Japanese data. The second set of NRPB risk estimates, which the authors preferred to
use at low environmental doses and dose rates, are, for the most part, in reasonable agreement
with EPA's current model estimates (to within about a factor of two).
6-25
-------
Table 6.8 Comparison of general population risk estimates for fatal cancers due to low
level, whole-body, low-LET radiation.
Source of
estimate
NAS72"
NAS72"
NAS80
NAS80
NAS80
NAS80
EPA84
EPA89e
TINSC77
Ui>OV^ / /
Pr88f
UNSC88f
Fatalities per
106 person-rad
117
621
158
403
67
169
280
392
75-175
/ j i / *}
1200
110-550
Risk projection
model
Absolute
Relative
Absolute
Relative0
Absolute
Relative0
Ave.(Rel.& Abs.)
Relative0
Relative0
Relative0
DREP
1.0
1.0
1.0
1.0
2.48d
2.48d
1.0
1.0
9 ^
-t-. J
1.0
2-10
St88f
450
Relative0
3.0s
Factor by which risk estimate is reduced from that obtained by linear
extrapolation of high dose epidemiological results.
As revised in NAS80.
For all cancers other than leukemia and bone cancer.
Based on comparison of linear coefficients for linear and linear-quadratic
models used to calculate radiogenic cancers other than leukemia and
bonecancer; the corresponding DREF is 2.26 for these two sites.
Refers to this document.
From analyses of A-bomb survivor data using DS86 dosimetry.
Except breast - a DREF of 2 is assumed for that site.
6-26
-------
Table 6-9. Site-specific mortality risk per million person-rad from low level, low-LET
radiation exposures of the general population.
Cancer EPA NRPBa NRPBb
Leukemia
Bone
Thyroid
Breast
Lung
Stomach
Intestine
Liver
Pancreas
Urinary
Other
Total
44.8
2.5
6.4 (2.1)c
55.4
70.1
46.0
22.9
49.6
34.6
17.7
42.3
392
84
15
7.5
110
350
73
110
45
—
—
500
1290
28
5
2.5
55
120
24
37
15
—
—
163
450
Relative risk model recommended by authors for use only at high dose rates.
Use at low dose rates would be equivalent to adopting a DREF of 1. (St88).
Preferred relative risk model projection for use at low dose rates; assumes
DREF=2 for breast and DREF=3 for all other sites.
Value in parentheses represents estimate for important case of iodine-131 (or
iodine-129) exposure.
6.2.12 Sources of Uncertainty in Low-LET Risk Estimates
The most important uncertainties in estimating risk from whole body, low-LET
radiation appear to relate to: (1) the extrapolation of risks observed in populations exposed to
relatively high doses, delivered acutely, to populations receiving relatively low dose chronic
exposures and (2) the projection of risk over a full lifespan - most critically, the extent to
which high relative risks seen over a limited follow-up
period among individuals exposed as children carry over into later years of life when baseline
cancer incidence rates are high.
Another significant uncertainty relates to the extrapolation of risk estimates from one
population to another (e.g., from the Japanese A-bomb survivors to the U.S. general
population). This source of uncertainty is regarded as important for estimating risk of
6-27
-------
radiogenic cancer in specific organs for which the baseline incidence rates differ markedly in
the two populations.
In addition to the model uncertainties alluded to above, errors in dosimetry and
random statistical variations will contribute to the uncertainty in the risk estimates. The
errors in T65 dosimetry were discussed Section 6.2.10. The residual error of DS86 dosimetry
is estimated to be a relatively minor contributor to the overall uncertainty (see below).
Statistical variability will be most important where relatively few excess cancers have so far
been observed: e.g., with respect to specific cancer sites or with respect to childhood
irradiation in the A-bomb survivors.
6.2.12.1 Low Dose Extrapolation
Results from animal and cellular studies often show decreasing effects (e.g., cancers,
mutations, or transformations) per rad of low-LET radiation at low doses and dose rates.
Based on a review of this literature, the National Council on Radiation Projection (NCRP80)
has concluded that "linear interpolation from high doses (150 to 350 rads) and dose rates (>5
rads min"1) may overestimate the effects of either low doses (0-20 rads or less) or of any dose
delivered at dose rates of 5 rad y"1 or less by a factor of two to ten." Judged solely from
laboratory experiments, therefore, about a factor of ten reduction from the linear prediction
would seem to constitute a plausible lower limit on the effectiveness of low-LET radiation
under chronic low dose conditions.
Epidemiological evidence would seem to argue against such a large DREF from
human cancer introduction, however. Data on the A-bomb survivors and patients irradiated
for medical reasons indicate that excess breast cancer incidence is proportional to dose and
independent of dose fractionation (NAS80, NEH85). The evidence on thyroid cancer
induction is equivocal: medical x-ray data suggest a linear dose response (NAS80, NIH85); on
the other hand, iodine-131 radiation appears to be at least 3 times less effective than an equal
dose of x-rays in inducing human thyroid cancer, one plausible explanation for which is a
reduced effectiveness at low dose rates (NCRP77).
The BEIR HI Committee's analysis of the A-bomb survivor data based on T65
dosimetry, suggested a quadratic component to the dose response function. After removing
the estimated neutron-induced leukemia, the Committee's linear-quadratic fit to the data
yielded a linear coefficient that was a factor of 2.3 times lower than the coefficient obtained
from a simple linear fit (NAS80). Thus, the analysis suggested a 2.3 times lower risk at low
doses (and dose rates) than estimated by linear extrapolation of the high dose data. Results of
the curve fitting for solid tumors were too unstable to estimate a shape for the dose response;
for simplicity, the Committee assumed that the shape of the linear-quadratic fit for solid
tumors was identical to that derived for leukemia. At low doses, the linear-quadratic model
predicts about 2.5 times fewer solid tumors than the corresponding linear model. However,
the DS86 data appear to be more consistent with a simple linear dose response for both
leukemia and solid tumors. Reflecting this finding, low dose extrapolations of the linear and
linear-quadratic fits to the DS86 data apparently differ from one another by less than a factor
of 2 (Sh88, Pi89). Thus, if one posits a linear-quadratic dose response model, the available
6-28
-------
human data would suggest that linear extrapolation from high doses and dose rates
overestimates risks at low doses and dose rates by about a factor of 2 or less.
6.2.12.2 Time and Age Dependent Factors
Because epidemiological follow-up of exposed population is generally incomplete, a
risk projection model must be used in estimating lifetime risks due to a given exposure. For
leukemia and bone cancer, where the expression time is limited to 25 years, absolute and
relative risk projection models yield the same number of radiogenic cancers. For other
cancers, the BEER, m Committee assumed that radiogenic cancers would occur throughout the
estimated lifetime. This makes the choice of projection model more critical because the
relative risk projection yields estimated lifetime risks 2-3 times larger than an absolute risk
projection. Recent follow-up of the A-bomb survivor population strongly suggests that the
relative risk projection model better describes the variation risk of solid tumors over time
(NIH85). However, there may be some cancers, apart from leukemia and bone cancers, for
which the absolute risk projection model is a better approximation. For other cancers, the
relative risk may have been roughly constant for the current period of follow-up but may
eventually decrease over time. The uncertainty relating to risk projection will naturally
decrease with further follow-up of irradiated study cohorts, but in view of the continuing
increase in attributable risk with age in the A-bomb survivors, it would appear that the
relative risk projection model does not overestimate the lifetime task in the general population
by more than about a factor of 2.
Similarly, there is yet insufficient information on radiosensitivity as a function of the
age at exposure, particularly on the ultimate effects of exposure during childhood. As the A-
bomb survivor population ages, more information will become available on the cancer
mortality of persons irradiated when they were young. Recent follow-up studies support the
view that relative risks are highest in those aged 0-9 years at exposure. Full inclusion of the
projected effects on this group was a major contributor to the increase in risk found with the
recent analysis based on DS86 dosimetry (Pr87, Pr88).
6.2.12.3 Extrapolation of Risk Estimates to U.S. Population
There is also an uncertainty associated with applying the results of an epidemiological
study on a population to another population having different demographic characteristics. A
typical example is the application of the Japanese data for A-bomb survivors to Western
people. Seymour Jablon has called this the "transportation problem," a helpful designation
because it is often confused with the risk projection problem described above. However,
there is more than a geographic aspect to the "transportation problem." Risk estimates for
one sex must sometimes be based on data for the other. In transporting risk estimates from
one group to another, one may have to consider habits influencing health status, such as
differences between smokers and nonsmokers, as described in Section 6.4 for the case of risk
estimated for radon progeny.
The BEIR HI Committee addressed this problem in its 1980 report and concluded,
based largely on the breast cancer evidence, that the appropriate way to transport the Japanese
risk to the U.S. population was to assume that the absolute risk over a given observation
6-29
-------
period was transferrable but that relative risk was not. Therefore, the Committee calculated
what the relative risk would be if the same number of excess cancer deaths was observed in a
U.S. population having the same age characteristics as the A-bomb survivors. A constant
absolute risk model, as postulated by the Committee, would imply that, whatever the factors
are that cause Japanese and U.S. baseline cancer rates to differ, they have no effect on the
incidence of radiation-induced cancers; i.e., the effects of radiation and these factors are
purely additive.
An alternative approach to the "transportation problem" was taken by the 1972 NAS
BEER-I Committee. This committee assumed relative risks would be the same in the United
States and Japan and transferred the observed percentage increase directly to the U.S.
population. Since the U.S. and Japanese baseline rates differ drastically with respect to
mortality from specific cancers, this approach implies some large differences in the predicted
number of specific cancers resulting from a given dose of radiation in the two countries. The
most important differences relate to cancers of the breast, lung, and stomach. Baseline rates
of breast and lung cancers are higher in the United States by factors of about 4 and 2,
respectively, while the risk of stomach cancer is about 8 times higher in Japan (Gi85). As
noted above, it appears that the absolute risk should be transported for breast cancer.
Evidence is lacking regarding the other cancer sites, however. If lung cancer risk were to be
transported with a relative risk model, retaining the absolute model for other cancers, the
estimated risk from a whole-body exposure would increase by about 20 percent; on the other
hand, applying the relative risk model to stomach cancer alone would lower the whole-body
risk by about 8 percent Based on these considerations, including the tendency for changes in
specific cancers to cancel one another, EPA believes that using the absolute risk
"transportation model" is unlikely to cause large errors in the total risk estimate. Thus, in the
case of uniform whole-body doses, the amount of uncertainty introduced by transporting
cancer risks observed in Japan to the U.S. population appears to be small compared to other
sources of uncertainty in this risk assessment.
6.2.12.4 Dosimetry and Sampling Errors
As discussed in Section 6.2.10, there were systematic biases in the T65 dosimetry
system for the Japanese A-bomb survivors, leading to a significant downward bias in the
estimates of risk due to low-LET radiation. Under DS86 dosimetry, systematic errors are
believed to be no more than about ± 15% (1 SD) (Ka89). Random errors in the individual
dose estimates are estimated to be ± 28% (1 SD), with an overall uncertainty in individual
doses of about ± 32% (Ka89). The random errors in dosimetry will tend to cancel, but they
are expected to bias the slope of the dose response curve downward, reducing the estimate of
risk (Ma59, Da75, Gi84). The magnitude of this bias has been estimated to be roughly 10%
(Pi89).
The precision of risk estimates are also limited by statistical fluctuations due to finite
sample size. The uncertainty in the low-LET risk coefficient for leukemia or all cancers due
to this cause is about ± 20% (90% confidence interval) (Sh89). Uncertainties due to sampling
error are larger where data are sparse, e.g. with respect to risks for specific age groups or
specific cancer sites (Sh88). Finally, there will be some error in ascertaining cancer cases,
most often an under-reporting of cases or mislabeling of cancer type. The latter type of error
6-30
-------
would not be expected to greatly affect the estimates of whole-body risk from ionizing
radiation. The former would tend to bias risk estimates downward somewhat, but it would be
difficult to quantify this effect.
6.2.12.5 Summary and Conclusions Regarding Uncertainties in Low-LET Cancer Risk
Estimates
Uncertainties in low-LET risk estimates arise both from data uncertainties pertaining to
ascertainment of radiation doses and cancer cases and from uncertainties in the proper choice
of model assumptions. The data uncertainties include both systematic errors (biases) and
random errors. Generally speaking, the modeling uncertainties are larger, but random
sampling errors may be a very important contributor to the uncertainty in risk for certain
types of radiogenic cancers or for certain irradiated subpopulations.
The EPA central estimate of average lifetime risk, approximately 400 fatal cancers per
106 person-rad, is taken from the NAS BEIR IE Committee report (NAS80), incorporating the
most conservative model assumptions utilized by the Committee, i.e., a linear dose response
and age-specific relative risks projected over a lifetime for solid tumors (L-RR model). For
reasons discussed above, it would now appear that estimates of average lifetime risk based on
the L-RR model assumptions must be revised upwards - to roughly 1,200 fatal cancers/106
person-rad. Although further analysis of the A-bomb survivor data may increase this
estimate, the conservatism inherent in the model's assumptions supports the view that the
1,200/106 value is an upper bound, pending release of the NAS BEER V report now in
preparation.
Animal data would suggest that the linear dose response may overestimate risk by
roughly a factor of 3. Likewise, while the epidemiological data clearly indicate an increase in
risk with age at expression, the (age-specific) constant relative risk projection may overstate
lifetime risk by about a factor of 2. Allowing even for the additional sources of uncertainty
discussed above, it would appear that the upper bound (L-RR) model estimate may be high by
a factor of 5 to 10. Therefore, as a lower bound estimate of the average lifetime risk, a value
which is one-tenth the upper bound, or 120 fatal cancers/106 person-rad, has been adopted.
The L-RR model estimate from BEIR HI, about 400 fatal cancers/106 person-rad, falls
near the geometric mean of what tentatively appears to be a reasonable range for the estimate
of risk, based on current information. EPA has chosen the BEIR HI,
L-RR model value as its "central estimate." It should be emphasized that this estimate cannot
be regarded as "conservative" in the sense of providing any significant margin of safety with
respect to public health protection. The decision by EPA to employ the central estimate of
400 fatalities/106 person-rad and a range of 120-1,200 fatalities/106 person-rad was reviewed
and approved by a special panel set up by the Agency's outside Radiation Advisory
Committee and by the Committee itself, as an interim measure for this proposed rulemaking.
The uncertainty in risks for specific cancer sites may be substantially larger than the
uncertainty in the whole-body risk. One reason is that the epidemiological data pertaining to
some sites may be very sparse. In addition, the uncertainty in projecting risk from one
6-31
-------
population to another (e.g., Japanese to U.S.) is important at sites for which incidence rates
differ markedly between populations.
6.3 FATAL CANCER RISK RESULTING FROM HIGH-LET RADIATION
This section explains how EPA estimates the risk of fatal cancer resulting from
exposure to high-LET radiations. Unlike exposures to x-rays and gamma rays where the
resultant charged particle flux results in linear energy transfers (LET) of the order of 0.2 to 2
keV per um in tissue, 5-MeV alpha particles result in energy deposition of more than 100
keV per um. High-LET radiations have a larger biological effect per unit dose (rad) than
low-LET radiations. How much greater depends on the particular biological endpoint being
considered. For cell killing and other readily observed endpoints, the relative biological
effectiveness (RBE) of high-LET alpha radiations is often 10 or more times greater than low-
LET radiations. The RBE may also depend on the dose level; for example, if linear and
linear-quadratic dose response functions are appropriate for high- and low-LET irradiations,
respectively, then the RBE will decrease with increasing dose.
6.3.1 Quality Factors and RBE for Alpha Particles
For purposes of calculating dose equivalent, each type of biologically important
ionizing radiation has been assigned a quality factor, Q, to account for its relative efficiency
hi producing biological damage. Unlike an RBE value, which is for a specific tissue and
well-defined endpoint, a quality factor is based on an overall assessment by radiation
protection experts of potential harm of a given radiation relative to x or gamma radiation. In
1977, the ICRP assigned a quality factor of 20 to alpha particle irradiation from radionuclides
(ICRP77). However, the appropriateness of this numerical factor for estimating fatal
radiogenic cancers is still unclear, particularly for individual sites.
The dose equivalent (in rem) is the absorbed dose (in rad) times the appropriate
quality factor for a specified kind of radiation. For the case of internally deposited alpha-
particle emitters, the dose equivalent from a one-rad dose is 20 rem. Prior to ICRP Report 26
(ICRP79), the quality factor assigned to alpha particle irradiation was 10. That is, the
biological effect from a given dose of alpha particles was estimated to be 10 times that from
an acute dose of low-LET x-rays or gamma rays of the same magnitude in rad. The ICRP
decision to increase this quality factor to 20 followed from its decision to estimate the risk of
low-LET radiations, in occupational situations, on the assumption that biological effects were
reduced at low doses and dose rates. There is evidence that the risks from high-LET
radiation are linear with dose and independent of dose rate (for low to moderate doses).
Implicit in ICRP's risk estimates for low dose/dose rate gamma radiation is a dose rate
reduction factor of about 2.5. The EPA (linear) risk model for low-LET radiation does not
employ a DREF; therefore, in order to avoid an artifactual inflation in high-LET risk
estimates, EPA has assumed an RBE of 8 (20/2.5) for calculating the risks from alpha
particles (see Section 6.3.3).
In 1980, the ICRP published the task group report "Biological Effects of Inhaled
Radionuclides," which compared the results of animal experiments on radiocarcinogenesis
following the inhalation of alpha-particle and beta-particle emitters (ICRP80). The task group
6-32
-------
concluded that: "...the experimental animal data tend to support the decision by the ICRP to
change the recommended quality factor from 10 to 20 for alpha radiation."
6.3.2 Dose Response Function
In the case of high-LET radiation, a linear dose response is commonly observed in
both human and animal studies. This response is not reduced at low dose rates (NCRP80).
Some data on human lung cancer indicate that the carcinogenic response per unit dose of
alpha radiation is maximal at low doses (Ar81, Ho81, Wh83); in addition, some studies with
animals show the same response (Ch81, U182). EPA agrees with the NAS BEIR IH
Committee that: "For high-LET radiation, such as from internally deposited alpha-emitting
radionuclides, the linear hypothesis is less likely to lead to overestimates of the risk and may,
in fact, lead to underestimates" (NAS80). However, at low doses, departures from linearity
are small compared to the uncertainty in the human epidemiological data, and EPA believes a
linear response provides an adequate model for evaluating risks in the general environment.
A possible exception to a linear response is provided by the data for bone sarcoma
(but not sinus carcinoma) among U.S. dial painters who ingested alpha-emitting Ra-226
(NAS80). These data are consistent with a dose-squared response (Ro78). Consequently, the
NAS BEIR HI Committee estimated bone cancer risk on the basis of both linear and quadratic
dose response functions. However, as pointed out in NAS80, the number of U.S. dial painters
at risk who received less than 1,000 rads was so small that the absence of excess bone cancer
at low doses is not inconsistent with the linear response model. Therefore, the consistency of
these data with a quadratic (or threshold) response is not remarkable and, perhaps, not
relevant to evaluating risks at low doses. In contrast to the dial painter data, the incidence of
bone cancer following short-lived radium-224 irradiation, observed in spondylitics by Mays
and Spiess (Ma83, NAS80) in a larger sample at much lower doses, is consistent with a linear
response. Therefore, for high-LET radiations, EPA has used a linear response function to
evaluate the risk of bone cancer.
Closely related to the choice of a dose response function is what effect the rate at
which a dose of high-LET radiation is delivered has on its carcinogenic potential. This is an
area of active current research. There is good empirical evidence, from both human and
animal studies, that repeated exposures to radium-224 alpha particles are 5 times more
effective in inducing bone sarcomas than a single exposure that delivers the same dose
(Ma83, NAS80). The 1980 NAS BEER Committee took this into account in its estimates of
bone cancer fatalities, which EPA is using.
6.3.3 Assumptions Made by EPA for Evaluating the Risk from Alpha-Particle
Emitters
EPA has evaluated the risk to specific body organs by applying an RBE of 8 for alpha
radiations to the risk estimates for low dose rate, low-LET radiations as described above. As
in the case of low-LET radiations, EPA risk estimates for high-LET radiations are based on a
linear dose response function. For bone cancer and leukemia, EPA uses the absolute risk
projection model described in the previous section. For other cancers, the Agency uses
relative risk projections.
6-33
-------
Lifetime risk estimates for alpha doses, as a function of age, sex, and cancer site, are
easily obtained by multiplying the appropriate entry in Table 6-6 or 6-7 by a factor of 8. The
whole-body risks from lifetime exposure of the general population are then calculated to be
3.1 X 10'3/rad (mortality) and
5.0 X 10'3/rad (incidence).
As outlined above, the risk estimate for bone cancer in the BEIR IE report is based
directly on data for high-LET (alpha) radiation. Some readers may note that the EPA high-
LET risk estimate, 20 bone cancer fatalities per 106 person-rad, is less than the 27 fatalities
listed in Table A-27 of NAS80 for alpha particles. This is because the analysis in Appendix
A of NAS80 (but not Chapter V of that report) assumes that in addition to a 2-year minimum
induction period, 25 years are available for cancer expression. This is usually not the case for
doses received beyond about age 50. Hence, the estimated lifetime risk is smaller when it is
based on a life table analysis that considers lifetime exposure in conjunction with competing
causes of death.
6.3.4 Uncertainties in Risks from Alpha-Particle Emitters
The uncertainties in risk associated with internally deposited alpha emitters are often
greater than for low-LET radiation. Human epidemiological data on the risks from alpha
emitter are largely confined to: (1) lung cancer induced by radon decay products (see below);
(2) bone cancer induced by radium; and (3) liver cancer induced by injected thorotrast
(thorium). Many of the risk estimates presented here for alpha irradiation assume an RBE of
8, as determined from high dose experiments on animals. The available evidence on cells,
animals, and humans points to a linear dose response relationship for the risk from alpha
emitters (NAS88). The extrapolation to low doses is therefore considered to be less important
as a source of uncertainty for alpha irradiation than for low-LET irradiation. There is,
however, considerable variability in the RBE determined from animal studies; the
extrapolation of these results to humans is also problematic.
For many alpha-emitting radionuclides, the most important source of uncertainty in the
risk estimate is the uncertainty in the dose to target cells. Contributing to this uncertainty are
uncertainty in the location of these cells, ignorance regarding the metabolism of the
radionuclide, nonuniformity of radionuclide deposition in an organ, and the short range of
alpha particles in tissue (see Chapter 5).
In the case of alpha irradiation of the lung by radon decay products, there are human
epidemiological data that allow direct estimation of the risk per unit exposure. Knowledge of
RBE and the actual dose to target cells is therefore not important except as the dose per unit
exposure might differ between mine and indoor environments. As a consequence, the
estimated uncertainty in average radon risk estimates is similar to that for low-LET radiation.
[As discussed in Section 6.4.5, the EPA is employing a central risk estimate for excess radon
exposure of 360 fatal lung cancers/106 WLM and an uncertainty range of 140-720 fatal lung
cancers/10s WLM.]
As discussed in Section 6-2, recent analyses of the Japanese A-bomb survivor data
indicate that risk estimates for whole-body, low-LET radiation predicated on the linear,
6-34
-------
relative risk model will have to be increased approximately three-fold, although individual
organ risks will generally change by differing factors. Since the organ specific, high-LET
risk estimates used here are 8 times those calculated for low-LET radiation, one would expect
a corresponding 3-fold increase in high-LET risk estimates. Moreover, application of a DREF
to the calculation of low-LET risks would not affect this conclusion, since, as discussed
above, this would imply a compensating increase in the RBE. Consequently, it might be
argued that current EPA estimates of risk due to alpha irradiation are too low.
While EPA intends to conduct a comprehensive review of both its low- and high-LET
risk estimates after the BEIR V report becomes available, we do not believe that current high-
LET risk estimates are biased low in a serious way. It should be noted, in this connection,
that the doses from internally deposited alpha emitters are usually concentrated in certain
organs - especially bone, bone marrow, and lung. Risks of bone cancer caused by bone
seeking radionuclides (NAS80; NAS88) or lung cancers caused by inhaled radon decay
products (see Section 6.4) are derived directly from epidemiological data on high-LET
radiation; consequently, these risk estimates will not be affected by changes in the Japanese
data. Epidemiological evidence indicates that the risk of radiogenic leukemia induced by
alpha emitters deposited in the bone is lower than would be estimated from the gamma ray
risk after adjusting for alpha RBE (NAS88); possibly this discrepancy relates to difficulty in
estimating dose to target cells in the bone marrow due to alpha particles originating in the
mineral phase of the bone. EPA's estimates of risk from alpha emitters deposited in the lung
in the form of insoluble particles are also conservative. Alpha radiation emitted from such
particles, for the most part, irradiate the pulmonary region of the lung (the alveoli). The risk
of lung cancer is calculated, in this case, by multiplying the pulmonary region dose by the
risk factor for the whole lung. Using the pulmonary dose as an effective lung dose will bias
the risk estimate high by an unknown but possibly large factor, especially since the great
majority of human lung cancers seem to originate in the tracheobronchial region of the lung.
The next section describes how EPA estimates the risk due to inhalation of alpha-
emitting radon progeny, a situation where the organ dose is highly nonuniform.
6.4 ESTIMATING THE RISK FROM LIFETIME POPULATION EXPOSURES FROM
RADON-222 PROGENY
The Agency's estimates of the risk of lung cancer due to inhaled radon progeny do not
use a dosimetric approach, but rather are based on what is sometimes called an
epidemiological approach: that is, on the excess human lung cancer in groups known to have
been exposed to radon progeny.
When radon-222, a radioactive noble gas, decays, a number of short half-life
radionuclides (principally polonium-218, lead-214, bismuth-214, and polonium-214) are
formed. These decay products, commonly referred to as "progeny" or "daughters," readily
attach to inhalable aerosol particles in air. When inhaled, the radon progeny are deposited on
the surfaces of the larger bronchi of the lung. Since two of these radionuclides decay by
alpha-particle emission, the bronchial epithelium is irradiated by high-LET radiation. A
wealth of data indicate that a range of exposures to the bronchial epithelium of underground
miners causes an increase in bronchial lung cancer, both in smoking and in nonsmoking
6-35
-------
miners, and in some members of the general public. Recently the National Academy of
Sciences, BEER IV Committee, and the International Commission on Radiological Protection
reviewed the question of radon risks and reported their conclusions (NAS88, ICRP87).
Although considerable progress has been made in modeling the deposition of radon
daughters in the lung, it is not yet possible to characterize adequately the bronchial dose
delivered by alpha radiation from inhaled radon-222 progeny (NAS88). This is in part due to
the uncertainty concerning the kinds of cells in which bronchial cancer is initiated and the
depth of these cells in the bronchial epithelium.
Aside from the uncertainties in the dose calculations, a purely dosimetric approach to
radon risk estimation appears untenable. Such an approach relates the risk from a given
absorbed dose to the lung resulting from radon progeny exposure to that from gamma or x-
ray exposure. This approach ignores the extensive epidemiological data on radon exposed
miners and bases risk estimates indirectly on epidemiological studies of populations exposed
to low-LET radiation. It must also, therefore, make use of an RBE for alpha particles
estimated from animal studies. Given the uncertainties in the latter epidemiological studies
and in the RBE, there would seem to be no advantage to this approach. Consequently, EPA
agrees with the BEER IV Committee conclusion that radon decay product dosimetry in the
lung is only useful for extrapolating radon risk estimates from one exposure situation to
another (NAS88).
6.4.1 Characterizing Exposures to the General Population vis-a-vis Underground
Miners
Exposures to radon progeny under working conditions are commonly reported in a
special unit called the working level (WL). One working level is any combination of short
half-life radon-222 progeny having 1.3 x 105 MeV per liter of potential alpha energy
(FRC67). This value was chosen because it is the alpha energy released from the total decay
of the short-lived radon progeny at radioactive equilibrium with 100 pCi/L of
radon-222. The WL unit was developed because the concentration of specific radon progeny
depends on ventilation rates and other factors. A working level month (WLM) is the unit
used to characterize a miner's exposure to one working level of radon progeny for a working
month of about 170 hours. Because the results of epidemiological studies are expressed in
units of WL and WLM, the following outlines how they can be interpreted for members of
the general population exposed to radon progeny.
There are age- and sex-specific respiratory rate and volume differences, as well as
differences in duration of exposure, in a general population as compared to a mining
population. In earlier reports, EPA used an "exposure equivalent," a modified WLM in which
adjustments were made for age-specific differences in airway dimensions and surface area,
respiratory frequency, and tidal volume. These factors were expected to influence aerosol
deposition and, therefore, radiation dose from radon daughters. This approach to quantifying
exposure, correcting for differences in these factors, was recommended by Evans (Ev69) and
is consistent with the original derivation of the working level (Ho57).
6-36
-------
The BEIR IV Committee, however, concluded that the tracheo-bronchial "dose per
WLM in homes, as compared to that in mines, differs by less than a factor of 2," and advised
that the dose and risk per WLM exposure in residences and in mines should be considered to
be identical until better dosimetric estimates are developed (NAS88). EPA will follow the
lead of the BEIR IV Committee in this regard and will not use the "exposure equivalent"
correction employed to compensate for age- and sex-specific tracheo-bronchial deposition in
earlier EPA reports. In this report, exposure of any individual to 1 WL for 170 hours is 1
WLM and for 1 year is 51.56 WLM. This change puts EPA risk estimates in standard units
generally used for this purpose, still without requiring dose calculations.
For indoor exposure, an occupancy factor of 0.75 is still employed. Discussion of the
support for this estimate can be found in EPA86.
6.4.2 The EPA Model
The initial EPA method for calculating radon risks has been described in detail
(EPA79, E179). As new data were reported, the EPA revised its model to reflect changes, as
contained in consecutive reports (EPA79, EPA82, EPA83a, EPA83b, EPA84, EPA85,and
EPA86). The Agency initially projected radon lung cancer deaths for both absolute and
relative risk models, but, since 1978, EPA has based risk estimates due to inhaled radon-222
progeny on a linear dose response function, a relative risk projection model, and a minimum
induction period of 10 years. A life table analysis has been used to project this risk over a
full life span. Lifetime risks were initially projected on the assumption that an effective
exposure of 1 WLM increased the age-specific risk of lung cancer by 3 percent over the age-
specific rate in the U.S. population as a whole (EPA79). In the most recent documents,
lifetime risks were calculated for a range of risk coefficients from 1 percent to 4 percent per
WLM (EPA86).
Although occupational exposures to pollutants other than
radon-222 progeny are probably not important factors in the observed lung cancer risk for
underground miners (E179, Th82, Mu83, Ra84, Se88), the use of occupational risk data to
estimate the risk of a general population is far from optimal, as it provides no information on
the effect of radon progeny exposures for children and women. While for most estimates, it
is assumed that the risk per unit dose received by children is no higher than that received by
adults, this assumption may not be correct.
The A-bomb survivor data indicate that, in general, the risk from childhood exposure
to low-LET radiation is greater than from adult exposure and continues for at least 33 years,
the time over which A-bomb survivors have been observed (Ka82). There are not, as yet,
adequate age-specific data on occurrence of lung cancer in those under 10 years of age at the
time of exposure (Ka82). Another limitation of the underground miner data is the absence of
women in the studied populations. The A-bomb survivor data indicate women are as
sensitive as men to radiogenic lung cancer from low-LET radiation even though, on the
whole, they smoke less (Pr83). These data are not conclusive, however.
6-37
-------
6.4.3 Comparison of Earlier Risk Estimates
Several estimates of the risk due to radon progeny have been published since the
original EPA model was developed. These risk estimates were reviewed recently in a number
of EPA reports (EPA84, EPA85, and EPA86).
The recent EPA risk estimates for lifetime exposure to a general population, along
with AECB, NAS, UNSCEAR, ICRP, and NCRP estimates of the risk of lung cancer
resulting from inhaled radon progeny, are listed in Table 6-10. The AECB estimate for
lifetime exposure to Canadian males is 830 fatalities per million person-WLM (Th82). In
Table 6-10, this estimate has been adjusted for the U.S. 1970 male and female population.
The National Institute for Occupational Safety and Health reviewed published data on
miner studies used as a basis for estimated risk coefficients and pointed out some of the
strengths and limitations of selected studies (NIOSH87).
The occupational exposure groups that constitute the epidemiological database for the
risk estimates are as follows:
1. U.S. Uranium Miners (NIOSH87)
(a) Strengths: A large, clearly defined, well-traced cohort with some smoking
histories and exposure records on the same persons. Standard sampling
techniques were used to make measurements.
(b) Limitations: There were few measurements in small mines, work histories
were self-reported, exposures were high, and potential error due to excursions
in exposure levels is high.
(c) Follow-up: 19 years in 1977.
2. Czechoslovakian Uranium Miners (NIOSH87)
(a) Strengths: Extensive exposure data with a large number of low level exposures
and limited exposure to other underground mining. Many possible confounding
factors have been investigated and eliminated.
(b) Limitations: Exposure estimates prior to 1960 based on radon gas
measurements. Person years at risk not determined in standard manner.
Smoking effect neglected. Elevated levels of arsenic in ore.
(c) Follow-up: 26 years in 1975.
3. Ontario Uranium Miners (NIOSH87)
(a) Strengths: Miners received low mean cumulative exposures. Prior mining
experience was carefully traced. Exposures prior to 1967 may be disputed.
6-38
-------
Table 6-10. Risk estimate for exposures to radon progeny.
Organization
EPA
NAS*
AECB"
ICRP
UNSCEAR
NCRP°
Model
Rel.
A-S Abs.
Rel.
-
-
Dec. Abs.
Fatalities per Exposure period
106 person- WLM
760 (460)a
730 (440)a
600 (300)a
150-450
200-450
130
Lifetime
Lifetime
Lifetime
Working Lifetime
Lifetime
Lifetime
Expression
period
Lifetime
Lifetime
Lifetime
30 years
40 years
Lifetime
*BEIR m
EPA and AECB based their estimates of risk for the general population on an
exposure equivalent, corrected for breathing rate (and other factors). For comparison
purposes, the values in parentheses express the risk in more customary units, in which
a continuous annual exposure to 1 WL corresponds to 51.6 WLM.
Adjusted for U.S. General Population: see text.
NCRP84: Table 10.2; assumes risk diminishes exponentially with a 20-year halftime,
and no lung cancer risk is expressed before age 40.
Sources:
EPA83b; NAS80; Th82; ICRP81; EPA86; UNSC77; NCRP84; USRPC80.
Models: Rel. - Relative Risk Projection
A-S Abs. - Age-Specific Absolute Risk Projection
Dec. Abs. - Decaying Absolute Risk Projection
6-39
-------
(b) Limitations: Median age of the cohort was 39 years in 1977. Thoron and
gamma exposures may have been high but not accounted for. Smoking history
is limited.
(c) Follow-up: 18 years in 1977.
4. Malmberget Iron Miners (NIOSH87)
(a) Strengths: Low exposure levels, long follow-up and stability of work force.
Complete ascertainment of vital status and confirmation of diagnosis. Risk
from confounders was examined and ruled out.
(b) Limitations: Relatively small cohort with limited exposure data and an
unclear cohort definition.
(c) Follow-up: 44 years in 1976
5. Eldorado - Uranium Miners (NAS88)
(a) Strengths: Very low exposure rates, miners screened for prior mining
experience, roughly equal groups of surface only and underground only
miners, Silica and diesel exhaust exposures low. Potential
confounders investigated.
(b) Limitations: Exposure estimates are disputed. Sixteen percent of the miners
excluded for incorrect or missing data. Average age in 1980 was 43 years.
(c) Follow-up: 14 years in 1980.
6.4.4 Recent Radon Risk Estimates
6.4.4.1 BEIRIV
At the beginning of 1988, the National Academy of Sciences released the BEIR IV
Committee report, reviewing information on the risks from radon and other alpha-emitting
radionuclides (NAS88). With the cooperation of the principal investigators, BEIR IV
examined in detail the mortality experience of four cohorts of underground miners (the U.S.,
Ontario, and Eldorado uranium miners and the Malmberget iron miners) and how the
mortality related to radon daughter exposure. The Committee calculated the relationship of
age-specific relative risk to exposure level and time-since-exposure (TSE) in two analyses.
The first used internal cohort comparisons and was a grouped-data analog of a Cox relative-
risk regression (NAS88). The second analysis compared the cohorts with external rates and
was a generalization of standard SMR methods. Separate parallel analyses were carried out to
establish a single combined value for each parameter.
The mathematical form of the Committee's preferred TSE model for the radon related
age-specific mortality rate at age a is
6-40
-------
r(a) = r0(a)[l + 0.025 y(a)(W1 + O.SWj)] (6-1)
where
r0(a) = age-specific lung cancer mortality rate
y(a) = 1.2, if a is less than 55 years
1.0, if a is between 55 and 64 years
0.4, if a is greater than 64 years
Wj = WLM incurred between 5 and 15 years prior to age a
W2 = WLM incurred more than 15 years prior to age a
The Committee model is, therefore, an age-specific, relative-risk projection model with a 5-
year latent period prior to expression of risk.
The BEIR IV Committee also estimated what the lung cancer risk coefficient would be
for an age-constant, relative-risk model. The results of this analysis are summarized in
Table 6-11.
Table 6-11. BEIR IV committee estimate of lung cancer risk coefficient for age-constant,
relative-risk model.
Cohort
U.S.
Ontario
Eldorado
Malmberget
Combined
Excess Risk
per WLM
0.6
1.4
2.6
1.4
1.34
95% Confidence
Limits
0.3 - 1.3
0.6 - 3.3
1.3 - 6.0
0.3 - 8.9
0.8 - 2.3
In its analysis, the BEIR IV Committee identified two major areas of uncertainty
affecting its conclusions: (1) uncertainty related to the Committee's analysis of cohort data
and (2) uncertainty related to projection of the risk to other groups. The Committee's TSE
model uses risk coefficients derived from analysis of data from four miner cohorts. Random
or systematic errors, particularly systematic errors, could bias the conclusions. Sources of
error in addition to basic sampling variation include: (1) errors in exposure estimates,
particularly since the magnitude of error may differ among the studies; (2) errors of
misclassification of cause of death; (3) errors in smoking status of individual miners, and (4)
modeling uncertainty--i.e., does the model properly address all parameters that are
determinants of risk?
6-41
-------
Having developed the TSE model for miners, the Committee anticipated the following
sources of uncertainty in projecting the model across other groups: (1) effect of gender (miner
data all for males); (2) effect of age (miner data contain no information on exposures before
about age 20); (3) effect of smoking (miner data contain poor information on smoking status);
(4) temporal expression of risk (not enough miners have died to establish accurately the
pattern of lifetime risk from radon exposure), and (5) extrapolation from mining to indoor
environments (what are significant differences in the air in mines compared to air indoors?).
After reviewing the various sources of uncertainty, the BEIR IV Committee concluded [p42],"
...The imprecision that results from sampling variation can be readily quantified, but other
sources of variation cannot be estimated in a quantitative fashion." Therefore, the Committee
chose not to combine the various uncertainties into a single numerical value" (NAS88).
The question of errors in exposure estimates is particularly interesting since the
modeling is strongly influenced by the U.S. uranium miner data. In fact, the model risk
estimates would be 33 percent higher if the U.S. cohort was removed. Exposure hi the U.S.
cohort is poorly known: cumulative WLM (CWLM) are calculated from measured radon
levels for only 10.3 percent of the miners, varying amounts of estimation are required for
about 36.1 percent of the miners, and guesswork is used for about 53.6 percent of the miners
(NAS88, Lu71). Only 26.1 percent of the U.S. uranium miner exposure data are based on
measured values (Lu71).
The Ontario cohort exposure estimates also are not well founded. Upper and lower
estimates were developed: the lower from measured values, the upper based on engineering
judgment (NAS88). Eldorado cohort estimates of CWLM were based almost entirely on
measured values, while Malmberget cohort estimates were based on a reconstruction of past
ventilation conditions (NAS88). Of the four cohorts, the United States has one of the poorest
bases for CWLM estimates. One serious problem is the potential error due to large
excursions in radon daughter concentrations (NIOSH87). The uncertainties in exposure
estimates are particularly significant in view of the rather large impact the U.S. cohort has on
the form of the model.
When the BEIR IV model is run with the 1980 lifetable and vital statistics at an
exposure level of 0.001 WLM per year, the reference risk can be calculated (see Table 6-12).
Table 6-12. BEER. IV Risk Model - Lifetime Exposure and Lifetime Risk.
Group Risk (lO'VWLM)
Male 530
Female 185
Combined 350
6-42
-------
6.4.4.2 ICRP 50
The International Commission on Radiological Protection, in its Publication 50,
addressed the question of lung cancer risk from indoor radon daughter exposures. The ICRP
Task Group took a direction quite different from the BEIR Committee. The Task Group
reviewed published data on three miner cohorts: U.S., Ontario, and Czech uranium miners.
The estimated risk coefficients by cohort are presented in Table 6-13.
Table 6-13. Estimated lung cancer risk coefficients from radon progeny exposure for three
miner cohorts.
Cohort Follow-up Relative model Absolute model
U.S. 1950-1977 0.3%- 1.0%
Czech 1948-1975 1.0%-2.0%
Ontario 1958-1981 0.5%-1.3%
Average 1%
2-8 cases/106 PWLMY
10-25 cases/106 PWLMY
3-7 cases/106 PWLMY
10 cases/106 PWLMY
Source: ICRP87.
The relative risk model then developed for a constant exposure rate is:
t-T
MO = >.0(t)[l + I r(tj ^tj dtj (6-2)
0
= the mortality rate at age t
where:
X,0(t) = the age-specific lung cancer rate at age t
r(X) = risk coefficient at age of exposure t^
^tg) = age-dependent exposure rate
1 = time lag (minimal latency) = 10 years
In the case of a constant exposure rate or constant annual exposure, the equation collapses
to:
6-43
-------
_
E(t - T)] (6-3)
where:
r = age averaged relative risk coefficient
E(t - x)= E[t - T]
= cumulative exposure to radon daughters to age t-x
Since ICRP recommends the use of the relative risk model, the ICRP 50 absolute risk
model will not be addressed further in this document.
To adapt the relative risk model derived from studies of underground miners for the
general population, the ICRP Task Group introduced several adjustments. The first was to
correct for co-carcinogenic influences in mines. To account for unidentified, unproven
carcinogens that might be present in mine environments but not elsewhere, only 80 percent of
the risk was attributed to radon. The second adjustment was for dosimetric corrections. The
dose to bronchial epithelium used by the Task Group for persons indoors was estimated to be
only 80 percent as large as that for persons in mines; therefore, the risk to the public from
radon was considered to be 80 percent of the risk of miners.
Adjusting the average relative risk coefficient of
1 percent per WLM by these two factors gives a risk coefficient of 0.64 percent per WLM:
1.0% x 0.8 x 0.8 = 0.64%. (6-4)
The third adjustment made by the Task Group is related to age. Since reports of
Japanese A-bomb survivors and some other radiation-exposed groups support an elevated
estimate of risk in children compared to adults, the Task Group increased the risk coefficient
of persons between birth and age 20 by a factor of 3.
The final relative risk coefficients in the ICRP 50 model are: 1.9 percent per WLM if
the age at time of exposure is between birth and 20 years, and 0.64 percent per WLM if age
at time of exposure exceeds 20 years.
When the ICRP 50 relative risk model is run with 1980 U.S. lifetable and vital
statistics at an exposure level of 0.001 WLM per year, the reference risk calculated is:
Group Risk (lO'VWLM)
Male 610
Female 205
Combined 420
6-44
-------
6.4.5 Selection of Risk Coefficients
To estimate the range of reasonable risks from exposure to radon-222 progeny for use
in the Background Information Document for Underground Uranium Mines (EPA85), EPA
averaged the estimates of BEIR m, the EPA model, and the AECB to establish an upper
bound of the range. The lower bound of the range was established by averaging the
UNSCEAR and ICRP estimates. The Agency chose not to include the NCRP estimate in its
determination of the lower bound because this estimate was believed to be outside the lower
bound. With this procedure, the EPA arrived at relative risk coefficients of 1.2 percent to 2.8
percent per WLM exposure equivalent (300 to 700 fatalities per million person-WLM
exposure equivalent) as estimates of the possible range of effects from inhaling radon-222
progeny for a full lifetime. Although these risk estimates did not encompass the full range of
uncertainty, they seemed to illustrate the breadth of much of current scientific opinion.
The lower limit of the range of 1985 EPA relative risk coefficients, 1.2 percent per
effective WLM, was similar to that derived by the Ad Hoc Working Group to Develop
Radioepidemiological Tables, which also used 1.2 percent per WLM (NIH85). However,
some other estimates based only on U.S. and Czech miner data averaged 1 percent per WLM
(Ja85) or 1.1 percent per WLM (St85). On the other hand, three studies - two on miners
(Ra84, Ho86) and one on residential exposure (Ed83, Ed84) - indicated a relative risk
coefficient greater than 3 percent per WLM, perhaps as large as 3.6 percent
The EPA therefore increased the upper limit of its estimated range of relative risk
coefficients. To estimate the risk due to radon-222 progeny, the EPA used the range of
relative risk coefficients of 1 to 4 percent per WLM. (See EPA86 for a more detailed
discussion.) Based on 1980 vital statistics, this yielded, for members of the general public, a
range of lifetime risks from 380 to 1,520 fatal cases per 106 WLM (expressed in exposure
equivalents). In standard exposure units, uncorrected for breathing rate and age, this
corresponds to 230 to 920 cases per 106 WLM. Coincidentally, the geometric mean estimate
obtained in this way with 1980 vital statistics, 4.6xlO"4/WLM in standard units of exposure, is
numerically the same as that obtained using a 3 percent relative risk coefficient and 1970 vital
statistics (see Table 6-7).
However, in light of the two recently published consensus- based reports, BEIR IV
and ICRP 50, and a recent report on the Czech miner groups (Se88), the Agency has
reviewed its basis for radon risk estimation. Comparable relative risk coefficients for miners
(age-constant relative risk) yield a coefficient of around 1 percent in ICRP 50, 1.34 percent in
BEIR IV, and 1.5 percent in the Czechs. This suggests that the range, 1 percent to
4 percent, used by EPA may be too wide. Nevertheless, note that only 5 of the 20 or so
studies for which there are some data are included in these estimates.
The BEIR IV Committee noted and modeled a drop in relative risk with increasing
time of exposure and a decreasing relative risk with increasing age after exposure (NAS88).
The Czech miners show a similar response pattern (Se88). Though the Committee did note a
dose rate effect in the U.S. uranium miner cohort, i.e., a decrease in risk per unit exposure at
high dose rates, it was not included in the model (NAS88). The possibility of a similar dose-
rate effect was found recently in a study on Port Radium uranium miners (Ho87).
6-45
-------
The ICRP 50 Task Group worked from a different database and developed a simpler
model with fewer age- and time-dependent parameters. The Task Group provided a 3 times
higher risk for exposure between birth and 20 years of age than after 20 years of age
(ICRP87). The finding in the recent Czech report that risk prior to age 30 is 2 to 2.5 times
greater than after age 30 lends some support to the ICRP conclusions (Se88).
Both BEER IV and ICRP 50 models treat radon and smoking risks as multiplicative.
This conclusion is based primarily on data from the U.S. uranium miner cohort. Although
apparently based on weaker evidence, the report on Malmberget miners and the recent report
on Czech miners both concluded that the interaction of smoking and radon exposure is small
(Ra84, Se88). The attributable risk per unit exposure in smokers and non-smokers was
essentially the same (Se88). The true interaction of radon and cigarette smoking is
controversial. Both antagonistic (Ax78, Lu79, Ax80) and multiplicative (Lu69, Wh83)
interactions have been reported in man, and animal studies can be found to justify any
position (Ch81, Ch85, Cr78). In prior calculations, EPA has always treated the interaction
between radon daughters and cigarette smoke as multiplicative. EPA will continue to treat
the radon daughter-smoke interaction as multiplicative at this time.
Important unresolved issues pertaining to the risks from inhaled radon progeny remain.
At the advice of the Radiation Advisory Committee of EPA's Science Advisory Board, EPA
will continue to use relative risk models but shall include both BEIR IV and ICRP 50 model
calculations to illustrate the difference in results from the two models. The ICRP 50 model
will be slightly modified. The risk reduction factor of 0.8 to compensate for differences in
dosimetry will be removed to place the ICRP 50 model and BEIR IV model on a comparative
basis. Calculations in the ICRP 50 model will be made using risk coefficients of 2.4 percent
per WLM from birth to age 20 and 0.8 percent per WLM for ages greater than 20 years,
yielding estimates listed in Table
6-14.
Table 6-14 summarizes risk estimates based on the BEIR IV and the ICRP 50 model,
modified as described above. For the calculations in this document, both models were
adjusted for the effect of background radon exposure (see section below).
6-46
-------
Table 6-14. Lifetime risk from radon daughter exposure of Lung cancer death (per 106
WLM).
Model
Group BEIRIV ICRP 50
Men
Women
530
185
760
255
Combined Population 350 500
(Range) - (170-840)
The ICRP Task Group concluded that, all things considered, the range of variation of
the mean relative risk coefficient is from about 0.3 up to 2 times the value stated (ICRP87).
The range of risk cited in Table 6-14 for the ICRP model reflects this uncertainty in the risk
coefficient. Since the BEIR IV Committee did not provide a numerical range of uncertainty,
no range is given for that model.
Correction of Radon Risk Estimates for the Effect of Background Exposure
A relative risk model for radon-induced lung cancer generally assumes the excess risk,
A^, from a given exposure, is proportional to the observed baseline risk of lung cancer in the
population, A,0. Thus, for a constant exposure rate, w, the excess risk at age, a, attributable to
previous exposure can be written:
A,(w,a) = A,0(a) |3(a)f(w,a) (6-5)
For example, in the case of an age-constant relative risk model with a 10-yr minimum
latency:
|3(a) = p = constant (6-6)
f(w,a) = (a-10)w (6-7)
Although \ is commonly assumed to be proportional to A,0, a more consistent (and
biologically plausible) way to formulate a relative risk model is to assume that the radon risk,
A^ is proportional to A,0'» the lung cancer rate that would prevail in the absence of any radon
exposure (Pu88):
A,(w,a) = \ (a)p(a)f(w,a) (6-8)
6-47
-------
Presuming that the risk model can be used to relate X,0(a) to X,0'(a), then
^(a) = ^ (a) [1 + P(a)f(w,a)] (6-9)
where w is the average exposure rate in the population. It follows from the previous equation
that
X,l(a) = X0(a)/[l + P(a)f(w,a)] (6-10)
The inferred baseline rate without radon exposure depends, of course, on both the risk
model and the presumed average background exposure rate. The excess risk associated with
an arbitrary exposure situation can be calculated using standard life table methodology.
The ICRP 50 committee did correct the baseline rate in this way in calculating lifetime
population risks, assuming an average exposure rate of 0.2 WLM/yr. The BEIR IV
Committee did not incorporate the correction, noting that it would be small (see NAS88, p.
53). In arriving at a final estimate based on the ICRP 50 and BEIR IV models (see Table 6-
15), EPA has incorporated a model-specific baseline correction, calculated on the assumption
of a 0.25 WLM/yr average radon exposure rate (Pu88). As seen from Tables 6-14 and 6-15,
this correction results in roughly a 15 percent reduction in each of the estimates of lifetime
risk for the general population.
Table 6-15. Lifetime risk from excess radon daughter exposure (adjusted for a background
exposure of 0.25 WLM/yr).
Risk of Excess Lung Cancer Deaths per 106 WLM
Group BEIR IV ICRP 50 Average
Men
Women
Population
Combined
(Range)
460
160
305
640
215
420
(140-720)
550
190
360
(140-720)
Summary of Baseline Corrected Radon Risk Estimates
Consistent with the recommendations of the Agency's Radiation Advisory Committee,
EPA has here averaged the risk estimates derived from the BEIR IV and ICRP 50 models.
These estimates are based on 1980 U.S. vital statistics and are adjusted for an assumed
background exposure of 0.25 WLM/yr. Thus, as shown in Table 6-15, the excess lifetime
risk in the general population due to a constant, low-level, lifetime exposure is estimated to
be 360 excess lung cancer deaths per 106 WLM, with a range of 140 to 720 excess lung
6-48
-------
cancer deaths per 106 WLM. (At lifetime exposures above about 100 WLM, numerical
estimates would be reduced because of "competing risk" considerations.)
The BEIR IV and ICRP models differ substantially with respect to their dependence
on age and time since exposure. Hence, in evaluating exposures at different ages or time
periods it is instructive to consider the predictions made by each model. Illustrative examples
of such calculations are given in Tables
6-16 and 6-17.
6-49
-------
Table 6-16. Lifetime risk for varying age at first exposure and duration of exposure
(Background = 0.25 WLM/yr).
Lifetime Risk of Lung Cancer per 10 WLM
Age(yr)
Birth
10
20
30
40
50
60
70
80
90
100
Male
Exposure
Duration(yr) BEIR IV ICRP 50
1
10
Lifetime
1
10
1
10
1
10
1
10
1
10
1
10
1
10
1
10
1
10
1
10
476
480
459
481
483
486
495
509
535
572
592
602
516
378
331
251
182
88
55
12
8
2
1
1382
1394
638
1398
1402
470
474
477
472
461
435
392
335
253
182
96
57
15
8
1
1
_
Female
BEIR IV ICRP50
184
185
159
186
186
188
190
195
205
217
217
208
170
114
95
69
52
32
21
7
4
1
511
515
213
516
517
173
173
172
168
161
148
130
109
79
58
34
22
8
4
-
_
6-50
-------
Table 6-17. Lifetime risk for varying age at first exposure and duration of exposure
(Background = 0.25 WLM/yr).
Excess Lung Cancer Deaths per 106
Persons Exposed at 1 WLM/yr
Male
Exposure
Age(yr)
Birth
10
20
30
40
50
60
70
80
90
100
Duration(yr)
1
10
Lifetime
1
10
1
10
1
10
1
10
1
10
1
10
1
10
1
10
1
10
1
10
BEIRIV
472
4723
32171
481
4814
486
4902
508
5299
571
5804
600
4909
374
2949
246
1406
84
323
11
30
2
2
ICRP50
1372
13725
44859
1398
13984
470
4691
476
46788
461
4267
391
3187
251
1623
94
439
14
45
1
2
_
-
Female
BEIRIV
183
1828
12352
186
1857
187
1891
195
2041
217
2142
208
1652
114
895
68
456
31
146
7
19
_
2
ICRP50
508
5085
16545
516
5159
172
1721
172
1676
161
1468
129
1051
79
546
34
192
8
30
.
1
_
-
6-51
-------
6.5 OTHER RADIATION-INDUCED HEALTH EFFECTS
The earliest report of radiation-induced health effects was in 1896 (Mo67), and it dealt
with acute effects in skin generally caused by very large x-ray exposures. Within the six-year
period following, 170 radiation-related skin damage cases had been reported. Such injury,
like many other acute effects, is the result of exposure to hundreds or thousands of rads.
Under normal situations, environmental exposure does not cause such large doses, so possible
acute effects will not need to be considered in assessing the risk to the general population
from routine radionuclide emissions.
Radiation-induced carcinogenesis was the first delayed health effect described: the
first case was reported in 1902 (Vo02), and 94 cases of skin cancer and 5 of leukemia were
reported by 1911 (Up75). Radiation-induced genetic changes were noted soon afterward. In
1927, HJ. Muller described x-ray-induced mutations in animals (in the insect, Drosophila),
and in 1928, LJ. Stadler reported a similar finding in plants (Ki62). At about the same time,
radiation effects on the developing human embryo were observed. Case reports in 1929
showed a high rate of microcephaly (small head size) and central nervous system disturbance
and one case of skeletal defects in children irradiated in utero (UNSC69). These effects, at
unrecorded but high exposures and at generally unrecorded gestational ages, appeared to
produce central nervous system and eye defects similar to those reported in rats as early as
1922 (Ru50).
For purposes of assessing the risks of environmental exposure to radionuclide
emissions, the genetic effects and in utero developmental effects are the only health hazards
other than cancer that are addressed in this Background Information Document (BID).
6.5.1 Types of Genetic Harm and Duration of Expression
Genetic harm (or the genetic effects) of radiation exposure is defined as stable,
heritable changes induced in the germ cells (eggs or sperm) of exposed individuals, which are
transmitted to and expressed only in their progeny and in future generations.
Of the possible consequences of radiation exposure, the genetic risk is more subtle
than the somatic risk, since it affects not the persons exposed, but relates only to subsequent
progeny. Hence, the time scales for expression of the risk are very different. Somatic effects
are expressed over a period on the order of a lifetime, while about 30 subsequent generations
(nearly 1,000 years) are needed for near complete expression of genetic effects. Genetic risk
is incurred by fertile people when radiation damages the nucleus of the cells which become
their eggs or sperm. The damage, in the form of a mutation or a chromosomal aberration, is
transmitted to, and may be expressed in, a child conceived after the radiation exposure.
However, the damage may also be expressed in subsequent generations or only after many
generations. Alternatively, it may never be expressed because of failure to reproduce or
failure of the chance to reproduce.
EPA treats genetic risk as independent of somatic risk even though somatic risk may
be caused by mutations in somatic cells because, whereas somatic risk is expressed in the
person exposed, genetic risk is expressed only in progeny and, in general, over many
6-52
-------
subsequent generations. Moreover, the types of damage incurred often differ in kind from
cancer and cancer death. Historically, research on genetic effects and development of risk
estimates have proceeded independently of the research on carcinogenesis. Neither the dose
response models nor the risk estimates of genetic harm are derived from data on studies of
carcinogenesis.
Although genetic effects may vary greatly in severity, the genetic risks considered by
the Agency in evaluating the hazard of radiation exposure include only those "disorders and
traits that cause a serious handicap at some time during lifetime" (NAS80). Genetic risk may
result from one of several types of damage that ionizing radiation can cause in the DNA
within eggs and sperm. The types of damage usually considered are: dominant and recessive
mutations in autosomal chromosomes, mutations in sex-linked (x-linked) chromosomes,
chromosome aberrations (physical rearrangement or removal of part of the genetic message on
the chromosome or abnormal numbers of chromosomes), and irregularly inherited disorders
(genetic conditions with complex causes, constitutional and degenerative diseases, etc.).
Estimates of the genetic risk per generation are conventionally based on a 30-yr
reproductive generation. That is, the median parental age for production of children is
defined as age 30 (one-half the children are produced by persons less than age 30, the other
half by persons over age 30). Thus, the radiation dose accumulated up to age 30 is used to
estimate the genetic risks. EPA assessment of risks of genetic effects includes both first
generation estimates and total genetic burden estimates.
In the EPA Background Information Document for Radionuclides (EPA84), direct and
indirect methods for obtaining genetic risk coefficients are described, and some recent
estimates based on these methods are tabulated. Briefly, the direct method takes the
frequency of mutation or occurrence of a heritable defect per unit exposure observed in
animal studies and extrapolates to what is expected for humans. Direct estimates are usually
used for first generation effects estimates. The indirect method, on the other hand, uses
animal data in a different way. The estimated human spontaneous mutation rate per gene site
is divided by the average radiation-induced mutation rate per gene observed in mouse studies,
to obtain the relative radiation mutation risk in humans. The inverse of this relative radiation
mutation risk is the expected "doubling dose" for radiation-induced mutations in man. The
doubling dose is the exposure in rads which will double the current genetic malformation
level in man and usually is used to estimate equilibrium effects or all future generation
effects.
A doubling dose estimate assumes that the total population of both sexes is equally
irradiated, as occurs from background radiation, and that the population exposed is large
enough so that all genetic damage can be expressed in future offspring. Although it is
basically an estimate of the total genetic burden across all future generations, it can also
provide an estimate of effects that occur in the first generation. Usually a fraction of the total
genetic burden for each type of damage is assigned to the first generation using population
genetics data as a basis to determine the fraction. For example, the BEIR HI Committee
geneticists estimated that one-sixth of the total genetic burden of x-linked mutations would be
expressed in the first generation and five-sixths across all subsequent generations. EPA
6-53
-------
assessment of risks of genetic effects includes both first generation estimates and total genetic
burden estimates.
The 1986 UNSCEAR report (UNSC86) reviewed data on genetic effects. While there
was much new information, changes in direct estimates of first generation risk were minimal,
reflecting primarily changes in estimates of survival of reciprocal translocations. There was
however, an appreciable change in the doubling dose estimate of genetic risk. Because of
Hungarian studies the birth prevalences of isolated and multiple congenital anomalies of in
man was estimated to be 597.4 per 104 live births (UNSC86). The UNSCEAR Committee
also estimated congenital anomalies and other multifactorial disorders to have a spontaneous
prevalence of 600,000 per 106 live births. The UNSCEAR Committee however, made no
estimate of the genetic radiation risk coefficients for these types of conditions (UNSC86).
The 1988 UNSCEAR Committee also reviewed genetic risks (UNSC88) and confirmed the
conclusions of the 1986 UNSCEAR Committee (Table 6-18).
The Agency concluded that the "spontaneous prevalence" of multifactorial disorders
described by the UNSCEAR Committees were not all "disorders and traits that cause a
serious handicap at sometime during lifetime." Since the multifactorial disorders compose a
large fraction of the genetic risk in the BEIR in report, the BEIR IQ risk estimates will be
used until the relevance of the Hungarian studies can be evaluated. The Agency also has
concluded estimates of detrement (years of life lost or impaired) as made by several
UNSCEAR Committees (UNSC82, 86, 88) should not be used to evaluate genetic risk at this
time. As these changes in genetic risk assessment mature, the Agency will review their
applicability.
6-54
-------
Table 6-18. UNSCEAR 1988 Risks of genetic disease per 1 million live-births in a
population exposed to a genetically significant dose of 1 rad per generation of
low-dose-rate, low-dose, low-LET irradiation.
Type of genetic
disorder
(100 rad doubling dose)
Current incident
per 106 liveborn
Effects of 1 rad per generation
First Generation Equilibrium
Autosomal dominant
and x-linked
Autosomal recessive
diseases
-Homozygous effects
-Partnership effects
Chromosomal diseases
due to structural
anomalies
Sub-total (rounded)
Early acting dominants
Congenital anomalies
Other multifactorial
diseases*
Heritable tumors
Chromosonal diseases
due to numerical
anomalies
* prevalence up to age 70
Source: UNSC88
10,000
25,000
400
13,000
unknown
60,000
600,000
unknown
3,400
15
no increase
negligible
2.4
100
11
18
115
not estimated
not estimated
not estimated
not estimated
not estimated
6-55
-------
6.5.2 Estimates of Genetic Harm Resulting from Low-LET Radiations
A number of committees have addressed the question of genetic risk coefficient
(NAS72, 80, 88; UNSC58, 62, 66, 72, 77, 82, 86, 88; Of80). The detailed estimates of the
BEIR m Committee (NAS80) are listed in Table 6-19, those of UNSCEAR (UNSC88) are
listed in Table 6-18, and a summary of estimates of the various committees is listed in
Table 6-20.
Although all of the reports cited above used somewhat different sources of
information, there is reasonable agreement in the estimates. However, all these estimates
have a considerable margin of error, both inherent in the original observations and in the
extrapolations from experimental species to man. Some of the committee reports assessing
the situation have attempted to indicate the range of uncertainty; others have simply used a
central estimate (see Table 6-20). The same uncertainties exist for the latter (central
estimates) as for the former.
Most of the difference is caused by the newer information used in each report. Note
that all of these estimates are based on the extrapolation of animal data to humans. Groups
differ in their interpretation of how genetic experiments in animals might be expressed in
humans. While there are no comparable human data at present, information on hereditary
defects among the children of A-bomb survivors provides a degree of confidence that the
animal data do not lead to underestimates of the genetic risk following exposure to humans.
(See "Observations on Human Populations," which follows.)
It should be noted that the genetic risk estimates summarized in Table 6-20 are for
low-LET, low-dose, and low-dose-rate irradiation. Much of the data was obtained from high
dose rate studies, and most authors have used a sex-averaged factor of 0.3 to correct for the
change from high-dose rate, low-LET to low dose rate, low-LET exposure (NAS72, 80,
UNSC72, 77). However, factors of 0.5 to 0.1 have also been used in estimates of specific
types of genetic damage (UNSC72, 77, 82).
Studies with the beta-particle-emitting isotopes carbon-14 and tritium yielded RBEs of
1.0 and 0.7 to about 2.0, respectively, in comparison to high-dose rate, high-dose exposure to
x-rays (UNSC82). At present, the RBE for genetic endpoints due to beta particles is taken as
1 (UNSC77, 82).
6.5.3 Estimates of Genetic Harm from High-LET Radiations
Although genetic risk estimates are made for low-LET radiation, some radioactive
elements, deposited in the ovary or testis, can irradiate the germ cells with alpha particles.
The relative biological effectiveness (RBE) of high-LET radiation, such as alpha particles, is
6-56
-------
Table 6-19. BEIR El estimates of genetic effects of an average population exposure of 1
rem per 30-yr generation (chronic x-ray or gamma radiation exposure).
Type of genetic
disorder
Current incidence Effect per 106 liveborn
Per 106 liveborn per rem per generation
Autosomal dominant
and x-linked
Irregularly inherited
Recessive
Chromosomal aberrations
10,000
90,000
1,000
6,000
First Generation*
5-65
(not estimated)
Very few
Fewer than 10
Equilibrium**
40-200
20-900
Very slow
increases
Increases
only
slightly
Total
107,000
5-75
60-1100
* First-generation effects estimates are reduced from acute fractionated exposure estimates
by a factor of 3 for dose rate effects and 1.9 for fractionation effects (NAS80, p. 117)
** Equilibrium effects estimates are based on low dose rate studies in mice (NAS80, pp. 109-
110).
Source: NAS80.
6-57
-------
Table 6-20. Summary of genetic risk estimates per 106 liveborn of low-dose rate, low-LET
radiation in a 30-yr generation.
Serious hereditary effects
Source
BEAR, 1956 (NAS72)
BEIR I, 1972 (NAS72)
UNSCEAR, 1972 (UNSC72)
UNSCEAR, 1977 (UNSC77)
ICRP, 1980 (Of80)
BEIR m, 1980 (NAS80)
UNSCEAR, 1982 (UNSC82)
UNSCEAR, 1986 (UNSC86)
UNSCEAR, 1988 (UNSC86)
First generation
-
49a (12-200)b
9a (6-15)
63
89
19a (5-75)
22
17
18
Equilibrium
(all generations)
500
300a(60-1500)
300
185
320
260a(60-1100)
149
104
115
a Geometric mean of the lower and upper bounds of the estimates. The geometric mean of
two numbers is the square root of their product.
b Numbers in parentheses are the range of estimates.
6-58
-------
defined as the ratio of the dose (rad) of low-LET radiation to the dose of high-LET radiation
producing the same specific patho-physiological endpoint.
In the Background Information Document for Radionuclides (EPA84), an RBE of 20
was assigned to high-LET radiation when estimating genetic effects. It was noted that studies
comparing cytogenetic endpoints after chronic low-dose-rate gamma radiation exposure, or
incorporation of plutonium-239 in the mouse testis,
have yielded RBEs of 23 to 50 for the type of genetic injury (reciprocal translocations) that
might be transmitted to liveborn offspring (NAS80, UNSC77, 82). Neutron RBE, determined
from cytogenetic studies in mice, also ranged from about 4 to 50 (UNSC82, Gr83a, Ga82).
However, an RBE of 4 for plutonium-239 compared to chronic gamma radiation was reported
for specific locus mutations observed in neonate mice (NAS80).
Most recently, the NAS BEIR IV Committee reviewed the effects of alpha-emitting
radionuclides and estimated the genetic effects (See Table 6-21). The BEIR IV genetic risk
estimates for alpha-emitters were based on the low-LET estimates given in Table IV-2 in the
1980 BEIR HI report, applying an RBE of 15 for chromosome aberrations and 2.5 for all
other effects.
Table 6-21. Genetic risk estimates per 106 live-born for an average population exposure of 1
rad of high-LET radiation in a 30-year generation.
Serious Hereditary Effects
First Generation Equilibrium
(all generations)
Range 28 - 298 165 - 2885
Geometric Mean 91 690
Source: NAS88
These risk estimates, to a first approximation, give an average RBE of about 2.7
relative to the BEIR HI low-LET estimates. This is numerically similar to the dose rate
effectiveness factor for high dose rate. Therefore, for simplicity, it would be possible to use
the same genetic risk coefficients per rad of high dose-rate, low-LET and per rad of high-LET
radiation.
6.5.4 Uncertainty in Estimates of Radiogenic Harm
Chromosomal damage and mutations have been demonstrated in cells in culture, in
plants, in insects, and in mammals (UNSC72,77,82), and in peripheral blood lymphocytes of
persons exposed to radiation (UNSC82, Ev79, Po78). However, they cannot be used for
6-59
-------
predicting genetic risk in progeny of exposed persons. Some believe such changes to be a
direct expression of damage analogous to that induced by radiation in germ cells. At least,
aberrations in peripheral lymphocytes show that radiation-induced chromosome damage can
occur in vivo in humans.
Since human data are so sparse, they can be used only to develop upper bounds of
some classes of genetic risks following radiation exposure. Most numerical genetic risk
estimates are based on extrapolations from animal data.
Data below (Table 6-22), collected by Van Buul (Va80), on induction of reciprocal
translocations in spermatogonia in various species, indicate that animal-based estimates for
this type of genetic effect may be within a factor of 4 of the human value. The 1986
UNSCEAR Committee (UNSC86) did report on radiation induction of reciprocal
translocations in other primates, but the range of responses and conclusions remain the same.
However, if there were no human data on this genetic injury, in the majority of cases,
assuming that animal results and human results would be similar would underestimate the risk
in humans.
Table 6-22. Radiation-induced reciprocal translocations in several species.
Species Translocations
(1C4 per rad)
Rhesus monkey 0.86 + 0.04
Mouse 1.29 ± 0.02 to 2.90 + 0.34
Rabbit 1.48 + 0.13
Guinea pig 0.91+0.10
Marmoset 7.44 + 0.95
Human 3.40 + 0.72
A basic assumption in the doubling-dose method of estimation is that there is a
proportionality between radiation-induced and spontaneous mutation rates. Some of the
uncertainty was removed in the 1982 UNSCEAR report with the observation that in two-test
systems (fruit flies and bacteria), there is a proportionality between spontaneous and induced
mutation rates at a number of individual gene sites. There is still some question as to
whether or not the sites that have been examined are representative of all sites and all gene
loci, with developing evidence that the mouse 7-locus system is more sensitive to radiation
than other members of the mouse genome (Ne88). Current research is focused on
transposable genetic elements and the relevance of "mobile-genetic-element-mediated
spontaneous mutations" to assumptions hi the doubling dose method (UNSC86). The Agency
will review its position as new evidence develops.
6-60
-------
There is some uncertainty as to which hereditary conditions would be doubled by a
doubling dose; future studies on genetic conditions and diseases can apparently, only increase
the total number of such conditions. Every report, from the 1972 BEER, and UNSCEAR
reports to the most recent, has listed an increased number of conditions and diseases that have
a genetic component and hence may be increased by exposure to ionizing radiations.
6.5.4.1 Observations on Human Populations
A study of the birth cohort consisting of children of the Japanese A-bomb survivors
was initiated in mid-1946. In a detailed monograph, Neel and Schull (Ne56) outlined the
background of this first study and made a detailed analysis of the findings to January 1954
when the study terminated. The study was designed to determine: (1) if during the first year
of life, any differences could be observed in children born to exposed parents when compared
to children born to suitable control parents, and (2) if differences existed, how they should be
interpreted (Ne56).
This study addressed a number of endpoints, including sex ratio, malformations,
perinatal data, and anthropometric data; subsequent studies have addressed other endpoints.
Recent reports on this birth cohort of 70,082 persons have reported data on six endpoints.
Frequency of stillbirths, major congenital defects, prenatal death, and frequency of death prior
to age 17 have been examined in the entire cohort. Frequency of cytogenetic aberrations (sex
chromosome aneuploidy) and frequency of biochemical variants (a variant enzyme or protein
electrophoresis pattern) have been measured on large subsets of this cohort.
There were small but statistically insignificant differences between the number of
effects in the children of the proximally and distally exposed with respect to these various
indicators. These differences are in the direction of the hypothesis that mutations were
produced by the parental exposure. Taking these differences then as the point of departure
for an estimate of the human doubling dose, an estimated doubling dose for low-LET
radiation at high doses and dose rates for human genetic effects of about 156 rem (Sc81) or
250 rem (Sa82) was obtained as an unweighted average. When each individual estimate was
weighted by the inverse of its variance, an average of 139 rem was found (Sc84). Because of
the assumptions necessary for these calculations, as well as the inherent statistical errors, the
errors associated with these estimates are rather large. As a result, a reasonable lower bound
to the human estimate overlaps much of the range based on extrapolation from mouse data.
The most recent report evaluated the following possible genetic effects: (1) untoward
pregnancy outcomes, (2) all causes of early mortality, (3) balanced chromosomal exchanges,
(4) sex-chromosome aneuploids, (5) early onset cancer, and (6) protein mutations. On the
basis of the findings of the study, the authors concluded that the gametic doubling dose
measured in humans for acute penetrating radiation exposure from atomic bombs is 150 rem
to 190 rem (Ne88).
6-61
-------
The EPA is using the geometric mean of the BEIR IE range of doubling doses: about
110 rads. EPA believes this estimate of doubling dose probably overstates the risk; however,
it is compatible with both human and mouse data and should not be changed at this time.
EPA estimates of genetic risks will be reviewed and revised, if necessary, when more
complete reports on the Japanese A-bomb survivors are published.
6.5.4.2 Ranges of Estimates Provided by Various Models
Following recommendations of the 1980 BEIR HI and earlier committees, EPA has
continued to use a linear nonthreshold model for estimating genetic effects, although some
data on specific genetic endpoints obtained with acute low-LET exposures are equally well
described by a linear-quadratic function. Moreover, in some of these cases, it has been found
that a reduction in dose rate (or fractionation of dose) produced a reduction in the quadratic
term seen at high doses with little or no effect on the linear component. Such observations
can be qualitatively explained, as previously discussed in reference to somatic effects
(Section 6.2.2), in terms of the dual radiation action theory of Kellerer and Rossi (Ke72), as
well as alternative theories, e.g., one involving enzyme saturation (Go80, Ru58).
Even though genetic risk estimates made by different committees based on the linear
non-threshold model vary, the agreement is reasonably good. Some of the committees made
estimates in terms of a range. These ranges are expressed as a single value by taking the
geometric mean of the range. This method was recommended and first used by UNSCEAR
(UNSC58) for purposes of expressing genetic risk estimates. While the authors of the reports
used different animal models, interpreted them in different ways, and had different estimates
of the level of human genetic conditions in the population, the range of risk coefficients is
about an order of magnitude (see Table 6-20). For the most recent, more comparable
estimates, the range is a factor of 2 to 4 (see ICRP, BEIR HI, and UNSC 1982 in Table 6-
17).
6.5.5 The EPA Genetic Risk Estimates
EPA has used the estimates from BEIR HI (NAS80) based on a "doubling dose" range
with a lower bound of 50 rem and an upper bound of 250 rem. The reasons are as follows:
mutation rates for all gene loci affected by ionizing radiation are not known nor have all loci
associated with "serious" genetic conditions been identified. Because the risk estimated by
the direct method is incomplete, even for the subject animal species, and does not include the
same types of damage estimated by doubling doses, EPA does not consider it further.
Moreover, the BEIR HI genetic risk estimates provide a better estimate of uncertainty than the
UNSCEAR 1982 and ICRP estimates because the BEIR IE Committee assigned a range of
uncertainty for multifactorial diseases (> 5 percent to < 50 percent) that reflects the
uncertainty in the numbers better than the other estimates (5 percent and
10 percent, respectively).
The BEIR HI estimates for low-LET radiations give a considerable range. To express
the range as a single estimate, the geometric mean of the range is used, a method first
recommended by UNSCEAR (UNSC58) for purposes of calculating genetic risk. The factor
of 3 increase in risk for high-dose rate, low-LET radiation, noted earlier, is also used. The
6-62
-------
weighted RBE for high-LET radiation as estimated in BEIR IV is about 3, which is
numerically the same as the dose rate factor noted above.
Genetic risk estimates used by EPA for high- and low-LET radiations are listed in
Table 6-23. As noted above (Section 6.5.1), EPA uses the dose received before age 30 in
assessing genetic risks.
The EPA estimates in Table 6-23 are limited, like all other human genetic risk
estimates, by the lack of confirming evidence of genetic effects in humans. These estimates
depend on a presumed resemblance of radiation effects in animals to those in humans. The
largest human source of data, the Japanese A-bomb survivors, appears at best to provide an
estimate of the doubling dose for calculating the genetic risk in man which is not statistically
significant (Ne88).
Table 6-23. Estimated frequency of genetic disorders in a birth cohort due to exposure of the
parents to 1 rad per generation.
Serious heritable disorders
(Cases per 106 liveborn)
Radiation First generation All generations
Low Dose Rate,
LOW-LET 20 260
High Dose Rate,
LOW-LET 60 780
High-LET 90 690
6-63
-------
In developing the average mutation rate for the two sexes used in the calculation of
the relative mutation risk, the BEIR ffl Committee postulated that the induced mutation rate
in females was about 40 percent of that in males (NAS80). Studies by Dobson, et al., show
that the basis for the assumption was invalid and that human oocytes should have a risk
equivalent to that of human spermatogonia. This would increase the risk estimate obtained
from doubling-dose methods by a factor of 1.43 (Do83, Do84, Do88). Recently Dobson et al.
(Do88) have shown that mouse oocytes are very sensitive to radiation, doses of 4 to 12 rads
killing 50 percent of the immature mouse oocytes. Immature oocytes in women are not so
easily killed. Dobson et al. (Do88) have also shown the existence of a special,
hypersensitive, non-DNA lethality target (apparently the plasma membrane) in immature
mouse oocytes. Irradiation with low energy neutrons, whose recoil protons have track lengths
less than a cell diameter, induces genetic effects in immature mouse oocytes and yields
effects similar to those observed in other cells (Do88). Immature human oocytes do not have
the same hypersensitive target as mouse oocytes and so should be as susceptible as
spermatogonia to genetic effects of radiation.
Unfortunately, BEIR HI and, since it is based on BEIR m, BEIR IV have embedded
sex-sensitivity differences in their risk estimates. In BEIR ffl: (1) autosomal dominants and
X-linked effects are based on a lower estimate where the oocyte has zero sensitivity and an
upper estimate where the oocyte is 44 percent as sensitive as spermatogonia (p. 118); (2)
irregularly inherited effects are based on an estimate where the oocyte is 44 percent as
sensitive as spermatogonia (pp. 114 and 110); and (3) chromosomal aberrations estimates are
based on oocytes and spermatogonia of equal sensitivity (p. 123, NAS80).
Since the sex-specific differences are in both BEIR IH and BEIR IV, no attempt is
made at this time to correct them. After BEIR V is published, EPA's genetic risk estimates
will be reviewed and may then be revised.
The combined uncertainties in doubling-dose estimates and the magnitude of genetic
contributions to various disorders probably introduce an overall uncertainty of about an order
of magnitude in the risk estimates. Moreover, the BEIR Committee, in deriving its estimate,
has assumed that almost all of the risk was due to irregularly inherited mutations which
would be eliminated slowly. They may include mild mutations which are but slightly
detrimental in their heterozygous state. However, they may be sustained by advances in
medical science, thus persisting and accumulating for generations. To what extent this occurs
will depend on medical practices in the future.
6.5.6 Effects of Multigeneration Exposures
As noted earlier, while the somatic effects (cancer) occur in persons exposed to
ionizing radiation, the genetic effects occur in progeny, perhaps generations later. The
number of effects appearing in the first generation is based on direct estimates of the
mutations induced by irradiation and should not change appreciably regardless of the
background or "spontaneous" mutation rate in the exposed population. The estimate for total
genetic effects, or the equilibrium estimate, is based on the doubling-dose concept. For these
estimates, the background mutation rate is important: it is the background rate mat is being
"doubled."
6-64
-------
If there is long-lived environmental contamination, such that 30 generations or more
are exposed (>1000 years), the background mutation rate will change and come into
equilibrium with the new level of radiation background. There will be an accumulation of
new radiation-induced mutations until the background mutation rate has reached equilibrium
with this continued insult.
While predicting 1,000 years in the future is chancy at best, if it is assumed that there
are no medical advances, and no changes in man or his environment, then an estimate can be
made. In Table 6-23, it is estimated that exposure to 1 rad per generation of low-dose-rate,
low-LET radiation will induce 260 cases of serious heritable disorders per 106 live births in
all generations. This is for a background mutation rate leading to 29,120 cases of serious
heritable disorders per 106 live births. The "all generations" estimate in Table 6-23 is equal
to the BEIR El "equilibrium" estimate in Table 6-20. The "all generations" estimate is used
for exposures to a single generation; the same number is employed as the "equilibrium"
estimate for multigeneration exposures (see NAS80, p. 126, note 16). Thus, the risk estimate
can be re-expressed as an estimate of the effects expected for a given change in the level of
background radiation (Table 6-24). Since these calculations are based both on the background
level mutations and the doubling dose, changes in either must be reflected in new
calculations.
Table 6-24. Increase in background or level of genetic effects after 30 generations or more.
Increase in background Increase in serious heritable
radiation (mrad/y) disorders per 106 live births
Low-dose rate, High-LET
low-LET radiation radiation
0.1
1.0
10.0
0.8
8.0
80
2.1
21.2
212
6.5.7 Uncertainties in Risk Estimates for Radiogenic Genetic Effects
As noted throughout the preceding sections, there are sources of uncertainty in the
genetic risk estimates. The overall uncertainty can be addressed only in a semi-quantitative
manner. The identified sources of uncertainty are listed in Table 6-25. Uncertainties listed in
this table are likely to be independent of each other and therefore unlikely to be correlated in
sign. Although the root mean square sum of the numerical uncertainties suggests the true risk
could be a factor of 4 higher or lower [(X/-T-) by a factor of 4], it is unlikely, in light of the
Japanese A-bomb survivor data, that the upper bound is correct.
6-65
-------
Table 6-25. Causes of uncertainty in the genetic risk estimates.
Source of Uncertainty
Degree of Uncertainty
in Risk Estimates
Selection of species to use in
developing a direct estimate
Selection of species and loci to
use in developing a doubling dose
+indeterminate (a)
Use of - division by a factor of 3 -
to convert acute, high dose, low-LET
estimates to chronic, low-LET estimates
Sensitivity of oogonia compared to
spermatogonia as described in BEIR-IQ
Background rate selected for use
with a doubling dose
Selection of RBE for high-LET
radiation compared to an RBE of 20
Underestimate of the doubling dose
required in man
x/-^ factor of 4
-100% to estimate
x/-=- factor of 3
-44% to 56%
x/4-,indeterminate
x/f a factor of 5
x/4- a factor of 2(b)
(b)
The risk estimate cannot go below zero, -100%; but it may not be possible to
determine the upper bound, indeterminate.
If the most recent analysis of the Japanese A-bomb survivors is correct, the lower
bound for an estimate of the doubling dose in man is at least 2 times greater than the
doubling dose estimate derived from the mouse.
6-66
-------
6.5.8 Teratogenic Effects
Although human teratogenesis (congenital abnormalities or defects) associated with x--
ray exposure has a long history, the early literature deals mostly with case reports. (St21,
Mu29, Go29). However, the irradiation exposures were high.
In 1930, Murphy exposed rats to x-rays at doses of 200 R to 1,600 R. Of 120
exposed females, 34 had litters, and five of the litters had animals with developmental defects
(Mu30). He felt that this study confirmed his clinical observations and earlier reports of
animal studies. Although there were additional studies of radiation-induced mammalian
teratogenesis before 1950, the majority of the studies were done after that time (see Ru53 for
a review), perhaps reflecting concerns about radiation hazards caused by the explosion of
nuclear weapons in 1945 (Ja70).
Much of the work done after World War n used mice (Ru50, Ru54, Ru56) or rats
(Wi54, Hi54). Early studies, at relatively high radiation exposures, 25 R and above,
established some dose-response relationships. More important, they established the timetable
of sensitivity of the developing rodent embryo and fetus to radiation effects (Ru54, Hi53,
Se69, Hi66).
Rugh, in his review of radiation teratogenesis (Ru70), listed the reported mammalian
anomalies and the exposures causing them. The lowest reported exposure was 12.5 R for
structural defects and 1 R for functional defects. He also suggested human exposure between
ovulation and about 7 weeks gestational age could lead to structural defects, and exposures
from about 6 weeks gestational age until birth could lead to functional defects. In a later
review (Ru71), Rugh suggested structural defects in the skeleton might be induced as late as
the 10th week of gestation and functional defects as early as the 4th week. It should be noted
that the gestation period in mice is much shorter than that in humans and that weeks of
gestation referred to above are in terms of equivalent stages of mouse-human development.
However, estimates of equivalent gestational age are not very accurate.
Rugh (Ru71) suggested there may be no threshold for radiation-induced congenital
effects in the early human fetus. In the case of human microcephaly (small head size) and
mental retardation, at least, some data support this theory (Ot83, Ot84).
However, for most teratogenic effects, the dose response at low doses is not known.
In 1978, Michel and Fritz-Niggli (ML78) reported induction of a significant increase in growth
retardation, eye and nervous system abnormalities, and post-implantation losses in mice
exposed to 1 R. The increase was still greater if there was concurrent exposure to
radiosensitizing chemicals such as iodoacetimide or tetracycline (Mi78).
In other reports of animal studies, it appeared as if teratologic effects, other than
perhaps growth retardation, had a threshold for induction of effects (Ru54, Ru53, Wi54).
However, Ohzu (Oh65) showed that doses as low as 5 R to preimplantation mouse embryos
caused increased resorption of implanted embryos and structural abnormalities in survivors.
Then in 1970, Jacobsen (Ja70) reported a study in which mice were exposed to 5, 20, or 100
R on the eighth day of pregnancy. He concluded that the dose response function for
6-67
-------
induction of skeletal effects was linear, or nearly linear, with no observable threshold. This
appears consistent with a report by Russell (Ru57), which suggested a threshold for some
effects whereas others appeared to be linearly proportional to dose.
One of the problems with the teratologic studies in animals is the difficulty of
determining how dose response data should be interpreted. Russell (Ru54) pointed out some
aspects of the problem: (1) although radiation is absorbed throughout the embryo, it causes
selective damage that is consistently dependent on the stage of embryonic development at the
time of irradiation, and (2) the damaged parts respond, in a consistent manner, within a
narrow time range. However, while low-dose irradiation at a certain stage of development
produces changes only in those tissues and systems that are most sensitive at that time, higher
doses may induce additional abnormalities in components that are most sensitive at other
stages of development, and may further modify expression of the changes induced in parts of
the embryo at maximum sensitivity during the time of irradiation. In the first case, damage
may be to primordial cells themselves, while in the second, the damage may lead indirectly to
the same or different endpoints.
The human embryo/fetus starts as a single, fertilized egg and divides and
differentiates to produce the normal infant at term. (The embryonic period, when organs
develop, is the period from conception through 7 weeks gestational age. The fetal period, a
time of in utero growth, is the period from 8 weeks gestational age to birth.) The different
organ and tissue prirnordia develop independently and at different rates. However, they are in
contact through chemical induction or evocation (Ar54). These chemical messages between
cells are important in bringing about orderly development and the correct timing and fitting
together of parts of organs or organisms. While radiation can disrupt this pattern,
interpretation of the response may be difficult. Since the cells in the embryo/fetus
differentiate, divide, and proliferate at different times during gestation and at different rates,
gestational times when cells of specific organs or tissues reach maximum sensitivity to
radiation are different Each embryo/fetus has a different timetable. In fact, each half
(left/right) of an embryo/fetus may have a slightly different timetable.
In addition, there is a continuum of variation from the hypothetical normal to the
extreme deviant which is obviously recognizable. There is no logical place to draw a line of
separation between normal and abnormal. The distinction between minor variations of normal
and frank malformation, therefore, is an arbitrary one, and each investigator must establish his
or her own criteria and apply them to spontaneous and induced abnormalities alike (HWC73).
The limitations of the human data available make the use of animals in both
descriptive and experimental studies inevitable. However, this gives rise to speculation about
the possible relevance of such studies to man. There are species differences in development
attributable partly to the differing complexity of the adult organs, but especially to differences
in growth rates and timing of birth in relation to the developmental events. For example, the
histological structure of the brain is, in general, surprisingly similar, both in composition and
in function, from one mammalian species to another, and the sequence of events is also
similar (Do73). However, the processes of brain development that occur from conception to
about the second year of life in man are qualitatively similar to those seen in the rat during
the first six weeks after conception (Do79, Do81).
6-68
-------
For example, a major landmark, the transition from the principal phase of
multiplication of the neuronal precursors to that of glial multiplication, occurs shortly before
mid-gestation in man, but at about the time of birth in the rat (Do73). In this respect, then,
the rat is much less neurologically mature at birth than the newborn human infant Many
other species are more mature at birth; the spectrum ranges from the late-maturing mouse and
rat to the early-maturing guinea pig, with non-human primates much closer to the guinea pig
than to man (Do79, Do81). As a consequence, it is unreasonable to compare a newborn rat's
brain, which has not begun to myelinate, with that of a newborn human which has, or with
that of a newborn guinea pig in which myelination has been completed (Do79, Do81).
Nevertheless, in the study of teratogenic effects of prenatal exposure to ionizing
radiation, in which the timing of the exposure in relation to the program of developmental
events dictates the consequences of that insult, it is necessary only to apply the experimental
exposure at the appropriate stage (rather than at a similar age) of embryonic or fetal
development in any species to produce similar results in all (Do79, Do81). The duration of
exposure must, however, match the different time scales in the different species. Unless these
elementary rules of cross-species adjustments are followed, extrapolation of even qualitative
estimates of effects will be of dubious relevance and worth.
Because of the problems in interpretation listed above, a pragmatic approach to
evaluation of studies is useful. The dose response should be given as the simplest function
that fits the data (often linear or linear with a threshold). No attempt should be made to
develop complex dose response models unless the evidence is unequivocal.
6.5.8.1 Teratologic Effects: Mental Retardation in Humans
The first report of congenital abnormalities in children exposed in utcro to radiation
from atomic bombs was that of Plummer (P152). Twelve children with microcephaly, of
which ten also had mental retardation, had been identified in Hiroshima in a small set of the
in utero exposed survivors. They were found as part of a program started in 1950 to study
children exposed in the first trimester of gestation. However, not all of the in utero exposed
survivors were examined. In 1955, the program was expanded to include all survivors
exposed in utero.
Studies initiated during the program have shown radiation-related (1) growth
retardation; (2) increased microcephaly; (3) increased mortality, especially infant mortality;
(4) temporary suppression of antibody production against influenza; and (5) increased
frequency of chromosomal aberrations in peripheral lymphocytes (Ka73).
Although there have been a number of studies of Japanese A-bomb survivors,
including one showing a dose- and gestational age-related increase in postnatal mortality
(Ka73), only the incidences of microcephaly and mental retardation have been investigated to
any great extent. In the most recent report, Otake and Schull (Ot83, 84) showed that mental
retardation was particularly associated with exposure between 8 and 15 weeks of gestation (10
to 17 weeks of gestation if counted from the last menstrual period). They further found the
data suggested little, if any, non-linearity and were consistent with a linear dose-response
relationship for induction of mental retardation that yielded a probability of occurrence of
6-69
-------
severe mental retardation of 4.16.+0.4 cases per 1,000 live births per rad of exposure (Ot84).
A child was classified as severely mentally retarded if he or she was "unable to perform
simple calculations, to make simple conversation, to care for himself or herself, or if he or
she was completely unmanageable or had been institutionalized" (Ot83, 84). There was,
however, no evidence of an effect in those exposed at 0 to 7 weeks of gestation (Ot83).
Exposure at 16 weeks or more of gestation was about a factor of 4 less effective, with only a
weak relationship between exposure and risk, and with few cases below 50 rads exposure
(Ot84).
Mental retardation can be classified as mild (IQ 50-70), moderate (IQ 35-49), severe
(IQ 20-34), and profound (IQ < 20) (WHO75). However, some investigators use only mild
mental retardation (IQ 50-70) and severe mental retardation (IQ < 50) as classes (Gu77b,
HaSla, St84). Mental retardation is not usually diagnosed at birth but at some later time,
often at school age. Since the mental retardation may have been caused before or during
gestation, at the time of birth, or at some time after birth, that fraction caused before or
during gestation must be estimated. In like manner, since mental retardation caused before
birth may be due to genetic conditions, infections, physiologic conditions, etc., the fraction
related to unknown causes during gestation must be estimated. This is the fraction that might
possibly be related to radiation exposure.
Estimates of the risk of mental retardation for a rad of embryo/fetus exposure in the
U.S. population can be derived using the absolute risk calculated by Otake and Schull for the
Japanese survivors (Ot84). Otake and Schull (Ot84) gave an estimate for one case entitled,
"The Relationship of Mental Retardation to Absorbed Fetal Exposure in the 'Sensitive' Period
When All 'Controls' Are Combined." This estimate of frequency of mental retardation, 0.416
per 100 rads, could be directly applicable to a U.S. population. In this case, the risk estimate
would be about four cases of severe mental retardation per 1,000 live births per rad of
exposure during the 8th and 15th week of gestation.
The ICRP published an excellent review of biology and possible mechanisms of
occurrence of radiation-induced brain damage, in utero (ICRP86). ICRP estimates: (1) for
exposures from the 8th through the 15th week after conception, the risk of severe mental
retardation is 4 x 10'3 per rad, with a confidence interval of 2.5 x 10"3 to 5.5 x 10"3, and (2)
for exposures from the 16th through the 25th week after conception, the risk of severe mental
retardation is 1 x 10"3 per rad. However, a threshold below 50 rad cannot be excluded
(ICRP86).
The 1986 UNSCEAR Committee also reviewed biology and possible mechanisms
(UNSC86). Although increased external granular layer pyknosis had been found in rats after
exposures of 3 rad and degraded behavioral performance had been reported in rats after four 1
rad doses, the UNSCEAR Committee concluded that"... no effects having clearly
pathological connotations have been reported for doses in the brain structures lower than 0.1
Gy (10 rad) low-LET radiation." (UNSC86).
If the ICRP estimate is applicable, the low-LET background radiation (about 15
mrads) delivered during the 8- to 15-week gestational age-sensitive period could induce a risk
of 6 x 10"5 cases of severe mental retardation per live birth. This can be compared to an
6-70
-------
estimate of a spontaneous occurrence of 0.6 x 10"3 to 3.1 x 10~3 cases of idiopathic severe
mental retardation per live birth (EPA84).
6.5.8.2 Teratologic Effects: Microcephaly in Humans
Plummer (P152) reported microcephaly associated with mental retardation in Japanese
A-bomb survivors exposed in utero. Wood (Wo65, 66) reported both were increased. The
diagnosis of reduced head circumference was based on "normal distribution" statistical theory
(Wo66); i.e., in a population, the probability of having a given head circumference is
expected to be normally distributed around the mean head circumference for that population.
For example, in a population of live-born children, 2.275 percent will have a head
circumference 2 standard deviations or more smaller than the mean, 0.621 percent will have a
head circumference 2.5 standard deviations or more smaller than the mean, and 0.135 percent
will have a head circumference 3 standard deviations or more smaller than the mean
(statistical estimates based on a normal distribution).
For most of the studies of the Japanese A-bomb survivors exposed in utero, if the
head circumference was two or more standard deviations smaller than the mean for the
appropriate controls in the unexposed population, the case was classified as having reduced
head circumference even if the data had not been adjusted for differences in stature (Ta67,
Mi72, Wo65). While a definitive relationship between reduced head circumference and
mental retardation has not been established, there is evidence that they are related.
Studies of the Japanese survivors show a relationship between reduced head size and
mental retardation, but all these studies are based on subsets of the total in utero population.
The fraction of mentally retarded with reduced head circumference has been reported as 50
percent (RERF78) to 70 percent (Wo66), while the fraction of those selected for reduced head
circumference who had mental retardation has been reported as 11 percent (Wo66) to 22
percent (Mi72). Thus, while the relationship appears to exist, it has not been quantified.
The majority of the cases of reduced head size are observed in those exposed in the
first trimester of gestation, particularly the 6th or 7th to 15th weeks of gestation (Mi59,
Wo66, Mi72, Wo65, Ta67). Most recently, it has been shown that reduction in head
circumference was a linear function of dose (Is84). However, the authors noted that the
analysis was based on T65 dosimetry, and the data should be reanalyzed after completion of
the dosimetry reassessment currently in progress.
These findings of reduction in head circumference, with a window of effect in the
same time period of gestation as mental retardation, help support the observations on mental
retardation. Although the exact dose response functions are still uncertain, data on both types
of effects have so far been consistent with a linear, no-threshold dose response during the
critical period.
6.5.8.3 Other Teratologic Effects
6-71
-------
Effects other than mental retardation and microcephaly have been noted in the Japan's
A-bomb survivors. Schull et al (Sc99) reported that in individuals exposed prenatally
between weeks 8 and 25 of gestation there is a progressive shift downward in IQ score with
increasing exposure and that the most sensitive group is between 8 and 15 weeks gestational
age at time of exposure. Much the same pattern was reported for average school
performance, especially in the earliest years of schooling (Ot88). Finally, a linear-
nonthreshold relationship between exposure and incidence of unprovoked seizures in later life
has been demonstrated to be consistent with the data for individuals exposed between 8 and
15 weeks gestational age (Du88).
Japanese A-bomb survivors exposed in utero also showed a number of structural
abnormalities and, particularly in those who were microcephalic, retarded growth (Wo65). No
estimate has been made of the radiation-related incidence or dose-response relationships for
these abnormalities. However, UNSCEAR (UNSC77) made a very tentative estimate based
on animal studies that the increased incidence of structural abnormalities in animals may be
0.005 cases per R per live bom, but stated that projection to humans was unwarranted. In
1986, UNSCEAR assumed the risk of an absolute increase of malformed fetuses of the order
of 5E-3 per rad seen in animals might apply to the human species as well, for exposure over
the period from 2 to 8 weeks post-conception (UNSC86). In any event, the available human
data cannot show whether the risk estimates derived from high-dose animal data overestimate
the risk in humans or if a threshold can be excluded.
It should be noted that all of the above estimates are based on high-dose-rate, low-
LET exposure. In 1977, UNSCEAR also investigated the dose rate question and stated:
"In conclusion, the majority of the data available for most species indicate a
decrease of the cellular and malformature effects by lowering the dose rate or
by fractionating the dose. However, deviations from this trend have been
well documented in a few instances and are not inconsistent with the
knowledge about mechanisms of the teratogenic effects. It is therefore
impossible to assume that dose rate and fractionation factors have the same
influence on all teratological effects." (UNSC77).
6.5.9 Nonstochastic Effects
Nonstochastic effects, those effects that increase in severity with increasing dose and
have a threshold, have been reviewed in the 1982 UNSCEAR report (UNSC82).
Nonstochastic effects following in utero exposure were reviewed in the 1986 UNSCEAR
report (UNSC86). In general, acute doses of 10 rads low-LET radiation and higher are
required to induce these effects in animals. It is possible that some of the observed effects of
in utero exposure are nonstochastic: e.g., the risk of embryonic loss, estimated to be 10"2 per
R (UNSC77) or per rad (UNSC86) following radiation exposure soon after fertilization.
However, there are no data to address the question of similar effects in humans. Usually,
nonstochastic effects are not expected at environmental levels of radiation exposure.
In 1986, the United Nations Scientific Committee on the Effects of Atomic Radiation
also reviewed the question of mental retardation as a part of the overall review of the
6-72
-------
biological effects of prenatal radiation exposure (UNSC86). UNSCEAR, like the ICRP,
concluded there was a risk of severe mental retardation of 4 x 10"3 per rad over the period of
8 to 15 weeks after conception and of 1 x 10"3 per rad over the period 16-25 weeks after
conception (UNSC86). UNSCEAR also estimated (1) a pre-implantation loss of 1 x 10~2 per
rad during the first two weeks after conception, (2) a malformation risk of 5 x 10~3 per rad
during weeks 2 to 8 after conception, and (3) a risk of leukemia and solid tumors expressed
during the first 10 years of life of 2 x 10A per rad (UNSC86).
The British National Radiation Protection Board (NRPB) reviewed available
information including the 1988 UNSCEAR report to develop new health effects models
(St88). The NRPB estimated a mental retardation risk of 4.5 X 10"3 cases per rad of exposure
during weeks 8 to 15 of gestation. The NRPB also estimated a cancer risk of 2.5 X 10"4
cases of leukemia and 3.5 X 10^ cases of solid tumors per rad of in utero exposure (St88).
EPA has adopted similar risk coefficients for estimating prenatal carcinogenic,
teratologic, and nonstochastic effects in man (see Table 6-26).
Table 6-26. Possible effects of in utero radiation exposure.
Type of Risk
Conceptus
Risk per Rad
Risk per Event in a
100 mrad per Year
Background
Fatal Cancer
Mental Retardation
(exposure at 8 - 15 weeks)
Mental Retardation
(exposure at 16 - 25 weeks)
Malformation
(exposure at 2 - 8 weeks)
Pre-implantation
Loss (exposure at
0-2 weeks)
6.0 x 10"4
4 x 1Q-3
1 x 10'3
5 x lO'3
1 x lO'2
4.5 x 10'5
6.0 x 10'5
1.5 x 10-5
5.8 x 10'5
3.8 x 10'5
6-73
-------
6.6 SUMMARY OF EPA'S RADIATION RISK FACTORS - A PERSPECTIVE
Table 6-27 summarizes EPA's estimate of risk from lifetime whole-body exposures to
high- and low-LET radiation and to radon decay products. The nominal risk factors reflect
EPA's best judgment as to the relationship between dose and risk based on review of all
relevant information available to the Agency. Likewise the cited ranges reflect EPA's current
best judgment as to the uncertainties in these risk factors.
To provide a perspective on the risk of fatal radiogenic cancers and the hereditary
damage due to radiation, EPA has calculated the risk from background radiation to the U.S.
population using the risk factors summarized in Table 6-23. The risk from background
radiation provides a useful perspective for the risks caused by emissions of radionuclides.
Unlike cigarette smoking, auto accidents, and other measures of common risks, the risks
resulting from background radiation are neither voluntary nor the result of self-induced
damage. The risk caused by background radiation is largely unavoidable; therefore, it is a
good benchmark for judging the estimated risks from radionuclide emissions. Moreover, to
the degree that the estimated risk of radionuclides is biased, the same bias is present in the
risk estimates for background radiation.
The absorbed dose rate from low-LET background radiation has three major
components: cosmic radiation, which averages about 28 mrad/yr in the United States;
terrestrial sources, such as radium in soil, which contribute an average of 28 mrad/yr
(NCRP87); and the low-LET dose resulting from internal emitters. The last differs among
organs, to some extent, but for soft tissues it is about 24 mrad/yr (NCRP87). Other minor
radiation sources such as fallout from nuclear weapons tests, cosmogenic radionuck'des,
naturally occurring radioactive materials in buildings, airline travel, and consumer products,
contribute about another 7 mrad for a total low-LET whole-body dose of about 87 mrad/yr.
The lung and bone receive somewhat larger doses, not included in the 87 mrad/yr estimate,
due to high-LET radiations (see below). Although extremes do occur, the distribution of this
background annual dose to the U.S. population is relatively narrow. A population-weighted
analysis indicates that 80 percent of the U.S. population would receive annual doses that are
between 75 mrad/yr and 115 mrad/yr (EPA81).
As outlined in Section 6.2, the BEER, in linear, relative risk models yield, for lifetime
exposure to low-LET radiation, an average lifetime risk of fatal radiogenic cancer of 3.9X10"4
per rad. Note that this average is for a group having the age- and sex-specific mortality rates
of the 1970 U.S. population. This risk estimate can be used to calculate the average lifetime
risk due to low-LET background radiation as follows. The average duration of exposure in
this group is 70.7 yr, and at 90 mrad/yr, the average lifetime dose is 6.4 rads. The risk of
fatal cancer per person hi this group is:
(3.9X10-4 rad'1) (8.7xlO'3 rad/y) (70.7 y) = 2.4 x 10'3 (6-11)
or about 0.24 percent of all deaths. The vital statistics used in EPA's radiation risk analyses
indicate that the probability of dying from cancer in the United States from all causes is
6-74
-------
Table 6-27. Summary of EPA's radiation risk factors.
Risk
Low LET CIO'6 rad'1)
Teratological:
Severe mental
retardation
Genetic:
Severe hereditary
defects, all
generations
Somatic:3
Fatal cancers
All cancers
Fatal cancers
High LET (It)'6 rad'1)
Genetic:
Severe hereditary
defects, all
generations
Somatic:
Fatal cancers
All cancers
Significant
Exposure Period
Weeks 8 to 10
of gestation
30 year
reproductive
generation
Lifetime
Lifetime
In utero
30 year
reproductive
generation
Lifetime
Lifetime
Radon decay products (10'6 WLM'1)
Fatal lung cancer Lifetime
Risk Factor
Nominal
Range
4,000 2,500- 5,500
260 60- 1,100
390 120- 1,200
620 190- 1,900
600 180- 1,800
690 160- 2,900
3,100
5,000
360
960- 9,600
1,500 - 15,000
140 - 720
The range assumes a linear, non-threshold dose response. However, it is plausible that a
threshold may exist for this effect.
6-75
-------
about 0.16, i.e., 16 percent. Thus, the 0.24 percent result for the BEIR El linear dose
response model indicates that about 1.5 percent of all U.S. cancer is due to low-LET
background radiation. The BEIR in linear-quadratic model indicates that about 0.1 percent of
all deaths are due to low-LET background radiation or about 0.6 percent of all cancer deaths.
Table 6-11 indicates a risk of S^xlO^rad"1 for alpha emitters in lung tissue. UNSCEAR
estimated that in "normal" areas the annual absorbed dose in the lungs from alpha emitters
other than radon decay products would be about 0.51 mrad (UNSC77). The individual
lifetime cancer risk from this exposure is:
(5.6 x 10-4 rad'1) (5.1 x 104 rad/y) (70.7y) = 2.0 x 10'5, (6-12)
which is about 1/100 of the risk due to low-LET background radiation calculated by means of
the BEIR HI linear model.
The 1982 UNSCEAR report indicates that the average annual absorbed dose to the
endosteal surfaces of bone due to naturally occurring, high-LET alpha radiation is about 6
mrad/yr, based on a quality factor of 20 and an absorbed dose equivalent of 120 mrem/yr
(UNSC82). Table 6-11 indicates that the individual lifetime risk of fatal bone cancer due to
this portion of the naturally occurring radiation background is:
(2.0 x 10'5 rad'1) (6 x 10'3 rad/y) (70.7/y) = 8.5 x 10'6. (6-13)
The exposure due to naturally occurring background radon-222 progeny in the indoor
environment is not well known. The 1982 UNSCEAR report lists for the United States an
indoor concentration of about 0.004 working levels (15 Bq/m3) (UNSC82). This estimate is
not based on a national survey and is known to be exceeded by as much as a factor of 10 or
more in some houses. However, as pointed out in UNSC82, the national collective exposure
may not be too dependent on exceptions to the mean concentration. The UNSCEAR estimate
for the United States now appears low (Ne86); the average residential exposure is probably
0.2-0.3 WLM/yr (in standard exposure units).
Assuming 0.25 WLM/yr is a reasonable estimate for indoor exposure to radon-222
progeny in this country, the mean lifetime exposure, indoors, is about 18 WLM. Based on
the geometric mean lifetime risk coefficient from Section 6.4.5, 360 cases/106 WLM, a
lifetime risk of 0.64 percent is estimated. For comparison, roughly 5 percent of all deaths in
1980 were due to lung cancer. Based on these assumptions, therefore, about one of eight
lung cancer deaths may be attributable to background radon exposure. This would correspond
to about 4 percent of all cancer deaths. This is 2.5 times the 1.61 percent of all cancer
fatalities estimated above for low-LET background radiation. The reader is cautioned,
however, that this risk estimate applies only to the United States population taken as a whole,
i.e., men and women, smokers and nonsmokers. Since the vast majority of the 1980 lung
cancer mortality occurred in male smokers, this risk estimate cannot be applied
indiscriminately to women or nonsmokers (see Section 6.4).
6-76
-------
The spontaneous incidence of serious congenital and genetic abnormalities has been
estimated to be about 105,000 per 106 live births, about 10.5 percent of live births (NAS80,
UNSC82). The low-LET background radiation dose of about 87 mrad/year in soft tissue
results in a genetically significant dose of 2.6 rads during the 30-year reproductive generation.
Since this dose would have occurred in a large number of generations, the genetic effects of
the radiation exposure are thought to be at an equilibrium level of expression. Since genetic
risk estimates vary by a factor of 20 or more, EPA uses a log mean of this range to obtain an
average value for estimating genetic risk. Based on this average value, the background
radiation causes about 690 genetic effects per 106 live births (see Section 6.5). This result
indicates that about 0.6 percent of the current spontaneous incidence of serious congenital and
genetic abnormalities may be due to the low-LET background radiation.
6-77
-------
Chapter 6 References
Ar81 Archer, V.E., Health Concerns in Uranium Mining and Milling, J. Occup.
Med., 23, 502-505, 1981.
Ar54 Arey, L.B., Developmental Anatomy, 6th ed., W.B. Saunders, Philadelphia,
1954.
Au67 Auxier, J.A., Cheka, J.S., Haywood, F.F., Jones, T.D. and J.H. Thorngate, Free-
Field Radiation Dose Distributions from the Hiroshima and Nagasaki
Bombings, Health Phys. I2(3):425-429, 1967.
Au77 Auxier, J.A., Ichiban - Radiation Dosimetry for the Survivors of the Bombings
of Hiroshima and Nagasaki, TDD 27080, Technical Information Center, Energy
Research and Development Administration, National Technical Information
Service, Springfield, Virginia, 1977.
Ba73 Baum, J.W., Population Heterogeneity Hypothesis on Radiation Induced
Cancer, Health Phys., 25(1):97-104, 1973.
Bo82 Bond, V.P. and J.W. Thiessen, Reevaluations of Dosimetric Factors, Hiroshima
and Nagasaki, DOE Symposium Series 55, CONF-810928, Technical
Information Center, U.S. Department of Energy, Washington, D.C., 1982.
Bu81 Bunger, B., Cook, J.R. and M.K. Barrick, Life Table Methodology for
Evaluating Radiation Risk: An Application Based on Occupational Exposure,
Health Phys., 40 (4):439-455.
Ch81 Chameaud, J., Perraud, R., Chretien, J., Masse, R. and J. Lafuma, Contribution
of Animal Experimentation to the Interpretation of Human Epidemiological
Data, in: Proc. Int. Conf. on Hazards in Mining: Control, Measurement, and
Medical Aspects, October 4-9, 1981, Golden, Colorado, pp. 228-235, edited by
Manual Gomez, Society of Mining Engineers, New York, 1981.
Ch83 Charles, M.E., Lindop, P.J. and AJ. Mill, A Pragmatic Evaluation of the
Repercussions for Radiological Protection of the Recent Revisions in Japanese
A-bomb Dosimetry, IAEA SM-266/52, Proceedings, International Symposium
on the Biological Effects of Low-Level Radiation with Special Regard to
Stochastic and Non-stochastic Effects, Venice, IAEA, Vienna, April 11-15,
1983.
Ch85 Chameaud J., Masse R., Morin M., and Lafuma J., Lung Cancer Induction by
Radon Daughters in Rats, in: Occupational Radiation Safety in Mining, Vol. 1,
H. Stokes, editor, Canadian Nuclear Assoc., Toronto, Canada, pp. 350-353,
1985.
6-78
-------
Co78 Cook, J.R., Bunger, B.M. and M.K. Barrick, A Computer Code for Cohort
Analysis of Increased Risks of Death (CAIRO), ORP Technical Report 520/4-
78-012, U.S. Environmental Protection Agency, Washington, D.C., 1978.
Cu79 Cuddihy, R.G., McClellan, R.O., and Griffith, W.C. Variability in Target
Deposition Among Individuals Exposed to Toxic Substances, Toxicol. Appl.
Pharmacol. 49: 179-187, 1979.
Da75 Davies, R.B. and B. Hulton, The Effects of Errors in the Independent Variables
in a Linear Regression, Biometrika, 62:383-391, 1975.
Da86 Darby, S.C., Epidemiological Evaluation of Radiation Risk Using Populations
Exposed at High Doses, Health Phys. 51. (3): 269-281, 1986.
Do73 Dobbing, J. and J. Sands, Quantitative Growth and Development of the Human
Brain. Arch. Dis. Child., 48:757-767 (1973).
Do79 Dobbing, J. and J. Sands, Comparative Aspects of the Brain Growth Spurt,
Early Human Dev., 3:109-126 (1979).
Do81 Dobbing, J., The later development of the brain and its vulnerability, pp. 744-
758, in: Scientific Foundations of Pediatrics, 2nd edition, J.A. Davis and J.
Dobbing, editors, William Heinemann Medical Books Ltd., London, 1981.
Do83 Dobson, R.L. and J.S. Felton, Female Germ Cell Loss from Radiation and
Chemical Exposures, Amer. J. Ind. Med., 4: 175-190, 1983.
Do84 Dobson, R.L. and T. Straume, Mutagenesis in Primordial Mouse Oocytes Could
Be Masked by Cell Killing: Monte Carlo Analysis, Environ. Mutagen. £, 393,
(1984) [Abstract].
Do88 Dobson, L., Straume T. and Kwan C, The Problem of Genetically Meaningful
Dose and Hypervulnerable Lethality Targets in Certain Oocytes. Thirty-Sixth
Annual Meeting of the Radiation Research Society, Philadelphia, 1988, Book
of Abstracts, p 137 Abstract EK-3.
Du88 Dunn, K., H. Yoshimaru, M. Otake, J.F. Annegers, and WJ. Schull. Prenatal
Exposure to Ionizing Radiation and Subsequent Development of Seizures.
Technical Report RERF TR 13-88, Radiation Effects Research Foundation,
Hiroshima, 1988.
Ed83 Edling C., Kling H., and Axelson O., Radon in Homes - A Possible Cause of
Lung Cancer, in: Lung Cancer and Radon Daughter Exposure in Mines and
Dwellings. Linkoping University Medical Dissertations No. 157, by Christer
Edling, Department of Occupational Medicine, Linkoping University,
Linkoping, Sweden, pp. 123-149, 1983.
6-79
-------
Ed84
E179
EPA78
EPA79
EPA81
EPA82
EPA83a
EPA83b
EPA84
EPA85
Edling C., Wingren G., and Axelson, O., Radon Daughter Exposure in
Dwellings and Lung Cancer, in: Indoor Air, Volume 2: Radon, Passive
Smoking, Particulates and Housing Epidemiology, B. Berglund, T. Lindvall and
J. Sundell, editors, Swedish Council for Building Research, Stockholm,
Sweden, pp. 29-34, 1984.
Ellett, W. H. and Nelson, N. S., Environmental Hazards From Radon Daughter
Radiation, in: Conference/Workshop on Lung Cancer Epidemiology and
Industrial Applications of Sputum Cytology, Colorado School of Mines Press,
Golden, Colorado, pp. 114-148, 1979.
U.S. Environmental Protection Agency, Response to Comments: Guidance on
Dose Limits for Persons Exposed to Transuranium Elements hi the General
Environment, EPA Report 520/4-78-010, Office of Radiation Programs,
Washington, D.C., 1978.
U.S. Environmental Protection Agency, Indoor Radiation Exposure Due to
Radium-226 in Florida Phosphate Lands, EPA Report 520/4-78-013, Office of
Radiation Programs, Washington, D.C., revised printing, July 1979.
U.S. Environmental Protection Agency, Population Exposure to External
Natural Radiation Background in the United States, Technical Note
ORP/SEPD-80-12, Office of Radiation Programs, Washington, D.C., 1981.
U.S. Environmental Protection Agency, Final Environmental Impact Statement
for Remedial Action Standards for Inactive Uranium Processing Sites
(40 CFR 192), Volume I, EPA Report 520/4-82-013-1, Office of Radiation
Programs, Washington, D.C., 1982.
U.S. Environmental Protection Agency, Draft Background Information
Document, Proposed Standards for Radionuclides, EPA Report 520/1-83-001,
Office of Radiation Programs, Washington, D.C., 1983.
U.S. Environmental Protection Agency, Final Environmental Impact Statement
for Standards for the Control of Byproduct Materials from Uranium Ore
Processing (40 CFR 192), Volume I, EPA Report 520/1-83-008-1, Office of
Radiation Programs, Washington, D.C., 1983.
Environmental Protection Agency, Radionuclides Background Information
Document for Final Rules. Volume I, EPA Report 520/1-84-022-1, US EPA,
Office of Radiation Programs.
Environmental Protection Agency, Background Information
Document—Standard for Radon-222 Emissions from Underground Uranium
Mines. EPA 520/1-85-010, Office of Radiation Programs, USEPA,
Washington, D.C., 1985.
6-80
-------
EPA86 Environmental Protection Agency, Final Rule for Radon-222 Emissions from
Licensed Uranium Mill Tailings, Background Information Document, EPA
520/1-86-009, Office of Radiation Programs, Washington, DC, 1986.
Ev79 Evans, H.J., Buckton, K.E., Hamilton, G.E.,et al.,Radiation-induced
Chromosome Aberrations in Nuclear Dockyard Workers, Nature, 277, 531-534,
1979.
FRC67 Federal Radiation Council, Radiation Guidance for Federal Agencies,
Memorandum for the President, July 21, 1967, Fed. Reg., 32, 1183-84,
August 1, 1967.
Ga82 Garriott, M.L. and D. Grahn, Neutron and Gamma-Ray Effects Measured by
the Micronucleus Test, MuL Res. Let., 105, 157-162, 1982.
Gi84 Gilbert, E.S., Some Effects of Random Dose Measurements Errors on Analyses
of Atomic Bomb Survivor Data, Rad. Res., 9£, 591-605, 1984.
Gi85 Gilbert, E.S., Late Somatic Effects, in: Health Effects Model for Nuclear Power
Plant Accident Consequence Analysis by J.S. Evans, D.W. Cooper and D.W.
Moeller, NUREG/CR-4214, U.S. Nuclear Regulatory Commission, 1985.
Go29 Goldstein, L. and D.P. Murphy, Etiology of El-health of Children Bom After
Maternal Pelvic Irradiation: n, Defective Children Born After Post Conception
Pelvic Irradiation, Amer. J. Roentgenol. Rad. Ther., 22: 322-331, 1929.
Go80 Goodhead, D.T., Models of Radiation Interaction and Mutagenesis, pp. 231-
247, in Radiation Biology in Cancer Research, R.E. Meyn and H. R. Withers,
eds., Raven, New York, 1980.
Go82 Goodhead, D.T., An Assessment of the Role of Microdosimetry in
Radiobiology, Rad. Res., 91, 45-76, 1982.
Gr83a Grahn, D., et al., Interpretation of Cytogenetic Damage Induced in the Germ
Line of Male Mice Exposed for Over 1 Year to 239Pu Alpha Particles, Fission
Neutrons, or 60Co Gamma Rays, Rad. Res., 95, 566-583, 1983.
Gr83b Grahn, D., Genetic Risks Associated with Radiation Exposures During Space
Flight, Adv. Space Res., 3(8), 161-170, 1983.
Gr85 Grosovsky, AJ. and J.B. Little, Evidence for Linear Response for the Induction
of Mutations in Human Cells by X-Ray Exposures below 10 Rads, Proc. Natl.
Acad. Sci. USA, 82, 2092-2095, 1985.
6-81
-------
Gu77a Gustavson, K.H, Hagberg, B., Hagberg, G. and K. Sars, Severe Mental
Retardation in a Swedish County, I, Epidemiology, Gestational Age, Birth
Weight and Associated CNS Handicaps in Children Born 1959-70, Acta
Paediatr. Scand., 66, 373-379, 1977.
Gu77b Gustavson, K.-H., Hagberg, B., Hagberg, G. and K. Sars, Severe Mental
Retardation in a Swedish County, EL Etiologic and Pathogenetic Aspects of
Children Born 1959-70, Neuropadiatrie, 1:293-304, 1977.
HaSla Hagberg, B., Hagberg, G., Lewerth, A. and U. Lindberg, Mild Mental
Retardation in Swedish School Children, I. Prevalence, Acta Paediatr. Scand.,
70, 441-444, 1981.
HaSlb Hagberg, B., Hagberg, G., Lewerth, A. and U. Lindberg, Mild Mental Retardation in
Swedish School Children, n, Etiologic and Pathogenetic Aspects, Acta Paediatr.
Scand., 70:445-452, 1981.
Ha82 Harley, N.H. and B.S. Pasternak, Environmental Radon Daughter Alpha Dose
Factors in a Five-Lobed Human Lung, Health Phys., 42, 789-799, 1982.
He83 Herbert, D.E., Model or Metaphor? More Comments on the BEIR HI Report,
pp. 357-390, in Epidemiology Applied to Health Phys., CONF-830101, DE-
83014383, NTIS, Springfield, Virginia, 1983.
Hi53 Hicks, S.P., Developmental Malformations Produced by Radiation, A Timetable
of Their Development, Amer. J. Roentgenol. Radiat Thera., 69, 272-293, 1953.
Hi54 Hicks, S.P., The Effects of Ionizing Radiation, Certain Hormones, and
Radiomimetic Drugs on the Developing Nervous System, J. Cell. Comp.
Physiol., 43 (Suppl. 1), 151-178, 1954.
Hi66 Hicks, S.P. and CJ. D'Amato, Effects of Ionizing Radiations on Mammalian
Development, Adv. Teratol., i, 195-266, 1966.
Ho77 Hofmann, W. and F. Steinhausler, Dose Calculations for Infants and Youths
Due to the Inhalation of Radon and Its Decay Products in the Normal
Environment, in: Proceedings of the 4th International Congress of the
International Radiation Protection Association, Paris, 2, 497-500, 1977.
Ho81 Hornung, R. W. and S. Samuels, Survivorship Models for Lung Cancer
Mortality in Uranium Miners - Is Cumulative Dose an Appropriate Measure of
Exposure?, in: Proc. Int. Conf. on Hazards in Mining: Control, Measurement,
and Medical Aspects, October 4-9, 1981, Golden, Colorado, 363-368, edited by
Manuel Gomez, Society of Mining Engineers, New York, 1981.
6-82
-------
Ho84
Ho86
Ho87
HWC73
ICRP75
ICRP77
ICRP79
ICRP80
ICRP81
ICRP87
Is84
Howe, G.R., Epidemiology of Radiogenic Breast Cancer, in: Radiation
Carcinogenesis: Epidemiology and Biological Significance, 119-129, edited by
J.D. Boice, Jr. and J.F. Fraumeni, Jr., Raven Press, New York, 1984.
Howe, G.R., Nair, R.C., Newcomb, H.B., Miller, A.B. and J.D. Abbatt, Lung
Cancer Mortality (1950-1980) in Relation to Radon Daughter Exposure in a
Cohort of Workers at the Eldorado Beaver Lodge Uranium Mine, JNCI, 77,
357-362, 1986.
Howe, G.R, Nair, R.C, Newcombe, H.B, Miller, A.B,Burch, J.D and Abbott,
J.D. Lung Cancer Mortality (1950-80) in Relation to Radon Daughter
Exposure in a Cohort of Workers at the Eldorado Port Radium Uranium Mine:
Possible Modification of Risk by Exposure Rate. JNCI, 79: 1255-1260 (1987).
Health and Welfare Canada, The Testing of Chemicals for Carcinogenicity,
Mutagenicity and Teratogenicity, Health Protection Branch, HWC, Ottawa,
1973.
International Commission on Radiological Protection, Committee JJ on
Permissible Dose for Internal Radiation, Task Group on Reference Man, ICRP
Publ. 23, Pergamon Press, 1975.
International Commission on Radiological Protection, Recommendations of the
International Commission on Radiological Protection, ICRP Publ. 26, Ann.
ICRP, I, (1), Pergamon Press, 1977.
International Commission on Radiological Protection, Limits for Intakes of
Radionuclides by Workers, ICRP Publication 30, Part 1, Ann. ICRP, 2 (3/4),
Pergamon Press, New York, 1979.
International Commission on Radiological Protection, Effects of Inhaled
Radionuclides, ICRP Publication 31, Pergamon Press, 1980.
International Commission on Radiological Protection, Limits for Intakes of
Radionuclides by Workers, ICRP Publication 32, Part 3, Ann. ICRP, 6 (2/3),
Pergamon Press, 1981.
International Commission on Radiological Protection, Lung Cancer Risk from
Indoor Exposures to Radon Daughters, ICRP Publication 50, Pergamon Press,
NY, 1987.
Ishimaru, T., Nakashima, E. and S. Kawamoto, Relationship of Height, Body
Weight, Head Circumference and Chest Circumference to Gamma and Neutron
Doses Among In Utero Exposed Children, Hiroshima and Nagasaki. Technical
Report RERF TR 19-84, Radiation Effects Research Foundation, Hiroshima,
1984.
6-83
-------
Ja80 Jacobi, W. and K. Eisfeld, Dose to Tissue and Effective Dose Equivalent by
Inhalation of Radon-222 and Radon-220 and Their Short-Lived Daughters, GFS
Report S-626, Gesellschaft fuer Strahlen und Unweltforschung mbH, Munich,
1980.
Ja85 Jacobi W., Paretzke H. G. and Schindel F., Lung Cancer Risk Assessment of
Radon-Exposed Miners on the Basis of a Proportional Hazard Model, in:
Occupational Radiation Safety in Mining, Volume 1, H. Stocker, editor,
Canadian Nuclear Association, Toronto, Ontario, Canada, pp. 17-24, 1985.
Ja70 Jacobsen, L., Radiation Induced Fetal Damage, Adv. Teratol., 4, 95-124, 1970.
Ja81 James, A. C. et al., Respiratory Tract Dosimetry of Radon and Thoron
Daughters: The State-of-the-Art and Implications for Epidemiology and
Radiobiology, in: Proc. Int Conf. on Hazards in Mining: Control,
Measurement, and Medical Aspects, October 4-9, 1981, Golden, Colorado, 42-
54, edited by Manuel Gomez, Society of Mining Engineers, New York, 1981.
Ka73 Kato, H., Late Effects in Children Exposed to the Atomic Bomb While In
Utero, Technical Report 18-73, Atomic Bomb Casualty Commission,
Hiroshima, 1973.
Ka82 Kato, H. and W.J. Schull, Studies of the Mortality of A-bomb Survivors, 7.
Mortality, 1950-1978: Part I, Cancer Mortality, Rad. Research 90, 395-432,
1982, (Also published by the Radiation Effect Research Foundation as: RERF
TR 12-80, Life Span Study Report 9, Part L)
Ka89 Kaul, D.C., Uncertainty Analysis of DS86 Dosimetry System, RERF Update I
(2), 4, 1989.
Ke72 Kellerer, A.M. and H.M. Rossi, The Theory of Dual Radiation Action, Curr.
Topics Rad., Res. Quart, 8, 85-158, 1972.
KeSla Kerr, G.D., Review of Dosimetry for the Atomic Bomb Survivors, in:
Proceedings of the Fourth Symposium on Neutron Dosimetry, Gessellschaft fur
Strahlen- und Umweltforschung, Munich-Neuherberg, Federal Republic of
Germany, June 1-5, i 501, Office for Official Publications of the European
Communities, Luxembourgh, 1981.
KeSlb Kerr, G.D., Findings of a Recent ORNL Review of Dosimetry for the Japanese
Atomic Bomb Survivors, ORNL/TM-8078, Oak Ridge National Laboratory,
Oak Ridge, Tennessee, 1981.
Ki62 King, R.C., Genetics, Oxford University Press, New York, 1962.
6-84
-------
La78 Land, C.E. and J.E. Norman, Latent Periods of Radiogenic Cancers Occurring
Among Japanese A-bomb Survivors, in: Late Biological Effects of Ionizing
Radiation, I, 29-47, IAEA, Vienna, 1978.
La80 Land, C.E., Boice, J.D., Shore, R.E., Norman, J.E. and M. Tokunaga, et al.,
Breast Cancer Risk from Low-Dose Exposures to Ionizing Radiation: Results
of Parallel Analysis of Three Exposed Populations of Women, J. Nati. Cane.
Inst, 65, 353-376, 1980.
La83 Land, C.E. and D.A. Pierce, Some Statistical Considerations Related to the
Estimation of Cancer Risk Following Exposure to Ionizing Radiation, pp. 67-
89, in Epidemiology Applied to Health Phys., CONF-830101, DE83014383,
NTIS, Springfield, Virginia, 1983.
Le62 Lea, D.E., Actions of Radiations on Living Cells, 2nd edition, Cambridge
University Press, 1962.
Lo81 Loewe, W.E. and E. Mendelsohn, Revised Dose Estimates at Hiroshima and
Nagasaki, Health Phys., 41, 663-666, 1981.
Lu71 Lundin, F E Jr., Wagoner, J K and Archer, V E Radon Daughter Exposure and
Respiratory Cancer. Quantitative and Temporal Aspects, Joint Monograph No.
1, N10SH-NIEHS, USPHS, DHEW, Washington, DC, 1971.
Ma59 Mandansky, A., The Fitting of Straight Lines When Both Variables Are Subject
to Error, J. Amer. Statis. Assoc., 54, 173-205, 1959.
Ma83 Mays, C.W. and H. Spiess, Epidemiological Studies in German Patients
Injected with Ra-224, pp. 159-266, in: Epidemiology Applied to Health
Physics, CONF-830101, DE-83014383, NTIS, Springfield, Virginia, 1983.
Mc78 McDowell, E.M., McLaughlin, J.S., Merenyi, D.K., Kieffer, R.F., Harris, C.C.
and B.F. Trump, The Respiratory Epithelium V. Histogenesis of Lung
Carcinomas in Humans, J. Natl. Cancer Inst., .61, 587-606, 1978.
Mi78 Michel C. and H. Fritz-Niggli, Radiation-Induced Developmental Anomalies in
Mammalian Embryos by Low Doses and Interaction with Drugs, Stress and
Genetic Factors, pp. 399-408, in: Late Biological Effects of Ionizing
Radiation, Vol. U, IAEA, Vienna, 1978.
Mi59 Miller, R.W., Delayed Effects Occurring Within the First Decade After
Exposure of Young Individuals to the Hiroshima Atomic Bomb, Technical
Report 32-59, Atomic Bomb Casualty Commission, Hiroshima, 1959.
Mi72 Miller, R.W. and WJ. Blot, Small Head Size Following In Utero Exposure to
Atomic Radiation, Hiroshima and Nagasaki, Technical Report 35-72, Atomic
Bomb Casualty Commission, Hiroshima, 1972.
6-85
-------
Mo67 Morgan, K.Z. and J.E. Turner, Principles of Radiation Protection, John Wiley
and Sons, Inc., New York, 1967.
Mo79 Mole, R.H., Carcinogenesis by Thorotrast and Other Sources of Irradiation,
Especially Other Alpha-Emitters, Environ. Res., 18, 192-215, 1979.
Mu29 Murphy, D.P., The Outcome of 625 Pregnancies in Women Subject to Pelvic
Radium or Roentgen Irradiation, Amer. J. Obstet. Gyn., .18, 179-187, 1929.
Mu30 Murphy, D.P. and M. DeRenyi, Postconception Pelvic Irradiation of the Albino
Rat (Mus Norvegieus): Its Effects Upon the Offspring, Surg. Gynecol. Obstet.,
50, 861-863, 1930.
Mu83 Muller, J., Wheeler, W.C., Gentleman, J.F., Suranyi, G. and R.A. Kusiak, Study
of Mortality of Ontario Miners, 1955-1977, Part I, Ontario Ministry of Labor,
Ontario, May 1983.
NAS72 National Academy of Sciences - National Research Council, The Effects on
Populations of Exposures to Low Levels of Ionizing Radiation, Report of the
Committee on the Biological Effects of Ionizing Radiations (BEIR I Report),
Washington, D.C., 1972.
NAS80 National Academy of Sciences - National Research Council, The Effects on
Populations of Exposure to Low Levels of Ionizing Radiation, Committee on
the Biological Effects of Ionizing Radiation, (BEIR ffl), Washington, D.C.,
1980.
NAS88 National Academy of Sciences - National Research Council,. Health Risks of
Radon and Other Internally Deposited Alpha-Emitters, BEIR IV, National
Academy Press, Washington, DC, 1988.
NASA73 National Aeronautics and Space Administration, Bioastronautics Data Book,
NASASP-3006, 2nd Edition, edited by J. R. Parker and V. R. West,
Washington, D.C., 1973.
NCHS73 National Center for Health Statistics, Public Use Tape, Vital Statistics -
Mortality Cause of Death Summary - 1970, PB80-133333, Washington, D.C.,
1973.
NCHS75 National Center for Health Statistics, U.S. Decennial Life Tables for 1969-71,
1(1), DHEW Publication No. (HRA) 75-1150, U.S. Public Health Service,
Rockville, Maryland, 1975.
NCRP75 National Council on Radiation Protection and Measurement, Natural
Background Radiation in the United States, NCRP Report No. 45, Washington,
D.C., 1975.
6-86
-------
NCRP77 National Council on Radiation Protection and Measurements, Protection of the
Thyroid Gland in the Event of Releases of Radioiodine, Report No. 55,
Washington, D.C., 1977.
NCRP80 National Council on Radiation Protection and Measurements, Influence of Dose
and Its Distribution in Time on Dose-Response Relationships for Low-LET
Radiation, NCRP Report No. 64, Washington, D.C., 1980.
NCRP84 National Council on Radiation Protection and Measurements, Evaluation of
Occupational and Environmental Exposures to Radon and Recommendations,
NCRP Report No. 78, Washington, D.C., 1984.
NCRP85 National Council on Radiation Protection and Measurements, Induction of
Thyroid Cancer by Ionizing Radiation, NCRP Report No. 80, Washington,
D.C., 1985.
NCRP87 National Council on Radiation Protection and Measurements, Ionizing
Radiation Exposure of the Population of the United States, NCRP Report
No. 93, Bethesda, MD, 1987.
Ne56 Neel, J. V. and W. J. Schull, The Effect of Exposure to the Atomic Bombs on
Pregnancy Termination in Hiroshima and Nagasaki, National Academy of
Sciences, Publ. 461, Washington, D.C., 1956.
Ne70 Nelson, K.B. and J. Deutschberger, Head Size at One Year as a Predictor of
Four-Year I.Q., Develop. Med. Child Neural., 12, 487-495, 1970.
Ne86 Nero, A.V., Schwehr, M.B., Nazaroff, W.W. and K.L. Revzan, Distribution of
Airborne Radon-222 Concentrations in U.S. Homes, Science, 234, 992-997,
1986.
Ne88 Neel, J. V., Schull, W. J., Awa, A. A., Satoh, C., Otake, M., Kato, H, and
Yoshimato, Y., Implications of the Hiroshima - Nagasaki Genetic Studies for
the Estimation of the Human "Doubling Dose" of Radiation. Presentation at
XVI the International Congress of Genetics, Toronto, 1988.
NIH85 National Institutes of Health, Report of the National Institutes of Health Ad
Hoc Working Group to Develop Radioepidemiological Tables, NTH Publication
No. 85-2748, U.S. Government Printing Office, Washington, DC 20402, p 92,
1985.
NIOSH87 National Institute for Occupational Safety and Health. Radon Progeny in
Underground Mines, DHHS (NIOSH) Publication NO. 88-101, USPHS, CDC,
NIOSH, Cincinnati, Ohio, 1987
NRC85 Nuclear Regulatory Commission, Health Effects Model for Nuclear Power
Plant Accident Consequence Analysis. NUREG/CR-4214. U.S. Nuclear
Regulatory Commission, Washington, DC, 1985.
6-87
-------
ORNL84 Oak Ridge National Laboratory, Age Dependent Estimation of Radiation Dose,
[in press], 1984.
Of80 Oftedal, P. and A.G. Searle, An Overall Genetic Risk Assessment for
Radiological Protection Purposes, J. Med. Genetics, .17, 15-20, 1980.
Oh65 Ohzu, E., Effects of Low-Dose X-Irradiation on Early Mouse Embryos, Rad.
Res. 26, 107-113, 1965.
Ot83 Otake, M. and W.H. Schull, Mental Retardation in Children Exposed In Utero
to the Atomic Bombs: A Reassessment, Technical Report RERFTR 1-83,
Radiation Effects Research Foundation, Hiroshima, 1983.
Ot84 Otake, M. and WJ. Schull, In Utero Exposure to A-bomb Radiation and
Mental Retardation: A Reassessment, Brit. J. Radiol., 57, 409-414, 1984.
Ot88 Otake, M., WJ. Schull, Y. Fujikoshi, and H. Yoshimaru. Effect on School
Performance of Prenatal Exposure to Ionizing Radiation: A Comparison of the
T65DR and DS86 Dosimetry Systems. Technical Report RERF TR 2-88,
Radiation Effects Research Foundation, Hiroshima, 1988.
Pi89 Pierce, D.A., Stram, D.O. and M. Vaeth, Allowing for Random Errors in
Radiation Exposure Estimates for the Atomic Bomb Survivor Data, RERF TR
2-8, Radiation Effects Research Foundation, Hiroshima, 1989.
P152 Plummer, G.W., Anomalies Occurring in Children Exposed In Utero to the
Atomic Bomb in Hiroshima, Pediat, 10, 687-692, 1952.
Po78 Pohl-Ruling, J., Fischer, P. and E. Pohl, The Low-Level Shape of Dose
Response for Chromosome Aberration, pp. 315-326, in: Late Biological
Effects of Ionizing Radiation, Volume n, International Atomic Energy Agency,
Vienna, 1978.
Pr83 Prentice, R.L., Yoshimoto, Y. and M.W. Mason, Relationship of Cigarette
Smoking and Radiation Exposure to Cancer Mortality in Hiroshima and
Nagasaki, J. Nat Cancer Inst., 70, 611-622, 1983.
Pr87 Preston, D.L. and D.A. Pierce, The Effect of changes in Dosimetry on Cancer
Mortality Risk Estimates in the Atomic Bomb Survivors. RERF 9-87,
Radiation Effects Research Foundation, Hiroshima, 1987.
Pr88 Preston, D.L. and D.A. Pierce, The Effect of Changes in Dosimetry on Cancer
Mortality Risk Estimates in the Atomic Bomb Survivors, Rad. Res. 114, 437-
466, 1988.
Ra84 Radford, E.P. and K.G. St. Cl. Renard, Lung Cancer in Swedish Iron Miners
Exposed to Low Doses of Radon Daughters, N. Engl. J. Med., 310, 1485-1494,
1984.
6-88
-------
RERF78
RERF83
RERF84
Ro78
Ro78
Ru50
Ru53
Ru54
Ru56
Ru57
Ru58
Ru70
Ru71
Radiation Effects Research Foundation. Radiation Effects Research
Foundation, 1 April 1975 - 31 March 1978. RERF Report 75-78, Hiroshima,
1978.
Radiation Effects Research Foundation, Reassessment of Atomic Bomb
Radiation Dosimetry in Hiroshima and Nagasaki, Proc. of the U.S.-Japan Joint
Workshop, Nagasaki, Japan, Feb. 16-17, 1982, Radiation Effects Research
Foundation, Hiroshima, 730, Japan, 1983.
Radiation Effects Research Foundation, Second U.S.-Japan Joint Workshop for
Reassessment of Atomic Bomb Radiation Dosimetry in Hiroshima and
Nagasaki, Radiation Effects Research Foundation, Hiroshima, 730, Japan, 1984.
Rossi, H.H. and C.W. Mays, Leukemia Risk from Neutrons, Health Phys., 34,
353-360. 1978.
Rowland, R.E., Stehney, A.F. and H.F. Lucas, Dose Response Relationships for
Female Radium Dial Workers, Rad. Res., 76, 368-383, 1978.
Russell, L.B., X-ray Induced Developmental Abnormalities in the Mouse and
Their Use in the Analysis of Embryological Patterns, I. External and Gross
Visceral Changes, J. Exper. Zool., 114. 545-602, 1950.
Rugh, R., Vertebrate Radiobiology: Embryology, Ann. Rev. Nucl. Sci., 3_, 271-
302, 1953.
Russell, L.B. and W.L. Russell, An Analysis of the Changing Radiation
Response of the Developing Mouse Embryo, J. Cell. Comp. PhysioL, 43
(Suppl. 1), 103-149, 1954.
Russell, L.B., X-Ray Induced Developmental Abnormalities in the Mouse and
Their Use in the Analysis of Embryological Patterns, JJ. Abnormalities of the
Vertebral Column and Thorax, J. Exper. Zool., 131, 329-390, 1956.
Russell, L.B., Effects of Low Doses of X-rays on Embryonic Development in
the Mouse, Proc. Soc. Exptl. Biol. Med., 95, 174-178, 1957.
Russell, W.L., Russell, L.B. and E.M. Kelly, Radiation Dose Rate and Mutation
Frequency, Science, 128:1546-1550, 1958.
Rugh, R., The Effects of Ionizing Radiation on the Developing Embryo and
Fetus, Seminar Paper No. 007, Bureau of Radiological Health Seminar
Program, U.S. Public Health Service, Washington, D.C., 1970.
Rugh, R., X-ray Induced Teratogenesis in the Mouse and Its Possible
Significance to Man, Radiol., 99, 433-443, 1971.
6-89
-------
Sa82 Satoh, C. et al., Genetic Effects of Atomic Bombs, in: Human Genetics,
Part A: The Unfolding Genome, A. R. Liss, Inc., New York, 267-276, 1982.
Sc81 Schull, W.J., Otake, M. and J.V. Neel, Genetic Effects of the Atomic Bombs:
A Reappraisal, Science, 213, 1220-1227, 1981.
Sc82 Schwarz, G., and Dunning, Jr., D.E., Imprecision in Estimates of Dose from
Ingested Cs-137 due to Variability in Human Biological Characteristics, Health
Phys. 43, 631-645, 1982.
Sc84 Schull, WJ. and J.K. Bailey, Critical Assessment of Genetic Effects of Ionizing
Radiation on Pre- and Postnatal Development, pp. 325-398, in: Issues and
Reviews in Teratology, Volume 2, H. Kalter, editor. Plenum Press, New York,
1984.
Sc88 Schull, W.J., M. Otaki, and H. Yoshimaru. Effects on Intelligence Test Score
of Prenatal Exposure to Ionizing Radiation in Hiroshima and Nagasaki: A
Comparison of the T65DR and DS86 Dosimetry Systems. Technical Report
RERF TR 3-88, Radiation Effects Research Foundation, Hiroshima, 1988.
Se69 Senyszyn, J.J. and R. Rugh, Hydrocephaly Following Fetal X-Irradiation,
Radio!., 93_, 625-634, 1969.
Se88 Sevc, J, Kunz, E, Tomasek, L, Placek, V, and Horacek, J Cancer in Man after
Exposure to Rn Daughters, Health Physics, 54: 27-46 (1988).
Sh87 Shimizu, Y., Kato, H., Schull, W.J., Preston, D.L., Fujita, S. and Pierce, D.A.,
Life Span Study Report 11. Part I. Comparison of Risk Coefficients for Site-
Specific Cancer Mortality Based on DS86 and T65DR Shielded Kerma and
Organ Doses, RERF TR 12-87, Radiation Effects Research Foundation,
Hiroshima, 1987. Also published as Rad. Res. 118. 502-524, 1989.
Sh88 Shimuzu, Y, Kato, H. and WJ. Schull, Life Span Study Report 11. Part 2.
Cancer Mortality in the Years 1950-85 Based on the Recently Revised Doses
(DS86), RERF TR 5-88, Radiation Effects Research Foundation, Hiroshima,
1988.
Sm78 Smith, P.G. and R. Doll, Radiation-Induced Cancers in Patients with
Ankylosing Spondylitis Following a Single Course of X-ray Treatment, in:
Proc. of the IAEA Symposium, Late Biological Effects of Ionizing Radiation,
i, 205-214, IAEA, Vienna, March 1978.
Sp56 Spector, W.S., editor, Handbook of Biological Data, Table 314, Energy Cost,
Work: Man, W. B. Sanders Co., Philadelphia, 1956.
Sp83 Spiers, F.W., Lucas, H.F., Rundo, J. and G.A. Anast, Leukemia Incidence in
the U.S. Dial Workers, in: Conference Proc. on Radiobiology of Radium and
6-90
-------
the Actinides in Man, October 11-16, 1981, Health Phys., 44(Suppl. l):65-72,
1983.
St21 Stettner, E., Bin weiterer Fall einer Schadingung einer menschichen Frucht
durch Roentgen Bestrahlung., Jb. Kinderheilk. Phys. Erzieh., 95, 43-51, 1921.
St81 Straume, T. and R. L. Dobson, Implications of New Hiroshima and Nagasaki
Dose Estimates: Cancer Risks and Neutron RBE, Health Phys., 41(4):666-671,
1981.
St84 Stein, Z.A. and M.W. Susser, The Epidemiology of Mental Retardation, in:
Epidemiology of Pediatric Neurology, B. Schoenberg, editor, Marcel Dekker,
Inc., New York, [in press], 1984.
St85 Steinhausler F., and Hofmann W., Inherent Dosimetric and Epidemiological
Uncertainties Associated with Lung Cancer Risk Assessment in Mining
Populations, in: Occupational Radiation Safety in Mining, Volume 1, H.
Stocker, editor, Canadian Nuclear Association, Toronto, Ontario, Canada, pp.
327-334, 1985.
St88 Stather, J.W., C.R. Muirhead, A.A. Edwards, J.D. Harrison, D.C. Lloyd, and
N.R. Wood. Health Effects Models Developed from the 1988 UNSCEAR
Report. NRPB-R226. National Radiation Protection Board, Chilton, England,
1988.
Ta67 Tabuchi, A., Hirai, T., Nakagawa, S., Snimada, K. and J. Fugito, Clinical
Findings on In Utero Exposed Microcephalic Children, Technical Report 28-67,
Atomic Bomb Casualty Commission, Hiroshima, 1967.
Th82 Thomas, D.C. and K.G. McNeill, Risk Estimates for the Health Effects of
Alpha Radiation, Report BSfFO-0081. Atomic Energy Control Board, Ottawa,
1982.
To80 Tobias, C. A., Blakely, E.A., Ngo, F.Q.H. and T.C.H. Yang, The Repair-
Misrepair Model, pp. 195-230, in: Radiation Biology and Cancer Research, R.
E. Meyn and H. R. Withers, eds., Raven, New York, 1980.
To84 Tokunaga, M., Land, C.E., Yamamoto, T., Asano, M., TaMoka, S., Ezaki, E.
and I. Nishimari, Incidences of Female Breast Cancer Among Atomic Bomb
Survivors, Hiroshima and Nagasaki, 1950-1980, RERF TR 15-84, Radiation
Effects Research Foundation, Hiroshima, 1984.
U182 Ullrich, R.L., Lung Tumor Induction in Mice: Neutron RBE at Low Doses,
NTIS-DE 82009642, National Technical Information Service, Springfield,
Virginia, 1982.
6-91
-------
UNSC58 United Nations, Report of the United Nations Scientific Committee on the
Effects of Atomic Radiation, Official Records: Thirteenth Session, Supplement
No. 17 (A/3838), United Nations, New York, 1958.
UNSC62 United Nations, Report of the United Nations Scientific Committee on the
Effects of Atomic Radiation, Official Records: Seventeenth Session,
Supplement No. 16 (A/5216), United Nations, New York, 1962.
UNSC66 United Nations, Report of the United Nations Scientific Committee on the
Effects of Atomic Radiation, Official Records: Twenty-First Session,
Supplement No. 14 (A/6314), United Nations, New York, 1966.
UNSC69 United Nations, Report of the United Nations Scientific Committee on the
Effects of Atomic Radiation, Supplement No. 13 (A/7613), United Nations,
New York, 1969.
UNSC72 United Nations Scientific Committee on the Effects of Atomic Radiation,
Ionizing Radiation: Levels and Effects, Volume II: Effects, Report to the
General Assembly. Sales No. E. 72. DC. 18., United Nations, New York, 1972.
UNSC77 United Nations Scientific Committee on the Effects of Atomic Radiation,
Sources and Effects of Ionizing Radiation, Report to the General Assembly,
with Annexes, Sales No. E.77 DC.l., United Nations, New York, 1977.
UNSC82 United Nations Scientific Committee on the Effects of Atomic Radiation,
Ionizing Radiation: Sources and Biological Effects, 1982 Report to the General
Assembly, Sales No. E.82. IX.8, United Nations, New York, 1982.
UNSC86 United Nations Scientific Committee on the Effects of Atomic Radiation,
Genetic and Somatic Effects of Ionizing Radiation, 1986 Report to the General
Assembly, Sales No. #. 86 DC 9, United Nationa, New York, 1986.
UNSC88 United Nations Scientific Committee on the Effects of Atomic Radiation,
Sources, Effects and Risks of lonization Radiation, 1988 Report to the General
Assembly, Sales No. #. 88. DC. 7, United Nationa, New York, 1988.
Up75 Upton, A.C., Physical Carcinogenesis: Radiation—History and Sources, pp.
387-403, in: Cancer 1, F.F. Becker, editor. Plenum Press, New York, 1975.
USRPC80 U.S.Radiation Policy Council, Report of the Task Force on Radon in
Structures, USRPC-80-002, Washington, D.C., 1980.
Va80 Van Buul, P.P.W., Dose-response Relationship for X-ray Induced Reciprocal
Translocations in Stem Cell Spermatogonia of the Rhesus Monkey (Macaca
mulatta), Mutat. Res., 73, 363-375, 1980. (Cited in UNSC82.)
6-92
-------
Vo02 Von Frieben, A., Demonstration lines cancroids des rechten Handruckens das
sich nach lang dauernder Einwirkung von Rontgenstrahlen entwickelt hatte.
Fortschr. Geb. Rontgenstr., 6:106 (1902) cited in Up75.
Wh83 Whittemore, A.S. and A. McMillan, A Lung Cancer Mortality Among U.S.
Uranium Miners: A Reappraisal, Technical Report No. 68, SIAM Inst. Math.
Soc., Stanford University, Stanford, 1983.
WHO75 World Health Organization, International Statistical Classification of Diseases,
Injuries, and Causes of Death, 9th Revision, Geneva, 1975.
Wi54 Wilson, J.G., Differentiation and the Reaction of Rat Embryos to Radiation, J.
Cell. Comp. Phvsiol., 43 (SuppL 1), 11-37, 1954.
Wo65 Wood, J.W., Johnson, K.G. and Y. Omari, In Utero Exposure to the Hiroshima
Atomic Bomb: Follow-up at Twenty Years, Technical Report 9-65, Atomic
Bomb Casualty Commission, Hiroshima, 1965.
Wo66 Wood, J.W., Johnson, K.G., Omari, Y., Kawamoto, S. and R.J. Keehn, Mental
Retardation in Children Exposed In Utero to the Atomic Bomb—Hiroshima and
Nagasaki, Technical Report 10-66, Atomic Bomb Casualty Commission,
Hiroshima, 1966.
6-93
-------
CHAPTER 7: INDIVIDUAL DOSE ASSESSMENT OF DISPOSAL OP TRANSURANIC
WASTE IN MINED GEOLOGIC REPOSITORIES
7.1 INTRODUCTION
This chapter deals with the disposal of transuranic (TRU) waste in geologic
repositories. Most TRU waste is from, the production of nuclear weapons. It consists of
materials contaminated mainly with radioactive isotopes of plutonium and americium, but
also contains other transuranic isotopes. The waste form varies widely, but most of the
waste can be described as contaminated plastic, wood, rubber, metal, cloth, paper, and
laboratory trash.
Risk assessments for spent nuclear fuel, high-level, and transuranic radioactive
wastes were conducted in support of the disposal standards proposed in 1982. Review of
the risk assessment by an EPA Science Advisory Board committee produced a number of
recommendations which called for the analyses to be less conservative, that is, to use
parameter values that were considered more likely and that would tend to produce lower
estimates of population risks. In addition, DOE had developed extensive data on the nine
specific locations to be evaluated for the first disposal facility. These data were available
for use in risk assessments for the final rule.
An important consideration in repromulgating disposal standards has been the
assessment of risks associated with the disposal of these wastes in mined geologic
repositories. The risk assessments carried out in support of the development of the
standards are intended to be "generic" hi nature. In developing the assessments, the
Agency considered a wide range of geologic environments and other related parameters.
In the early stages of the EPA work, the DOE, which is responsible for developing a
geologic disposal facility, had not yet developed extensive data associated with its
principal candidate sites. Therefore, the risk assessments conducted in support of the
1982 proposed standards used data from the general literature on potential waste disposal
environments as well as the limited data that had been obtained by DOE up to about
1980. Individual dose assessments in the current effort utilize recent data from DOE
efforts to develop mined geologic repositories for high-level and transuranic radioactive
wastes.
The performance of the generic TRU waste disposal facility has been evaluated
using the same methodology as the risk assessments for the spent nuclear fuel
repositories presented in 1985 (EPA85) and updated in 1992 (RAE92). In this chapter,
performance is quantified in terms of risks to an individual consuming contaminated
ground water near the disposal facility. Radionuclide releases are assumed to occur
through normal ground-water flow and gaseous transport, if applicable.
The risk assessment for the TRU waste disposal system is based upon the same
conceptual models and data used in the spent nuclear fuel repository analyses. The data
used in the TRU risk assessment are presented in detail later in this chapter. Because of
the generic nature of the assessment, the results of the risk calculations are not intended
to project actual risks expected at particular sites; such projections will be possible only
7-1
-------
after the potential sites are more fully characterized. Instead, the assessments are
intended to provide estimates of potential disposal facility performance. The Agency has
attempted to ensure that its generic calculations are based upon reasonable parameter
values.
7.2 TIME FRAME
Recommendations in the published literature vary widely on the time frame over
which radioactive waste disposal alternatives should be evaluated or compared. If the
future could be predicted with accuracy, then a very long time frame would provide a
more complete evaluation than a short time frame and, ultimately, a more complete
comparison among alternative disposal systems. For the containment requirements now
in effect under 40 CFR Part 191, compliance must be demonstrated over a period of
10,000 years. That demonstration requires an analysis of the movement of radionuclides
out of the repository into the environment. Such an analysis is also a first step in
demonstrating compliance with the individual dose or ground-water protection
requirements. For these requirements, the second step in the analysis involves following
the radionuclides through the environment via pathways by which an individual could be
exposed to radiation. Once the analysis for 1,000 years has been developed, very little
additional effort would be needed to extend the projections to 10,000 years.
In the course of performing numerous risk assessments of radioactive waste
disposal systems, the Agency has concluded that the risks identified over relatively short
time spans, such as a few hundred to one thousand years, do not adequately portray
important differences among alternative sites or disposal systems. This is because the
ground-water travel times would probably be sufficiently long at most sites that no
significant radionuclide releases would be predicted over this time period. If the analyses
were carried further into the future, there could be substantial differences among the sites
because of their different hydrologic or geochemical characteristics. With these
considerations in mind, the risk assessments carried out in support of this rulemaking
have been based upon a time frame of 10,000 years. This time frame appears long
enough to identify important differences among sites and among other aspects of the
disposal systems. Many of the computer simulations have been extended to 100,000 years
in order to provide better insight on the long-term performance of disposal alternatives.
Part of the risk assessment is concerned with the uncertainties in the calculated
results. There are several sources of uncertainty, including spatial and temporal
variations in site parameter values, an incomplete knowledge of the natural site
characteristics, and the prediction of possible future events at the disposal site. Since site
conditions far in the future are more difficult to reliably predict, the uncertainties in
modeling system performance may increase with the length of the simulation period.
Uncertainties may also increase as the radionuclide transport distance increases. In other
words, it is easier to reliably predict the transport of radionuclides over a short distance
than over a long distance. The variabilities in transport parameter values over a long
flow path are generally greater than those for a shorter flow path. Thus, the
uncertainties depend both on the distance to the accessible environment and the length of
the simulation period.
7-2
-------
7.3 MEASURES OF RISK
In examining the long-term effects of a radioactive waste repository, the Agency
has considered both population and individual doses. These two risk measures provide
very different perspectives. For example, a relatively unproductive ground-water supply
could be contaminated with radionuclides released from a repository at some point in the
future, but because of the limited availability of water from this supply, only a few
individuals would be exposed to the radioactivity. In this case, the individual doses might
be high, although the overall effect on the population could still be quite low. On the
other hand, a future release from a radioactive waste repository could lead to very low
level contamination of water supplies that serve a large population. In this case, the dose
to any individual in that population might be small, while the cumulative population dose
could be substantial in terms of total cancers and genetic effects in the population.
Because of these differences, the Agency has developed and applied techniques that can
estimate both population and individual risks.
This analysis addresses only individual risks and ground-water protection as the
population risks are addressed in the now reinstated containment requirements. An
estimate of an individual's risk is determined by estimating the annual radiation dose
from consuming two liters per day of ground water contaminated with radionuclides from
the repository. This exposure is presented as a function of time after disposal for an
individual using ground water at a. particular distance (typically two kilometers) directly
down gradient from the edge of the repository. This report uses the maximum individual
dose from ingestion of drinking water as a measure to compare the effectiveness of
various types of geologies and engineered designs.
Ground water protection is evaluated in terms of the concentrations of
radionuclides down gradient from the repository. The concentrations are calculated for
Ra-226, total alpha-emitters, and beta and gamma-emitting radionuclides.
7.4 COMPUTER CODE UTILIZED
The computer code used in this assessment is NEFTRAN-S. This code was
developed by Sandia National Laboratories under contract to the NRC. The model is
described and documented in SAND90. NEFTRAN-S calculates cumulative radionuclide
releases, ground-water concentrations, and individual doses. Because it calculates
radionuclide decay and ingrowth during transport, it is useful for assessment periods in
excess of 10,000 years. The code uses the distributed velocity method to calculate
radionuclide transport in a network of one-dimensional legs. The flow network is
designed to represent ground-water flow in the vicinity of a repository.
The distributed velocity method used in NEFTRAN-S is a method for modeling the
dispersive or diffusive transport of radionuclides in porous media. Rather than using a
single transport velocity, a range of transport velocities is used. The radionuclide
inventory is partitioned and each portion of the inventory is transported at a different
velocity. This simulates the effect of dispersion or diffusion by allowing portions of the
7-3
-------
inventory to be transported at higher or lower velocities than the centroid of the
contaminant plume. The distributed velocity method has some computational advantages
over other numerical approximation methods when applied to dispersive radionuclide
transport. More detail on the mathematical basis of the distributed velocity method is
given in reference SAND90.
One of the most useful aspects of the NEFTRAN-S code is its capacity to perform
probabilistic analyses. The code uses a Monte Carlo sampling method to randomly select
input values from probability distributions. For each random selection of input values,
the transport model is executed and the results from each such sample are saved. After a
number of samples have been evaluated, the results can be expressed as a probability
distribution. This procedure is used in the risk assessment to evaluate the effects of
parameter uncertainties.
7.5 GENERIC DISPOSAL SYSTEM FOR TRANSURANIC RADIOACTIVE WASTE
7.5.1 System Models
The waste disposal system considered in this risk assessment is based on national
plans to develop mined geological repositories for disposal of TRU radioactive wastes.
Such repositories consist of underground mines or excavations with working levels
between 300 and 1,000 meters below the surface. Like the spent fuel repository
assessment presented in 1985, the TRU waste disposal facility assessment will focus on
disposal in four different host rock types.
Transuranic radioactive wastes differ from spent nuclear fuel and high-level
radioactive waste in both radionuclide content and waste form. Transuranic wastes
consist of a variety of waste forms, including plastic, rubber, wood, glass, cloth, and metal.
The waste is generated from reprocessing, fabrication, and research at DOE facilities.
The principal radioactive constituents of transuranic waste are plutonium and americium.
Present plans call for disposal in a mined geologic facility, with the wastes packaged in
metal drums or boxes and stacked in the mined waste disposal rooms. After emplacement
of the wastes, the disposal facility would be backfilled to enhance its mechanical stability
and to retard the movement of fluids. The shafts and boreholes which connect the
disposal facility to the surface would be backfilled and sealed.
The structure of the analysis can be represented as shown in Figure 7.5-1. The
components of the system to the right of the vertical dotted lines represent the "accessible
environment." The components on the left side of the diagram represent the release and
transport mechanisms from the repository to the accessible environment.
In order for radionuclides to reach the accessible environment, they must be
released from the waste form. Since much of the radioactivity in TRU waste is present as
surface contamination, the waste form is not expected to significantly limit the
radionuclide release rates. The release rates are likely to be controlled by radionuclide
solubility. Upon leaving the waste form, the radionuclides enter the backfilled openings of
7-4
-------
REPOSITORY
WASTE
FORM
>,
REPOSITORY
TRANSPORT
PATHWAYS
PATHWAYS
TO SURFACE
PATHWAYS
TO AQUIFER
ACCESSIBLE
ENVIRONMENT
ATMOSPHERE
LAND
SURFACE
AQUIFER BEYOND
CONTROLLED AREA
INDIVIDUAL
RAE -104276
Figure 7.5-1. Components in the risk assessment of radioactive waste releases.
-------
the disposal facility. In general, radionuclides may travel from the disposal facility to the
accessible environment in three ways: 1) direct pathways to the land surface, 2) vertical
migration in slowly moving ground water to an aquifer and then to the surface, and 3)
transport of radioactive gases to the ground surface from a disposal facility in the
unsaturated zone.
Direct pathways to the surface, caused by unusual events or human intrusion, are
not evaluated in the current risk assessment. Gaseous transport is also excluded from the
current analysis because TRU waste does not have a significant potential to produce
radioactive gases.
Release pathways involving ground water present the possibility that individuals
may be exposed to radioactivity by ingesting contaminated water from the aquifer. The
possibility of such a scenario could be minimized by siting the disposal facility in an area
where human use of ground water is unlikely. It is assumed in these analyses that the
disposal facility is sited in a remote area where ground water is not used to support the
needs of a large population.
The movement of radionuclides from the waste form, through the disposal facility,
and ultimately to the accessible environment depends on a number of possible scenarios
that might alter the conditions of the underground environment. The risk analyses
reported here consider only normal ground-water flow.
7.5.2 Disposal Facility Parameters
Certain assumptions need to be made regarding the geometry and physical
characteristics of the disposal facility. However, an examination of the Agency's risk
analysis models indicates that they are not highly sensitive to these engineering
assumptions. The layout of the generic TRU waste disposal facility is based on the design
of the Waste Isolation Pilot Plant (WIPP)- These parameters, which are summarized in
Table 7.5-1, are taken from La89.
Two general categories of TRU waste are planned to be disposed in the facility.
Contact-handled (CH) wastes are those with surface dose rates less than 200 mrem per
hour. These wastes, which account for the majority of TRU waste, will be disposed in
steel drums. The drums will be placed in the disposal rooms and stacked in three tiers.
Waste packages with surface dose rates exceeding 200 mrem per hour are classified as
remote-handled (RH) wastes. These wastes can not be handled directly because of the
high radiation levels. Remote-handled wastes will be disposed in special canisters which
are placed in boreholes in the walls of the disposal rooms or drifts. Although the surface
radiation levels of some TRU wastes are high, there is relatively little heat generated by
the radioactive decay so no special facility design features are assumed necessary to
dissipate the small anticipated thermal loading.
Each waste room measures about 92 meters long, 10 meters wide, and 4 meters
high. Seven disposal rooms make up one panel. There are a total of ten panels in the
underground disposal facility. The ten panels are enclosed in an approximately square
7-6
-------
Table 7.5-1. Repository parameters used in TRU waste risk assessment.
Parameter
Dimensions of repository
Length
Width
Height
Repository area
Excavated area
Total mined-out volume
Number of waste drums
Number of waste panels
Number of waste rooms per panel
Waste room dimensions
Length
Width
Height
Value
700m
700m
4 m
490,000 m2
110,000 m2
440,000 m3
583,000
10
7
92m
10m
4 m
Source: La89
7-7
-------
area of 490,000 square meters. The total excavated area is about 110,000 square meters.
After each panel is filled with waste and backfill, the panel is sealed to isolate it
hydrologically from the other waste panels. When all waste is emplaced in the facility,
the vertical access shafts to the waste horizon will be backfilled and sealed.
A total controlled area of 100 square kilometers at the disposal site will provide a
distance of approximately five kilometers from the center of the site to the accessible
environment. For consistency with the repository risk assessments performed for HLW
repositories, the TRU waste risk assessment uses a distance of two kilometers from the
edge of the disposal facility to the accessible environment. For modeling purposes, the
cross-sectional area of the ground-water flow path in the aquifer is defined by the length
of the disposal facility (perpendicular to the flow path) multiplied by the thickness of the
uppermost aquifer. Since the TRU waste disposal facility is approximately square, the
orientation of the facility relative to the aquifer flow direction is not important.
The mined volume of the facility, as well as the porosity of the backfill, must be
considered in calculating the amount of radionuclides that might dissolve in the ground
water that could gradually seep into the disposal facility after its closure. Because such
dissolution might be limited by solubility factors, this water volume is significant to some
models.
7.5.3 Waste Form Parameters
The principal radionuclides in TRU waste presently planned to be emplaced into
the disposal facility are plutonium, americium, and uranium. The waste also contains
smaller amounts of short-lived radionuclides such as strontium and cesium. The TRU
waste inventory used in the generic risk assessment is taken from recent projections of
the waste inventory for the WIPP site. The estimates are documented in the
Environmental Impact Statement for the WIPP facility (DOE89) and in the System
Analysis of the WIPP (La89). The estimates include existing waste in storage at DOE
facilities and waste expected to be generated through the year 2013.
About 96 percent of the TRU waste volume is classified as CH waste. The CH
waste contains about 95 percent of the total radioactivity. The principal radionuclides in
CH waste are, in order of decreasing activity, Pu-238, Pu-241, Am-241, Pu-239, and
Pu-240. The RH waste accounts for about 4 percent of the volume and 5 percent of the
total activity. The principal radionuclides in RH waste are, in order of decreasing
activity, Pu-241, Pu-239, Pu-238, Sr-90, Cs-137, and Pu-240. The total combined
inventory of CH and RH wastes is shown in Table 7.5-2.
It is estimated that the TRU waste inventory will be contained in 385,000 drums
and 19,500 boxes (La89). Of the boxes, 13,500 are Standard Waste Boxes, each with a
volume of 1.78 cubic meters. The remaining 6,000 boxes are of the "old" type. Each "old"
box measures 4x4x7 feet and has a volume of 3.2 cubic meters. Most of the old boxes
(4,500) are wooden and the rest are metal. The total waste volume is equivalent to the
volume of 583,000 drums.
7-8
-------
Table 7-5-2. Radionuclide inventory in the generic TRU waste
repository.
Radionuclide
Sr-90
Cs-137
Th-232
U-233
U-235
U-238
Np-237
Pu-238
Pu-239
Pu-240
Pu-242
Am-241
Cm-248
Initial Quantity in
Repository
(curies)8
70,400
59,600
0.274
7,800
42
22.3
8.02
3,980,000
519,000
136,000
23.3
782,000
0.188
Half-Life
(years)
29
30
1.4E+10
159,000
7.04E+8
4.47E+9
2,140,000
88
24,400
6,540
376,000
432
348,000
Ingestion Dose
Conversion Factor1*
(mrem/Ci)
1.30E+08
4.61E+07
4.77E+08
1.06E+09
l.OOE+09
9.46E+08
4.01E+09
3.85E+09
4.46E+08
4.45E+08
4.24E+08
4.43E+09
1.60E+10
NOTE: For convenience, some radionuclides which were found to be very small
contributors to the total risks were omitted. Omissions are described in the
text.
aSource: La89.
bSource: EPA89.
7-9
-------
The waste containers for the majority of the TRU waste are not designed to
provide radionuclide containment for a long period of time. Under the conditions likely to
be present in the disposal facility after closure, the steel drums would be subject to rapid
corrosion. The wooden boxes would also degrade rapidly because of the presence of
moisture or through bacterial decomposition. In modeling the long-term performance of
the disposal facility, no credit is taken for the integrity of the waste containers. At the
time of disposal, all of the TRU waste inventory is assumed to be in contact with moisture
and to begin being released from the waste form.
The TRU waste is such that the present waste form is not expected to limit the
releases of radionuclides. Radionuclide release is assumed to be controlled by solubility.
At the bedded salt site, for example, the brine in the pore spaces of the salt can be highly
corrosive (La89), so radionuclide solubilities ha the brine could be quite high. The
solubility is assumed to be the same for all radionuclides. A solubility of l.OE-06
mole/liter is used in the base case analysis (La89) for all sites and other values are
evaluated in the sensitivity study.
Some of the radionuclides included in the waste inventories in DOE89 and La89
were not included in the risk assessment because of their short half-lives. On the basis of
half-life, Co-60, Ru-106, Sb-125, Ce-144, and Eu-155 have been eliminated from the risk
assessment. The remaining short-lived radionuclides, Pu-241, Cm-244, and Cf-252, decay
to long-lived radionuclides that must be included in the assessment.
The inventory shown in the table was used in NEFTRAN-S computer calculations.
The NEFTRAN-S code has the capability to perform the ingrowth calculations.
7.5.4 Release Mechanism
In this analysis, only undisturbed normal ground-water flow is analyzed. The
results quantify the individual dose and ground-water protection levels.
7.5.4.1 Normal Ground-Water Flow
All scenarios involving ground water are modeled using a Darcian flow system.
The ground-water transport pathways all involve a vertical and a horizontal leg. The
vertical leg is from the disposal facility vertically to an aquifer. The horizontal leg is the
distance from the edge of the repository to the accessible environment. The five values
needed to predict Darcian flow for each leg are distance, hydraulic conductivity, porosity,
hydraulic gradient, and cross-sectional area. The first four are used to find travel time by
the expression:
T = (d-p)/(i-K)
where:
T is the fluid travel time (years)
d is the length of the leg (meters)
7-10
-------
p is the effective porosity
i is the hydraulic gradient
K is the hydraulic conductivity (meters/years)
Volumetric flow of water is found by:
V = K-i-A
where:
V is the volumetric flow (cubic meters/year)
i is the hydraulic gradient, and
A is the cross-sectional area of the pathway (square meters)
Additional equations used to implement the conceptual models are discussed in the
computer code manual for NEFTRAN-S (SAND90). Specific parameter values
characterizing the disposal facility are presented later in this section.
Unlike disposal facilities in basalt, granite, and tuff, a disposal facility in salt will
have no normal ground-water flow through the undisturbed host rock. During the
construction and operation of the disposal facility, water in the surrounding rock would be
expected to gradually drain so that the rock will enter an unsaturated condition near the
openings. After the end of the operational period and sealing of the disposal facility,
water would be expected to gradually seep back into pores and fractures in the rock. The
creep closure of the salt will cause the hydraulic conductivity near the waste horizon to
gradually decrease back to its value before excavation of the waste facility. This is
assumed to effectively prevent any flow of water from the disposal facility, except in cases
where inadvertent human intrusion or faults provide high permeability flow paths.
7.5.4.2 Gaseous Releases of Radionuclides
Some waste disposal sites present the possibility that radionuclides may be
released in gaseous form. For gaseous release to occur the geologic medium must be
porous and unsaturated. The presence of air-filled pore spaces allows diffusive and/or
advective transport of the radioactive gases. Decomposition of organic material in TRU
waste may produce gases such as methane, hydrogen, hydrogen sulfide, or other gases.
However, these gases are not likely to contain radioactive isotopes. Since TRU
radionuclides are not likely to exist in gaseous form, the transport of radioactive gases
from the disposal facility is not considered a viable release mechanism.
7.5.5 Generic Site Media Analyzed
Generic site models have been developed for four geological media: bedded salt,
basalt flows, unsaturated volcanic tuff formations, and granite. Three of the four generic
site models were developed based on actual sites determined to be representative of the
media: (1) the bedded salt deposits in the Palo Duro Basin in Texas and the Paradox
Formation in Utah, (2) the basalt flows on the Hanford reservation in Washington, and (3)
7-11
-------
the unsaturated volcanic tuff formations at Yucca Mountain in Nevada. No specific
granitic formations were used as the basis for the granite model, but rather the granite
moc el presented here attempts to roughly represent geological and hydrological conditions
tha^ might be encountered in certain regions of the country: (1) the north-central portion
of the United States, and (2) the New England area (EPA85).
The four generic conceptual site models are configured similarly (Figure 7.5-2). A
number of parameters are used to describe the geologic and hydrologic conditions assumed
fo: each of the model repository sites. These are included in the description of the
analysis performed for each generic site medium. The conceptual framework of the
li hology for the basalt and salt sites is that the repository horizon is situated between an
"upper aquifer" and a "lower aquifer." The tuff model assumes unsaturated conditions at
the repository level with only a lower aquifer. Since it is assumed that in the granite
model the repository is located within a granitic plutonic body, there is only an upper
aquifer in the granite model.
To simulate conditions present at a real site, the aquifers do not represent single
hydrostratographic units but rather they represent "synthetic aquifers" whose properties
are defined to approximate the combined properties of a number of transmissive units
above and below the repository horizon. For example, if a number of such transmissive
units are present above the repository at a particular site and if the application of a
generic model described here is intended to represent conditions similar to those at the
site, then one can calculate the combined volumetric flows in the upper units and define
appropriate hydrologic parameters so that the equivalent aquifer conveys the same total
flow. Similarly, by varying one or more additional parameters, it is possible to simulate
the effective fluid velocity in any one of the actual units. This will be illustrated in
subsequent sections when specific lithologies are discussed.
The ground water pathway in the generic risk analyses is modeled with the
NEFTRAN-S code. At saturated sites the upper aquifer is assumed to be the aqueous
pathway of radionuclide transport. An upward gradient is assumed to exist between the
repository horizon and the upper aquifer. Thus, greater emphasis is generally placed on
the properties of the upper aquifer. At potential repository sites, however, the
hydrogeologic environment may be different from that assumed in the generic model. For
example, there may be no significant aquifer below the repository (as in a number of
crystalline rock sites), or above the repository (as in the case of a repository in the
unsaturated zone), or there may be a prevailing gradient that is downward from the
upper aquifer, in which case the lower aquifer would appear to be the more likely release
pathway. These cases can all be accommodated within the modeling of NEFTRAN-S.
The four generic conceptual site models are discussed in the following four sections.
Each section presents a description of the conceptual site model in terms of the
parameters used to evaluate the model through the NEFTRAN-S code. The results of the
individual dose and ground water protection assessments and sensitivity analyses are also
presented in each section. The results of the four sets of analyses are compared in
Section 7.5.6.
7-12
-------
SURFACE DEPOSITS
UPPER AQUIFER
UPPER CONFINING BEDS
LOWER CONFINING BEDS
LOWER AQUIFER
BASEMENT ROCKS
RAE-104250
Figure 7.5-2. General cross-sectional structure for risk analysis.
7-13
-------
7.5.5.1 Site Analysis - Basalt
7.5.5.1.1 Introduction
Basalt deposits in the Pacific Northwest have been under investigation for a
number of years as potential host formations for a nuclear waste repository. Basalt is a
dense, dark, fine-grained rock formed by the solidification of volcanic lava. The basalt
deposits in the Northwest are flood basalts. They were extruded over extremely large
areas and formed a layered structure of individual flows tens to hundreds of feet thick,
separated by relatively minor sedimentary deposits and fractured or highly porous zones
between the basalt flows. The dense interiors of the basalt flows are the potential
repository host rocks evaluated in this section. Basalt deposits are permeated by
fractures, but at the depths being considered for a repository, these fractures are expected
to be quite tightly closed, thereby restricting the volume and the velocities of any
ground-water movement. Nevertheless, there is expected to be some ground-water
migration through a basalt repository and it is possible this might be accelerated by
repository-induced effects. The layered structure of the basalt deposits provides for
horizontal ground-water movement through relatively permeable zones between flows, In
addition, the fracturing in a basalt deposit is expected to be somewhat greater than that
in a well-chosen repository site in granite. This does not mean that such fracturing would
necessarily lead to unacceptable repository performance, but only that it must be an
important consideration in choosing a site and in estimating the performance of a
repository at that site.
Before 1987, the Department of Energy had investigated the possibility of siting a
repository in basalt at the Hanford Reservation in southeastern Washington State. The
relatively advanced stage of the Department of Energy investigations at Hanford has
provided considerable data on the characteristics of potential sites and repository host
flows. However, much of the work carried out at the Hanford Reservation had been the
subject of severe criticism by the Nuclear Regulatory Commission and others, and
therefore the Agency incorporated into its analyses input not only from the Department of
Energy and its contractors but also from technical professionals from other organizations.
Based on such data, the Agency believes that it is possible to define conceptual models of
a basalt repository that should be adequate to make rough approximations of the potential
performance that might be expected from such repositories and to identify some of the
parameters that are most critical in determining that performance.
Section 7.5.5.1.2 discusses the important input parameters that have been used in
the Agency's risk analyses for basalt. The data are based primarily on the Hanford site.
Since most of those data can best be presented in the form of tables and figures, there is a
minimum of text discussing additional details in this section. Also presented are data
from the EPA population risk assessment (EPA82). Section 7.5.5.1.3 provides the results
of the "base case" analyses of individual risks and ground water contamination. Section
7.5.5.1.4 provides the results of the sensitivity and uncertainty analyses.
7-14
-------
7.5.5.1.2 Input Parameters for Basalt
The conceptual model developed to support the evaluation of TRU waste disposal
at a generic basalt site is depicted in Figure 7.5-3. A conceptual model for a generic
basalt site was originally described in "Population Risks from Disposal of High-Level
Radioactive Wastes in Geologic Repositories" (EPA82). The original conceptual model was
modified based on characteristics of the Hanford site, as described in Appendix A of "Risk
Assessment of Disposal of High-Level Radioactive Wastes in Geologic Repositories"
(EPA85).
The geologic and hydrogeologic parameters which define the model are given in
Table 7.5-3. The table provides values for the parameters which are required as input to
the NEFTRAN-S code. The table first gives the parameter values used in the EPA
evaluation of population risks (EPA82). The table then gives the parameter values
obtained from the description of the Hanford site (EPA85). Finally, the table gives the
parameter values used in the current evaluation.
The aquifer parameters were taken from the previous generic analyses performed
in 1980 (EPA82). The parameter values which differ from those in the 1980 risk
assessment include the conductivity, porosity, and vertical gradient in the host rock.
These values differ significantly from the values used in the 1980 assessments and are
based on site characterization data collected by DOE at the Hanford reservation.
Geochemical parameters are also necessary to evaluate the transport of
radionuclides through geologic media. For each radionuclide in the waste inventory,
retardation values are required. These values are dependent on the geologic medium in
which the waste is disposed. The retardation values used in the analysis of the basalt
site are given in Table 7.5-4.
Releases from the source were characterized in terms of radionuclide solubility in
ground water. The solubilities used for the basalt assessments are shown in Table 7.5-5.
A single set of waste form and repository configuration parameters was assumed
for all sites modeled. These parameters include the radionuclide inventory and the
dimensions and capacity of the underground repository facility. These parameters are
discussed for all sites in Section 7.5.2 and 7.5.3.
Analyses were conducted to evaluate sensitivities and uncertainties in the
parameter values. In the sensitivity studies, single parameters were varied discretely
from the base case values. In the uncertainty analysis, statistical distributions were
defined for the key input parameters and those parameters were varied in a Monte Carlo
analysis. Three key parameters were identified for the sensitivity and uncertainty
analyses. The parameters characterize the release from the waste form and the rate of
transport through the ground-water system. The specific parameters selected for the
analyses are the radionuclide solubilities, the vertical hydraulic conductivity in the host
rock, and the radionuclide retardation factors. While other related parameters could have
been included in the sensitivity and uncertainty analyses, those identified represent the
7-15
-------
Figure 7.5-3. Cross-sectional structure for basalt repository
(not to scale).
7-16
-------
Table 7.5-3. Site parameters used in the individual dose and groundwater
protection assessment of basalt.
Parameter
Average porosity of
backfill in repository
Distance from
repository to overlying
aquifer (meters)
Hydraulic conductivity
of the host rock
between the repository
and the aquifer, after
thermal effects
(meters/yr)
Porosity of the host
rock between the
repository and aquifer
Hydraulic gradient
between the repository
and aquifer
Thickness of aquifer
(meters)
Hydraulic conductivity
of the aquifer
(meters/yr)
Porosity of the aquifer
Horizontal gradient in
aquifer
Horizontal distance
along the aquifer to the
accessible environment
(meters)
Early Generic EPA
Model"
0.2
100
0.00032
0.0001
0.025
30
31.5
0.15
0.01
1600
Hanford Site*
Not available
20
0.032
0.0001
0.014
30
315
0.01
0.0003
2000
Current Model
0.2
20
0.032
0.0001
0.014
30
31.5
0.15
0.01
2000
aEPA-520/3-80-006 (EPA82)
bEPA-520/l-85-028 (EPA85)
7-17
-------
Table 7.5-4. Radionuclide retardation factors for basalt.
Element
Strontium
Cesium
Lead
Radium
Actinium
Thorium
Protactinium13
Uranium
Neptunium
Plutonium
Americium
Curium
Range of Retardation Factors*
Low
50
100
20
50
20
500
20
20
10
100
60
100
"Base Case"
200
1,000
50
500
50
5,000
50
50
100
500
500
500
High
2,000
10,000
500
5,000
1,000
10,000
1,000
1,000
500
5,000
50,000
10,000
aFrom 1983 WISP report (NAS83).
TJecause values were not given in WISP report, values of uranium were used based on
chemical similarities.
7-18
-------
Table 7.5-5. Radionuclide solubilities for basalt.8
Nuclide
Ac-227
Am-241
Cm-248
Cs-137
Np-237
Pa-231
Pb-210
Pu-238
Pu-239
Pu-240
Pu-242
Ra-226
Sr-90
Th-229
Th-230
Th-232
U-233
U-234
U-235
U-236
U-238
Solubility (Ci/m3)
1.64E+01
8.28E-01
1.06E-03
1.19E+01
1.67E-04
1.09E-02
1.60E+01
4.08E+00
1.49E-02
5.47E-02
9.51E-04
2.24E-01
1.23E+01
4.87E-02
4.65E-03
2.55E-08
2.26E-03
1.46E-03
5.08E-07
1.53E-05
8.01E-08
aBased on l.OE-06 mole/liter (La89).
7-19
-------
key parameters for characterizing the magnitude of the radionuclide releases and the
transport through the host rock and aquifer.
Ranges of retardation factors are given in NAS83. In the risk assessments, these
ranges have been extended to include minimum retardation factors of one. A retardation
factor of one represents a limiting case in which the site geochemistry is such that the
repository host formation and the aquifer provide no radionuclide retardation. These
conditions provide a bounding analysis of disposal system performance and indicate the
importance of retardation in EPA's modeling of generic repository performance.
The parameter ranges are shown in Table 7.5-6. The ranges encompass the values
used in previous Agency assessments. The probability distributions are given for use in
the NEFTRAN-S uncertainty analysis. Due to the wide range of values, log uniform
distributions were used for all of the parameters. This is preferable to using uniform
distributions because the use of log uniform distributions causes the median values of the
parameters to be closer to their base case values and is therefore more appropriate for
parameters that vary over several orders of magnitude.
7.5.5.1.3 Base Case Results from the Assessment of the Generic Basalt Site
Figure 7.5-4 shows the results of the deterministic assessment of individual dose
versus time using the NEFTRAN-S computer code. The analysis assumes undisturbed
ground-water flow vertically through the repository horizon to the upper aquifer and then
laterally through the aquifer. The assessment assumes an individual drinking water
consumption of 2 liters per day at a point 2000 meters down gradient. Sensitivity of
individual dose to solubility, retardation, and hydraulic conductivity are discussed in
Section 7.5.5.1.4.
No radionuclides reach the 2000-meter boundary prior to approximately year
50,000. Thus, individual dose prior to year 50,000 is zero. At approximately year 50,000,
the most mobile radionuclides, with retardation factors of 50, reach the 2000-meter
boundary. Also, some of the radioactive decay products arrive at this time. Dose
increases abruptly to approximately 1080 mrem/yr. The rapid increase is due to the
relatively low dispersivity used in the model. A higher dispersivity would have led to a
more gradual increase in the dose. Major contributing radionuclides include U-233 (650
mrem/yr), Pa-231 (150 mrem/yr), Ac-227 (130 mrem/yr) and U-234 (100 mrem/yr). Dose
remains fairly constant until year 93,000, when the arrival of Np-237 abruptly increases
the dose to 17,000 mrem/yr.
Ground water protection was evaluated through three measures. First is the
concentration of Ra-226. Second is the total concentration of all alpha-emitting
radionuclides, excluding radon. Third is the drinking water dose resulting from all beta
and gamma-emitting radionuclides. Each of these measures was evaluated through the
NEFTRAN-S analysis.
Ra-226 is part of the Pu-238 decay series. Figure 7.5-5 shows the concentration of
Ra-226 as a function of time, calculated 2000 meters down gradient. Radium first
7-20
-------
Table 7.5-6. Parameter ranges and distributions for basalt.
Parameter
Solubility
(mole/liter)
Vertical hydraulic
conductivity (m/yr)
Retardation factors
Minimum
l.OE-09
3.2E-04
(a)
Maximum
l.OE-03
3.2E-01
(b)
Distribution
Type
Log Uniform
Log Uniform
Log Uniform
^he sensitivity and uncertainty analyses used retardation factors of one, as well as the "low"
values from Table 7.5-4.
bSee Table 7.5-4.
7-21
-------
§
TJ
'5
1
1.00E+08
1.00E+07
1.00E+06 -
1.00E+05
1.00E+04 -
1.00E+03
1.00E+02
1.00E+01 •
1.00E+00
100
Base case
1,000 10,000
Time (years)
100,000
RAE -104643
Figure 7.54. Individual dose for basalt.
-------
1.00E+05
1.00E+04 -
"C"
{« 1.00E+03 -
£ 1.00E+02 -
O 1.00E+01 -
•g 1.00E+00
0>
C 1.00E-01
O
° 1.00E-02 -
1.00E-03
10,000
Ra-226 - base case
100,000
Time (years)
RAE -104647
Figure 7.5-5. Ra-226 groundwater concentrations - base case basalt.
-------
appears at approximately 50,000 years. Its concentration increases sharply to
approximately 0.3 pCi/liter and remains fairly constant for the remainder of the
100,000-year simulation period.
Figure 7.5-6 shows the total concentration of alpha-emitting radionuclides as a
function of time, calculated 2000 meters down gradient. Concentration is zero until
50,000 years, when it rises sharply to almost 1100 pCi/liter. Major contributors to the
total concentration are U-233 (866 pCi/liter), U-234 (132 pCi/liter), Pa-231 (54 pCi/liter)
and U-236 (21 pCi/liter). Concentration remains fairly constant until year 93,000, when
the arrival of Np-237 begins to increase the total concentration to 6600 pCi/Hter at year
96,000.
The total concentration of beta-emitting radionuclides is measured in terms of the
dose which would result from the consumption of two liters per day of the contaminated
ground water. There are four beta-emitting radionuclides: Sr-90, Cs-137, Ac-227
(generated through the decay of Pu-239 and U-235), and Pb-210 (generated through the
decay of Pu-238). As shown in Figure 7.5-7, the total dose is zero until 50,000 years. The
dose then increases and remains at 100 mrem/yr to 200 mrem/yr. The dose results
mainly from the concentration of Ac-227, although Pb-210 contributes somewhat. Sr-90
and Cs-137 do not contribute to the dose because of their short half-lives. The dose from
beta-emitting radionuclides is less than 20 percent of the dose from all radionuclides
shown in Figure 7.5-4.
7.5.5.1.4 Sensitivity and Uncertainty Analyses for the Generic Basalt Assessments
The previous section discussed the results of evaluating individual doses and
ground water concentrations using the base case parameter values given in Table 7.5-3.
This section discusses the sensitivity of individual dose and ground water concentrations
to variations in radionuclide solubility, hydraulic conductivity in the vertical transport leg,
and radionuclide retardation factors.
Individual Dose - Radionuclide solubility controls the rate at which radionuclides
enter into the ground water flow. Higher solubilities result in higher concentrations of
radionuclides per unit of water. Figure 7.5-8 shows the sensitivity of individual dose to
variations in solubilities. The base case solubility was l.OE-06 mole/liter. Individual
doses were calculated with higher (l.OE-03 mole/liter) and lower (l.OE-09 mole/liter)
solubilities. Varying the solubility does not effect the time of arrival of the first measured
dose. It does, however, significantly affect the magnitude of the dose. Increased solubility
results in a much greater and sharper initial dose. The magnitude of this peak dose is
approximately 2.0E05 mrem/yr. At high solubility the dose falls off more rapidly with
time due to depletion of the inventory.
Figure 7.5-9 shows the effect of varying the hydraulic conductivity of the basalt in
the vertical transport leg. Increasing the vertical hydraulic conductivity increases the
volume of flow through a given cross-sectional area and decreases the travel time. Since
the vertical distance from the repository to the aquifer is only 20 meters, the decrease in
the travel time is negligible. The increased flow, however, results in a greater release of
7-24
-------
s
s-
-------
s>
1.00E+05
1.00E+04
§ 1.00E+03
•o
1
•o
> 1.00E+02
1.00E+01
10,000
Total beta - base case
100,000
Time (years)
RAE -104655
Figure 7.5-7. Total beta groundwater dose - base case basalt.
-------
1.00E+08
1.00E+07
^ 1.00E+06
| 1.00E+05 i
$ 1.00E+04 t
Q
n 1.00E+03 -
•O
> 1.00E+02 -f
•o
~ 1.00E+01 t
1.00E+00
100
1,000 10,000
Time (years)
l\
II
II
II
100,000
Base Case
High Solubility
Low Solubility
RAE -104646
Figure 7.5-8. Sensitivity of dose to solubility - basalt.
-------
~4
t
1.00E+08
^ 1.00E+07 -
1 1.00E+06 i
£
£ 1.00E+05
fl>
O 1.00E+04 f
Q
1 1.00E+03 -•
> 1.00E+02 -
'•&
- 1.00E+01 -
1.00E+00
100
1,000 10,000
Time (years)
100,000
• Base Case
High Conductivity
Low Conductivity
RAE -104644
Figure 7.5-9. Sensitivity of dose to vertical hydraulic conductivity - basalt.
-------
radioactivity from the repository, and thus an increased dose. Decreasing the hydraulic
conductivity several orders of magnitude has a more significant effect on the time of the
initial dose, as well as on the magnitude of the dose.
Variations in radionuclide retardation have the greatest effect on individual dose
(Figure 7.5-10). Decreasing the retardation factors increases the mobility of the
radionuclides and thus decreases the travel time to the 2000-meter boundary. Thus, with
lower retardations, the Np-237 peak arrives much earlier, at 11,000 years. All the Np-237
has passed by year 40,000. The retardations, however, are not low enough to see the
arrival of plutonium, americium, or curium in 100,000 years. With no retardation,
however, all radionuclides are transported at the same velocity. Thus, doses arrive much
earlier, at 1,000 years, and doses are much greater due to the contributions from
plutonium, americium, and curium.
In addition to the deterministic sensitivity studies of individual dose, a
probabilistic uncertainty study was conducted. For the three parameters of interest -
solubility, vertical hydraulic conductivity, and retardation - parameters ranges were
assigned instead of single values. The ranges used for the three parameters are given in
Table 7.5-6. Two analyses were made: one using the low retardation values (Table 7.5-4)
as a minimum and a second assuming zero retardation. Using the Monte Carlo sampling
routine of the NEFTRAN-S code, peak doses were calculated for the 10,000-year period
following disposal. No dose was reported in the low-retardation analysis. In actuality,
this means there is a very low probability of dose in 10,000 years, given the parameter
uncertainty. The results of zero minimum retardation analysis are shown by the
histogram in Figure 7.5-11. Considering the parameter uncertainty as represented by the
input parameter ranges, there is a 0.12 probability of zero dose in 10,000 years. However,
these results, which include zero retardation, represent bounding conditions for repository
performance. They are included to represent the importance of retardation in EPA's
model of generic repository performance.
Ra-226 Concentrations - The sensitivities of Ra-226 concentrations in ground water
2000-meters down gradient are similar to those described for individual doses. Figure
7.5-12 shows that decreased solubility results in decreased concentrations, while increased
solubility results in increased concentrations. At high solubility the Ra-226 concentration
falls off rapidly due to depletion of the source. As shown in Figure 7.5-13, increasing the
hydraulic conductivity in the vertical leg has little effect on the initial arrival time but a
significant effect on the magnitude of the concentration of Ra-226, due to increased flow.
Decreasing the hydraulic conductivity has a more significant effect on the arrival time.
Finally, varying the retardation of all radionuclides has a significant effect on both the
arrival time and the magnitude of Ra-226 concentrations, as shown in Figure 7.5-14.
Total Alpha Concentrations - The sensitivities of concentrations of alpha-emitting
radionuclides in ground water 2000 meters down gradient are also similar to those
described for individual doses. Figure 7.5-15 shows that decreased solubility results in
decreased concentrations, while increased solubility results in increased concentrations.
Figure 7.5-16 shows that increasing the hydraulic conductivity in the vertical leg has little
effect on the initial arrival time but a significant effect on the magnitude of the
concentration, due to increased flow through the repository. Decreasing the hydraulic
7-29
-------
l.UUb+Utf -
^ 1.00E+07 -
^ 1.00E+06
£
E, 1.00E+05 -
o
0 1.00E+04 -
O
§ 1.00E+03 -
•o
| 1.00E+02 -
1.00E+01 -
•i nnPo/vt .
r\
\ -A.
/ \
V
1 - ., 1. . .- ...
..,„
I
Base Case
Low Retardation
No Retardation
100
1,000 10,000
Time (years)
100,000
RAE -104645
Figure 7.5-10. Sensitivity of dose to retardation - basalt.
-------
INDIVIDUAL DOSE (mrem/yr)
(12% of the samples show zero discharge)
RAE-104686
Figure 7.5-11. Distribution of individual dose due to parameter
uncertainty (zero miTiiTtmm retardation) - basalt.
7-31
-------
1.00E+04 -
•C- 1.00E+02 -
g 1.00E+00
7T 1.00E-02 -
0
|5 1.00E-04 '
§ 1.00E-06 -
O
O 1.00E-08 '
1 nnir.-tn .
\
! /-"
jf
Ra-226 - base case
Ra-226 - high solubility
""" Ra-226 - low solubility
10,000
100,000
RAE -104649
Time (years)
Figure 7.5-12. Sensitivity of Ra-226 groundwater concentrations to solubility - basalt.
-------
1.00E+03 -
^ 1.00E+03 -
£ 1.00E+03 -
0; 1.00E+03 -
0 1.00E+03 -
*5 1.00E+03 --
-------
1.00E+05
1.00E+04 -
| 1.00E+03 -
O 1.00E+02 -
CL
§ 1.00E+01 -
2 1.00E+00 -
c
0 1.00E-01 -
c
O
O 1.00E-02 -
1 nnp.n3 -
: ^ — !.._„.
/
/
/
/
i
i
i
i
i
i
i i
Ra-226 - base case
Ra-226 - low retardation
Ra-226 - no retardation
100
1,000 10,000
Time (years)
100,000
RAE -104650
Figure 7.5-14. Sensitivity of Ra-226 groundwater concentrations to retardation - basalt.
-------
^
a
i.uuc+uo -
^ 1.00E+07 -
£ 1.00E+06 -
0. 1.00E+05
0 1.00E+04 -
2
£ 1.00E+03 -
C 1.00E+02 -
° 1.00E+01 -
1 nnfjjvi .
'
.
\
\
V (*
s
10,000
Time (years)
Total alpha • base case
Total alpha • high solubility
— Total alpha - low solubility
100,000
RAE -104653
Figure 7.5-15. Sensitivity of total alpha groundwater concentrations to solubility - basalt.
-------
1.00E+08
1.00E+07
O. 1.00E+05
O 1.00E+04 -
«
|j 1.OOE+03 4-
o>
g 1.00E+02 4-
o
o
1.00E+01
1.00E+00
10,000
100,000
Time (years)
• Total alpha - base case
• Total alpha - high
conductivity
Total alpha - low
conductivity
RAE -104652
Figure 7.5-16. Sensitivity of total alpha groundwater concentrations to vertical
hydraulic conductivity - basalt.
-------
conductivity has a more significant effect on the arrival time. Finally, varying the
retardation of all radionuclides has a significant effect on both the arrival time and the
magnitude of the concentration of alpha-emitting radionuclides, as shown in Figure
7.5-17.
Total Beta and Gamma Concentrations - The sensitivities of dose from beta and
gamma-emitting radionuclides 2000 meters down gradient are shown in Figures 7.5-18
through 7.5-20. The only radionuclides in this category are Ac-227 and Pb-210, both of
which are decay products from other radionuclides in the inventory. Although Sr-90 and
Cs-137 are beta-emitters, they do not contribute to the concentrations because of their
short half-lives. Figure 7.5-18 shows that decreased solubility results in decreased
concentrations and thus dose, while increased solubility results in increased
concentrations and dose. As shown in Figure 7.5-19, increasing the hydraulic conductivity
in the vertical leg has little effect on the initial arrival time but a significant effect on the
magnitude of the dose, due to increased flow. Decreasing the hydraulic conductivity has a
more significant effect on the arrival time. Finally, varying the retardation of all
radionuclides has a significant effect on both the arrival time and the magnitude of the
dose from beta-emitting radionuclides (Figure 7.5-20).
7.5.5.2 Site Analysis - Granite
7.5.5.2.1 Introduction
Granitic rocks are widely distributed throughout the United States and thus offer
the possibility of being found in connection with other desirable characteristics for a
repository site. At depth they can be extremely "tight", the naturally occurring fractures
being kept almost completely closed by the high lithostatic pressure. Mined openings in
granitic rock are expected to be highly stable for well chosen sites and there is
considerable experience in such underground excavations from various kinds of hard rock
mines and tunnels. The likelihood of associated valuable resources is low; when present
they are often limited to veins at the boundaries of the granitic bodies. Water wells are
occasionally drilled into granitic rock but because of the general trend of decreasing
permeability with depth, such wells rarely exceed several hundred feet. An important
distinction between granitic rocks and most of the other host rocks being considered for a
repository is that they are often found as a bedrock formation or as an intrusive plutonic
body, and thus the possibility of extensive aquifers at a depth below the repository is
much less likely. This decreases the possibility of a productive and high pressure source
of water that could cause upward flow and carry radionuclides towards the surface. On
the other hand, the certain presence of fractures and the water saturated condition
expected at depth virtually guarantee that there would be some ground-water movement
through a repository in granite. It may occur at extremely low volumetric flow rates and
velocities, but it would be present and must be taken into account in estimating the
performance of a repository.
The Department of Energy had previously carried out a screening of the entire
United States and had identified the North Central and Northeastern regions of the
7-37
-------
~-4
s
1 .UUCTVU -
1.00E+07
& 1.00E+06 -
£ 1.00E+05 -
•«••
0 1.00E+04 -
S 1.00E+03 •
1.00E+02 -
o
° 1.00E+01
1 nnr^rtn .
r\
•V-,
i , i
J
Total alpha - base case
Total alpha - low retardation
•" Total alpha - no retardation
100
1,000 10,000
Time (years)
100,000
RAE -104654
Figure 7.5-17. Sensitivity of total alpha ground water concentration to
retardation- basalt.
-------
oo
CO
1.00E+05
| 1.00E+04 +
,§
w
O 1.00E+03 +
|
.2 1.00E+02 +
*o
10,000
Time (years)
100,000
• Total beta - base case
Total beta - high solubility
Total beta • low solubility
RAE -104657
Figure 7.5-18. Sensitivity of total beta groundwater dose to solubility - basalt.
-------
1.00E+05
^ 1.00E+04
£
^ 1.00E+03 4
4)
§
•o
75 1.00E+021
1.00E+01
1.00E+00
10,000
100,000
Time (years)
Total beta - base case
Total beta • high
conductivity
— Total beta - low conductivity
RAE -104656
Figure 7.5-19. Sensitivity of total beta groundwater dose to vertical hydraulic
conductivity - basalt.
-------
1.00E+05
1.00E+04
t
s>
o>
O 1.00E+03
TJ
75
TJ
> 1.00E+02
TJ
1.00E+01
100
1,000 10,000
Time (years)
Total beta - base case
Total beta • low retardation
Total beta - no retardation
100,000
RAE -104658
Figure 7.5-20. Sensitivity of total beta groundwater dose to retardation • basalt.
-------
United States as the most likely to contain statable repository environments in granitic
rocks. Based on data collected to date by the Department of Energy and others, it is
possible to define certain idealized conceptual models of repository sites in each of the two
regions so as to make first approximations of the potential performance of such
repositories and to identify parameters that are most critical to long-term performance.
This section contains a summary of models and parameters used by the Agency as
"generic" sites that are based on simplified models of the general geologic and hydrologic
conditions reported at promising locations in each of these two regions. It is important to
emphasize that these models are generic in nature, as are all the models of all media
analyzed, and are not intended to represent performance at a particular site.
Section 7.5.5.2.2 discusses the important input parameters that have been used in
the Agency's risk analysis for granite. These parameters are based on data from the
generic North Central site and the generic Northeastern site. Section 7.5.5.2.3 provides
the results of the base case analyses. Section 7.5.5.2.4 provides the results of the
sensitivity analyses and uncertainty analyses.
7.5.5.2.2 Input Parameters for Granite
The conceptual model developed to support the evaluation of TRU waste disposal
at a generic granite site is depicted in Figure 7.5-21. A conceptual model for a generic
granite site was originally described in "Population Risks from Disposal of High-Level
Radioactive Wastes in Geologic Repositories" (EPA82). The original conceptual model was
modified based on characteristics of the generic granite sites in the northeastern and
north-central United States, as described in Appendix A of "Risk Assessment of Disposal
of High-Level Radioactive Wastes hi Geologic Repositories" (EPA85).
The geologic and hydrogeologic parameters which define the model are given in
Table 7.5-7. The table provides values for the parameters which are required as input to
the NEFTRAN-S code. The table first gives the parameter values used in the EPA
evaluation of population risks (EPA82). The table then gives the parameter values
obtained from the descriptions of the generic granite sites in the north-central and
northeastern United States (EPA85). Finally, the table gives the parameter values used
in the current evaluation.
Two sources were used to compile the parameter information shown in the table.
The aquifer parameters were taken from the previous generic analyses performed by the
Agency in 1980 (EPA82). The aquifer parameters are similar for all sites evaluated, thus
emphasizing the performance capabilities of the host rock. The host rock parameters,
including the distance between the repository horizon and the aquifer, hydraulic
conductivity, porosity, and vertical gradient, were based on a study of two actual
representative granite sites in the northeastern and north-central United States (EPA85).
Geochemical parameters are also necessary to evaluate the transport of
radionuclides through geologic media. For each radionuclide in the waste inventory,
retardation values are required. These values are dependent on the geologic medium in
which the waste is disposed. The retardation values used in tne granite analysis are
7-42
-------
SURFACE DEPOSITS
s^^?./;^^^??.->^.^.a^,ivJ>oc=^^??.^^»!V?V
Li . >fc. ilr^ -^.*V ^*_*V .
v/A
•""<** rj
V^Sv*.*
*V
-------
Table 7.5-7. Site parameters used in the risk assessment of granite.
Input Parameter
Average porosity of
backfill in repository
Distance from
repository to
overlying aquifer
(meters)
Hydraulic
conductivity of the
host rock between
the repository and
the aquifer, after
thermal effects
(meters/year)
Porosity of the host
rock between the
repository and
aquifer
Hydraulic gradient
between the
repository and
aquifer
Thickness of aquifer
(meters)
Hydraulic
conductivity of
aquifer (meters/year)
Porosity of the
aquifer
Horizontal gradient
in aquifer
Horizontal distance
along the aquifer to
the accessible
environment (meters)
Early Generic EPA
Model8
0.2
230
3.20E-5
10-4
0.1
30
31.5
0.15
0.01
1600
Representative Granite Sites1*
North Central
Not Available
448
3.2E-2
w-4
0.01
10
5.7
0.018
0.005
0°
New England
Not Available
370
3.2E-2
10"*
0.01
80
315
0.039
0.01
2000
Current
Model
0.2
400
3.2E-2
io-4
0.01
30
31.5
0.15
0.01
2000
aEPA-520/3-80-006 (EPA82)
bEPA-520/l-85-028 (EPA85)
cThe model for this site assumes that surface water on the site is in contact with the ground water; therefore, there
is no horizontal distance to the accessible environment, but only a vertical flow path.
7-44
-------
Table 7.5-8. Radionuclide retardation factors for granite.
Element
Strontium
Cesium
Lead
Radium
Actiniumb
Thorium
Protactinium
Uranium
Neptunium
Plutonium
Americium
Curium
Range of Retardation Factors3
Low
10
100
10
50
10
500
10
10
10
10
500
200
"Base Case"
200
1,000
50
500
50
5,000
50
50
100
200
3,000
2,000
High
2,000
10,000
200
5,000
500
10,000
500
500
500
5,000
50,000
10,000
aFrom 1983 WISP report (NAS83).
Because values were not given in WISP report, values of uranium were used based on
chemical similarities.
7-45
-------
Releases from the source were characterized in terms of radionuclide solubility in
ground water. The solubilities used for the granite assessments are shown in Table 7.5-9.
A single set of waste form and repository configuration parameters was assumed
for all sites modeled. These parameters include the radionuclide inventory and the
dimensions and capacity of the underground repository facility. These parameters are
discussed for all sites in Section 7.5.2 and 7.5.3.
Analyses were conducted to evaluate sensitivities and uncertainties in the
parameter values. In the sensitivity studies, single parameters were varied discretely
from the base case values. In the uncertainty analysis, statistical distributions were
defined for the key input parameters and those parameters were varied in a Monte Carlo
analysis. Three key parameters were identified for the sensitivity analysis. The
parameters characterize the release from the waste form and the rate of transport
through the ground-water system. The specific parameters selected for the analysis are
the radionuclide solubilities, the vertical hydraulic conductivity in the granite host rock,
and the radionuclide retardation factors. While other related parameters could have been
included in the sensitivity and uncertainty analyses, those identified represent the key
parameters for characterizing the magnitude of the radionuclide releases and the
transport through the host rock and aquifer.
The parameter ranges for the granite analyses are shown in Table 7.5-10. The
ranges encompass the values used in previous Agency assessments. The probability
distributions are given for use in the NEFTRAN-S uncertainty analysis. Due to the wide
range of values, log-uniform distributions were used for all of the parameters. This is
preferable to using uniform distributions because a log-uniform distribution causes the
median parameter value to be close to the base case value and is therefore more
appropriate for parameters that vary over several orders of magnitude.
7.5.5.2.3 Base Case Results from the Assessment of Generic Granite Sites
Figure 7.5-22 shows the results of the deterministic assessment of individual dose
versus time for the granite site using the NEFTRAN-S computer code. The analysis
assumes an undisturbed vertical ground-water flow through the repository horizon to the
upper aquifer and then laterally through the aquifer. Dose was evaluated at a point 2000
meters down gradient. The assessment also assumes an individual drinking water
consumption of 2 liters per day. Sensitivity of individual dose to solubility, retardation,
and hydraulic conductivity are discussed in Section 7.5.5.2.4.
No radionuclides reach the 2000-meter boundary prior to approximately year
53,000. Thus, individual dose prior to year 53,000 is zero. At approximately year 53,000,
the most mobile radionuclides, with retardation factors of 50, reach the 2000-meter
boundary. Also, some of the radioactive decay products arrive at this time. Dose
increases abruptly to approximately 570 mrem/yr. The rapid increase in dose is due to
the relatively low dispersivity used as input to the model. A higher dispersivity would
have led to a more gradual increase. From year 53,000 to year 61,000, dose increases to
900 mrem/yr. From year 61,000 to year 100,000, dose increases to 980 mrem/yr. Major
7-46
-------
Table 7.5-9. Radionuclide solubilities for granite.8
Nuclide
Ac-227
Am-241
Cm-248
Cs-137
Np-237
Pa-231
Pb-210
Pu-238
Pu-239
Pu-240
Pu-242
Ra-226
Sr-90
Th-229
Th-230
Th-232
U-233
U-234
U-235
U-236
U-238
Solubility (Ci/m3)
1.64E+01
8.28E-01
1.06E-03
1.19E+01
1.67E-04
1.09E-02
1.60E+01
4.08E+00
1.49E-02
5.47E-02
9.51E-04
2.24E-01
1.23E+01
4.87E-02
4.65E-03
2.55E-08
2.26E-03
1.46E-03
5.08E-07
1.53E-05
8.01E-08
aBased on l.OE-06 mole/liter (La89).
7-47
-------
Table 7.5-10. Parameter ranges and distributions for granite.
Parameter
Solubility
(mole/liter)
Vertical hydraulic
conductivity (m/yr)
Retardation factors
Minimum
l.OE-09
3.2E-05
(a)
Maximum
l.OE-03
3.2E-01
(b)
Distribution
Type
Log Uniform
Log Uniform
Log Uniform
'The sensitivity and uncertainty analyses used retardation factors of one, as well as the "low"
values from Table 7.5-8.
bSee Table 7.5-8.
7-48
-------
-si
fc
1.00E+08
1.00E+07 -
!«.
£ 1.00E+06 -
£
£ 1.00E+05
$ 1.00E+04 -
O
Q
•= 1.00E+03 t
3
;> 1.00E+02
- 1.00E+01 t
1.00E+00
100
Base case
1,000 10,000
Time (years)
100,000
RAE -104627
Figure 7.5-22. Individual dose for granite.
-------
contributing radionuclides at year 61,000 include U-233 (488 mrem/yr), Pa-231 (177
mrem/yr), Ac-227 (152 mrem/yr) and U-234 (68 mrem/yr).
Ground water protection was evaluated through three measures. First is the
concentration of Ra-226. Second is the total concentration of all alpha-emitting
radionuclides, excluding radon. Third is the drinking water dose resulting from all beta
and gamma-emitting radionuclides. Each of these measures was evaluated through the
NEFTRAN-S analysis.
Ra-226 is part of the Pu-238 decay series. Figure 7.5-23 shows the concentration of
Ra-226 as a function of time, calculated 2000 meters down gradient. Ra-226 first arrives
at year 54,000, with a ground water concentration of l.OE-07 pCi/liter. From 54,000 years
to 60,000 years its concentration increases to 5.4E-07 pCi/liter. Then its concentration
increases sharply to approximately 2.0E-02 pCi/liter at year 67,000, and then increases
steadily to 0.7 pCi/liter at the end of the 100,000-year simulation period.
Figure 7.5-24 shows the total concentration of alpha-emitting radionuclides as a
function of time, calculated 2000 meters down gradient. Concentration is zero until
53,000 years. It then rises sharply to almost 790 pCi/liter at year 61,000. The
concentration then decreases slowly but steadily to a concentration of 750 pCi/liter at year
100,000. Major contributors to the total concentration at year 61,000 are U-233 (630
pCi/liter), U-234 (88 pCi/liter), Pa-231 (49 pCi/liter) and U-236 (16 pCi/liter).
The total concentration of beta-emitting radionuclides is measured in terms of the
dose which would result from the consumption of two liters per day of the contaminated
ground water. There are four beta-emitting radionuclides: Sr-90, Cs-137, Ac-227
(generated through the decay of Pu-239 and U-235), and Pb-210 (generated through the
decay of Pu-238). As shown in Figure 7.5-25, the total dose is zero until 53,000 years.
The dose then increases rapidly and then slowly, varying from 100 mrem/yr to 225
mrem/yr. The dose results mainly from the concentration of Ac-227, although Pb-210
contributes somewhat. Sr-90 and Cs-137 do not contribute to the dose because of their
short half-lives. The dose from beta-emitting and gamma-emitting radionuclides is about
20 percent of the total dose from all radionuclides.
7.5.5.2.4 Sensitivity and Uncertainty Analyses in the Generic Granite Assessments
The previous section discussed the results of evaluating individual doses and
ground water concentrations using the base case parameter values given in Table 7.5-7.
This section discusses the sensitivity of individual dose and ground water concentrations
to variations in radionuclide solubility, hydraulic conductivity in the vertical transport leg,
and radionuclide retardation factors.
Individual Dose - Radionuclide solubility controls the rate at which radionuclides
enter into the ground water flow. Higher solubilities result in higher concentrations of
radionuclides per unit of water. Figure 7.5-26 shows the sensitivity of individual dose to
variations in solubilities. The base case solubility was l.OE-06 mole/liter. Individual
doses were calculated with higher (l.OE-03 mole/liter) and lower (l.OE-09 mole/liter)
7-50
-------
en
?
O
*^^*
Concentration i
I.UUC-HJ9 ~
1.00E+04 -
1.00E+03 -
1.00E+02 -
1.00E+Q1 -
1.00E+00 -
1.00E-01 -
1.00E-02 -
1.00E-03 -
1.00E-04 -
1.00E-05 -
1.00E-06 -
1.00E-07 -
1.00E-08 -
10,0
; C
1
\
00 100,0
Ra-226 • base case
Time (years)
RAE-104631
Figure 7.5-23. Ra-226 groundwater concentration - base case granite.
-------
S
to
1.00E+05
o> 1.00E+04 -
& 1.00E+03
1.00E+02
o
8
0 1.00E+01
1.00E+00
10,000
• Total alpha • base case
100,000
Time (years)
RAE - 10463S
Figure 7.5-24. Total alpha groundwater concentrations - base case granite.
-------
£
o
•o
75
1.00E+04
1.00E+03
1.00E-H02 -
g 2 1.00E+01
TJ
1.00E+00
10,000
100,000
Time (years)
Total beta - base case
RAE-104639
Figure 7.5-25. Total beta groundwater dose - base case granite.
-------
1.00E+08
1.00E+07 -
£ 1.00E+06 -
2 1.00E+05 -
$ 1.00E+04-
o
Q
75 1.00E+03-
3
> 1.00E+02 -
tJ
"" 1.00E+01 -.
1.00E400
100
1,000 10,000
Time (years)
100,000
Base Case
Low Solubility
~ High Solubility
RAE -104628
Figure 7.5-26. Sensitivity of dose to solubility - granite.
-------
solubilities. Varying the solubility does not effect the time of arrival of the first measured
dose. It does, however, significantly affect the magnitude of the dose. Increased solubility
results in a much greater and sharper initial dose. The magnitude of this peak dose is
approximately 21,000 mrem/yr. At high solubility the dose falls off more rapidly with
time due to depletion of the inventory.
Figure 7.5-27 shows the effect of varying the hydraulic conductivity of the granite
in the vertical transport leg. Increasing the vertical hydraulic conductivity increases the
volume of flow through a given cross-sectional area and decreases the travel time. Since
the vertical distance from the repository to the aquifer is 400 meters, the decrease in the
travel time due to an increase in vertical hydraulic conductivity is more apparent than in
the basalt analysis. The increased flow through the repository horizon results in a
greater release of radioactivity from the repository, and thus an increased dose.
Decreasing the hydraulic conductivity several orders of magnitude to 3.2E-05 m/yr
resulted in zero dose during the 100,000-year assessment period.
Variations in radionuclide retardation have the greatest effect on individual dose
(Figure 7.5-28). Decreasing the retardation factors increases the mobility of the
radionuclides and thus decreases the travel time to the 2000-meter boundary. Thus, with
lower retardations, the Np-237 peak arrives much earlier, at 10,500 years. All the Np-237
has passed by year 46,000. The plutonium radionuclides, with a lower retardation factor
of 10, arrive with the uranium and neptunium at 10,500 years. The retardations are not
low enough to see the arrival of americium or curium in 100,000 years. With no
retardation, however, all radionuclides are transported at the same velocity. Thus, doses
arrive much earlier, at 1,000 years, and doses are much greater due to the contributions
from plutonium, americium, and curium.
In addition to the deterministic sensitivity studies of individual dose, a
probabilistic uncertainty study was conducted. For the three parameters of interest -
solubility, vertical hydraulic conductivity, and retardation - parameters ranges were
assigned instead of single values. The ranges used for the three parameters are given in
Table 7.5-10. Two analyses were conducted: one using the low retardation values (Table
7.5-8) as a minimum and a second assuming zero as a minimum retardation. Using the
Monte Carlo sampling routine of the NEFTRAN-S code, peak doses were calculated for
the 10,000-year period following disposal. No dose was reported in the low-retardation
analysis. In actuality, this means there is a very low probability of dose in 10,000 years,
given the parameter uncertainty. The results of the zero minimum retardation analysis
are shown by the histogram in Figure 7.5-29. Considering the parameter uncertainty as
represented by the input parameter ranges, there is a 0.62 probability of zero dose in
10,000 years. However, these results, which include zero retardation, represent bounding
conditions for repository performance. They are included to represent the importance of
retardation in EPA's model of generic repository performance.
Ra-226 Concentrations - The sensitivities of Ra-226 concentrations in ground water
2000 meters down gradient are similar to those described for individual doses. Figure
7.5-30 shows that decreased solubility results in decreased concentrations, while increased
solubility results in increased concentrations. As shown in Figure 7.5-31, increasing the
hydraulic conductivity in the vertical leg has a minor effect on the initial arrival time but
7-55
-------
_^
£
"ZT"
1.
¥
o
0
^M
•O
>
1
I.OOt+OB
1.00E+07
1.00E+06 -
1.00E+05
1.00E+04 -
1.00E+03 -
1.00E+02 -
1.00E+01 -
1.00E+00 .
*•
1 — i,,.. , ...
f
' < i)
^
100
1,000 10,000
Time (years)
Base Case
High Conductivity
100,000
RAE -104629
Figure 7.5-27. Sensitivity of dose to vertical hydraulic conductivity - granite.
-------
1.00E+07 -
->• 1.00E+06 -
1.00E+05
0 1.00E+04 -
3 1.00E+03 -
TJ
=5 1.00E+02 -
c
1.00E+01 -
1.00E+00 -
1
v****** * *****
/
J
-A.
|_
M
1
\
\
100
1,000 10,000
Time (years)
100,000
Base Case
Low Retardation
No Retardation
RAE-104630
Figure 7.5-28. Sensitivity of dose to retardation - granite.
-------
>.
o
HI
O
HI
DC
H.
HI
>
r-
5
HI
CC
i
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
^-
i_
—
i.
-
'-
'.
~
-
'r
~.
—
r m
- •B_||_JIBIH__^|BH
i i r^ i i i i i i i i i i i i i i
LLJLUUJIJULULL1LLIUJUJ
opppppppp
INDIVIDUAL DOSE (mrem/yr)
(62% of the samples show zero discharge)
RAE-104687
Figure 7.5-29. Distribution of individual dose due to parameter uncertainty
(zero minimum retardation) - granite.
7-58
-------
s
co
i .uuc+uo -
1.00E+07 -
•C- 1.00E+06
J| 1.00E+05 -
g 1.00E+04 -
S 1.00E+03 -
g 1.00E+02 -
'£ 1.00E+01 -
2
g 1.00E+00 -
g 1.00E-01 -
g 1.00E-02 -
° 1.00E-03 -
1.00E-04 -
1.00E-05
X"*"*
I
'/
//
Xs / .-'""'
I //
Ra-226 - base case
Ra-226 - high solubility
Ra-226 • low solubility
10,000
100,000
Time (years)
RAE -104634
Figure 7.5-30. Sensitivity of Ra-226 groundwater concentrations to solubility - granite.
-------
tt
•mm
O
a.
o
0)
o
1.00E+05
1.00E+04
1.00E+03
1.00E+02
1.00E+00
1.00E+05
1.00E+05
1.00E+05
1.00E+05
1.00E+05
1.00E+05
10,000
Ra-226 - base case
----- Ra-226 - high conductivity
100,000
Time (years)
RAE -104632
Figure 7.5-31. Sensitivity of Ra-226 groundwater concentrations to vertical
hydraulic conductivity - granite.
-------
a significant effect on the magnitude of the concentration of Ra-226, due to increased flow.
Decreasing the hydraulic conductivity results in no Ra-226 concentration during the
100,000-year assessment period. Finally, varying the retardation has a significant effect
on both the arrival time and the magnitude of Ra-226 concentrations, as shown in Figure
7.5-32.
Total Alpha Concentrations - The sensitivities of concentrations of alpha-emitting
radionuclides in ground water 2000 meters down gradient are also similar to those
described for individual doses. Figure 7.5-33 shows that decreased solubility results in
decreased concentrations, while increased solubility results in increased concentrations.
Figure 7.5-34 shows that increasing the hydraulic conductivity in the vertical leg has a
minor effect on the initial arrival time but a significant effect on the magnitude of the
concentration, due to increased flow through the repository. Decreasing the hydraulic
conductivity results in no alpha concentration during the 100,000-year assessment period.
Finally, varying the retardation has a significant effect on both the arrival time and the
magnitude of the concentration of alpha-emitting radionuclides, as shown in Figure
7.5-35.
Total Beta and Gamma Concentrations - The sensitivities of dose from beta and
gamma-emitting radionuclides 2000 meters down gradient are shown in Figures 7.5-36
through 7.5-38. Only Ac-227 and Pb-210 contribute to the concentrations. The other two
beta-emitters, Sr-90 and Cs-137, do not arrive due to their short half-lives. Figure 7.5-36
shows that decreased solubility results in decreased concentrations and thus dose, while
increased solubility results in increased concentrations and dose. As shown in Figure
7.5-37, increasing the hydraulic conductivity in the vertical leg has a minor effect on the
initial arrival time but a significant effect on the magnitude of the dose, due to increased
flow. Decreasing the hydraulic conductivity results in zero dose. Finally, varying the
retardation has a significant effect on both the arrival time and the magnitude of the dose
from beta-emitting radionuclides (Figure 7.5-38).
7.5.5.3 Site Analysis - Bedded Salt
7.5.5.3.1 Introduction
For almost 30 years, salt deposits have been considered prime candidates for a
nuclear waste repository. There are a number of reasons for this. Salt deposits are
common in several regions of the United States and they are found at depths considered
to be suitable for a repository. By their very presence, they indicate relative geologic
stability and hydrologic isolation, since if ground water had ready access to them the salt
would have been dissolved and carried away. While it is the case that almost all known
salt beds are undergoing gradual dissolution by ground water, the rates of such
dissolution processes are generally so slow that these deposits are expected to remain
substantially intact for millions of years. In addition, there is extensive experience in
constructing underground mines in salt. Another advantage is that gradual creep of the
salt will aid in the resealing and the reestablishment of total isolation of a repository
placed in such an environment. On the other side, there is the disadvantage that if some
7-61
-------
1 .WUCfUO
1.00E+07 -
^ 1.00E+06 -
0> 1.00E+05 -
£ 1.00E+04
O 1.00E+03 -
3 1.00E+02 -
g 1.00E+01 -
*•• 1.00E+00 -
4= 1.00E-01 -
75
S 1.00E-02 -
g 1.00E-03 -
0 1.00E-04 -
LJ
1.00E-05 -
1.00E-06 -
1.00E-07 -
— —• — """ — 'N
^.— •"""" !.__
/"*
/ .„.."•
I .••"'*'"' j
I f f
I /
I J
• I
I
I
,i , r
100 1,000 10,000 100,0
Ra-226 • base case
Ra-226 - low retardation
Ra-226 - no retardation
Time (years)
RAE -104633
Figure 7.5-32. Sensitivity of Ra-226 groundwater concentrations to
retardation - granite.
-------
1.UUC+UO -
_ 1.00E+07
0>
£ 1.00E+06 -
O
S 1.00E+05 -
c
.2 1.00E+04 -
±2 -1 OOF4.03 -
C I •WkTww
c 1.00E+02 -
O
O
1.00E+01 -
1.QOE4-OQ -
r
\
N
I
I
, \
•*.
10,000
Time (years)
100,000
Total alpha - base case
Total alpha • high solubility
— Total alpha - low solubility
RAE - 10463S
Figure 7.5-33. Sensitivity of total alpha groundwater concentrations to
solubility • granite.
-------
1.00E+05
^ 1.00E+04 -
o
O 1.QOE+03 -
O
i
1.00E+02
O
C 1.00E+01 i
S
10,000
100,000
Time (years)
Total alpha - base case
Total alpha - high
conductivity
RAE -104637
Figure 7.5-34. Sensitivity of total alpha groundwater concentratons to vertical
hydraulic conductivity - granite.
-------
1.00E+07 -
t?
£ 1.00E+06 -
0. l.OOE+05 -
c
.2 1.00E+04 -
t- 1.00E+03 -
0)
C 1.00E+02 -
O
O
1.00E+01 -
LOOE+OO -
I
r\_ „..
f — — — —
1
«*i
\
\
i
i
100
1,000 10,000
Time (years)
100,000
Total alpha - base case
Total alpha - low retardation
Total alpha - no retardation
RAE -104636
Figure7.5-35. Sensitivity of total alpha groundwater concentrations to
retardation - granite.
-------
1.UUC+U4 ~
j= 1.00E+03
2
£
^ 1.00E+02
"w
"O
!> 1.00E+01 -
T»
c
1.00E+00 -
l\
.
_
1 |
1 1
A/-' s.'.r"'i ' '•'"'" *"
f"
f .
Total beta • high solubility
- Total beta - low solubility
10,000
100,000
Time (years)
RAE -104642
Figure 7.5-36. Sensitivity of total beta groundwater dose to solubility - granite.
-------
-q ^
JT 1.00E+03 -
8>
0)
g 1.00E+02 -
•o
"3
3
•5 1.00E+01 -
TJ
••••
1.00E+00 -
f "^^
/ __^ —
1 S
1
1
i
i
1
1
I
1
1
1
/
10,000
Time (years)
Total beta - base case
Total beta - high
conductivity
100,000
RAE -104641
Figure 7.5-37. Sensitivity of total beta groundwater dose to vertical hydraulic
conductivity - granite.
-------
1.00E+08
_ 1.00E+07 -
^ 1.00E+Q6 -
£
£ 1.00E+05
g 1.00E+04
•D
TO 1.00E+03 4
•5 1.00E+02 -
S
£ 1.00E+01 -
100
Total beta»base case
----- Total beta - low retardation
Total beta • no retardation
1,000 10,000
Time (years)
100,000
RAE -104640
Figure7.5-38. Sensitivity of total beta groundwater concentrations to
retardation - granite.
-------
unforeseen circumstances arise that bring ground water into contact with the salt near
the repository, the effects might be severe with the relatively rapid dissolution of the salt.
Also, salt deposits are located in sedimentary basins that often contain other valuable
resources such as oil, gas, and potash. As a result, the adoption of a site for a nuclear
waste repository may either preempt access to the resources present at the site or lead to
future risks from efforts to obtain those resources.
The Department of Energy has investigated potential repository sites in bedded
salt deposits in the Paradox Basin in Utah and in the Palo Duro Basin in Texas. Based
on data collected by the Department and others, it is possible to define conceptual models
of repository sites in each of the two basins so as to make rough first approximations of
the potential performance of such repositories and to identify some of the parameters that
are most critical in determining that performance. The models and parameters used by
the Agency as the "generic" salt site are based on simplified models of the geologic and
hydrologic systems in each of these two basins.
Section 7.5.5.3.2 discusses the important input parameters that have been used in
the Agency's risk analyses for bedded salt. Since most of these data can best be presented
in the form of tables and figures, there is minimal textual discussion of additional details.
Also presented are data from the Agency's population risk assessment (EPA82). Section
7.5.5.3.3 provides the results of the "base-case" analyses. Section 7.5.5.3.4 provides the
results of the sensitivity and uncertainty analyses.
7.5.5.3.2 Input Parameters for Bedded Salt
The conceptual model developed to support the evaluation of disposal at a generic
salt site is depicted in Figure 7.5-39. The Agency's original conceptual model for a generic
salt site was described in "Population Risks from Disposal of High-Level Radioactive
Wastes in Geologic Repositories" (EPA82). The original conceptual model was modified
based on characteristics of the Palo Duro Basin and Paradox Basin salt formations, as
described in Appendix A of "Risk Assessment of Disposal of High-Level Radioactive
Wastes in Geologic Repositories" (EPA85).
The geologic and hydrogeologic parameters, and their values, which define the
generic salt model are given in Table 7.5-11. The table lists the parameter values used in
the EPA evaluation of population risks (EPA82), the parameter values obtained from the
descriptions of the Palo Duro Basin and Paradox Basin sites (EPA85), the parameter
values used in the current evaluation.
Two sources were used to compile the parameter information shown in the table.
The aquifer parameters were taken from the previous generic analyses performed by the
Agency in 1980 (EPA82). The aquifer parameters are similar for all sites evaluated, thus
emphasizing the performance capabilities of the host rock. The host rock parameters,
including the distance between the repository horizon and the aquifer, hydraulic
conductivity, porosity, and vertical gradient, were based on a study of two representative
7-69
-------
SURFACE DEPOSITS
UPPER CONFINING BEOS
30m
900m
SALT
REPOSITORY
LOWER CONFINING BEDS
RAE-104253
Figure 7.5-39. Cross-sectional structure of the model generic
salt repository (not to scale).
7-70
-------
Table 7.5-11. Site parameters used in the risk assessment of bedded salt.
Input Parameter
Initial average
porosity of backfill in
repository
Distance from
repository to
overlying aquifer
(meters)
Hydraulic
conductivity of the
host rock between
the repository and
the aquifer, after
thermal effects
(meters/year)
Porosity of the host
rock between the
repository and
aquifer
Hydraulic gradient
between the
repository and
aquifer
Thickness of aquifer
(meters)
Hydraulic
conductivity of
aquifer (meters/year)
Porosity of the
aquifer
Horizontal gradient
in aquifer
Horizontal distance
along the aquifer to
the accessible
environment (meters)
Early Generic
EPA Model8
0.2
100
0
0.01
0.1
30
31.5
0.15
0.01
1600
Representative Bedded Salt Sitesb
Palo Duro Basin
Not Available
1105
0
n/a
0.26
300
1.6
0.05
0.005
2000
Paradox Basin
Not Available
666
0
n/a
0
18
7.6
0.2
0.02
2000
Current
Model
0.2
900
0
n/a
0.1
30
10
0.15
0.01
2000
aEPA-520/3-80-006 (EPA82).
bEPA-520/l-85-028 (EPA85).
7-71
-------
salt sites in the Palo Duro Basin and the Paradox Basin (EPA85) and on the generic
analysis.
Geochemical parameters are also necessary to evaluate the transport of
radionuclides through geologic media. For each radionuclide in the waste inventory,
retardation values are required. These values are dependent on the geologic medium in
which the waste is disposed. The retardation values used in the salt site analysis are
given in Table 7.5-12.
Releases from the source were characterized in terms of radionuclide solubility in
ground water. The solubilities used for the salt assessments are shown in Table 7.5-13.
A single set of waste form and repository configuration characteristics were
assumed. Many of the parameters in the analysis are common to the generic repositories
in all media analyzed by the Agency including, for example, the radionuclide inventory
and the dimensions and capacity of the underground repository facility. All such
parameters are discussed for all sites in Sections 7.5.2 and 7.5.3.
Analyses were conducted to evaluate sensitivities and uncertainties in the
parameter values. In the sensitivity studies, single parameters were varied discretely
from the base case values. In the uncertainty analysis, statistical distributions were
defined for the key input parameters and those parameters were varied in a Monte Carlo
analysis. Three key parameters were identified for the sensitivity analysis. The
parameters characterize the release from the waste form and the rate of transport
through the ground-water system. The specific parameters selected for the analysis are
the radionuclide solubilities, the hydraulic conductivity in the vertical transport leg, and
the radionuclide retardation factors. While other related parameters could have been
included in the sensitivity and uncertainty analyses, those identified represent the key
parameters for characterizing the magnitude and timing of the radionuclide releases and
the transport through the host rock and aquifer.
The parameter ranges for the salt analyses are shown in Table 7.5-14. The ranges
encompass the values used in previous Agency assessments. The probability distributions
are given for use in the NEFTRAN-S uncertainty analysis. Due to the wide range of
values, log-uniform distributions were used for all of the parameters. This is preferable to
using uniform distributions, since log-uniform distributions cause the median parameter
values to be close to the base case values and are therefore more appropriate for
parameters that vary over several orders of magnitude.
7.5.5.3.3 Base Case Results from the Assessment of Generic Salt Sites
Individual doses and ground water contamination were evaluated for undisturbed
conditions only. Under undisturbed conditions, the hydraulic conductivity of salt is
essentially zero, resulting in no ground water flow. Therefore, under undisturbed
conditions there is no radionuclide release, no dose to individuals, and no contamination
of ground water.
7-72
-------
Table 7.5-12. Radionuclide retardation factors for salt.
Element
Strontium
Cesium
Lead
Radium
Actinium
Thorium
Protactinium
Uranium
Neptunium
Plutonium
Americium
Curium
Range of Retardation Factors8
Low
1
1
5
5
10
300
10
10
10
10
300
200
"Base Case"
10
10
20
50
20
1,000
20
20
50
200
1,000
1,000
High
100
2,000
100
500
60
5,000
60
60
300
10,000
5,000
3,000
aFrom 1983 WISP report (NAS83).
Because values were not given in WISP report, values of uranium were used based on
chemical similarities.
7-73
-------
Table 7.5-13. Radionuclide solubilities for salt.*
Nuclide
Ac-227
Am-241
Cm-248
Cs-137
Np-237
Pa-231
Pb-210
Pu-238
Pu-239
Pu-240
Pu-242
Ra-226
Sr-90
Th-229
Th-230
Th-232
U-233
U-234
U-235
U-236
U-238
Solubility (Ci/m3)
1.64E+01
8.28E-01
1.06E-03
1.19E+01
1.67E-04
1.09E-02
1.60E+01
4.08E+00
1.49E-02
5.47E-02
9.51E-04
2.24E-01
1.23E+01
4.87E-02
4.65E-03
2.55E-08
2.26E-03
1.46E-03
5.08E-07
1.53E-05
8.01E-08
aBased on l.OE-06 mole/liter (La89).
7-74
-------
Table 7.5-14. Parameter ranges and distributions for salt.
Parameter
Solubility
(mole/liter)
Retardation factors
Hydraulic
Conductivity (m/yr)
Minimum
l.OE-09
(a)
0
Maximum
l.OE-03
(b)
3.0E-06
Distribution
Type
Log Uniform
Log Uniform
Log Uniform
aThe sensitivity and uncertainty analyses used retardation factors of one, as well as the "low"
values from Table 7.5-12.
bSee Table 7.5-12.
7-75
-------
7.5.5.3.4 Sensitivity and Uncertainty Analyses for the Generic Salt Assessments
At a salt site, releases are controlled by the ground water flow regime. Therefore,
only variations in the hydraulic conductivity were considered. Neither radionuclide
solubility nor retardation are of concern. Increasing the hydraulic conductivity from zero
to 3.0E-06 m/yr (La89, RAE92), a high value for salt, resulted in a vertical travel time of
30 million years, and consequently no dose or ground water contamination occurs during
the 100,000-year assessment period.
7.5.5.4 Site Analysis - Tuff
7.5.5.4.1 Introduction
Welded tuff has recently received increased attention as a potential host rock for a
high-level waste repository. It is unique among geologic media considered in this analysis
in that the repository horizon at the tuff site is assumed to be unsaturated. This is
because the generic tuff site is modeled after the unsaturated tuff site in southern
Nevada. The welded tuffs that serve as the generic host rock for this analysis consist of
airborne volcanic debris fused into a mass with high porosity and low permeability. They
appear to have the necessary engineering properties for repository construction. Because
the tuff is composed of fragments of porous volcanic rock, the residence time of water
moving through it is relatively long and the likely mineral assemblages are expected to
provide favorable retardation. Tuff shares with granite and basalt a relatively low
occurrence of oil, gas, or valuable minerals that might be exploited by future drilling to
any considerable depth. Similarly, the depth of the water table is a deterrent to the
drilling of water wells or the development of underlying aquifers.
Two distinctive and important features emerge from the analyses conducted to date
by the Agency and Sandia National Laboratories (SAND84-1492) for a repository located
above the water table in an unsaturated zone. First, unlike any other medium, upward
aqueous flow is improbable as long as the rock remains unsaturated. Second, as long as
infi1t.rat.inn at the ground surface is low enough to maintain an unsaturated condition,
water in a flow path such as a fault zone or drill hole may preferentially move into the
matrix pore space by capillary attraction rather than downward along the flow path
(SAND84-1492).
The Department of Energy is currently investigating the area including Yucca
Mountain in southern Nevada as a possible candidate site for a high-level waste
repository. Other tuff sites may be found, but the relative abundance of hydrogeologic
data for this location, coupled with the very low precipitation in the region, make it
appropriate to use the general site characteristics to define a conceptual model of a
repository in unsaturated tuff. Analysis of a tuff repository is a departure from the
Agency's original risk analyses of generic repositories, which did not consider this
lithology. However, a tuff repository was included in the analyses in 1985. The addition
of tuff as a possible isolation medium was presented in 1985 on the basis of its apparent
7-76
-------
performance and the additional insight it brings to evaluating performance of a repository
located in an unsaturated medium.
The preliminary tuff data used in the current analysis are largely derived from
studies of the Yucca Mountain area but should be regarded as representative of a generic
hypothetical site. It is the Agency's opinion that the parameter values used present a
valid but conservative estimate of the performance of a repository in unsaturated tuff.
Section 7.5.5.4.2 discusses the important input parameters that have been used in
the Agency's risk analyses for tuff. Since most of the input parameter data can best be
presented in the form of tables and figures, there is minimal textual discussion of
additional details. The section includes a discussion of the data used to characterize
gaseous releases and transport, a release scenario unique to an unsaturated site. Section
7.5.5.4.3 provides the results of the base case assessment. Section 7.5.5.4.4 presents the
results of the sensitivity and uncertainty analyses.
The conceptual model for the generic tuff site assumes that the waste disposal
horizon will be located above the water table in unsaturated rock. Therefore, the
potential exists for gaseous transport of radionuclides upward through the unsaturated
rock to the surface in addition to the aqueous transport of radionuclides downward to the
aquifer. The generic analysis performed in support of the 1985 standard did not consider
gaseous releases. It has since been realized that gaseous releases may be significant at
an unsaturated site where spent nuclear fuel is disposed. The Agency has investigated
gaseous releases from a tuff site (RAE92a) and has found that C-14 (as carbon dioxide)
and possibly 1-129 are the only radionuclides likely to be released in a gaseous state in
significant quantities. C-14 and 1-129 are not present in the initial TRU waste inventory
and are not generated through the decay of any of the radionuclides in the inventory.
Therefore, gaseous transport of these radionuclides from a TRU waste repository is not
considered.
7.5.5.4.2 Input Parameters for Tuff
The NEFTRAN-S model for tuff assumes downward flow from the repository
through the unsaturated zone to the underlying aquifer. As long as the water infiltration
rate is less than the saturated hydraulic conductivity, flow is driven by gravity and a
downward gradient of one is assumed. Between the repository and the saturated zone,
natural variations in hydrologic properties are simplified to a single set of vertical leg
parameters. Potential releases to the accessible environment are modeled through the
uppermost aquifer, located about 200 meters below the repository.
Hydraulic conductivity is used in conjunction with Darcy's Law to estimate
volumetric flow rates through various components, such as pathways from the repository
down through the unsaturated zone to the aquifer and horizontally within the aquifer.
For further elaboration on the mathematical equations, one may consult EPA82 and the
references cited there. Only Darcian flow has been treated in the analyses and work by
DOE at specific sites tends to confirm that this approach is adequate (SAND90a). The
porosity is used to convert Darcian flow velocities into average effective fluid velocities in
7-77
-------
the direction of movement. In particular, the Darcian flow velocity is divided by the
volumetric moisture content to obtain an effective fluid velocity. This is used to
determine the time of arrival of contaminated ground water at the discharge point to the
accessible environment.
Figure 7.5-40 shows the geologic cross section used to define the simplified model
used in the Agency's analyses for tuff. Table 7.5-15 shows the geometric and hydrologic
input parameters. The radionuclide retardation factors and solubilities are shown in
Tables 7.5-16 and 7.5-17, respectively. The generic assessment of TRU waste disposal in
tuff use the waste form and repository parameters presented in Sections 7.5.2 and 7.5.3.
The parameter values used in the risk assessment were varied in a sensitivity
study. The parameters selected for the sensitivity and uncertainty analyses are listed in
Table 7.5-18, along with the ranges and distribution types. As explained earlier, log
uniform distributions were selected for the radionuclide solubilities, the retardation
factors, and the infiltration rate. The infiltration rate is analogous to the vertical
hydraulic conductivity in the saturated site assessments because it determines the travel
time from the repository to the aquifer.
7.5.5.4.3 Base Case Results from the Assessment of the Generic Tuff Site
Assuming the base case parameter values given in Table 7.5-15, the pore velocity
in the vertical transport leg is approximately 0.013 meters/year. With a distance of 200
meters between the repository and the aquifer, the unretarded travel time in the vertical
leg is approximately 15,600 years. The lowest base-case retardation factor is 40, for
uranium, actinium and protactinium. These radionuclides would be the first to reach the
aquifer, but only after 624,000 years. Thus, there are no doses or radionuclide
concentrations 2000 meters down gradient during the 100,000-year assessment period.
7.5.5.4.4 Sensitivity and Uncertainty Analyses for the Generic Tuff Assessments
Sensitivity analyses were conducted, varying solubility, infi1t.rat.imi rates, and
retardation. No radionuclides travel 2000 meters down gradient in the 100,000-year
assessment period unless retardation factors are reduced from the base case values.
The effect on individual dose of reducing retardation factors is shown hi Figure
7.5-41. Lowering the retardation factors results in a dose of 35 to 45 mrem/yr from the
arrival of radionuclides at year 76,000. The dose results mainly from U-233 (60%),
Pa-231, Ac-227, and U-234. With zero retardation, dose results as early as year 15,600.
Dose increases sharply to 7700 mrem/yr. Americium, neptunium and curium are depleted
early. After year 70,000 dose drops as the plutonium radionuclides are depleted.
A probabilistic assessment of peak dose over 10,000 years was conducted using
NEFTRAN-S and the parameter ranges shown in Table 7.5-18. Two analyses were
conducted: one using the low retardation values (Table 7.5-16) as a minimum on the
7-78
-------
syK'fV--il^--ti?-'Q^-TQ •**
:<^i^i^*£<7lSy3W^^^
WELDED TUFF
WELDED AND NON-WELDED TUFF
200m
200m
2400m
RAE-104254
Figure 7.5-40. Cross-sectional structure of the model generic
tuff repository (not to scale).
7-79
-------
Table 7.5-15. Site parameters used in the assessment of tuff.
Input Parameter
Average porosity of
backfill in repository
Distance from repository
to underlying aquifer
(meters)
Saturated hydraulic
conductivity of the host
rock between the
repository and the
aquifer, after thermal
effects (meters/year)
Infiltration rate (m/yr)
Porosity of the host rock
between the repository
and aquifer
Unsaturated hydraulic
gradient between the
repository and aquifer
Thickness of aquifer
(meters)
Hydraulic conductivity of
aquifer (meters/year)
Porosity of the aquifer
Horizontal gradient in
aquifer
Horizontal distance along
the aquifer to the
accessible environment
(meters)
Distance from repository
to ground surface
(meters)
Early Generic EPA
Model*
0.2
100
0.001
n/a
0.10
1
1,000
30
0.002
0.00034
2,000
—
Current Model
0.2
200
0.004
0.0005
0.06
1
2,400
200
0.002
0.0004
2,000
200
aEPA-520/3-85-028 (EPA85).
7-80
-------
Table 7.5-16. Radionuclide retardation distributions for tuff.
Element
Strontium
Cesium
Lead
Radium
Actiniumb
Thorium
Protactinumb
Uranium
Neptunium
Plutonium
Americium
Curium
Range of Retardation Factors3
Low
20
60
20
50
5
500
5
5
10
50
300
100
"Base Case"
200
500
50
500
40
5,000
40
40
100
200
1,000
500
High
10,000
10,000
500
5,000
200
10,000
200
200
500
5,000
50,000
10,000
aFrom 1983 WISP report.
kValues not given in WISP report; values of uranium were used based on chemical
similarities.
7-81
-------
Table 7.5-17. Radionuclide solubilities for tuff.
Nuclide
Ac-227
Am-241
Cm-248
Cs-137
Np-237
Pa-231
Pb-210
Pu-238
Pu-239
Pu-240
Pu-242
Ra-226
Sr-90
Th-229
Th-230
Th-232
U-233
U-234
U-235
U-236
U-238
Solubility (Ci/m3)
1.64E+01
8.28E-01
1.06E-03
1.19E+01
1.67E-04
1.09E-02
1.60E+01
4.08E+00
1.49E-02
5.47E-02
9.51E-04
2.24E-01
1.23E+01
4.87E-02
4.65E-03
2.55E-08
2.26E-03
1.46E-03
5.08E-07
1.53E-05
8.01E-08
aBased on l.OE-06 mole/liter (La89).
7-82
-------
Table 7.5-18. Parameter ranges and distributions for tuff.
Parameter
Solubility
Infiltration rate
(mm/yr)
Retardation factors
Minimum
l.OE-09
0.1
(a)
Maximum
l.OE-03
4.0
(b)
Distribution
Type
Log Uniform
Log Uniform
Log Uniform
^he sensitivity and uncertainty analyses used retardation factors of one, as well as the "low"
values from Table 7.5-16.
bSee Table 7.5-16.
7-83
-------
1.00E+07
^ 1.00E+06
§ 1.00E+05
h.
T 1.00E+04
8
2 1.00E+03-
co
§ 1.00E+02-
'•5
- 1.00E+01 4
1.00E+00
100
1,000 10,000
Time (years)
100,000
Low retardation
No Retardation
RAE -104659
Figure7.5-41. Sensitivity of dose to retardation - tuff.
-------
range of retardation and a second using a value of one as the minimum retardation
value.Using the Monte Carlo sampling routine of the NEFTRAN-S code, peak doses were
calculated for the 10,000-year period following disposal. The results of the low minimum
retardation analysis are shown by the histogram in Figure 7.5-42. Given the parameter
uncertainties, there is a 0.9 probability of zero dose during the 10,000-year assessment
period. The results of the zero minimum retardation analysis are shown by the histogram
in Figure
7.5-43. Given the parameter uncertainties, there is a 0.62 probability of zero dose during
the 10,000-year assessment period. However, these results, which include zero
retardation, represent bounding conditions for repository performance. They are included
to represent the importance of retardation in EPA's model of generic repository
performance.
Figures 7.5-44 and 7.5-45 show the effects of varying retardation on concentrations
of Ra-226 and total alpha-emitting radionuclides in the ground water. As expected,
reducing retardation results in earlier and greater concentrations. Reducing retardation
also results hi earlier and greater doses from beta and gamma-emitting radionuclides
(Ac-227 and Pb-210) in the ground water (Figure 7.5-46).
7.5.6 Comparison of Media Results
The results of the deterministic calculations for the four repository media are compared in
this section. Radionuclide concentrations and doses were calculated for a point 2,000
meters down gradient from the repository. Individual doses were based on the
consumption of two liters per day of ground water. Using the base case parameter values,
the undisturbed ground water flow scenario shows no individual doses or ground water
contamination at any of the sites during the first 10,000 years after waste disposal. In
addition, the bedded salt and tuff sites show no doses or ground water contamination
during the first 100,000 years.
The radionuclide doses at the basalt and granite sites are zero until about year 50,000.
As shown in Figure 7.5-47 the individual doses at the basalt and granite sites are similar.
Figure 7.5-48 shows the Ra-226 concentration in the ground water at the basalt and
granite sites. The Ra-226 concentration rises more slowly at the granite site, but the
concentration levels are comparable. Figure 7.5-49 shows the total concentrations of
alpha-emitting radionuclides. like the individual doses, the concentrations at the basalt
and granite sites are very similar. The results for the beta and gamma-emitting
radionuclides are shown in Figure 7.5-50. The doses first appear at about year 50,000
and are nearly the same.
7.6 UNCERTAINTY IN THE RISK ASSESSMENT
The generic assessment presented here encompasses many uncertainties which are
due to a number of factors such as the following:
• The long time frame over which predictions are needed;
7-85
-------
o
UJ
1
0.9
0.8
0.7
O 0.6
Ul
£ 0.5
| 0.4
§ 0.3
UJ
OC 0.2
0.1
0
1-
1 1 h
Ul
UJ
P
Ul
o
LU
UJ
o
LU
S
UJ
P
LLI
INDIVIDUAL DOSE (mrem/yr)
(90% of the samples show zero discharge)
RAE-104689
Figure 7.5-42. Distribution of individual dose due to parameter uncertainty
(low minimum retardation) - tuft
7-86
-------
1
0.9
> 0.8
§ 0.7
3
O 06
UJ
£ 0.5
HI
> 0.4
3 0.3
HI
CC 0.2
0.1
0
H h
,5
P
T- CM
Ut UJ
P P
H 1 1 1 1-
-------
1 .UUI.TW«S -
1.00E+04
^
~ 1.00E+03
3 1.00E+02
O
'•»= 1.00E+01 -
0 1.00E+00
O 1.00E-01
1.00E.02 -
-.'*'\
,-•-••""'""""
,'-''
f
1 ! /
1.00E+04
Ra-226 - low retardation
Ra-226 - no retardation
1.00E+05
Time (years)
RAE - 104660
Figure 7.5-44. Sensitivity of Ra-226 groundwater concentrations to retardation - tuff.
-------
00
to
l_
3.
c
o
s
«->
0)
O
O
o
I.WWE.TV<* -
1.00E+04 -
1.00E+03 -
1.00E+02 -
1.00E+01 -
1 nnp^nn .
' f \
I ~ "
I
I
I
i _i i—
I *^
I
I
I
I
I
1.00E+04
Time (years)
0 Total alpha - low retardation
• ~ Total alpha - no retardation
1.00E+05
RAE -104661
Figure7.5-45. Sensitivity of total alpha groundwater concentrations to
retardation
-------
1.00E+07
^ 1.00E+06-
]2 1.00E+05-
£ 1.00E+04-
gj 1.00E-I-03-
o
•o
(0
•o 1.00E+01
>
£ LOOE+OO-f
c
"" 1.00E-01-
1.00E-02
10,000
Low retardation
No retardation
100,000
Time (years)
RAE -104662
Figure7.5-46. Sensitivity of total beta groundwater dose to retardation - tuff.
-------
1.00E+06
~ 1.00E+05
t
| 1.00E+04
g 1.00E+03
•O
1
% 1.00E+02
s
£ 1.00E+01
1.00E+00
100
1,000 10,000
Time (years)
100,000
Basalt
Granite
RAE-104666
Figure 7.5-47. Comparison of media - individual dose.
-------
Z6-L
Concentration (pCi/liter)
I.UUC+UO
1.00E+05 -
1.00E+04 •
1.00E+03 -
1.00E+02 -
.OOE+Ol
1.00E+00
1.00E-01 -
1.00E-02 -
1.00E-03 -
1.00E-04 -
1.00E-05
1.00E-06 -
1.00E-007 -
1 1
i
I
I
I
J
1
Basalt
Granite
100
1,000 10,000
Time (years)
100,000
RAE -104668
Figure 7.5-48. Comparison of media - Ra-226 concentrations.
-------
CO
co
1.00E+06
^ 1.00E+05 -
.1
C 1.00E+04 -
.2 1.00E+03 -
I
§ 1.00E+02 -
O
O 1.00E+01 -
1.00E+00
100
1,000 10,000
Time (years)
1
100,000
Basalt
Granite
RAE -104667
Figure 7.5-49. Comparison of media - concentration of alpha-emitting radionuclides.
-------
1.00E+06
>> 1.00E+05 -
£ 1.00E+04
&
O 1.00E+03
r1 75
£ -o 1.00E+02
•MM
c 1.00E+01
1.00E+00
100
1,000 10,000
Time (years)
100,000
Basalt
Granite
RAE -104669
Figure 7.5-50. Comparison of media - dose from beta-emitting radionuclides.
-------
• The simplified nature of the models in comparison with the real physical
situation; and
• The generic nature of the modeling.
The purpose of the generic risk assessments is to make rough approximations of
the capabilities of geologic disposal media to contain radioactive waste. Therefore, despite
these uncertainties, the Agency believes that the estimates generated herein provide an
adequate technical basis for the associated regulations.
In order to lend perspective to the uncertainties in the generic calculations, the
Agency has proceeded as follows. First, in estimating parameters or in choosing models to
represent various processes, an attempt has been made to conservatively predict factors
that contribute to risks from the repository. This is the same philosophy that was
adopted in risk assessments for the proposed rule, although the degree of over-estimation
has been reduced in response to recommendations by the Agency's Science Advisory
Board. Again, a conservative approach was taken in the selection of many parameters
values but sufficient site-specific work has been done by previous studies to provide a high
degree of certainty to some parameter values.
Second, use has been made of sensitivity analyses in order to understand how
much the results of the assessment change with variations in certain model components
or parameters.
Third, in cases where it has been difficult to model the characteristics of a site or a
process on a generic basis, several choices of parameter values have been made to
understand the range of potential risk results. In the generic assessments, alternative
cases were used to model ground-water flow.
Two parameters identified as having a high degree of uncertainty are solubility
and retardation. Previously used values for these parameters vary over several orders of
magnitude. An area of relatively high uncertainty is the effect of retardation on
radionuclide migration. Two alternate cases were used to examine uncertainties in
retardation. Both cases use the values from the WISP report (NAS83) for retardation
factors. In addition to the recommended nominal values, the high and low sets of values
from NAS93 and a retardation factor of 1 were used in the assessment.
The radionuclide solubility is included as a sensitivity parameter in the
transuranic waste assessments presented. Neither alternate case had an effect on
releases into ground water because the TRU nuclides did not have low enough retardation
to travel the flow path in 10,000 years.
It is important to distinguish between the type of uncertainty included in the
generic analysis reported here and the uncertainties that would remain with real sites
when they are characterized and modeled in connection with the decision on where to put
a repository. Many of the uncertainties associated with generic assessments and included
here might better be characterized as variabilities. Among actual specific sites there
7-95
-------
might be a wide variation in the value of the parameter (property) in question. The
attempt was made in these generic risk assessments to incorporate such variations, which
correspond to an uncertainty in the final results, to determine how well they characterize
the performance of the repository.
An assessment of an actual site would include additional uncertainties associated
with data collection, site complexity, and difference of opinion about a specific site's
characteristics. Section 7.6.1 discusses uncertainties associated with site-specific risk
assessments and methods for evaluating these uncertainties. Section 7.6.2 discusses the
use of expert judgement in addressing uncertainties in performance assessment
calculations.
7.6.1 Evaluation of Uncertainties in Site-Specific Risk Assessments
Sources of Uncertainty - Many investigators have grouped uncertainties associated
with the assessment of geologic disposal into three categories: scenario uncertainty,
model uncertainty, and parameter value or data uncertainty, as summarized in
Figure 7.6-1 (Cranwell and Helton, 1981; Hunter et al., 1986; Davis et al., 1990). The
following summarizes the description of these categories of uncertainties.
Scenario uncertainty arises from the subjectivity inherent in predicting future
conditions. Davis et al. (1990) define a scenario as a "combination of anticipated or
unanticipated events and processes, either natural or human induced...[that]...could result
in the release of radionuclides from the underground facility, their migration through the
geosphere and biosphere, and their eventual exposure to humans." Cranwell et al. (1990)
define a procedure for scenario development and selection consisting of the following six
steps:
1. Identify all possible events and processes relevant to the long-term
performance of a geologic waste disposal facility.
2. Group similar events and processes to create a smaller, more manageable
set.
3. Screen the set of events and processes based on established criteria, such as
a very low probability of occurrence or negligible impact on waste isolation.
4. Systematically combine the remaining events and processes into scenarios,
the combined set of scenarios being mutually exclusive and collectively
exhaustive.
5. Screen the set of scenarios based on established criteria, such as a very low
probability of occurrence or negligible impact on waste isolation.
6. Select the set of scenarios that will be used in evaluating repository
performance.
7-96
-------
SOURCES OF UNCERTAINTIES
Scenario Development/Screening
•Completeness
•Unconservative Initial Screening
Modeling (Conceptual, Mathematical, Computational)
• Lack of Process Knowledge
• Oversimplification of System Processes and Dynamics
•Computational Limitations
•Coding Error
Parameter Value/Data
•Quality of Existing Data
•Representativeness of Existing Data
•Selection of Representative Values
RAE -103927
Figure 7.6-1. Sources of Uncertainties.
7-97
-------
Uncertainty related to scenario development has been categorized in several
different ways. It is widely recognized that uncertainties will arise with respect to the
completeness of the set of scenarios, since it is unrealistic to expect that every possible
event or process could be identified in the first step of the scenario development process.
Uncertainties can also arise in the subjective screening of the set of events and processes
and the set of scenarios, Steps 3 and 5 in the scenario development procedure. Potentially
important events and processes or scenarios could be screened out if initial estimates of
probabilities of occurrence or consequences are not conservative, i.e., the credibility or
consequences of scenarios are underestimated.
Uncertainties also stem from model development and implementation. Models of
the waste isolation system must be developed in order to rigorously evaluate each
scenario in terms of its probability of occurrence and the resulting consequence (its impact
on waste isolation). Four types of models must be developed: conceptual, mathematical,
numerical, and computer.
A conceptual model describes the system in terms of the processes taking place, the
variables related to the identified processes, and the temporal and spatial variations in
the processes (Davis et al., 1990). The inherent complexity of natural and engineered
systems requires that assumptions be made in developing conceptual models. It is likely
that processes operating within system components will not be completely understood.
Processes may vary within a system component (spatially) at any given time and may
vary over time (temporally). Also, the interactions among system components and the
processes operating within and among the components are complicated and must be
simplified. Such simplifying assumptions result in uncertainties related to the conceptual
models.
In order to quantitatively evaluate natural and engineered systems, mathematical
models must be developed from the conceptual models. Also, numerical models must be
developed as real solutions to the mathematical models. Uncertainties associated with
the development of mathematical and numerical models include those associated with
insufficient knowledge of the processes operating within the system, insufficient
knowledge with respect to temporal and spatial dependencies of processes operating
within the system, and the limitations inherent in attempting to represent complex and
interdependent system processes by mathematical expressions.
Mathematical models are implemented through the development of computer
models. Sources of uncertainty associated with computer codes include coding errors,
computational limitations, and user error. Computational errors can be caused, for
example, by truncation errors, discretization error, inappropriate convergence and
stability error. Another potential source of computational error is the use of numerical
algorithms with data beyond the required range for a particular algorithm.
Uncertainties associated with parameter value and data sets can result from
measurement error and from the misinterpretation of the collected data. Also, insufficient
knowledge of the system can lead to data uncertainty. The raw data must be reduced to a
form suitable for model input. Lack of representativeness, due to unknown spatial
7-98
-------
variations, and invalid assumptions about the system can lead to the improper selection of
the input data sets.
The preceding paragraphs discuss various sources of uncertainties that may arise
in conducting performance assessments of a geologic waste disposal system. It is useful to
categorize uncertainties according to similarities as well as by source. There are two
general categories of uncertainty: random uncertainty and knowledge uncertainty (Wu et
al., 1991). Random, or stochastic, uncertainty results from stochastic variability of some
random variable. Knowledge uncertainty results from imperfect knowledge about some
fixed value. Wu et al. states that the essential difference between random uncertainty
and knowledge uncertainty is that knowledge uncertainty may be reduced by increased
data sampling or experimentation, whereas random uncertainty will not be reduced.
Random, or stochastic, uncertainty is often referred to as Type 1 uncertainty, and
knowledge uncertainty is often referred to as Type 2 uncertainty (Hofer and Hoffman,
1987).
Uncertainty Evaluation Methods - Each source of uncertainty (scenario
uncertainty, model uncertainty, and parameter value and data uncertainty) should be
addressed in an appropriate manner. The intent is to minimize or eliminate uncertainties
where possible and to define or quantify where they are unavoidable. The following
discussion outlines techniques proposed and employed for evaluating uncertainties in
repository performance assessments.
Uncertainties associated with scenario development and model development are
generally subjective. Scenario uncertainties are best addressed through the
implementation of defensible scenario development methodologies that are well structured
and documented. Formalized expert judgement and peer review processes also aid in the
minimization and quantification of scenario uncertainties. Even though uncertainties are
unavoidable in predicting future conditions, quantification allows the impacts of scenario
uncertainties to be evaluated.
Uncertainties associated with model development are addressed in various ways.
Uncertainties associated with conceptual models, developed subjectively through the
interpretation of data and the hypothesis of system processes, can be addressed similarly
to scenario uncertainties. Expert judgement and peer review are primary methods for
minimizing and quantifying conceptual model uncertainties. Formal expert judgement
processes are discussed in detail in Section 7.6.2. Expert judgement would also play an
important role in developing the corresponding mathematical models. In addition,
mathematical model uncertainty can be evaluated and minimized through validation
exercises. Computational model uncertainty can be addressed through verification and
benchmarking exercises. Both mathematical and computational models would be
evaluated through peer review. If more than one plausible alternative model has been
identified, the sensitivity of results to alternative models should be evaluated.
Parameter value and data uncertainties are generally evaluated in a more rigorous
and quantitative manner than either scenario uncertainties or model uncertainties, since
parameter value and data uncertainties are more readily quantifiable, although expert
judgement plays an important role in developing model parameter value input.
7-99
-------
Evaluating data and parameter value uncertainty consists of two fundamental steps.
First, the uncertainties associated with each parameter are defined quantitatively. Data
and parameter value uncertainties consist of both stochastic (Type 1) and knowledge
(Type 2) uncertainties. Deterministic assessments can be made by using single parameter
values as input, such as mean or bounding parameter values. But the results of single
deterministic calculations are limited in that the uncertainties in the input parameters
will not be reflected in the output. Therefore, value ranges are developed for parameters
for use in calculations instead of single values. The value ranges are often presented
quantitatively in the form of probability distribution functions. Probability distribution
functions, developed through data analysis and expert judgement, allow the uncertainties
in parameter values to be incorporated into analyses. Typical forms of probability
distribution functions are shown in Figure 7.6-2.
Next, the propagation of parameter value uncertainties through system models is
evaluated. The following methods may be employed to evaluate the propagation of data
and parameter value uncertainties.
One well-known and often-used method for evaluating parameter uncertainties is
the Monte Carlo sampling technique. Commonly, mathematical models and associated
models are deterministic, allowing only single, discrete values to be used for input
parameters. Deterministic models usually produce single, discrete values for each output
parameter value, with no indication of the level of confidence in the output given the
uncertainties about the input.
The Monte Carlo technique allows uncertainty in parameter values to be directly
taken into account and reflected in the calculated model output. For each input
parameter, a sample set is formed consisting of the possible values for the variable. Most
likely, the sample set will be in the form of a probability distribution function where the
probability of the parameter having any one value is given. If the value of a parameter
were known with certainty, the resulting probability distribution function would be a
discrete single-value function. More likely, a particular parameter value is random or
unknown with the possible values spread over some range. The uncertainty in the
parameter value is reflected in the probability distribution function.
The Monte Carlo technique is based on the iterative recalculation of a
deterministic model. Instead of a single deterministic calculation of model output, the
model is run a large number of times, or iterations. The number of iterations is generally
dependent upon the complexity of the model. Prior to an iteration, an input data set, or
input vector, is created. The input vector has N dimensions, where N is equal to the
number of model input variables. To create an input vector, a value for each input
parameter is selected based on the probability distribution function for that input
parameter. Using the Monte Carlo technique for each input parameter, values with a
higher relative probability of occurring will be selected and used as input to the model
more often than values having a lower relative probability of occurring. The use of higher
probability parameter values more often as input to the deterministic model will be
reflected in the output sample set. The value of the model output parameter is then
calculated based on the input vector. Thus, if there are N iterations, there will be N input
vectors and N calculated output values in the output parameter set. The sample statistics
for the output parameter, such as the mean and the various percentiles, are then
7-100
-------
P(A)
Mm«an
Normal PDF
P(B)
Skewed PDF
P(C)
P(D)
Cmax
Uniform PDF
D.
Step PDF
.75
P(E)
.25
Discrete PDF
RAE-103928
Figure 7.6-2. Typical hypothetical parameter probability density
functions (PDFs).
7-101
-------
calculated. The relationship between the Monte Carlo sampling of input variables and
the distribution of output values is illustrated in Figure 7.6-3.
A critical aspect of the Monte Carlo technique is the input parameter sampling
scheme employed. The goal is to have the set of N selected values for a particular input
parameter reflect closely, in a statistical sense, the uncertainty in that particular input
parameter. Several sampling schemes have been developed. Random sampling, as the
name implies, involves the selection of values for a particular parameter at random within
the predefined probability function for the parameter. A large number of iterations is
necessary to ensure that the selected values adequately represent the parameter. Sample
representativeness is enhanced by using a more structured sampling scheme. For
example, stratified sampling involves the systematic partitioning of the range of values for
a particular parameter into some number of strata. Samples are then drawn from each
stratum, ensuring that the entire range of values is represented. Latin hypercube
sampling is a special case of stratified sampling where the range of values for a particular
parameter is partitioned into a number of cells equal to the number of iterations, each cell
having equal probability. One sample is then drawn from each cell.
The response surface methodology is another technique for evaluating the
propagation of uncertainties in risk assessments. The response surface methodology
involves the evaluation of uncertainties through the simplification of the deterministic
model. The aim is to replace the complex mathematical system model by a relatively
simple linear or nonlinear analytical function that is dependent on only a subset of the
original input variables.
Some number of input parameter sets, or input vectors, are defined and the
original complex model is then evaluated deterministically for each of the input vectors.
The simplified analytical function is derived through a regression analysis technique such
as the least squares method. The simplified function is then used to evaluate the
correlation of uncertainties in the input parameters with uncertainty in the output
parameter, i.e., sensitivity analyses.
A third technique employed in risk assessments for evaluating the propagation of
uncertainties is the differential analysis approach. The differential analysis approach
involves replacing the complex deterministic system model with a first order or second
order Taylor series expansion. The mean and the variance of the output variable can be
approximated by evaluating the series terms, using a mean value for each variable in the
input vector. This method is similar to the response surface methodology in that it is very
useful in conducting sensitivity analyses. This method is limited, however, in that it
allows only a localized evaluation of variance in the output parameter.
As an example of how uncertainties are treated in current site assessments, the
performance assessment program for the Waste Isolation Pilot Plant (WIPP), currently
under development, is described in detail in the report entitled "Preliminary Comparison
with 40 CFR Part 191, Subpart B for the Waste Isolation Pilot Plant," (SAND91). This
report, updated periodically, describes the WIPP methodology for evaluating compliance
with the disposal requirements of 40 CFR Part 191. The approach to addressing
uncertainties in the assessment of performance is described in the first volume of the
study, "Methodology and Results."
7-102
-------
>-OutputM
(e.g. Radionuciide
Release)
Input Varibles
X:
Input Vectors
Vector 1 Vector 2 Vector j
•11
L21
31
•12
L 22
x=,
L2j
L3j
RAE -103929
Figure 7.6-3. Monte Carlo Technique (adapted from Hunter et al., 1986).
7-103
-------
Model A
>,
Model B
•^
Model C
^-Output Mj
Based on Set of
n
*
a
I
ma
mb
Output M
RAE -103930
Figure 7.6-3. Monte Carlo Technique (adapted from
Hunter et aL, 1986) continued.
7-104
-------
The WIPP program generally employs the widely accepted approach to
performance assessment. Scenarios which are credible and which may impact
performance of the facility during the next 10,000 years are systematically identified. The
likelihood and consequences of each scenario are evaluated and quantified and the results
are combined into a complementary cumulative distribution function to demonstrate
compliance with the performance standards.
The Monte Carlo technique was adopted in the WIPP program to evaluate the
impact of parameter value uncertainties on the calculated consequences. Five reasons are
given in the WIPP compliance report to support this decision. First, it is felt that the
Monte Carlo approach best accommodates large uncertainties in input parameters and
complex interdependences among the parameters. Second, the direct result of the Monte
Carlo approach is the generation of a distribution function for the output parameter.
Third, the Monte Carlo method is straightforward in that complex ancillary calculations
are not necessary, just the repetitive recalculation of the deterministic models with a
predetermined set of input variable vectors. Fourth, the Monte Carlo technique is
amenable to evaluating the propagation of uncertainties through systems of
interdependent models. Finally, the Monte Carlo technique allows a direct evaluation of
the impacts of input parameter uncertainties on the certainty in the output parameter,
i.e., sensitivity studies.
The WIPP program uses the Monte Carlo technique to develop a family of
complementary cumulative distribution functions (CCDFs). Each CCDF is the result of
evaluating the deterministic models with a single input variable vector. Probability
distributions are developed for input parameters, based on the existing data base and
expert judgement (Section 7.6.2). Input vectors are defined by sampling from the input
parameter value distributions. Each input vector is then used to calculate scenario
probabilities and consequences which are assembled into a CCDF. One hundred
iterations (recalculations), for example, will require one hundred input vectors and will
generate one hundred CCDFs for the output parameter. From this set of CCDFs, the
mean CCDF, or any other statistically significant CCDF (median, percentiles) can be
derived and evaluated against the containment standard. It is recognized that this is a
simplistic overview of a sophisticated program. The WIPP performance assessment and
compliance program is discussed in detail in the WIPP compliance document (SAND91).
This overview, however, indicates the rigor with which uncertainties can be quantitatively
evaluated in site-specific assessments.
7.6.2 Expert Judgement
It is generally accepted that the use of expert judgement is required in the process
of evaluating the long-term containment potential of a geologic waste disposal facility. It
is expected that the use of expert judgement will be an integral component of the
demonstrations of compliance with the Environmental Standards for Disposal (40 CFR
Part 191, Subpart B). It is important to discuss the manner in which expert judgement
should be used. Expert judgement refers to opinions based on the knowledge of the expert
whose experience is relevant to the issue at hand.
7-105
-------
The use of expert judgement in the area of nuclear reactor safety is
well-documented in the U.S. Nuclear Regulatory Commission's study, "Severe Accident
Risks: An Assessment for Five U.S. Nuclear Plants" (NUREG-1150). Severe accident
risks were evaluated for five commercial nuclear power plants. Risks were estimated
based on the types and frequencies of accidents that could lead to severe core damage and
melting, the response of the containment structure to severe accident loading, the
radioactive release that could result from containment failure, and the offsite effects of
the potential releases (NRC90). The study was undertaken to provide a risk perspective
for the radioactive release resulting from a core meltdown. The long-term objective of the
assessment was to provide probabilistic risk assessment models for generic use in
research and security risk estimates. The use of expert judgement in the NUREG-1150
study is discussed in Ortiz et al. (1989), where it is referred to as "part of the largest
elicitation task to date" and "an advance over those processes developed in previous
probabilistic risk assessments."
Thirty-two issues in the following categories were evaluated:
1. System analysis issues.
2. In-vessel accident progression issues.
3. Containment load issues.
4. Molten core containment issues.
5. Structural issues.
6. Source-term issues.
A panel of twelve experts evaluated the first category and a total of 38 experts
evaluated the other five categories. Each expert provided information regarding
parameter value distributions. Consensus distributions were constructed by aggregating
the experts' distributions through simple averaging or sampling from the experts'
distributions through using the Monte Carlo technique.
Expert judgement is also being employed in the Waste Isolation Pilot Plant (WIPP)
performance assessment program. The WIPP program is employing a formal elicitation
process to evaluate parameters considered significant in assessing performance, but with
which there is considerable associated uncertainty (SAND91). A formal expert judgement
elicitation process is employed in the WIPP program when "data are lacking, either
because of the complexity of processes or the time and resources it would take to collect
data and/or when data have a major impact on the performance assessment" (SAND91).
An example of this is discussed in Trauth et al. (1991). Radionuclide
concentrations in the brine located hi the rooms and drifts of the WIPP repository are
considered to be a critical parameter in assessing the performance of the facility.
However, there is significant uncertainty in these radionuclide concentrations. A formal
elicitation process was used to develop concentration distributions. The distributions are
7-106
-------
used in the performance assessments instead of point values, allowing the impacts of
uncertainties in the values of the parameter and associated parameters to be evaluated.
In addition to the evaluation of probability distributions of significant system
parameter values, a formal expert judgement process is being employed in the WIPP
program for the identification and evaluation of future human-intrusion scenarios, an
inherently qualitative task. The formal elicitation process allows the defensible
quantification of the issue as is necessary for inclusion in the quantitative performance
assessments. Sixteen experts, external to the WIPP program and representing a diversity
of physical and social sciences, were systematically identified and organized into
four-member teams. Each team was charged with identifying reasonable, foreseeable
"futures" for human society and quantifying the likelihoods of occurrence of these futures.
Also, the teams were asked to evaluate how the futures could result in the intrusion of
the WIPP repository and to quantify the likelihoods of such intrusions.
To date, the teams have evaluated future states of human society and identified
reasonable "futures." The likelihoods of these futures were identified through a formal
elicitation process. The teams have also identified possible modes of intrusion associated
with the "futures" and developed quantitative probabilistic estimates of the frequencies of
these intrusions. The evaluation of human-intrusion scenarios through a structured
expert-judgement process is documented in Hora et al.(1991).
7-107
-------
REFERENCES
Cr90 Cranwell, R.M., R.V. Guzowski, J.E. Campbell, and N.R. Ortiz, "Risk
Methodology for Geologic Disposal of Radioactive Waste - Scenario Selection
Procedure," NUREG/CR-1667, SAND80-1429, Sandia National Laboratories,
Albuquerque, NM, 1990.
Cr81 Cranwell, R.M., and J.C. Helton, "Uncertainty Analysis for Geologic
Disposal of Radioactive Waste," in D.C. Kocher, ed., Proceedings of the
Symposium on Uncertainties Associated with the Regulation of the Geologic
Disposal of High-Level Radioactive Waste, Gatlinburg, IN, 1981.
DOE89 U.S. Department of Energy, "Draft Supplement Environmental Impact
Statement: Waste Isolation Pilot Plant," DOE/EIS-0026, 1989.
Da90 Davis, P.A., L.L. Price, KK. Wahi, M.T. Goodrich, D.P. Gallegos, E.J.
Bonano, and R.V. Guzowski, "Components of an Overall Performance
Assessment Methodology," NUREG/CR-5256, SAND88-3020, Sandia
National Laboratories, Albuquerque, NM, 1990.
EPA82 U.S. Environmental Protection Agency, "Population Risks from the Disposal
of High-Level Radioactive Wastes in Geologic Repositories,"
EPA-520/3-80-006, 1982.
EPA85 U.S. Environmental Protection Agency, "Risk Assessment of Disposal of
High-Level Radioactive Wastes in Geologic Repositories," EPA 520/1-85-028,
1985.
EPA89 U.S. Environmental Protection Agency, "Risk Assessment Methodology,
Environmental Impact Statement, NESHAPS for Radionuclides,"
Background Information Document, Volume 1, EPA/520/1-89-005,
September 1989.
Ho87 Hofer, E. and F.O. Hoffman, "Selected Examples of Practical Approaches for
the Assessment of Model Reliability - Parameter Uncertainty Analysis," in
Proceedings of an NEA Workshop on Uncertainty Analysis for Performance
Assessments of Radioactive Waste Disposal Systems. Nuclear Energy
Agency, Organization for Economic Co-Operation and Development, Paris,
France, 1987.
Ho91 Hora, S.C., D. von Winterfeldt, and KM. Trauth, "Expert Judgement on
Inadvertent Human Intrusion into the Waste Isolation Pilot Plant,"
SAND90-3063, Sandia National Laboratories, Albuquerque, New Mexico,
1991.
Hu86 Hunter, R.L., R.M. Cranwell, and M.S.Y. Chu, "Assessing Compliance With
the EPA High-Level Waste Standard: An Overview," NUREG/CR-4510,
SAND86-0121, Sandia National Laboratories, Albuquerque, NM, 1986.
7-108
-------
La89 Lappin, A.R., "System Analysis, Long Term Radionuclide Transport, and
Dose Assessments, Waste Isolation Pilot Plant (WIPP), Southeastern New
Mexico," Sandia National Laboratories, SAND89-0462, 1989.
NAS83 National Academy of Sciences - National Research Council, "A Study of the
Isolation System for Geologic Disposal of Radioactive Wastes," Report of the
Waste Isolation System Panel, Board on Radioactive Waste Management,
Washington, D.C., 1983.
NRC90 U.S. Nuclear Regulatory Commission, "Severe Accident Risks: Assessment
for Five U.S. Nuclear Power Plants," NUREG-1150, Washington, D.C., 1990.
Or89 Ortiz, N.R., T.A. Wheeler, M.A. Meyer, and R.L. Keeney, "Use of Expert
Judgement in NUREG-1150," SAND88-2253C, Sandia National
Laboratories, Albuquerque, NM, 1989.
RAE92 Risk Assessment for TRU Waste Disposal in Bedded Salt; Prepared by
Rogers & Associates Engineering Corporation, under contract with Sandy
Cohen & Associates, Inc., Contract No. 68D90170, Work Assignment 2-29,
March 1992.
RAE92a Rogers & Associates Engineering Corporation and SC&A, Inc., "Issues
Associated with Gaseous Releases of Radionuclides for a Repository in the
Unsaturated Zone," July 1992.
SAND84 Sinnock, S., Y. Lin, and J. Brannen, "Preliminary Bounds on Expected
Postclosure Performance of the Yucca Mountain Repository Site, Southern
Nevada," Sandia National Laboratories, SAND84-1492, December 1984.
SAND90 Campbell, J.E., C.D. Leigh, D.E. Longsine, "NEFTRAN-S: A Network Flow
and Contaminant Transport Model for Statistical and Deterministic
Simulations Using Personal Computers," SAND90-1987 • UC-507.
SAND90a Leigh, C., "Technical Basis for a Conceptual Model in Unsaturated Tuff for
the NEFTRAN-S Code," Sandia National Laboratories, SAND90-1986,
UC-502, May 1991.
SAND91 WIPP Performance Assessment Division, "Preliminary Comparison with
40 CFR Part 191, Subpart B for the Waste Isolation Pilot Plant, December
1991," Sandia National Laboratories, SAND91-0893, 1991.
Tr91 Trauth, K.M., S.C. Hora, and R.P. Rechard, "Expert Judgement as Input to
Waste Isolation Pilot Plant Performance-Assessment Calculations -
Probabilities of Significant System Parameters," SAND91-0625C, Sandia
National Laboratories, Albuquerque, NM, 1991.
7-109
-------
Wu91 Wu, Y.T., A.G. Journel, L.R. Abramson, and P.K. Nair, "Uncertainty
Evaluation Methods for Waste Package Performance Assessment,"
NUREG/CR-5639, Division of High-Level Waste Management, Office of
Nuclear Material Safety and Safeguards, U.S. Nuclear Regulatory
Commission, Washington, D.C., 1991.
7-110
-------