8
Types of Nuclear Reactors
8.1 Survey of Reactor Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
8.2 Light Water Reactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
8.3 Burners, Converters, and Breeders . . . . . . . . . . . . . . . . . . . . . . 186
8.4 The Natural Reactor at Oklo . . . . . . . . . . . . . . . . . . . . . . . . . . 191
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
8.1 Survey of Reactor Types
8.1.1 Uses of Reactors
The first nuclear reactors were built to produce 239
Pu for bombs. Subsequently,
reactors have been used for many other purposes, of which electricity
generation is now, by far, the most prominent. Further uses have been to propel
ships (mostly naval vessels), to produce radioisotopes, and, to a limited
extent, to supply heat. Many additional reactors have been built for teaching
or research, much of the latter involving the study of the properties of materials
under neutron bombardment and the intrinsic properties of neutrons
and other subatomic particles.
In some cases, applications have been combined. For example, the N reactor
at Hanford and the Chernobyl-type reactors were used for both 239
Pu
production and electricity generation. There has also been limited use of waste
heat from reactors to produce hot water or steam for industrial applications,
for heavy water production, and for desalination. These reactors have been
primarily in the USSR and Canada [1]. The total thermal capacity of these
facilities is much less than that of electricity-generating nuclear reactors, although
the use of heat from reactors could increase in the future, for example
if used for hydrogen production (see Section 16.6).
The focus in this book is on the use of reactors for electricity generation,
but the issue of 239
Pu production still arises. If the reactor fuel contains 238
U,
then 239
Pu will inevitably be produced following neutron capture in 238
U. In
breeder reactors, the production of 239
Pu is a central goal, with the 239
Pu intended
as fuel for further reactors. In nonbreeders, which means almost all of
the world’s operating reactors, 239
Pu is a by-product, but its fission, nonetheless,
often contributes significantly to the reactor’s total energy output. (We
171
172 8 Types of Nuclear Reactors
will return in Chapter 17 to the link between power reactors and the possible
use of their 239
Pu for bombs.)
8.1.2 Classifications of Reactors
Thermal Reactors and Fast Reactors
In previous chapters, reference has been made to the thermalization of neutrons
in reactors (i.e., to the slowing of the neutrons to thermal energies).
Reactors designed to operate in this fashion are termed thermal reactors.
However, it is also possible to operate a reactor with “fast” neutrons. There
is no moderator in such a reactor and few neutrons reach thermal energies,
but many lose energy by inelastic scattering and fission occurs at energies
ranging from the MeV region down to the keV region and below. These reactors
are called fast-neutron reactors or just fast reactors. The only prominent
example of an operating fast power reactor is the liquid-metal reactor (see
Section 8.3.3), although other types of fast reactors have been built for experimental
purposes.
Homogeneous and Heterogeneous Reactors
We have been tacitly assuming that the reactors under consideration are what
are sometimes known as heterogeneous, in which the fuel, coolant, and moderator
(if any) are distinct physical entities. All reactors used today for power
generation are of this form. However, in the early days of nuclear power, there
was considerable exploration of an alternative configuration, the homogeneous
reactor, defined as “a reactor whose small-scale composition is uniform and
isotropic” [2, p. 378]. Homogeneity can be achieved if the fuel is in liquid form,
where the liquid is circulated for heat transfer to a steam generator.
One variant of this reactor type was known as the aqueous homogenous
reactor because the fuel was mixed with water (H2O or D2O). In the so-called
homogeneous reactor experiment, two small reactors, called HRE-1 and HRE-
2, were built at Oak Ridge National Laboratory (ORNL) in the 1950s.1
For
HRE-2, the fluid was uranyl sulfate (UO2SO4) in heavy water (D2O), with the
uranium highly enriched in 235
U. This program had the potential of developing
a thermal breeder reactor, but although HRE-2 operated uninterruptedly for
over 100 days at 5 MWe, some difficulties developed, and the program was
dropped in favor of alternative liquid-fuel projects.
One alternative was the molten-salt reactor. The fluid was a mixture of
fluoride compounds, including the fissile component 235
UF4 and the fertile
component ThF4. After initial operation, 233
UF4 was successfully tried as an
1
For a description of the history of this program and the ORNL program on moltensalt
breeders, see Chapter 6 of Ref. [3]. Technical aspects of fluid-fuel reactors are
also discussed in Ref. [4, pp. 403–413].
8.1 Survey of Reactor Types 173
alternative to 235
UF4. Like HRE-2, this reactor was designed to be a thermal
breeder reactor. Again, there was initial success in the reactor operation, but
a decision was made in the 1960s to abandon development of thermal breeders
in favor of fast breeder reactors.2
A further homogenous reactor approach, a
liquid metal thermal breeder using uranium compounds in molten bismuth,
was also investigated but was abandoned without construction even of a test
reactor.
At present, there are no electricity-generating homogenous reactors, and
we will not consider them further in the chapter. Instead, we will restrict consideration
to heterogeneous reactors, which are so dominant that it is unusual
to include the specification “heterogeneous.” Nonetheless, some interest remains
in homogenous reactors, particularly in molten-salt reactors (see, e.g.,
Ref. [5]).3
In fact, one of the reactors selected for possible long-term development
under the Generation IV program is a molten-salt reactor (see Chapter
16). Overall, although we focus in this chapter on reactors of the sort in
actual use or in immediate prospect, it is well to remember that on a longer
time scale, a wide array of variants are possible. Some of these are being tentatively
explored in the thinking underway in the United States and elsewhere
on nuclear options for the future, as discussed in Chapter 16.
8.1.3 Components of Conventional Reactors
Overall
Any generating plant consists of an array of structural components and a
system of mechanical and electrical controls. In a nuclear plant, there are
special demands on structural integrity and reliability. In addition, a nuclear
reactor is characterized by the use of specialized materials, some aspects of
which were already discussed in Chapter 7. In standard reactors, these are the
fuel itself, the coolant, the moderator, and neutron-absorbing materials used
to control the power level. A main distinction between different reactor types
lies in the differences in the choices of fuel, coolant, and moderator.
Fuels
There are few nuclides that can be used as reactor fuels. The paucity of
possible candidates can be seen by examining the properties of the naturally
occurring heavy elements:
2
The fast breeder reactor program was subsequently sharply reduced, with the
centerpiece of the U.S. fast breeder reactor program, the Clinch River Breeder
Reactor, abandoned in 1975.
3
There has been speculation about a quite different sort of molten-salt reactor,
driven by a proton accelerator. If pursued, this would represent a radical departure
from the sorts of reactors that have been built to date.
174 8 Types of Nuclear Reactors
N Uranium (atomic number Z = 92). This is the main fuel in actual use,
especially 235
U which is fissile. In addition, 238
U is important in reactors,
primarily as a fertile fuel for 239
Pu production, and 233
U could be used as
a fissile fuel, formed by neutron capture in 232
Th.
N Protactinium (atomic number Z = 91). The longest-lived isotope of protactinium
(231
Pa) has a half-life of 3.3 × 104
yr, and therefore there is
essentially no Pa in nature.
N Thorium (atomic number Z = 90). Thorium is found entirely as 232
Th,
which is not fissile (for thermal neutrons). It can be used as a fertile fuel
for the production of fissile 233
U.
Between thorium (Z = 90) and bismuth (Z = 83), the isotope with the
longest half-life is 226
Ra (T = 1600 years); therefore, there are no fuel candidates,
quite apart from the issue of fissionability. By the time the atomic
number is as low as 83, the threshold for fission is much too high for a chain
reaction to be conceivable (see, e.g., Ref. [6, p. 574]). Thus, uranium and
thorium are the only natural elements available for use as reactor fuels. In
addition, 233
U and 239
Pu can be produced from capture on 232
Th and 238
U in
reactors. This means that the nuclides listed in Table 6.1 exhaust the practical
possibilities for reactor fuels. Of these, only 235
U is both fissile and found in
nature in useful amounts.
Restricting consideration to uranium fuel, there are a number of options
as to the form of the fuel. Reactors can operate over a considerable range of
enrichments in 235
U. Enrichment to a concentration in the neighborhood of
4% is now somewhat typical in the light water reactors that account for most
of today’s nuclear generation (see Section 8.1.4), with a trend over time toward
higher enrichments and greater burnup of the fuel. In today’s heterogeneous
reactors, the fuel is solid. For the most part, it is in an oxide form, as UO2,
but metallic fuel is a possibility and has been used in some reactors. The
fuel usually is in cylindrical pellets with typical dimensions on the scale of
1 cm, but some designs for future reactors are based on fuel in submillimeter
microspheres embedded in graphite, with the goal of enhanced ruggedness at
high temperatures.4
Moderators
As discussed in Section 7.2, a moderator is required if the reactor is to operate
at thermal neutron energies. This means that most operating reactors
use moderators, with the fast breeder reactor the exception. The options for
moderating materials are limited:
N Hydrogen (Z = 1). The isotopes 1
H and 2
H are widely used as moderators,
in the form of light (ordinary) water and heavy water, respectively.
4
In particular, this fuel is for high-temperature gas-cooled reactors (see Section
16.4.3).
8.1 Survey of Reactor Types 175
N Helium (Z = 2). The isotope 4
He is not used, because helium is a gas
and excessive pressures would be required to obtain adequate helium densities
for a practical moderator; 3
He would be similarly excluded, but, in
addition, it is a strong neutron absorber and would be unsuitable as a
moderator.
N Lithium (Z = 3). The isotope 6
Li (7.5% abundant) has a large neutronabsorption
cross section, making lithium impractical as a moderator.
N Beryllium (Z = 4). 9
Be has been used to a limited extent as a moderator,
especially in some early reactors. It can be used in the form of beryllium
oxide, BeO. However, beryllium is expensive and toxic.
N Boron (Z = 5). The very large neutron-absorption cross section in 10
B
(20% abundant) makes boron impossible as a moderator.
N Carbon (Z = 6). Carbon in the form of graphite has been widely used
as a moderator. It is important that the graphite be pure (i.e., be free of
elements that have high absorption cross sections for neutrons).
There are no advantages in considering elements heavier than carbon. The
effectiveness for moderation decreases with increasing mass, and there are no
counterbalancing advantages in other properties. Again, therefore, there is a
limited list of candidates: light water, heavy water, graphite, and beryllium.
Any of these can be used with uranium enriched in 235
U. With natural uranium,
it not possible to achieve a chain reaction with a light water moderator,
but it is practical to use heavy water or graphite, both of which have high
moderating ratios (see Table 7.2).
Coolants
The main function of the coolant in an electricity generating plant is to transfer
energy from the hot fuel to the electrical turbine, either directly or through
intermediate steps. During power plant operation, cooling is an intrinsic aspect
of energy transfer. However, in a nuclear reactor, cooling has a special
additional importance, because radioactive decay causes heat production to
continue even after the reactor is shut down and electricity generation has
stopped. It is still essential to maintain cooling to avoid melting the reactor
core, and in some types of reactor accidents (e.g., the accident at Three Mile
Island) cooling is the critical issue.5
The coolant can be either a liquid or a gas. For thermal reactors, the most
common coolants are light water, heavy water, helium, and carbon dioxide.
The type of coolant is commonly used to designate the type of reactor. Hence,
the characterization of reactors as light water reactors (LWRs), heavy water
reactors (HWRs), and gas-cooled reactors (GCRs).
5
If the fuel is designed to operate at high enough temperatures, cooling can be
provided by radiation from the fuel, and maintaining the flow of coolant would not
be essential under accident conditions. However, at present no operating reactors
are designed for such high temperatures.
176 8 Types of Nuclear Reactors
The coolant may also serve as a moderator, as is the case for LWRs and
HWRs. In gas-cooled reactors, the density of the coolant is too low to permit
it to serve as the chief moderator, and graphite is used. Fast reactors, in
which fission is to occur without moderation to thermal energies, usually use
a coolant that has a relatively high atomic mass number (A).6
Generically,
these reactors are termed liquid-metal reactors, because the coolant is a liquid
metal, most commonly sodium (A = 23).
Control Materials
As discussed in Section 7.5.2, control materials are needed to regulate reactor
operation and provide a means for rapid shutdown. Boron and cadmium are
particularly good control materials because of their high cross sections for the
absorption of thermal neutrons. These control materials are usually used in the
form of rods. Control rods for pressurized water reactors (PWRs) commonly
use boron in the form of boron carbide (B4C) or cadmium in a silver–indium
alloy containing 5% cadmium. Control rods for boiling water reactors (BWRs)
commonly use boron carbide [7, p. 715]. In addition, boron may be introduced
into the circulating cooling water to regulate reactor operation.
8.1.4 World Inventory of Reactor Types
Reactor Sizes
The earliest reactors had generating capacities well below 100 MWe, but there
was a rapid transition to 1000-MWe reactors and larger. The move to a larger
size was motivated by the desire to capture economies of scale. Some analysts
suggest that this escalation proceeded too rapidly, especially in the United
States, and was responsible for some of the difficulties encountered in achieving
short construction times and reliable operation.
The mean capacity of all reactors in operation worldwide in 2003 was
about 820 MWe (see Table 2.1). At the extremes, a class of older British
gas-cooled reactors have capacities of 50 MWe, whereas four PWRs that
went into operation in France in 2000 each have a capacity close to 1450
MWe [8].7
While most reactors built in recent years—including in France,
Japan, South Korea, the United States, and the United Kingdom (for the one
reactor in 1995)—are large, considerable attention is being given to smaller
reactors. Although going to smaller reactors means sacrificing economies of
6
It is also possible to use helium as a coolant because there the helium gas is not
dense enough to be an effective moderator.
7
Four 50-MWe units at the Calder Hall power plant in the United Kingdom were
officially shut down on March 31, 2003 after operations that began in the late 1950.
(Three of these units had actually suspended operations in 2001, but remained
listed as “operating” in standard tables [9]). Four other 50-MWe reactors remain
in operation at the Chapelcross plant.
8.1 Survey of Reactor Types 177
scale, some advantages can be regained if a number of identical units are
placed at the same site. (Questions of future reactor size are discussed further
in Section 16.1.3.)
Types of Reactors
A variety of different reactors are in use in the world today, although there
was greater diversity in the early days of reactor design. Table 8.1 lists the
types of nuclear power plant in operation in late 2003 as well as those reported
to be under construction or on order [10]. The dominant reactor is the
light water reactor (LWR), which uses ordinary water as both the coolant and
moderator and enriched uranium in UO2 pellets as the fuel. There are two
types of light water reactor: the pressurized water reactor (PWR) and the
boiling water reactor (BWR). Together, they account for 88% of the world’s
present generating capacity and 85% of the capacity nominally being built
or on order. The main types of reactors in past or present use for electricity
generation are as follows:
PWR. The pressurized water reactor accounts for almost two-thirds of all
capacity and is the only LWR used in some countries, for example France,
the former Soviet Union, and South Korea.
BWR. The boiling water reactor is a major alternative to the PWR and
both are used, for example, in the United States and Japan.
Table 8.1. World totals for nuclear reactors in commercial operation and under
construction, November 2003, classified by reactor type: number of reactors and
capacity (in GWe).
Capacity
Number (GWe)
First
Type Oper Cons Oper Cons Usual Fuela
Moderator Coolant Developed
PWR 263 18 236.0 16.0 UO2 enr H2O H2O USA
BWRb
92 5 80.6 6.4 UO2 enr H2O H2O USA
PHWRc
39 8 19.3 3.1 UO2 nat D2O D2O Canada
GCR 26 0 10.9 0.0 U, UOd
2 C CO2 UK
LGR 17 1 12.6 0.9 UO2 enr C H2O USA/USSR
LMFBR 3 0 1.0 0.0 UO2 + PuO2 None Liq Na Various
TOTAL 440 32 360 26
a
Fuel designations: enr = enriched in 235
U, nat = natural.
b
The listing for BWRs includes two ABWRs in operation and four under construc-
tion.
c
Includes one 148-MWe HWLWR in Japan.
d
Natural U used for GCR; enriched UO2 used for AGCR.
Source: Capacity data are from Ref. [10]. Fuel and country data are from Ref. [11,
p. 67].
178 8 Types of Nuclear Reactors
ABWR. The advanced boiling water reactor incorporates improvements
over earlier BWRs. It is in use in Japan, with additional units under
construction in Japan and Taiwan.
PHWR. The pressurized heavy water reactor uses heavy water for both the
coolant and moderator and operates with natural uranium fuel. It was
developed in Canada and is commonly referred to as the CANDU.8
Other
countries with CANDU units in operation include India, South Korea,
and Argentina.
GCR. The gas-cooled, graphite-moderated reactor uses a CO2 coolant and
a graphite moderator. Its use is limited to the United Kingdom; it is sometimes
known as the Magnox reactor. A larger second-generation version
is the advanced gas-cooled, graphite-moderated reactor (AGCR).
LGR. The light-water-cooled graphite-moderated reactor uses water as a
coolant and graphite (in addition to water) for moderation. The world’s
major currently operating LGRs are the RBMK reactors in the former
Soviet Union (11 in Russia and 2 in Lithuania).9
There were four such
units at the Chernobyl plant at the time of the accident there, but they
have all been shut down.
HTGR. The high-temperature gas-cooled reactor uses helium coolant and
a graphite moderator. The only HTGR that had been operating in the
United States (Fort St. Vrain) has been shut down, and there are no
HTGRs being used elsewhere for electricity generation, although active
studies of variants of the HTGR are underway.
LMFBR. The liquid-metal fast breeder reactor uses fast neutrons and needs
no moderator. A liquid metal is used as coolant, now invariably liquid
sodium. There are only two LMFBR reactors in operation (one each in
France and Russia).10
HWLWR. The heavy-water-moderated, light-water-cooled reactor is an unconventional
variant of the heavy water reactor, and only one has been
in recent operation, a 148-MWe plant in Japan. A new 700-MWe version
of the HWLWR is being designed in Canada, the ACR-700 (see Section
16.2.2).
The dominance of light water reactors, both for plants in operation and
those under construction, is seen in Table 8.1. These reactors were first developed
in the United States, in both the PWR and BWR configurations,
and have become the reactors of choice in almost all other major nuclear
8
This acronym stands for Canadian deuterium uranium and has an obvious double
meaning.
9
In addition, there are four 11-MWe LGRs in Russia.
10
In addition, the 246-MWe Monju reactor in Japan is listed by the IAEA as connected
to the grid, but it has been shut down since 1995 after operating for only
a few months [8].
8.1 Survey of Reactor Types 179
countries. The main exceptions are Canada, the United Kingdom, the former
Soviet Union (FSU), and India. Even in the United Kingdom and the FSU,
the most recently completed reactors are PWRs.
The number of reactors under construction or on order as of late 2003 (32)
was small compared to the number in operation (440). The average capacity
of these reactors is about 830 MWe—very close to the average for operating
plants. They range in size from four 202-MWe PHWRs being built in India [12]
to two ABWRs in Japan with capacities near 1300 MWe.
History of Commercial Reactor Development
After World War II, the leading countries in nuclear reactor development were
the United States, the United Kingdom, Canada, and the Soviet Union. Each
went in a different direction.
The first U.S. power reactors, beyond plutonium-producing or experimental
reactors, were built for submarines, not for civilian electricity generation.
The earliest were a PWR for the submarine Nautilus, commissioned
in 1955, and a sodium-cooled reactor for the submarine Seawolf. The Seawolf
reactor had difficulties, and sodium-cooled reactors were abandoned by the
navy in 1956 in favor of light water reactors [13, p. 423]. The navy PWR
program provided the foundation for the development of PWRs for electricity
generation, starting with the 60-MWe reactor at Shippingport, Pennsylvania,
in 1957.
As was noted in Section 2.3.2, during the 1950s a varied array of reactors
were ordered in the United States. These even included a small fast breeder
reactor (Fermi I) in Michigan, which went into operation for a few years
starting in 1966. However, after 1967, the only commercial power reactors put
into operation in the United States have been PWRs and BWRs, with the sole
exception (in 1979) of the trouble-plagued Fort St. Vrain HTGR in Colorado,
which has since been shut down. The commercial BWRs were an outgrowth
of a program of experimental BWR development carried out in the mid-1950s
at Argonne National Laboratory.
The United Kingdom and Canada followed routes that did not require enriched
uranium. The United States had a monopoly on uranium enrichment at
the time, and although it presumably would have provided enriched uranium
to such close allies, there may have been a reluctance on their part to become
dependent. The United Kingdom program began very early, with two 50-MWe
reactors at Calder Hall in 1956. These were GCRs, with graphite moderation
and CO2 cooling. They differed from most later reactors in the world in that
they used uranium metal for the fuel, not uranium dioxide (UO2). They gained
the name Magnox, because the fuel pin cladding material was a magnesium
alloy called Magnox [11, p. 165]. The GCRs that were built later had increasing
size, up to 420 MWe. From the mid-1970s, with one exception, the few
new plants brought on line in the United Kingdom have been AGCRs in the
180 8 Types of Nuclear Reactors
600–700 MWe range. Like the Magnox reactors, they use graphite moderation
and CO2 cooling, but their fuel is enriched UO2.
On the whole, after a fast start, the British reactor program has moved
fitfully, with indecision abetted by North Sea oil and natural gas. After 1989,
only one new reactor was put into commercial operation in the United Kingdom,
the 1188-MWe Sizewell B reactor in 1995 [8]. Interestingly, it is a PWR,
selected after prolonged study, adding further to the dominance of LWRs in
the world nuclear picture.
The Canadian nuclear program offers the main alternative to the LWR
among reactors now in operation or under construction. This program has involved
only one type of reactor, the pressurized heavy water reactor (PHWR)
known as the CANDU (Canadian deuterium uranium). Use of a deuterium
moderator enabled Canada to use natural, rather than enriched, uranium.
This was an attractive option for a country that had sophisticated scientific
and engineering capabilities, including experience with heavy water reactors
gained during World War II, but no enrichment facilities. The larger CANDU
reactors (greater than 600 MWe) had the best cumulative capacity factors of
any reactor type, as of the end of 2001 [14].
The PHWR has made substantial inroads outside of Canada, in particular
in India, South Korea, and Argentina, with smaller programs elsewhere. In
Canada, these reactors vary relatively little in size, starting at 525 MWe and
most recently built at 881 MWe. India has emulated the Canadian example
of reliance on PHWRs, although initially at the smaller size of about 200
MWe. The first of these were constructed under Canadian supervision, but,
subsequently, India has assumed independent responsibility. South Korea has
fewer PHWRs than India (four compared to fourteen) but they are larger—
each about 650 MWe.
The other major dissenter from LWRs had been the Soviet Union, but
this is now changing. The Soviet Union began with six 100-MWe light-watercooled,
graphite-moderated reactors (LGR) put into operation from 1958 to
1963 [8]. These were followed by larger LGR reactors with capacities of about
950 MWe, the so-called RBMK reactors among which were the Chernobyl
reactors. These LGRs were built to produce both plutonium and electricity,
as was the now-closed Hanford-N reactor in Washington. The Soviet Union
also developed its own PWRs, the WWER series, and these are now the
most numerous reactors in Russia and are widely used in much of Eastern
Europe [8].
Overall, what worldwide growth there is in nuclear power is now primarily
in the form of LWRs, with the Canadian PHWR as the only other significant
player. It is not clear whether this is because of intrinsic technical
and economic advantages of water-cooled reactors or because of historical
and commercial forces. For the future, there is considerable interest in new
HTGRs as a relatively near-term option and in liquid-metal-cooled reactors
for the longer term (see Chapter 16), but the dominance of water-cooled reactors,
and particularly LWRs, has not yet been seriously challenged.
8.2 Light Water Reactors 181
8.2 Light Water Reactors
8.2.1 PWRs and BWRs
The two types of LWR in use in the world are the pressurized water reactor
(PWR) and the boiling water reactor (BWR). The difference between them,
as embodied in the names, is in the condition of the water used as coolant
and moderator. In the PWR, the water in the reactor vessel is maintained
in liquid form by high pressure. Steam to drive the turbine is developed in a
separate steam generator. In the BWR, steam is provided directly from the
reactor. These differences are brought out in the schematic representation of
the two reactor types in Figure 8.1.
Under typical conditions in a PWR, temperatures of the cooling water into
and out of the reactor vessel are about 292◦
C and 325◦
C, respectively, and the
pressure is about 155 bar [7, p. 713].11
For the BWR, typical inlet and outlet
temperatures are 278◦
C and 288◦
C, respectively, and the pressure is only
about 72 bar. The high pressure in the PWR keeps the water in a condensed
phase; the lower pressure in the BWR allows boiling and generation of steam
within the reactor vessel.
Neither the PWR nor BWR has an overwhelming technical advantage over
the other, as indicated by the continued widespread use of both. Among the
major LWR users, the United States, Japan, and Germany use both types,
while France, South Korea, and Russia use PWRs almost exclusively in the
LWR part of their programs. Overall, the number of PWRs in operation is
significantly greater than the number of BWRs, and PWRs also have a lead
in reactors listed as under construction. The future is not clear-cut, however.
For example, in Japan, all three reactors under construction in 2003 were
BWRs, including two ABWRs. In the following discussion, we will emphasize
the PWR in giving specific illustrations but will consider both to some extent.
8.2.2 Components of a Light Water Reactor
The containment structure and enclosed components for a typical PWR and
a typical BWR are shown schematically in Figures 8.2 and 8.3.12
The most
conspicuous difference between them is the absence of a steam generator in
the BWR. At the heart of the reactors, literally and figuratively, is the reactor
core, contained within the reactor pressure vessel. The pressure vessel encloses
three vital components:
N The fuel itself, contained in many small fuel rods comprising the reactor
core.
N The surrounding water, acting as coolant, moderator, and heat-transfer
agent.
11
1 bar = 105
newton/m2
= 0.987 atm.
12
These diagrams are copied from a draft version of Ref. [15].
182 8 Types of Nuclear Reactors
Fig. 8.1. Schematic representation of BWR and PWR systems, emphasizing the
difference in the means for providing steam to the steam turbine. [Adapted from
figures provided by the U.S. Council on Energy Awareness.]
8.2 Light Water Reactors 183
Fig. 8.2. Schematic diagram of containment building with enclosed reactor vessel
and steam generator for an illustrative PWR: the 781-MWe Surry plant. The output
of the steam generator drives the turbines, external to the containment building.
(From [15, p. 4–4].)
N Control rods, used to establish the reactivity at the desired level and shut
the reactor down in case of an emergency.
The reactor pressure vessel is a massive cylindrical steel tank. Typically for
a PWR, it is about 12 m (40 ft) in height and 4.5 m (15 ft) in diameter [16,
p. 304]. It has thick walls, about 20 cm (8 in.), and is designed to withstand
pressures of up to 170 atm.
A second major component, or set of components, is the system for converting
the reactor’s heat into useful work. In the BWR, steam is used directly
from the pressure vessel to drive a turbine. This is the step at which electricity
is produced. In the PWR, primary water from the core is pumped at high pressure
through pipes passing through a heat exchanger in the steam generators.
Water fed into the secondary side of the steam generator is converted into
steam, and this steam is used to drive a turbine. The secondary loop is also
184 8 Types of Nuclear Reactors
Fig. 8.3. Schematic diagram of containment building with enclosed reactor vessel
for an illustrative BWR: the 1142-MWe Grand Gulf plant. Steam from the reactor
vessel drives the turbines, external to the containment building. (From [15, p. 4–40].)
closed. The exhaust steam and water from the turbine enter a condenser and
are cooled in a second heat exchanger before returning to the steam generator.
The cold side of the condenser heat exchanger represents the tertiary loop
for the PWR. In principle, this loop need not be closed, and the condenser
cooling water could circulate to and from a river or the ocean. More commonly,
this condenser output is circulated through a cooling tower, where it is cooled
by evaporation and the ultimate heat sink becomes the air. The part of the
water that is lost as steam from the cooling tower is made up by water drawn,
say, from a river.
Cooling towers became represented as ominous-looking symbols during the
Three Mile Island accident, but they themselves are quite benign. The steam
8.2 Light Water Reactors 185
that may be seen rising from a cooling tower is isolated from water passing
through the reactor core by the heat exchangers and, thus, is not a source of
radioactive emissions, even if the primary cooling water is slightly radioactive.
Such cooling towers are not unique to nuclear power plants and are used in
other facilities, including coal-fired plants, where it is necessary to dissipate
large amounts of heat.
The pressure vessel and the steam generators are contained within a massive
structure, the containment building, commonly made of strongly reinforced
concrete. In some designs, the concrete containment is lined with steel;
in others, there is a separate inner steel containment vessel. The containment
is intended to retain activity released during accidents and is believed capable
of protecting a reactor against external events including some airplane impacts
(see Section 17.5.5). In the Three Mile Island accident, the containment
very successfully retained the released radioactivity, although it may be noted
that the physical structure was not put fully to the test because there was no
explosion or buildup of high pressures. At Chernobyl, there was no containment,
with disastrous results. In principle, were the possibility of an accident
made negligibly small with improved reactor design and if terrorist attacks
were not a concern, a containment would be unnecessary. Nonetheless, it is
widely considered to be an important safety feature, providing an additional
layer of protection.
8.2.3 PWR Reactor Cores
We consider here the specific characteristics of a reactor core based on a
Westinghouse Corporation PWR design, but the gross features are similar for
all large LWRs.13
The reactor fuel is in the form of cylindrical uranium oxide
(UO2) pellets, about 0.8 cm in diameter and 1.35 cm in length. The pellets are
placed in tubes—called fuel rods or fuel pins—made of zircaloy, a zirconium
alloy (98% Zr, 1.5% Sn, and small amounts of other metals [16, p. 234])
selected on the basis of structural strength and low neutron absorption. The
zircaloy cladding is thin, 0.06 cm. The fuel rod is typically 3.7 m (12 ft.) in
length and 1.0 cm in diameter. There is some free space within the fuel rod
to allow for the expansion of the fuel pellets and to accommodate gaseous
fission products such as xenon and krypton. The fuel expansion is due to both
increased temperature and the replacement in the fuel of one uranium atom
by two fission-product atoms. Noble gases produced as fission products may
be trapped as bubbles in the fuel or may escape from the fuel into the gap
between the fuel and cladding.
A 17 × 17 array of fuel rods forms a “fuel bundle” or assembly. Although
this would allow 289 fuel rods in an assembly, positions are left open in some
13
Most of the detailed numbers in this paragraph are based on Westinghouse Corporation
information, as reported in Ref. [16], especially Table 9.1. They are specific
to this particular Westinghouse design; others differ in detail.
186 8 Types of Nuclear Reactors
assemblies for the insertion of control rods or measuring instrument rods.
The total core contains 193 assemblies and 50,952 fuel rods. Some 53 of the
assemblies have spaces for clusters of 20 control rods, which can be moved
in and out within the assembly. These control rods are made from a silver–
indium–cadmium alloy.
Fuel assemblies are replaced periodically, but not all in the same period of
reactor refueling. Thus, although a particular fuel assembly may remain in the
reactor for 3 years, one-third of the core can be changed every year. (Recently,
cores have been designed to have a longer time between fuel changes.) As the
235
U is consumed in the reactor, the reactivity of the fuel decreases. This is
compensated for in several ways. Burnable poisons in the fuel are consumed,
control rods that limit the reactivity are partly withdrawn, or the soluble
poison concentration in the cooling water is reduced.
8.3 Burners, Converters, and Breeders
8.3.1 Characterization of Reactors
As discussed in Section 7.1.1, the condition for a chain reaction is that for
every neutron initiating fission in one generation, one or more neutrons initiate
fission in the next generation. If, in addition, another fissile nucleus is produced
for every 235
U atom consumed, then there is no decrease in the amount of
nuclear fuel available. This is the principle of the breeder reactor.
The conversion ratio C (or breeding ratio B) is defined as the ratio of the
rate of production of fissile nuclei to the rate of consumption of fissile nuclei
[see Eq. (7.18)].14
For uranium fuel, this is the ratio of 239
Pu produced to 235
U
consumed. If the conversion ratio is small, the reactor is sometimes called a
burner; if the conversion ratio is between about 0.7 and 1.0, it is commonly
called a converter; and if it exceeds unity, the reactor is called a breeder (see,
e.g., Ref. [16, p. 458]).
8.3.2 Achievement of High Conversion Ratios in Thermal Reactors
Difficulty of Reaching a Conversion Ratio of Unity with 235
U
As discussed in Section 7.4, the limiting condition for a breeder reactor is
that the conversion ratio, C, be at least 1. This means that the number ηo of
neutrons produced for each neutron absorbed in 235
U, must be two or more.15
For thermal neutrons absorbed in 235
U, ηo = 2.075 (see Table 7.1). Were
there no losses, this would suffice for breeding: 1 neutron for continuing the
14
Sometimes, a distinction is made in terminology, with conversion ratio used when
C < 1 and breeding ratio used when C > 1.
15
The relation among ηo, η, and the commonly cited parameter ν, is discussed in
Section 7.1.2.
8.3 Burners, Converters, and Breeders 187
chain reaction, 1 neutron for production of 239
Pu, and 0.08 neutrons free to
be “wasted.” Such efficient utilization cannot be achieved, however, because
there is absorption in the moderator and other nonfuel materials, as well as
escape of neutrons from the core. Therefore, a 235
U-fueled thermal breeder
reactor is not practical. Nonetheless, the production of 239
Pu is significant
in uranium-fueled thermal reactors because its fission increases the overall
energy output from the reactor fuel beyond that gained from the 235
U alone.
Potential of 233
U for a Thermal Breeder Reactor
The number of neutrons produced is significantly higher for 233
U (ηo = 2.296)
than for 235
U, and there have been serious suggestions for developing 233
U
thermal breeders. These date to as early as 1945, in work done by Eugene
Wigner’s group in Chicago [3, Chapter 6]. A cycle is envisaged in which 233
U
is produced initially in a reactor with 235
U as the fissile fuel and 232
Th as
the fertile fuel. Subsequently, a 233
U–232
Th cycle could, in principle, be selfsustaining.
Not only is ηo higher for 233
U than for 235
U, but the capture
cross section is significantly higher for 232
Th than for 238
U at thermal energies,
making the conversion ratio higher than for a cycle based on 238
U (see
Table 7.1).
However, although thermal breeders based on 233
U are, in principle, possible
and preliminary exploratory work toward their development was done
at Oak Ridge National Laboratory in the 1950s, thermal breeders were abandoned
in favor of the fast breeder reactor. It is conceivable that interest in
thermal breeders could revive, but, to date, the few breeder reactors that have
gone into commercial operation have all been fast breeders.
High Conversion Ratios Without Breeding
Before turning to fast breeder reactors, it may be noted that even if breeding
is not achieved with thermal reactors, a high conversion ratio can still be
desirable. One motivation could be plutonium production. Another motivation
is the extension of fuel resources. As the conversion ratio increases, the energy
output increases for a given original 235
U content. A high conversion ratio
means a high ratio of capture in 238
U to absorption in 235
U. This must be
accomplished without losing criticality. Greater losses of neutrons to 238
U can
be compensated for by smaller losses in the moderator and reactor structure.
The use of carbon instead of light water as a moderator is favorable on
two counts if a high conversion ratio is desired (in addition to the advantage
that with a carbon moderator it is possible to use natural uranium). Because
carbon is a less effective moderator than water, more collisions are required to
reach thermal energies; therefore, there is more possibility of neutron capture
in 238
U at intermediate energies. Further, because of the low neutron-capture
cross section in 12
C (see Section 7.2.2), the loss of thermal neutrons to ab-
188 8 Types of Nuclear Reactors
sorption will be less for carbon than for light water.16
Together, this means a
higher conversion ratio.
The same general arguments apply to heavy water reactors. The conversion
ratio is higher in a heavy water reactor than in a light water reactor
due to less effective moderation in heavy water and a lower neutron-capture
cross section. For both graphite-moderated and heavy water reactors, there
have been suggestions that the 232
Th–233
U cycle be used, to further increase
conversion and extend the life of the uranium fuel, even without breeding.
It may be noted that reactors designed with production of plutonium for
weapons in mind, either as the main or as an auxiliary function, have been
mostly graphite moderated. Examples include the Windscale plant in England,
the plutonium production reactors at Hanford, and the RBMK reactors
built in the USSR. The five heavy water reactors at the Savannah River (South
Carolina) complex for plutonium production are the major exception.
8.3.3 Fast Breeder Reactors
Plutonium as Fuel for Fast Breeders
A thermal breeder reactor is not possible using 239
Pu due to the high ratio α
of the capture cross section to the fission cross section for thermal neutrons.17
However a fast breeder reactor is possible. It relies on a chain reaction in which
the neutrons are not thermalized but instead produce fission at relatively high
energies. If 239
Pu is the fissile fuel, the cycle uses 238
U as the fertile fuel and
239
Pu is both consumed and produced in the reactor. The cycle is started
using 239
Pu produced in a uranium-fueled reactor.
Although most of the fission neutrons are emitted with energies above
1 MeV, they can lose energy through inelastic scattering in 239
Pu and 238
U.
Fission is therefore produced at energies extending over a very broad energy
region, from above 1 MeV to below 1 keV. In the high-energy part of this
region the conditions are very favorable for breeding. For example, for 1.0-MeV
neutrons on 239
Pu, σf = 1.7 b, and the ratio (α) of the capture cross section
to the fission cross section is less than 0.03. The low value of this ratio means
that almost all absorption in 239
Pu leads to fission. With about three neutrons
per fission at 1 MeV, breeding with 239
Pu is readily achieved, with 1 neutron
for continuing the chain reaction, 1+
for breeding, and 1−
for losses. At lower
16
In terms of the formalism introduced in Chapter 7, this means that carbon leads
to a lower resonance escape probability p and a higher thermal utilization factor
f than does light water [see the four-factor formula, Eq. (7.5)]. Criticality can
still be maintained (k = 1), and C will be greater (see Eq. (7.23)).
17
For 239
Pu, α = 0.360 and ηo = 2.115 at thermal neutron energies. This may be
compared to the values for 233
U: α = 0.086 and ηo = 2.296. For breeding, the
crucial condition is that ηo > 2. Therefore 233
U is significantly better than 239
Pu
at thermal energies.
8.3 Burners, Converters, and Breeders 189
neutron energies, the cross section σf remains high, above 1.5 b for the most
part, but the ratio α rises—reaching about 0.3 at 40 keV—greatly reducing the
number of “surplus” neutrons [17, p. 753]. The actual conversion ratio depends
on the neutron reactions over the full range of “incident” neutron energies,
with the lower-energy neutrons contributing even if not as effectively as the
high-energy neutrons.
To avoid thermalization of the neutrons, fast breeder reactors use a coolant
with a relatively high mass number A. Liquid metals have the best combination
of high A and favorable heat-transfer properties, and the fast breeder
reactors in actual use have been liquid-metal fast breeder reactors (LMFBR).
The standard choice for the coolant is liquid sodium (23
Na).
The fuel is made of pellets of mixed plutonium and uranium oxides, PuO2
(about 20%) and UO2 (about 80%). Uranium depleted in 235
U is commonly
used, it being available as a residue from earlier enrichment. The fission cross
section for 239
Pu is between 1.5 and 2.0 b over virtually the entire fast-neutron
region (from 10 keV to 6 MeV), whereas for 238
U it is below 0.2 b for En < 1.4
MeV and falls rapidly at lower En (see Figure 6.1 and Ref. [17]). Therefore,
fission in 239
Pu is much more probable than fission in 238
U. The most probable
fast-neutron reactions in 238
U are inelastic scattering, which produces lower
energy neutrons, and capture, which produces 239
Pu.
Status of Fast Reactor Programs
The main incentive for the development of fast breeder reactors is the extension
of uranium supplies. A fast breeder economy would extract much
more energy per tonne of uranium than is obtained from other reactors (e.g.,
the LWRs). Further, with more energy per unit mass, it becomes economically
practical to use more expensive uranium ores, increasing the ultimate
uranium resource. A secondary incentive is the easing of the waste disposal
problem if plutonium and uranium (and possibly other actinides) are reused
in a closed cycle rather than disposed of as waste.
However, during the 1980s and 1990s, growth of nuclear power fell far
short of earlier expectations, there was little pressure on uranium supplies,
and interest in fast breeder reactors declined. Further, the initial fast breeder
reactors proved to be more expensive than alternatives, such as the LWR or
HWR. Particularly in the United States, there was also the concern that the
large-scale use and availability of 239
Pu might increase dangers from terrorism
and nuclear weapon proliferation (see Section 9.4.2 and Chapter 17). These
factors made breeder programs a vulnerable target, at a time when there was
significant opposition to any projects to advance nuclear power. Nonetheless,
some development of breeders has continued, in part to maintain the technology
as insurance against future needs.
In a later turn of the argument, it has been pointed out that a liquidmetal
fast reactor (LMR) can be used to destroy unwanted plutonium and
other heavy elements which are in weapons stockpiles or nuclear wastes. In
190 8 Types of Nuclear Reactors
this reversal of motivation, the LMR would be used to consume plutonium,
rather than to produce plutonium as a fuel. There is flexibility in this, because
as LMR technology and facilities are developed, they could be turned to either
purpose. However, if the driving fear is concern about misuse of plutonium, it
may appear more desirable to dispose of plutonium from weapons stockpiles
in ways that do not involve expanding a technology that is closely related to
potential plutonium production.
France had led in the development and deployment of breeder reactors,
with two completed reactors, the 233-MWe Phenix, put into operation in
1973, and the 1200-MWe SuperPhenix at Creys-Malville, which first generated
electricity in 1986 but was finally shut down in 1998 after a troubled history
of recurring technical difficulties. Small breeder reactors in Great Britain and
Kazakhstan were also shut down in the 1990s, and there now remain only two
LMFBRs operating to produce electricity, Phenix in France and a 560-MWe
reactor in Russia. However, interest in breeder reactors remains in a number
of countries, with ongoing and new activity, and it would be premature to
write breeder reactors off as an option for the future.18
The United States breeder reactor program has been marked by indecision
and opposition, with successive projects started and abandoned. It had
started auspiciously, with the operation of the Experimental Breeder Reactor
(EBR-I) in Idaho. On December 20, 1951 EBR-I generated the first electricity
from a nuclear reactor produced anywhere—enough for four light bulbs.
Its output was shortly thereafter increased to 0.1 MWe [19]. Two larger, fast
reactors made important research contributions in subsequent years—the Experimental
Breeder Reactor II (EBR-II) in Idaho and the Fast Flux Test
Reactor (FFTR) in Washington—but both of these projects have been terminated.
The most ambitious breeder proposal in the United States was for
the Clinch River Breeder Reactor (CRBR)—a project that was under active
consideration in the 1970s but was terminated by Congress in 1983.
Following the end of the CRBR project, a major fast reactor development
program was undertaken at Argonne National Laboratory as part of the integral
fast reactor plan (see Section 16.5). Advocates of this program stressed its
potential to offer a high degree of safety against reactor accidents and to consume
nuclear wastes in an on-line process. The breeding potential was often
secondary in these arguments, and the planned LMR need not have operated
as a breeder, namely with a conversion ratio greater than unity. Nonetheless,
the basic configuration of the system was similar to that of a breeder reactor.
Culminating several years of debate, most funding for this project was
terminated in the mid-1990s.
18
Current breeder projects include the 246-MWe Monju breeder reactor in Japan,
which is now shut down due to operating difficulties but which may be restarted, a
750-MWe LMFBR being planned in Russia (Beloyarsk-4) with completion scheduled
for 2009, a prototype 500-MWe breeder being planned in India (Kalpakkam),
and a 65-MWt fast neutron reactor being built in China [18].
8.4 The Natural Reactor at Oklo 191
Despite these difficulties, the long-term argument for breeder reactors remains
and breeder development may intensify, especially if the proliferation
problems can be satisfactorily addressed. Some of the reactors now being considered
under the program of the Generation IV International Forum (see
Chapter 16) are designed to operate with a fast-neutron spectrum and thus
have the potential of being operated as breeder reactors.
8.4 The Natural Reactor at Oklo
A remarkable discovery was made in 1972 by French scientists analyzing uranium
extracted from the Oklo uranium mine in Gabon. The uranium was
depleted in 235
U, sometimes by large amounts, although, normally, the isotope
ratios in uranium are nearly constant over the surface of the Earth. It
was soon suspected and then demonstrated that this isotopic anomaly was due
to a natural uranium chain reaction occurring more than a billion years ago.
Conclusive evidence in support of this explanation was provided by the relatively
high abundance of intermediate-mass nuclei, the rare earths, which are
characteristic fission products but are not normally found in large abundance
in nature (see, e.g. Ref. [20]).
The scenario, as it has been recreated, puts the event about 1.8 billion years
ago. At that time, the isotopic abundance of 235
U exceeded its present value
by the factor exp(∆λ t), where ∆λ is the difference in the decay constants
of the two isotopes and t is the time since the event. The decay constants of
235
U and 238
U are 0.985 × 10−9
yr−1
and 0.155 × 10−9
yr−1
, respectively (see
Table 2.1), giving ∆λ t = 1.49 for t = 1.8 × 109
yr. Therefore. the isotopic
abundance of 235
U was 4.4 times greater at the time of the Oklo event than
it is today, putting the enrichment at slightly above 3%. (This is strikingly
close to the enrichment used in modern LWRs.) The intrusion of water, acting
as a moderator, apparently initiated a chain reaction, which appears to
have simmered for at least several hundred thousand years. In this model of
what took place, the reaction did not occur earlier because the concentrated
uranium deposits had been only recently formed by the leaching of rocks and
the precipitation of the dissolved uranium.
Aside from having posed an intriguing scientific puzzle, with a very interesting
explanation, the Oklo event is considered by some to have significance
as a test of the motion of fission products through the ground. For the most
part, these products have moved very little over a period of more than 1 billion
years. This could have implications for the rate of movement of fission
products in buried nuclear wastes. The Oklo example cannot be used as an
all-embracing guide because differences in the chemical form of the product
and in the type of rock formation may vitiate an extrapolation from Oklo to
the behavior of an individual modern waste disposal site. However, Oklo illustrates
that at least under some circumstances, radionuclides do not migrate
appreciably from their initial location.
192 8 Types of Nuclear Reactors
References
1. H. Barnert, V. Krett, and J. Kupitz, “Nuclear Energy for Heat Applications,”
IAEA Bulletin 33, no. 1, 1991: 21–24.
2. Alvin M. Weinberg and Eugene P. Wigner, The Physical Theory of Neutron
Chain Reactors (Chicago: University of Chicago Press, 1958).
3. Alvin M. Weinberg, The First Nuclear Era: The Life and Times of a Technological
Fixer (New York: American Institute of Physics Press, 1994).
4. J. Smith, “Novel Reactor Concepts,” in Nuclear Power Technology, Vol. 1 of
Reactor Technology, W. Marshall, ed. (Oxford: Clarendon Press, 1983), pp. 390–
415.
5. Uri Gat and H. L. Dodds, “The Source Term and Waste Optimization of Molten
Salt Reactors with Processing,” in Future Nuclear Systems: Emerging Fuel Cycles
& Waste Disposal Options, Proceedings of Global ’93 (La Grange Park, IL:
American Nuclear Society, 1993), pp. 248–252.
6. Emilio Segr`e, Nuclei and Particles, 2nd edition (Reading, MA: W.A. Benjamin,
1977).
7. Ronald Allen Knief, Nuclear Engineering: Theory and Technology of Commercial
Nuclear Power, 2nd edition (Washington, DC: Hemisphere Publishing Company,
1992).
8. “World List of Nuclear Power Plants,” Nuclear News 46, no. 3, March 2003:
41–67.
9. “United Kingdom: Calder Hall Bows out at the Age of 47,” Nuclear News 46,
no. 6, May 2003: 48.
10. International Atomic Energy Agency, “Nuclear Power Plants Information, Operational
& Under Construction Reactors By Type” (updated 11/24/03). [From:
http://www.iaea.org/cgi-bin/db.page.pl/pris.reaopucty.htm]
11. D.J. Bennet and J.R. Thomson, The Elements of Nuclear Power, 3rd edition
(Essex, UK: Longman, 1989).
12. International Atomic Energy Agency, “Latest News Related to PRIS
and the Status of Nuclear Power Plants” (updated May 2003). [From:
http://www.iaea.org/programmes/a2/]
13. Richard G. Hewlett and Jack M. Holl, Atoms for Peace and War, 1953–1961
(Berkeley, CA: University of California Press, 1989).
14. International Atomic Energy Agency, Nuclear Power Reactors in the World,
Reference Data Series No. 2, April 2003 edition (Vienna: IAEA, 2003).
15. U.S. Nuclear Regulatory Commission, Reactor Risk Reference Document,
NUREG-1150, Vol. 1, Draft for Comment (Washington, DC: NRC, 1987).
16. F.J. Rahn, A.G. Adamantiades, J.E. Kenton, and C. Braun, A Guide to Nuclear
Power Technology (New York: Wiley, 1984).
17. Victoria McLane, Charles L. Dunford, and Philip F. Rose, Neutron Cross Section
Curves, Volume II of Neutron Cross Sections (New York: Academic Press,
1988).
18. World Nuclear Association, Information and Issues Briefs; Particular Countries.
[From: http://www.world-nuclear.org/info/info.htm]
19. Argonne National Laboratory West, “ANL-W History—Reactors (EBR 1):
EBR-1.” [From: http://www.anlw.anl.gov/anlw history/reactors/ebr i.html]
20. Michel Maurette, “Fossil Nuclear Reactors,” Annual Review of Nuclear Science
26, 1976: 319–350.