Reactor Types

With different fuels, moderators, control systems, cooling arrangements, spatial
configurations and so on, possible designs of nuclear reactor number in the hundreds.
Early reactor designers had a field day, letting their imaginations run riot; some of their
suggestions made colleagues' hair stand on end. Others seemed more feasible: amenable
engineering, using manageable materials, controllable, safe, and – ultimately - even
economic to build and operate.
As we shall see, the main lines of development of such commercial actors sprang from
the three partners in the Second World War 'atom-bomb' programme, the 'Manhattan
Project'. The. UK in  due course developed gas-cooled, graphite-moderated reactors; the
USA developed reactors cooled and moderated by ordinary 'light' water; and Canada
developed reactors moderated by heavy water, variously cooled. Both the UK and the
USA also began development of reactors using fast neutrons, with liquid metal coolant
and no moderator. Before we describe these and other reactors in detail, it may be useful
to identify some general aspects of reactor design.
To generate a given output of energy a reactor may have a very large volume of core,
with a comparatively low heat output per unit volume power density; alternatively it may
have a much more compact core with a higher power density. Natural uranium reactor
fuel has a low concentration of fissile nuclei; a reactor using such fuel must have a larger
volume of core than one using enriched uranium  - or plutonium  - fuel. A large reactor
costs more to build than a smaller one of the same output. On the other hand, natural
uranium fuel is much cheaper than enriched uranium fuel. What you lose by building the
larger, more expensive reactor you may subsequently save on fuel costs.
The energy output from a reactor can be measured directly as heat. this heat is used in a
'power reactor' to generate electricity, only a fraction of the total heat energy ultimately
reappears as electrical energy; the rest is discharged to the surroundings as
low-temperature at. In general, the higher the temperature the reactor can achieve, the
larger the fraction of energy that can be converted to electricity. As a rule only some 25
to 32 per cent of the total heat output is converted to electricity in systems now operating.
A system which converts 30 per cent of the heat to electricity is said to be 30 per cent
efficient - mainly because the remaining 70 per cent of the heat is not used. (This is not to
say that it cannot be used, merely that it is not.) Reactor energy outputs are accordingly
described either as heat  - for instance, 'megawatts thermal', MWt - or as electricity  - for
instance, 'megawatts electric', MWe. (A megawatt is one million watts.) A satisfactory
rule of thumb is to assume that, for a given power reactor, the output in MWe is between
one quarter and one third of the output in MWt. Unless use is made of the
low-temperature heat, the fraction MWe/MWt is a measure of the system's efficiency.
If a reactor core operates at a higher temperature, it produces steam of higher quality, and
generates electricity more efficiently. On the other hand, core materials which withstand
these higher temperatures are likely to be more expensive. Similarly, reactor fuel which 27
can be left in the core for a longer period at a higher temperature reduces the amount of
fuel needed for refuelling; but such fuel also costs more. A reactor which can be refuelled
'on load' -without having to shut down  - is less inconvenient for an electricity system, but
such refuelling arrangements are in general more expensive to build than those for
'off-load' refuelling.
The cooling system of a reactor may operate at a pressure anywhere from atmospheric up
to - at present - about 150 atmospheres. The higher the pressure, the heavier and stronger
must be the pressure system. This has implications not only for costs but also for safety,
since a rupture of the pressure system might have serious consequences, as we shall see.
Some designs enclose the reactor core in a pressure vessel of heavy welded steel; other
use pre-stressed concrete. Still others distribute the core materials in an array of much
smaller pressure tubes.
Interruption of cooling may be easier to control in a reactor of low power density than in
one of high power density, in which sharp temperature rises may occur with extreme
rapidity. Flaws or malfunctions in the pressure system may be easier to overcome in a
reactor with low coolant pressure than in one with high coolant pressure; a large welded
steel pressure vessel of complex geometry seems inherently more vulnerable to major
disruption than a pressure vessel of pre-stressed concrete, or a system made up of many
smaller pressure tubes.
 One large reactor  may cost less than two small ones producing the same total output  -
but, as we shall discuss, not necessarily, if the large reactor must add many extra items of
equipment for reasons of safety, standby and maintenance. All reactor designs have
shared one characteristic: a rapid increase in the size of successive reactors of the same
basic design. In reactors, perhaps more than in most other engineering technologies, a
change of scale is frequently not merely quantitative but qualitative, introducing a whole
new set of unknowns into the engineering. In the table at the end of this chapter we have
listed typical design parameters of different reactor types. In the succeeding sections we
shall describe them in more detail.
Experimental and Research Reactors
The first nuclear reactor was constructed amid tight wartime secrecy, in a disused squash
court under Stagg Field football stadium at the University of Chicago. Construction of the
reactor began in November 1942 and took less than a month. Bricks were machined out
of graphite. In some of the bricks were embedded balls of uranium metal or compressed
uranium oxide powder; uranium oxide had to be used because at that time only 5600 kg
of pure uranium metal were available. The graphite bricks were laid layer after layer onto
a growing pile of roughly spherical configuration inside a wooden supporting structure.
At intervals inside the pile were neutron-absorbing cadmium strips to ensure that stray
neutrons did not initiate a premature chain reaction. Instruments to measure
neutron-density were also included, and checked regularly to see how the pile was
progressing towards critical dimensions.28
By the time the 57th layer of bricks had been added it was clear that only the inserted
neutron absorbers were keeping the pile from criticality. By this time it was more than 6
metres high, with length and breadth to match, and contained about 36 tonnes of uranium
and over 340 tonnes of graphite.
On 2 December 1942 the scientists and technicians gathered on the balcony of the squash
court, watching the instrument readings, while Enrico Fermi called out instructions and a
young physicist named George Weil slowly pulled out the final control rod. Shortly after
2.30 p.m. the instruments recorded a steadily rising increase in neutron density in the pile.
The pile had 'gone critical': the first self-sustaining nuclear chain reaction was taking
place.
The heat generated in the pile was initially kept down to about 0-5 watts. But on 12
December the reaction-rate was allowed to increase until the heat generated - the 'power
level' - reached 200 watts. Further reactivity was available, but by this power level the
radiation from the pile was potentially harmful to personne l. Accordingly, in the spring of
1943, Chicago Pile No. 1 - CP-1, as it came to be called - was un-piled. Shortly
thereafter, rebuilt with added uranium and graphite inside adequate radiation shielding, at
a site outside Chicago, and re-christened Chicago  Pile No. 2, it could be operated at an
average power of 2 kilowatts (2 M) and intermittently up to 100 kWt. For some years,
until the name became totally inappropriate, any nuclear reactor was called an 'atomic
pile', after the first reactor.
CP-1 was the first true nuclear reactor. But even before it was built, thirty piles of less
than the necessary size and shape were built and tested. Such assemblies, which cannot
generate their own neutron supply without a supplementary source of neutrons, are called
'sub-critical assemblies'. Since the early 1940s countless hundreds of sub-critical
assemblies have been built and dismantled in many countries; and many true reactors
have been built for experimental or research purposes. The variety and range of designs
of experimental and research reactors is extensive, depending on the purpose for which a
particular reactor is constructed. Experimental and research reactors have a number of
uses. To bombard a sample of material with neutrons, the sample can be inserted  through
a suitable channel into a reactor core. The intention may be simply to study the effect of
neutron bombardment on the material - perhaps a material to be used in building a
reactor. Or the intention may be to convert some of the sample's stable nuclei by
absorption of neutrons into radioisotopes, for medical, industrial, agricultural or research
purposes. Some reactors have a 'thermal column': a graphite panel through the reactor
shielding, which allows a stream of thermal neutrons to emerge for research work outside
the reactor. (Although thermal neutrons are often called 'slow neutrons', their speed is
nonetheless about 2200 metres per second - considerably faster than a high-velocity
bullet.)
Some research reactors are designed to further the study of 'reactor physics' itself: neutron
densities, temperatures, the production of plutonium-239 from uranium-238, the build-up
of fission products, the effect of these fission products on reactivity, the performance of
new designs of fuel assemblies, the effects of 'unscheduled events' inside the reactor, and 29
so on.
Obviously, research reactors are also important for the training of qualified scientists and
technicians in the often extremely subtle and intricate - and potentially dangerous -
characteristics of reactor design and operation. Many countries now boast major centres
for reactor research and development. Nor are such reactors found only in heavily
industrialized countries. One of the longest-serving research reactors in the world, in
operation since 1959, is the 1 MW Trico reactor in Zaire.
A popular design of research reactor is the 'pool type'. It has a core of highly enriched
uranium at the bottom of a deep tank of water. The water acts as moderator, reflector,
coolant and shielding. It also allows a direct view of the core while the reactor is critical.
In no other reactor design is this possible. Because some of the radioactive emissions
from the reactor travel faster than the speed of light in water, the water in a pool-type
reactor glows with an eerie blue light called Cerenkov radiation.
The great majority of reactors, of whatever size or for whatever purpose, have been
experimental, in that every new reactor modification and development has had to be
designed and engineered on the basis  of previous experience, which in this field is often
inadequate or irrelevant or both. The US Atomic Energy Commission went so far as to
list all the reactors it licensed, for whatever purpose, as 'experimental' until 1971.
Plutonium Production Reactors
All uranium reactors produce plutonium, by neutron bombardment of uranium-238. The
first large-scale reactors were built expressly for this purpose: to produce plutonium for
nuclear weapons. A pilot model was built in 1943 at Oak Ridge, Tennessee. It could not
be constructed on the simple building-block principle that sufficed for CP-1; it would
have been not a little inconvenient to dismantle the entire reactor to recover the
plutonium. Furthermore, the rate of transmutation of uranium into plutonium depends on
the neutron density, which in turn depends on the rate of the chain reaction. If the
reaction is fast enough to create plutonium at a useful rate the heat generated becomes a
major problem. Complete fission of all the nuclei in one kilogram of uranium-235
releases about one million kilowatt-days of energy; each uranium-235 fission is likely at
most to initiate one further fission with one neutron and create one uranium-239 (and
hence plutonium-239) nucleus with another. That is, to create one kilogram of plutonium-
239 requires the fission of about one kilogram of uranium-235 - and the dissipation of all
that heat.
Accordingly, the Oak Ridge reactor was built in the form of a cube of graphite perforated
from one side to the other with parallel horizontal channels. Into these channels were slid
cylindrical slugs of natural uranium clad in aluminium. When a fuel slug had been
sufficiently irradiated it was pushed through the reactor, falling out of the graphite core
and into a tank of water, for subsequent processing. The fuel slugs fitted loosely in the
channels, leaving room for a flow of cooling air to remove the heat from the reaction
(eventually 3.8 MWt).30
Even while the Oak Ridge pilot model was still under construction, work began on the
first full-scale reactor, which was built on the bank of the Columbia river near the town
of Richland in Washington state. Construction of the first full-scale reactor, as tall as a
five-storey building, only took from June 1943 until September 1944. By early 1945
three full-scale reactors were in operation. The entire industrial installation, named the
Hanford reservation, was in due course to occupy nearly 1600 square kilometres, and
include nine production reactors, plus a vast array of ancillary plant. The Hanfor d
production reactors were similar in design to the Oak Ridge reactor; but their heat output
was so intense that cooling by gas  - helium was the original choice  - was found too
difficult. Cooling was accomplished by pumping water from the Columbia River directly
through a reactor core and back into the river.
After the end of the Second World War plutonium production reactors were built in the
UK, France and the Soviet Union. The UK production reactors were built on the Cumbria
coast, on the site of a disused ordnance factory which was renamed Windscale. Like the
Hanford reactors those at Windscale used natural uranium clad in aluminium, lying in
horizontal channels in a graphite core. The absence of a suitable water supply meant that
the Windscale reactors were cooled with air, blown by powerful fans through the cooling
channels in the graphite, and discharged through a stack 126 metres tall back to the
atmosphere. This once-through air-cooling was a far from desirable arrangement, whose
drawback was subsequently demonstrated in the dramatic accident in 1957 which
destroyed the Windscale No. 1 reactor.
If the purpose of a reactor is to produce fissile plutonium-239, the rate of plutonium
production can be 'optimized' by choice of core geometry, at the cost of other
performance characteristics. The reactor fuel must be changed at relatively short intervals
- months rather than years. By this time the fissile plutonium-239 in the fuel is playing a
significant part in the chain reaction, undergoing fission and thus being removed as well
as formed. Furthermore, some of the plutonium-239 absorbs one or more additional
neutrons without undergoing fission, becoming plutonium-240, plutonium-241, and
plutonium-242. Plutonium-240, which accumulates comparatively rapidly, is susceptible
to spontaneous fission, but is unlikely to fission when struck by a neutron, and cannot
therefore participate in a chain reaction. It is moreover virtually impossible to separate
from plutonium-239. Too high a fraction of the 240 isotope makes plutonium somewhat
unpredictable as a weapons material: hence the need to remove irradiated fuel before too
much plutonium-240 has been created. However, 'reprocessing' of fuel to extract
plutonium is an expensive and complex operation. Refuelling more often than is strictly
necessary to maintain a reactor's reactivity can only be justified within the remarkable
elasticity of military budgeting,
Gas-cooled Power Reactors
Magnox Reactors31
The first power reactors were of course, like the plutonium production reactors, military:
power plants for submarines, and multi-purpose reactors producing both plutonium and
electricity. (A 'nuclear submarine' is so called as much for its motive power as for its
cargo.) The first 'power reactors' so identified were started up in the USA and the Soviet
Union in 1954. The US reactor had an output of 2.40 MWe, and the Soviet APS-1 reactor
at Obninsk, now usually declared to have been the world's first power reactor, an output
of 5 MWe.

However, for obvious reasons, the general public heard little about the first US and 

Soviet power reactors. By default, if not by common consent, the world's 'first nuclear 

power station' was Calder Hall in the UK, whose first reactor started up in 1956. Calder 

Hall's claim to precedence is entirely defensible, if only because the first Calder Hall 

reactor, like its three successors, was a full order of magnitude larger than the Obninsk 

reactor, with an output of 50 MWe. On 17 October 1956 Her Majesty Queen Elizabeth II 

switched power from Calder Hall into the UK's National Grid: in a blaze of international 

publicity the age of 'nuclear power' - that is, electricity, not military might - was born.

The four Calder Hall reactors, on a site adjoining Windscale, were 'power' reactors only 

secondarily. Despite the fanfare and the Royal premiere the Calder Hall reactors, and the 32

four similar reactors built at Chapelcross across the Scottish border, were built and

optimized in order to produce weapons-plutonium to augment the output from the 

Windscale reactors. Nonetheless the Calder Hall and Chapelcross nuclear stations

became the cornerstone of the UK nuclear power programme. Their design

characteristics were developed and extended through the first generation of UK

commercial nuclear stations, eventually comprising a total of twenty-eight reactors,

including one in Italy and one in Japan. The nuclear patriarch of this family, the first 

Calder Hall reactor, is still going strong more than twenty-five years after its  first 

start-up. Many of the factors which affected its design and construction still preoccupy 

nuclear engineers.

Like the Windscale reactors this Calder Hall reactor uses natural uranium fuel and

graphite moderator. But their spatial arrangement is very  different, as are many other 

details. The fundamental difference is that the Calder Hall reactor has a closed-circuit 

cooling system, making it possible to recover heat from the reactor at a temperature and 

pressure high enough to be useful. The pressurized closed circuit system also ensures 

more efficient cooling, which in turn allows the chain reaction to operate and produce 

plutonium faster.

The heart of the Calder Hall design is a huge welded steel pressure vessel, enclosing the 

graphite reactor core which is pierced from the top to bottom by fuel channels. The 

Calder Hall fuel is clad, not in aluminium, but in a special magnesium alloy called

'Magnox', which is much less inclined to absorb neutrons, and is stronger and less 

susceptible to corrosion in the high temperature and neutron flux inside the reactor core. 

The entire family of reactors using such fuel has always been referred to as Magnox 

reactors.

The core contains an array of instruments which transmit readings of temperatures, 

neutron densities and other relevant data to the control room. Each sector of the core also 

has channels for several types of control rods which enter the reactor from above, held 

out on electromagnetic grapples, so that any reactor fault will shut off the magnets and let

the rods fall into the core to halt the fission reaction.

The pressure vessel, its contents and its attachments expand and contract with

temperature changes. The combination of the resulting thermal stresses, the gravitational 

stresses set up by the weight of the components, the vibration of moving parts and

fast-flowing coolant, and the somewhat unpredictable effects of prolonged intense

neutron irradiation presented the Calder Hall designers with a challenge whose equivalent 

still faces every nuclear engineer. The steel pressure vessel is itself enclosed inside a 

biological shield of concrete more than two metres thick. Assorted pipes and services 

pass through the biological shield; but, since the penetrating gamma rays and neutrons 

travel in straight lines, an appropriate arrangement of zigzags cuts off all out-coming 

radiation. The total weight of the reactor and its ancillary structures is considerable -

some 22,000 tonnes - and the site requirements are stringent; any subsidence might crack 

the concrete, reducing the effectiveness of the shielding.33

The hot coolant gas passes out of the reactor building through four cooling ducts into four 

towering 'heat exchangers' - a fancy word for boilers. Inside each is a labyrinth of tubing 

containing water; the hot carbon dioxide passes round the tubing, giving up its heat to the 

water which turns to steam and is used to drive turbo-generators. When the gas has given 

up its useful heat it emerges from the lower end of the heat exchanger, and passes into a 

gas circulator. This blows it back into the bottom of the reactor pressure vessel and up 

again through the fuel channels. Since the four loops of the cooling circuit are

pressurized, special provisions have to be made for changing fuel elements and for other 

maintenance inside the reactor core. Access to the channels for refuelling and servicing is 

from above, through holes in the horizontal roof of concrete shielding which is called the 

'pile cap'. On the pile cap, the working area above the reactor, are mobile 'charging' or 

'refuelling' machines, massive and complex assemblies.

The Calder Hall Magnox design - primarily for plutonium production  - is shut down and 

depressurized for refuelling. In the later Magnox designs for commercial nuclear stations 

it is not necessary to interrupt the operation of the reactor for refuelling; it can be carried 

out continually, a few channels per week, while the reactor is supplying power, 'on load'.

To change the fuel in the reactor, the 'discharge machine' is positioned over an access 

port, clamped onto the surface of the pile cap and pressurized. The shielding plug is 

removed, grapples extended down through a standpipe into the core, and the irradiated 

fuel elements lifted out of a channel and stored inside the thick walls of  the discharge 

machine - all by remote control, because of the radiation hazards. The shielding plug is 

replaced, the discharge machine depressurized and moved, and the charge machine,

loaded with fresh fuel, moved into position. The whole cycle is repeated, again by remote 

control, to lower new elements into place: clamping, pressurization, unplugging,

replugging, depressurizing and unclamping.

In either case, the irradiated fuel, intensely radioactive with fission products, is moved, 

inside the discharge  machine, to be dropped into a 'cooling pond': a deep tank of water, 

which serves to shield and cool the fuel while the more short-lived fission products 

within it decay to a less dangerous level of activity. After a suitable interval - usually 150 

days - the irradiated spent fuel is transported to Windscale for 'reprocessing'.

In all, eight Magnox stations, each with two identical reactors, were built for the Central 

Electricity Generating Board (CEGB), and one for the South of Scotland Electricity

Board.  Design details varied considerably from station to station, although all

incorporated on-load refuelling by means of a single charge -discharge machine. The 

Berkeley station's reactors use cylindrical steel pressure vessels, whereas the Bradwell 

station's reactors, built at the same time, use spherical ones. The Hunterston station is 

refuelled not from above but from below, where the temperature is lower. The different 

stations have different arrangements of heat exchangers and generating sets, different 

reactor buildings and so on. Instead of a water-filled cooling pond for its spent fuel the 

Wylfa station has three gas-filled storage magazines.Perhaps the most important variation 

in the Magnox stations is the power rating, which was increased progressively. To

accommodate this increase in power and size, the last two CEGB stations introduced a 34

major modification in design. Welding of a steel pressure vessel of more than a certain 

size to the stringently high standards necessary for a reactor becomes prohibitively 

difficult. Accordingly, the CEGB's Oldbury station embodied an entirely new approach. 

The pressure vessel was fabricated not from welded steel but from pre-stressed concrete, 

a much more manageable material for large and complex structures. In the Oldbury 

design not only the reactor core but also the heat exchangers and gas circulators are 

enclosed within the concrete pressure vessel. The pre-stressed concrete serves both as 

pressure vessel and as biological shield, the gas ducts are completely eliminated, 

removing one of the major escape routes for radioactivity in the event of an accident. The 

pre-stressed concrete design made possible a more than twofold increase in reactor size 

for the final  CEGB Magnox station, at Wylfa in Wales.

The power density of the Magnox stations, which averages about 0.9 kilowatts per litre, 

is, by nuclear standards, low. Because of its low density and consequent low thermal 

capacity, gas is a less efficient coolant than liquid; accordingly, the rate of heat

generation in a gas-cooled core must be kept low. (This in turn imposes an overall limit 

on maximum feasible heat output since high output entails a very large volume of core, 

with accompanying engineering complications.) Another characteristic of interest is the 

'specific power': power generated per unit mass of fuel. The specific power of Calder Hall 

fuel is about 2.40 kilowatts per kilogram of uranium; that of Wylfa fuel is about 3.16 

kilowatts per kilogram of uranium. Specific power is also sometimes called 'fuel rating'. 

The 'burn-up' of fuel is the cumulative heat output per unit mass; it is commonly 

measured in megawatt-days per tonne of uranium. The burn-up is of course a measure of 

how many fissions have occurred inside a given amount of fuel.

One of the main objectives of fuel designers is to achieve higher burn-up - that is, to be 

able to leave fuel in a reactor longer, before it becomes too distorted and too burdened 

with fission products to function properly. The limitations on burn-up of Magnox fuel are 

numerous. Natural uranium metal has a complicated crystal structure, and undergoes a 

variety of unwelcome changes at high temperatures and intense neutron fluxes. A

burn-up of about 5000 megawatt-days per tonne of uranium is about the best that can be 

comfortably attained by Magnox fuel. This limitation was one of several factors which 

eventually terminated the Magnox programme and provoked a search for another

approach.

The only other major nuclear power programme to opt for gas-cooled reactors was the 

French. The small French power reactors at Marcoule and Avoine started up in 1958. The 

second unit at Avoine - Chinon-2, a 200-MWe reactor - went critical in 1964, and France 

thereafter built, in all, seven gas-cooled power reactors with graphite moderator and  a 

70-MWe gas-cooled reactor with heavy water moderator. But French interest then swung 

abruptly away from gas-cooled to light water designs, in partnership with American

reactor-builders .

Advanced Gas-cooled Reactors (AGRs)

Even while the first Magnox stations were barely under construction, work commenced 35

on a second-generation design of gas-cooled power reactor: the advanced gas-cooled 

reactor (AGR). The aim was to achieve higher gas temperatures to improve the efficiency 

of electricity generation; higher fuel ratings to make the reactor more compact- and 

higher burn-up to reduce the frequency of refuelling. The temperatures achievable with 

Magnox fuel are limited by the characteristics of Magnox alloy and of uranium metal. 

Uranium metal undergoes a crystalline change of phase at 665 C, accompanied by

marked expansion; its behaviour even below this temperature is complex, since it

expands at different rates in different directions with increasing temperature. The melting 

point of Magnox is about 645 C; as well as melting at this temperature Magnox may also 

catch fire.

Accordingly, higher-temperature fuel must use some other form of uranium. The form 

most commonly chosen is uranium dioxide, UO2, often just called uranium oxide.

Whereas uranium metal melts at 1130 C, uranium oxide melts only at 2800 C. However, 

uranium oxide has a low thermal conductivity, much lower than that of uranium metal. 

When uranium metal is undergoing a fission reaction, its high thermal conductivity

means that the temperature is more or less uniform through the whole thickness of a fuel 

rod, even if this is several centimetres. The same is not true of uranium oxide. If solid 

uranium oxide is undergoing fission, the heat generated in the interior does not readily 

make its way to the surface, the interior is much hotter than the surface. Uranium oxide 

fuel elements must have a smaller diameter than metallic uranium elements, even though 

the melting point of uranium dioxide is so much higher.

The basic building block of uranium oxide  fuel is usually a pellet made by compressing, 

baking or otherwise persuading uranium oxide powder to assume the form of a small hard 

cylinder, about the size of a Liquorice Allsort. A column of such pellets, anything up to 

several metres in length depending on the fuel design, is stacked inside a thin-walled 

metal tube, to make a 'fuel pin'. The tube must be of a material which can withstand high 

temperature. Some oxide fuels use an alloy of zirconium called - unsurprisingly -  

zircaloy, which has advantages but is expensive; the usual alternative is stainless steel, as 

is the case with fuel for the advanced gas-cooled reactor. Stainless steel involves a further 

problem; it is strong and well-behaved structurally, but it has an unhealthy appetite for 

neutrons. Accordingly, the percentage of uranium-235 in the uranium oxide must be 

increased above its natural level: that is, the uranium oxide must be enriched. In AGR 

fuel the uranium is usually enriched to about 2 per cent.

The first AGR incorporating this type of fuel was a small 28 MWe prototype built at 

Windscale, which started up in 1962. It was followed in due course, and after many 

misadventures, by a programme of five full-scale twin-reactor stations, and then by two 

more twin-reactor stations. 

The basis of the overall AGR design is the pre-stressed concrete pressure vessel first 

developed for the last two Magnox stations. Like a Magnox reactor, an A G R has a core 

of machined graphite, under a dome like a huge steel bell-jar, with a large number of 37

openings at the top, through which pass the standpipes for access to the fuel channels. 

Outside the dome - but still inside the pressure vessel - are the heat exchangers or boilers, 

and below them the gas circulators.

The amount of fuel in an AGR is considerably less than in a Magnox reactor of 

comparable output, while the fuel rating is considerably higher. The coolant gas emerges 

from the fuel channels at a temperature of  around 650 C, more than 300 C higher than 

normal Magnox operating temperature.

An AGR is refuelled by a single refuelling machine, which pulls an entire fuel string of 

eight elements out of the reactor at once. Accordingly, the refuelling machine is itself the 

height of a four-storey building, and the reactor building must be built like an aircraft 

hangar to accommodate it. A single machine serves both reactors at a given station, 

moving between them, the fuel store and the spent fuel pond on a gantry or rails. The 

AGRs were designed to be refuelled on load; but this has been one of many technical 

problems which still remain short of fully satisfactory resolution.

High Temperature Gas-cooled Reactors (HTGRs)

Anyone desiring a heat source will have two objectives in mind: the total amount of heat 

output per unit time (that is, the total power), and the temperature at which the heat is 

made available. There is an unimaginable amount of heat in the ocean; but its low 

temperature makes it of little overt use.  While reactor designers were scaling up reactor 

sizes, to increase their power output, they were also pressing on towards much higher 

temperatures. Even an AGR operating flat out is only a so-so source of heat, as far as 

temperature is concerned. It can be used to raise passable steam to run a turbo-generator 

and produce electricity, but only at a moderate efficiency. More elegant industrial

applications are ruled out by the low temperature of the heat.

The limitation on the temperature at which heat is generated is nothing - or almost 

nothing  - to do with the chain reaction system itself. Under the right circumstances a 

chain reaction can run at temperatures anywhere up to those in the heart of a nuclear 

explosion  - millions of times higher than those in fossil-fuel boilers. However, long 

before such temperatures are reached it becomes peculiarly difficult to keep the whole 

assemblage in any semblance of order. We have already noted the awkwardness of 

uranium metal and Magnox cladding at temperatures over 60TC; other reactor materials 

present similar problems, albeit at variously higher temperatures. Clearly if really high 

temperatures are to be permissible, without having the entire reactor core bulge and warp 

itself hopelessly out of shape, or undergo unpleasant chemical reactions, a new approach 

is necessary.

The new approach which has received the most attention is one which dispenses entirely 

with metals in the reactor core, in favour of sophisticated combinations of refractory 

ceramic materials able to withstand without protest temperatures well into the thousands 

of degrees Celsius. Highly enriched uranium in tiny particles is blended intimately with 

the ceramic. Some designs proposed include also particles of another element called 

thorium. Thorium has nuclear properties rather like those of uranium-238. Natural

thorium is almost entirely thorium-232. A nucleus of thorium-232 can absorb a neutron to 

become thorium-233, which then emits two beta particles to become uranium-233, which 

is fissile. The process is directly analogous to that by which uranium-238 is transformed 

into plutonium-239. Uranium-233, like uranium-235, undergoes fission when struck by a 

slow neutron and in turn produces  more neutrons, to sustain a chain reaction. Uranium-

238 and thorium-232, although not fissile materials, are called 'fertile' materials, because