In an attempt to overcome some of the disadvantages of water, particularly its j low boiling point and consequent high operating pressure, reactor systems

f. have been proposed employing various organic fluids. In practice, only one

§ group of compounds, the polyphenyls, have proved sufficiently resistant to neu-

| tron radiation to be of interest. In general, these coolants are mixtures of

f polyphenyls chosen so that they remain liquid at room temperature. It is possi­

ble to operate these coolants as liquids in excess of 300°C at operating pres­sures of about 10 bars, compared with 155 bars required for water. In their pure form, these coolants are essentially noncorrosive to common reactor materials.

The big problem with organic coolants is that although they are relatively re­sistant to degradation by irradiation and thermal degradation, these processes

still occur to a significant extent. Radiolysis causes the formation of hydrogen and gives rise to a breakdown phenomenon called hydrogen embrittlement of the fuel canning. Also, the irradiation leads to the formation of polymers (mate­rials of very high molecular weight), which deposit as a solid on the fuel ele­ments. Although reactors have been operated with such coolants, they have not found general acceptance in commercial systems.

5.3.1 The NRX Incident

The NRX reactor in Chalk River, Canada, is an experimental reactor, in some re­spects a forerunner of the present CANDU reactors. It was designed to operate at a full power of 40 MW(t), and the layout of the fuel channels is illustrated in Figure 5.21. Single fuel rods are cooled by light water flowing in an annulus be­tween the rod and a pressure tube, which in turn passes through a calandria tube mounted in a tank of heavy water, which acts as the moderator.

On December 12, 1952. the reactor was undergoing tests at low power. The circulation flow of the light-water coolant was reduced in many of the rods since not much heat was being generated in the fuel. Noting that several red lights indicating withdrawn control rod positions suddenly came on, the super­visor went to the basement and found that an operator was opening valves that caused the control rod banks to rise to their fully withdrawn positions. He im­mediately closed all of the incorrectly opened valves, after which the rods should have dropped back in. Some of them did, but for unexplained reasons, others dropped in only enough to cause the red lights to turn off. The latter rods were almost completely withdrawn.

From the basement, the supervisor phoned his assistant in the control room, intending to tell him to start the test over again and to insert all the control rods

by pushing certain buttons. A misunderstanding resulted in the wrong button’s being pressed, but the operator in the control room soon realized that the reac­tor power was rising rapidly and he pressed the “scram” button to trip the reac­tor. The control rods should then have dropped in under the action of gravity, but many of them did not, and the power continued to climb. After a hurried consultation, it was decided to dump the heavy-water moderator from the ca- landria tank; this shut down the reactor, but not very quickly since it took some time to drain. The reactor power had peaked between 60 and 90 MW(t).

The increase in power, coupled with the low flow in some of the fuel chan­nels, caused boiling of the light water, which increased the internal pressure and caused the coolant pipes to rupture. The situation was exacerbated by the fact that loss of the light water from the fuel channels gave an increase in reac­tivity and increased the initial power pulse. Some fuel melting was experienced, and the heavy-water calandria tank was punctured in several places. About I million gallons of water containing about 10,000 curies of radioactive fission products had been dumped into the basement of the building.

The core and the calandria, which were damaged beyond repair, were re­moved and buried, and the site was decontaminated. An improved calandria and core were installed about 14 months after the incident.

The main lesson learned from this incident was that absolute security of control rod operations is mandatory, and modern systems go to great lengths to achieve this. The incident was made worse by the fact that this kind of system has a positive void coefficient, so that the natural event (i. e., the boiling of the water due to heat being input into it) leads to an increase in neutron population.

In Section 1.4 we described a typical fission reaction, producing atoms of bar­ium-141 and krypton-92 by the fission of a uranium-235 atom. In practice, fis­sion products range in atomic mass from about 80 to 160. For each kilogram of fissile material converted, a certain percentage is converted to one pair of fis­sion products, a certain percentage to another, and so on. The percentages of the fission products formed may be plotted as a function of atomic number (Figure 8.1). Typically, there are about 40 possible fission reactions producing about 80 different species of fission product. The half-lives of these species vary from a fraction of a second to 30 years or more. The short half-life materials are not important since they decay rapidly inside the reactor and during the storage period after removal from the core.

In discussing the significance of radioactive fission products in the environ­ment, it is usual to focus attention on those that are likely to be the most trou­blesome—in particular, the isotopes that if released would be absorbed and concentrated in specific rgans of the body. For example, various radioactive iso­topes of iodine that are formed in the fission reaction, or are subsequently formed by decay of other fission products, can concentrate in the thyroid gland. The iodine isotopes of main interest are I 131 (half-life, 8 clays), I 152 (half-life, 2.3 h), and I129 (half-life, 20 million years). In general, the longer the half-life, the less intense the radiation. In an accidental release it might be expected that io­dine would deposit on grassland, be eaten by cows, appear in the milk, and be taken up by people drinking milk, especially children. For this reason, the be­havior of iodine has received detailed attention in nuclear safety studies, and there are plans that in the highly improbable event of a serious release, milk from affected areas will be collected and disposed of.

A useful concept in considering the hazards of radioactive fission products is that of biological half-life. This is the time needed for any particular radioactive element, taken into the body, to be reduced to half its level of natural excielii >n

processes. The significance of this concept can be appreciated by comparing two of the most important fission products, caesium-137 and strontium-90. These isotopes have radioactive half-lives of approximately 30 years. However, the biological half-lives are veiy different, around 70 days for caesium and 50 years for strontium. The long biological half-life of strontium is due to the fact that it accumulates in the bone structure. Thus, strontium is considered a more serious hazard than caesium.

A material of great interest in radiological protection is plutonium-239, which also has a long biological half-life (200 years in the bone structure and 500 days in the lung). Since the radioactive half-life of plutonium is about 25,000 years, the effective half-life in the body is dominated by the biological half-life.

Another important radioisotope, tritium, is emitted in small quantities from water reactors and reprocessing plants. It is formed by the process of ternary fission, in which three, rather than the usual two, fission products are formed. The third fission product is often tritium, and since its molecular size is very small, it can diffuse through the canning material into the coolant circuit. It emits beta radiation and has a radioactive half-life of 12.6 years. Its biological half-life is around 12 days.

Discharge of fission products into the environment is very strictly controlled, and the authorized release rates for specific isotopes are calculated on the basis of the permissible dose to individuals, which is, of course, well below that which might cause any significant health effect.

Nuclear reactions also produce heavy elements (actinides) whose atomic weight is equal to or higher than that of the uranium isotope from which they are formed. Examples of these actinide elements are the plutonium isotopes, the most important of which is Pu239, a major fissile material. Other plutonium isotopes formed by neutron capture are Pu240, Pu241, and Pu242. Other actinide elements formed in the nuclear reaction include americium-241, americium — 243, and curium-244. The actinide elements are important in nuclear waste be­cause of their relatively long half-lives, ranging from 17 years for Cm244 to

25,0 years for Pu239. Thus, these actinides require long-term storage, on time scales of 1000 years or more, after discharge from the reactor (see Figure 8.2).

It is often asked: At what time may radioactive waste products be considered safe? A common answer is that the products may be assumed to be safe when their toxic hazard is comparable to that of the original ore from which the fuel was derived and, ultimately, the wastes were generated. A plot of the ratio of the hazard of radioactive waste to that of the original ore is shown in Figure 8.2 for several cases:

1. Fuel discharged from a light-water reactor without reprocessing. Here, the hazard of the waste falls below that of the original ore after 10,000 years.

2. Waste arising from normal reprocessing in which 0.5% of the uranium and plutonium are assumed to be contained in the waste. The hazard level falls to around that of the original ore after about 500 years. It is assumed that 99.5% of the plutonium extracted is used in fast reactors and that the fuel from these reactors would be reprocessed, giving a somewhat similar curve in terms of hazard as a function of time.

3. Waste from thermal reactor fuel in which the plutonium from the reprocess­ing plant has been incorporated. The hazard is intermediate between those in the first two cases above and falls to that of the original ore after about 1000 years.

Figure 8.2 shows that the hazard falls rapidly after about 100 years for all the cases, reflecting the decay of significant amounts of the shorter-lived fission products. The heat generation rate from the waste products (Figure 7.5) follows curves similar to those in Figure 8.2.

We see, therefore, that after an extended period of time the hazard level of the waste from a given reactor will fall below that of the natural ore sources from which the reactor fuel was derived. Thus the nuclear program would, in the long nm (after the cessation of operation of fission power plants), margin­ally reduce the amount of radioactivity on the earth. However, we must face the need for safe isolation of the waste products during their highly active initial phase, lasting about 1000 years. On a geological time scale, these periods are very short and would not present any difficulty provided care was taken in the placement of the material. In Chapter 1 we discussed the naturally occurring re­actor at Oklo. In that case, some of the fission products stayed in the vicinity of the reactor and did not migrate away from it, even though no special precau­tions were taken to contain them.

The release of energy from naturally occurring radioactive isotopes is far too slow to make them a practical energy source in themselves. However, a much more rapid release of energy is possible through the process of nuclearfission, which is illustrated in Figure 1.7. A neutron from a decay process may collide with a heavy nucleus (e. g., uranium), causing it to split into small nuclei (fission products) while releasing several more neutrons. These neutrons, in turn, can cause further uranium atoms to split. For a small piece of uranium this process win not be self-sustaining, because the neutrons escape from the surface. However, the bigger the piece of uranium, the greater the chance of the neu­trons being absorbed, and a self-sustaining sequence (called a chain reaction) can be set up if a large enough mass (i. e., a critical mass) is available.

The release of energy in the fission process may be illustrated by considering

Neutron

Neutrons

Fi^^e 1.7: Diagrammatic view of fission process.

the fission of a 235U atom, which splits up into barium and krypton atoms and releases three more neutrons:

235u + n —— > 141Ba + 92Kr + 3n

If we could weigh components in this reaction we would find that those on the right-hand side of the equation weighed 0.091% less than those on the left-hand side. Thus, during the reaction, approximately 0.1% of the original mass is con­verted into energy. This energy appears as kinetic energy of the fission products and neutrons, which then collide with surrounding atoms and increase their thermal vibration, that is, release heat. For each kilogram of ^U totally fis­sioned by the above reaction, 80 million million (8 x 10^) joules are released. This is equivalent to the energy available from 3000 tons of coal.

Uranium-235 is described as a fissile isotope; unfortunately, naturally occurring uranium consists mainly (99.3%) of a nonfissile isotope, ^U. Thus, only a small part of natural uranium can be burned in the fission process to pro­duce energy. The proportion of №U in natural uranium is 0.71% by weight, and thus 1 kg of natural uranium is equivalent in energy potential to about 20 tons of coal. However, the energy potential of uranium can be increased about 100­fold by conversion of the nonfissile 238U into another fissile material, namely, plutonium-239 (a9J>u). We shall return to this below.

The three neutrons emitted in the above fission reaction have an initial velocity of typically 20,000 km/s (about 6% of the velocity of light). Although these fast neutrons can interact with other atoms of ^U, their chance of doing so can be increased by about 1000-fold if their velocity can be reduced, say, to 2km/s. These slower-moving neutrons would have a velocity similar to that of atoms vibrating due to thermal motions, and hence they are often called thermal neutrons. Nuclear reactors using the fast neutrons are often termed fast reactors, and those using the slower neutrons are termed thermal reactors.

Fast neutrons are converted into thermal neutrons as a result of a series of collisions with surrounding atoms. If a fast neutron nfts a large atom, it tends to bounce off and lose only a small amount of its energy. However, if it hits a small atom such as hydrogen or carbon, it will lose a significant fraction of its kinetic energy. (An analogy may be made to the motion of balls on a biliard table. If a ball hits the massive cushion of the table, it bounces off with very little loss of ve­locity, or kinetic energy. If it hits a stationary ball, it may lose a large proportion of its kinetic energy, which is transferred to the other ball in the collision.) Thus, to convert a fast neutron to a slow or thermal neutron requires about 2^Ю suc­cessive collisions with uranium atoms but only about 20 collisions with the light­est atom, hydrogen. In the collision process the neutrons are sometimes absorbed without leading to a subsequent fission. Moreover, each successive collision may lead either to a fission reaction or to the neutron’s combining with the atom with which it is colliding to make another isotope. Thus, there is an advantage in sur­rounding the uranium with lighter material that can lead to the conversion of fast neutrons to thermal neutrons, which can then pass back into the uranium; this process is known as moderation, and the light material used is termed a moder­ator. Moderators used in therrmal reactors have included hydrogen (in the form of its oxide, water), the hydrogen isotope deuterium (also in the form of its oxide, heavy water), and carbon (usually in the form of graphite). The best moderator is heavy water, which absorbs neutrons only weakly. However, heavy water is an expensive material and it is often preferable to use ordinary (light) water even though it absorbs neutrons much more strongly.

Because 238U absorbs neutrons, it is not possible to produce a self-sus­taining chain reaction by simply assembling a large enough mass of natural ura­nium, which is 99.3% 238U. However, if pieces of natural uranium are distributed within heavy water or graphite, the neutrons produced in the fission reaction are converted to thermal neutrons (which, as mentioned above, are 1000 times more effective than fast neutrons in continuing the chain reaction), and a self-sustaining chain reaction is possible. This idea was first demonstrated by Enrico Fermi at Stagg Field, Chicago, on December 2, 1942; Fermi employed pieces of uranium distributed in a “pile” of graphite. Light water cannot be used to sustain a chain reaction with natural uranium because of the high absorption of neutrons by hy­drogen. However, nuclear reactors may be constructed with light water as a mod­erator provided the concentration of 235U is increased from 0.71 to about 3%.

As we shall see later, various generic types of nuclear reactors have arisen from the various possible combinations of fuel and moderator. These can be classified as follows:

1. Heavy water-moderated, heavy water-cooled reactors. These are the basis of the Canadian line of development and are called CANDU reactors.

2. Graphite-moderated, gas-cooled natural uranium reactors. These are the basis of the British Magnox reactors.

3. Light water-moderated, light water-cooled reactors with fuel enriched in uranium-235. These are the basis of the U. S. boiling-water reactor (BWR) and pressurized-water reactor (PWR) development.

The further development of the British system, the advanced gas-cooled re­actor (AGR), uses graphite as a moderator and a somewhat enriched fuel to compensate for the fact that the fuel is contained in stainless steel, which ab­sorbs a significant fraction of the neutrons.

We shall describe and discuss all the above reactors, and in particular their cooling problems, in the following chapters. However, before doing so, it may be interesting to glance back to prehistoric times. The light water-cooled and — moderated reactors have been around far longer than one might imagine. In fact, the invention of Enrico Fermi was preempted by nature approximately 2 billion years earlier. In 1972, evidence was found of the dormant remains of a natural fission reactor located at Oklo, in the West Mrican Republic of Gabon. This naThral reactor operated for a period of hundreds of thousands of years. Its existence was discovered by an intriguing piece of detective work by French nuclear scientists.

In May 1972, H. Bouzigues obtained a curious result during a routine analy­sis of standard samples of uranium ore from Gabon. He found that they con­tained about 0.4% less 235U than expected. This was not due to an error in his analysis or to a natural variation. On this planet, at any particular time, the ratio of 235U to 238u is fixed; some other explanation had to be found for the discrep­ancy. A careful investigation carried out by the French Commissariat a l’Energie Atomique (CEA) traced the abnormal ore to one particular location in Oklo. It was concluded that the deficiency in 235u could be explained only by the oc­currence of a natural fission reaction at the site. At the time this natural reactor was operating, the ore was buried deep underground and natural groundwater served as a moderator and to some extent as a coolant. Such a reactor would not be possible with the present-day concentration of 235u in naturally occurring uranium, as we explained above. However, it should be remembered that the half-life of 235u is about 700 million years and that of 238u is about 4.5 billion years. Thus, in prehistoric times, the concentration of 235u was much higher than it is today. When the earth was formed some 4.6 billion years ago, the con­centration of 235u in natural uranium was about 25%, and it had decreased to about 3% at the time when the Oklo reactor was operating.

It is thought that the natural reactor at Oklo operated under considerable pressure and temperature and that the rate of reaction was controlled by vari­ations of the water (moderator) density. Cooling was provided mainly by con­duction, with some limited circulation by permeation. The power level is estimated to have been somewhat less than 100 kW and the total energy re­leased over the period of operation to have been about 4.7 x 1017 joules (15,000 MW-years), representing the fission of about 6 metric tons of 235u.

This amount of energy is about that released in a modern pressurized-water reactor in 4 years.

It is possible that a combination of local circumstances may have led to other naturally occurring reactors. Though the search continues, none has been lo­cated so far. Such naturally occurring reactors have been impossible for the past 2 billion years because the 235U concentration has been below the required 3%. A detailed review of the Oklo reactor phenomenon is given by Cowan (1976). It is interesting to note that 2 tons of plutonium-239 would have been produced at the Oklo natural reactor, though the amount that remains is infinitesimal be­cause of the comparatively short half-life (25,000 years) of 239pu. Thus, it cannot be claimed that plutonium is an entirely human creation.

2.1 INTRODUCTION

A modern large nuclear power plant is a very complex piece of engineering with a wide diversity of components. In the design of such plants, careful considera­tion must be given to the effect of breakdowns of these components. In this chapter we shall be primarily concerned with those component breakdowns, or combinations of component breakdowns, that can give rise to an interruption in normal cooling. When such an interruption occurs, the fission reaction is rapidly terminated, but as we saw earlier (Section 2.2, particularly Table 2.2), heat gener­ation continues after shutdown of the fission reaction due to the continuing decay of the fission products that have been generated. All reactor systems are provided with alternative means of cooling in order to remove this fission product decay heat in the event that the normal cooling system fails to operate. In Chapter 6 we shall consider the consequences of the alternative cooling system itself failing to operate, although this is a very remote possibility.

The design of a nuclear power station must encompass a number of opera­tional states that can occur during normal operation of the reactor or as a result of some kind of fault. These operational states may be classified as described below and are summarized in Table 4.1.

1. Normal operation and operational transients. In addition to the normal op­erational state, as described, for instance, in Chapter 2 for the various reactor systems, the designer must think about transients that occur during opera­tion. The term transient implies a nonsteady state of operation encountered in proceeding normally from one steady operating state to another. An ex­ample would be bringing the reactor from a “cold” condition up to full — power operation. This kind of operational transient must be taken into account in the design and the methods for achieving it worked out in the op­erating instructions for the reactor. For instance, to avoid damaging the struc­ture of the reactor, there may be limits to the rate at which the temperature of the structure can be increased or decreased. To ensure economic opera-

tion of the reactor, many components that might require frequent mainte­nance are duplicated, and the rules for operation of the reactor must be care­fully worked out to ensure that safe operation can be maintained even when some of the components are out of service. In order to operate the reactor economically, consideration must be given not only to the steady state but also to all the things that are likely to happen as a matter of course in the op­eration of a complex engineering plant.

2. Upset conditions. The word upset is used to describe all the kinds of faults that are not expected during operation but that can be reasonably expected to occur during the lifetime of a plant as a result of a variety of external events. Consider, for example, the case of lightning striking the power lines leaving the plant. A plant generating 1000 ^W of electricity suddenly has no means of exporting this electricity to the grid. When electricity is no longer taken from the generator attached to the steam turbine, the turbine will in­crease in speed unless rapid action is taken to prevent such an occurrence. This action is to stop the flow of steam to the turbine and divert it directly into the condenser. The steam flow is reduced as rapidly as possible by using the control rods to stop the fission reaction—’’tripping” the reactor. A turbine trip of this kind might be expected to occur for one reason or an­other about once every year, and it is important to design properly to ac­commodate it.

An interesting consequence of such a trip is that the power station, instead of being an exporter of electricity, immediately becomes an importer of elec­tricity in order to drive the coolant pumps, instmmentation, and emergency cooling systems for the reactor. If the external power line has been broken, it is likely that no electricity can reach the site. Since the reactor has been tripped, it is no longer generating electricity and emergency power genera­tion systems must be provided. These are usually diesel-driven generators, and normally several of them are installed in case one is being serviced or

Table 4.1 • Classification of Reactor Operating States and Frequency of Occurrence

Operating states for which the system is designed to cope:

Normal operation

O^perational ^^«ents

Upsets

Emergencies

limiting fault conditions (including design basis accident, DBA)

Unprotected or beyond design basis accidents

fails to operate. This example of an upset transient is one of many that must be accounted for in design; others include loss of cooling water to the con­denser due to the failure of a cooling-water pump, loss of feedwater to the steam generator, and reactor coolant pump trips.

3. Emergency events. Although operational transients are certain to occur and up­sets are practically certain to occur during the lifetime of a plant, a number of events can be postulated that might have, say, a 1-in-10 chance of occurring in the lifetime of a particular plant. If we consider a sample of 10 plants, it is prac­tically certain that one of these events would occur within 1 of the plants dur­ing its lifetime. A large country such as the United States has more than 100 reactors in operation; therefore, emergency conditions are likely to occur within one of the plants every few years. The reactor design must cope with such emergencies, although some damage to plant components may be ex­pected as a result of the incidents. An emergency event would occur, for in­stance, as a result of breaks in small pipes in the reactor circuits, relief valves being stuck open, or fires within the plant electrical systems.

4. Limiting fault condition. It is possible to conceive of events, such as an earthquake, the complete severance of a main inlet pipe, or the complete severance of a steam line from the steam generator to the turbine, that would represent a severe accident to a reactor. Even though some accidents might occur only once in 10,000 years of reactor operation (though with 100 reac­tors operating, such an event might occur once every 100 years), reactors must be designed to meet these so-called limiting fault conditions safely. Al­though an emergency event (as described above) would not give rise to any release of activity off the reactor site, a limiting fault condition could give rise to extensive failure of the fuel canning and some consequent release of radioactivity off the site. The regulations set down by the national licensing bodies limit this release of radioactivity to a level that would not represent any significant risk to the public.

The reactor must be designed to meet the above operating states. Certain faults—for example, those related to coolant circulation pumps or gas circula­tors—can give rise to an interruption in normal cooling. ^ben such an inter­ruption occurs, the reactor is shut down by its automatic safety systems. But as we saw earlier, heat generation continues after shutdown of the fission reaction due to the continuing decay of the fission products that have been generated: the decay heat. So all reactor systems are provided with alternative means of cooling in order to remove this decay heat in the event that the normal cooling system fails.

The two most important safety systems are those associated with stopping (“tripping”) the fission reaction within the reactor (the control rods) and those associated with providing an alternative cooling system, the so-called emer­gency core cooling system (ECCS). These engineered safety systems need to be brought into operation reliably when required.

This is done as a result of instrnmentation signals received from sensors lo­cated around the plant that indicate when an unsatisfactory condition is being approached. They then initiate the action of the safety systems. This total reac­tor protection system has to be highly reliable. Such reliability is achieved through:

1. Duplication. Several sensors are used to measure critical parameters and sev­eral signal processors used to evaluate the signals. If all the sensors and processors are working, some of them are redundant. In a typical protection system, there are four identical sensors whose readings are compared. If two sensors of the four give identical signals requiring the activation of the safety system, action is taken. This allows for failure of two of the four systems.

2. Diversity. Different systems parameters are monitored to provide an indica­tion of the same form of fault condition. Thus, two completely different sig­nals, e. g., pressure and temperature, can be used to trip the reactor and/or initiate the emergency core cooling system for the same fault.

In some designs—for example, the British P^WR Sizewell B—the reactor pro­tection system itself consists of two diverse systems: the primary protection sys­tem and the secondary protection system. The primary protection system is a microprocessor-based system that provides reactor trip and actuation of the en­gineered safety systems. The secondary protection system utilizes magnetic logic relays to initiate the reactor trip and engineered safety systems indepen­dent of the primary protection system.

As to the physical trip systems, advanced gas-cooled reactors have two sep­arate systems for terminating the fission reaction: the first based on the control rods and the second on the injection of nitrogen, which is a neutron absorber, into the reactor gas.

Pressurized water reactors (P^^s) will shut down automatically if cooling water is lost from the core since this water is also the moderator. There are of course other systems. The normal one is based on control rods and the injec­tion of boric acid, a neutron absorber, into the reactor cooling water. In addi­tion, in some designs (Sizewell B) there is a completely separate second system for injecting very quickly large quantities of boric acid to deal with particularly severe faults.

Despite all these attempts to reduce the probability of failure of the protec­tion system, it is difficult to demonstrate that such systems have a better relia­bility than 1 failure for every 10,000 times they are called into operation.

However, since the protection system itself is seldom called into operation (i. e., about once a year to meet an upset), the chance of failure is still remote.

It is now important to identify the vital supporting role of the operator. For faults of the type described above, no claims whatever are made on the opera­tor for the detection of the fault and the safe shutdown of the reactor. This is achieved entirely automatically with high reliability. The role of the operator is really a management task of information gathering, planning, and decision making and only occasionally calls for more active control when routine oper­ation is disrupted. Operators are highly trained—on simulators and on actual plant—and are regularly tested for competence.

It is a requirement in the design of the most recent British reactors that the operator should not need to intervene to control an abnormal condition for a period of at least 30 minutes after it begins. The automatic systems are designed to achieve this. During this period the operator needs essentially to monitor the proper functioning of the safety systems. He takes action only if the response of these systems is judged inadequate for some reason.

Operational transients, upsets, emergency events, and limiting fault ccindi — tions, as defined above, represent the range of conditions against which the plant is designed. The most serious of these conditions, the limiting fault condi­tion, is often referred to as the design basis accident (DBA). It is possible to conceive of accidents that are more serious than the DBA and against which the reactor is relatively unprotected. Examples of such accidents are as follows:

1. Events that can be postulated but that are considered to be so unlikely that there is no justification for protecting the reactor against them. These might in­clude the occurrence of a large earthquake in a zone where earthquakes do not normally occur and the direct crash of a large aircraft into the reactor with simultaneous destruction of the containment and the protection systems.

2. The occurrence of an upset, emergency condition, or limiting fault condition with the simultaneous failure of the protection system and/or the safety sys­tems (for example, the emergency core cooling systems, ECCS). As indicated in Table 4.1, limiting fault conditions might occur once every 10,000 years. If the probability of failure of the emergency core cooling system was once in every thousand demands, a very severe accident leading possibly to the melting of the reactor core would occur once every 10 million years of reac­tor operation (i. e., 10-4 events per reactor year multiplied by 10-3 failures of the ECCS per event).

3. Although the designer tries to envisage all conceivable operational transients, upsets, emergency conditions, and limiting fault conditions, it is nevertheless possible that some event may happen that was not thought of. The most un­predictable events are those that involve a sequence of multiple failures cou­pled with unanticipated responses from the reactor operators. It was this kind of sequence that occurred in the Three Mile Island accident, which we shall describe in more detail in Chapter 5.

Having discussed in this and previous chapters the basic principles of con­trolling the nuclear reaction and cooling the fuel, we need to introduce a third basic principle, that of containing the radioactivity. Collectively these three basic principles of reactor safety can be remembered as the Three Cs.

The term containment can be used to describe both a system for preventing the release of radioactivity to the general environment and the building in which a reactor is housed. Containment of radioactivity involves a multibarrier approach. What are these barriers?

1. Most radioactive fission products are retained where they are formed within the fuel, so the fuel matrix itself provides the first barrier.

2. The fuel is sealed in metal tubes—stainless steel for AGRs, zirconium alloy for water reactors. These are strong enough in all normal circumstances to contain all the fission products that escape from the fuel matrix. This is the second barrier.

3. The reactor pressure vessel that contains the core and the high-pressure coolant forms the third barrier.

4. And for many reactors there is the further barrier of the containment build­ing itself—often a prestressed or reinforced-concrete-sealed and pressure-re­taining building capable of withstanding external impacts and internal explosions.

The whole purpose of the safety systems provided on a reactor is to ensure that these separate barriers are not challenged and all remain intact. The safety limits are thus defined with this specific objective in mind.

In providing a framework for what follows in this chapter, it is useful at this stage to consider some basic principles related to the energy aspects of an acci­dent. We may write the following simple energy balance for the reactor system:

Energy in — energy out = energy stored

As the reactor is brought up to power, some of the fission energy is stored in the reactor components as they are brought up to temperature. In particular, the fuel elements themselves store energy due to the large temperature gradient re­quired to transfer the heat from them, as indicated in Figure 3.1. Once the reac­tor reaches steady state operation, energy is no longer stored and energy in equals energy out. Energy is also stored within the primary circuit coolant as a result of its heat capacity and, for a pressurized coolant, as a result of its high pressure. Any transient process causing a departure from steady state condi­tions will also cause a change in the stored energy. However, the above equa­tion will continue to apply during the transient. Let us take two examples to illustrate this point:

1. If there is a failure of the secondary coolant passing, say, to a steam genera­tor, then the output of energy from the system is reduced and the thermal energy storage of the system must increase, leading to an increase of tem­perature in all the primary circuit components. In some systems (such as pressurized-water reactors) this will also lead to an increase in the primary system pressure, and the consequences of this must be carefully evaluated.

2. If there is a failure in heat extraction from the fuel elements by the primary coolant, then the energy produced by the fuel elements must be stored within the fuel elements themselves, giving rise to a rapid increase in tem­perature.

The concept of the energy balances associated with transient conditions can also be applied to the case of heat release via breaks in the reactor circuit. These will lead to a loss of primary circuit coolant, reducing the amount (or in­ventory) of this coolant in the circuit. If the coolant is released from the circuit in the form of a vapor, it takes with it much more energy than if it is released in the form of a liquid, and this is advantageous in reducing the energy storage (e. g., the amount of heat stored in the fuel elements) during the transient situa­tion. We shall return to this point in discussing the specific case of pressurized — water reactors (P^WRs) in Section 4.3.

A very extreme case of heat retention in the fuel is that where no heat at all is removed from the fuel following a transient leading to a reactor trip. This case is illustrated in Figure 4.1. Initially, the fuel temperature becomes equalized, which gives rise to an initial relatively rapid increase in the fuel surface temper­ature as shown. The fuel element continues to heat up because of heat released during fission product decay, and this rate decreases with time as the fission products gradually disappear. The fuel will ultimately reach its melting point, however, and in designing for the various levels of transient condition it is ob­viously important to prevent this from occurring. The rate of rise of temperature of the fuel will depend on the initial heat rating, which determines the amount of fission products present at any given time. The temperature transients also differ from fuel to fuel and from reactor to reactor.

Having introduced the general principles of design in response to various transients, the types of occurrence and system design in response to them for

water reactors (P^^, B^^, and CANDU), gas-cooled reactors, and fast reactors will be discussed. First, however, we will illustrate some of the points by using a homely example—making tea with a domestic electric kettle.

The reactor pressure vessel represents the second containment barrier or line of defense. It consists of a massively thick ferritic steel structure. What are the possi­ble challenges to the integrity of this vessel? First, it is designed, constructed, and inspected to the highest quality standards. Failure of the vessel due to internal or external loadings within the design basis is considered incredible. However, var­ious failure modes or mechanisms can be postulated in severe beyond-design — basis accidents. Thus the vessel might fail due to

• gross overpressurization

• displacement or damage from the support structures

• creep failure due to overheating or vessel wall thinning

• shock loadings due to internal fuel-coolant interactions or hydrogen explosions

Overpressurization could occur as a result of the reactors failing to trip or shut down in response to an operational transient. An example would be some fault that removes the coupling of the reactor to the heat sink, which unless the reactor is shut down will lead to a rapid rise of primary circuit pressure. In prac­tice initially the negative temperature coefficient, followed by the lifting of the relief valves and subsequent voiding of the core, limits the reactor power and ultimately terminates the fission reaction. As a result the primary pressure peaks well under that which might cause failure of the vessel (-400 bars).

Displacement or removal of the vessel from its supports could occur due to an earthquake of exceptional magnitude or, alternatively, as a result of internal shock loadings from within the vessel itself (see below).

As indicated above, the process of core degradation and melting could result at some point in molten fuel or other materials entering the lower plenum of the reactor vessel. In the TMI-2 accident some 20 tonnes of molten fuel ended up in this location.

What the consequences are depends on how the core support plate fails and when. The lower part of the vessel may still contain a pool of water notwith­standing the high temperatures existing in the upper part of the vessel. If the mass of molten material above the lower core plate jets into the pool of water. a “steam explosion” may occur and may damage the vessel. Such events are discussed in Section 6.3. Alternatively the jet of heavy molten fuel may penetrate to the vessel wall and result in rapid heating of the wall. Wall thinning will result, and if the vessel is still at high pressure, plastic collapse may occur. This may happen quite rapidly—within minutes. Rather than forming a jet, the molten fuel may enter the plenum at the periphery of the core support plate or through the baffle plate and pour down the side of the lower head as in the TMI-2 accident. This more gentle process may not provoke a fuel-coolant interaction. However, if a pool of molten material does form in the lower head, overheating, wall thinning, and ultimately creep failure may result. Failure of the penetrations for the in-core instrumenta­tion may also occur. Evidence from inspection of the lower head of the TMI-2 vessel, however, suggests that some considerable cooling was available via cracks in the fuel debris and via the gap between the debris and the vessel wall. Al­though some damage was observed of the vessel wall and the penetrations, no failure of the pressure boundary occurred.

If the failure of the bottom core support plate results in jets of molten fuel entering a pool of water in the lower plenum, a fuel-coolant interaction “steam explosion” may result. This is particularly the case if the pressure in the primary system is low. This process may cause damage inside and outside the vessel. In the worst case the vessel may be lifted off its supports and/or the head of the vessel may be blown off, damaging the containment. To ensure that the con­tainment is not damaged, it is important to show either that missiles with suffi­cient kinetic energy are not formed or alternatively that the containment structures can accommodate the missile without damage to the containment function. Typically if the fuel-coolant interaction produced an explosive energy input of 1 GJ, then a considerable portion of the kinetic energy of the molten core slug projected upward through the vessel is absorbed by plastic deforma­tion of the internal core structures and stretching of the vessel bolts. Perhaps only 5-10% of the total energy will be imparted to the upper head. This process occurred during the course of the SL-1 accident (Section 5.2.1).

Finally, the intense, and possibly explosive, interaction of molten fuel with water will cause the fuel to be dispersed as small particles. These could form a debris bed in the bottom of the vessel. Depending on the size of the particles and the ability of the operators to continue to feed water to the vessel, it may be possible to cool this debris bed over a long period. This would terminate this class of accident without release of radioactive material into the containment. If it is not possible to cool the fuel debris, the bottom head will fail, releasing the molten fuel into the cavity in which the vessel sits.

2.5.1. Liquid Metal-Cooled Fast Breeder Reactors

The most prevalent design for a fast reactor system is that employing sodium as the coolant. The advantages of liquid sodium in cooling reactors are discussed in Chapter 3. Briefly, sodium is an excellent heat transfer agent and can cope with the very high volumetric power densities encountered in reactors of this type (typically five times those of a P^WR see Table 2.3). The sodium-cooled fast reactor, which Is illustrated schematically in Figure 2.15, consists of a pool of sodium contained in a primary vessel in which the core is submerged.

Sodium is pumped through the core (the pumps being submerged in the sodium pool, as illustrated). The hot sodium then passes through an intermedi­ate heat exchanger, where heat is transferred from the primary coolant to a sec­ondary sodium stream; the secondaiy stream passes through the steam generator, where steam is raised for electricity generation. In contrast to the AGR and PWR, this reactor has three heat transfer stages: from the fuel elements to the primary sodium coolant, between the primary sodium coolant and a sec­ondary coolant, and between this secondary coolant and evaporating water in the steam generator. This somewhat complex system ensures that the primary coolant stays in the primary vessel and that any radioactive substances in the primary vessel are not transferred to the steam generator, where the potential exists for a chemical interaction between the sodium and the water (due to minute leakages).

Since the reactor utilizes fast neutrons, there is no moderator. The layout of the U. K. 250-^MW (electrical) prototype sodium-cooled fast reactor (PFR) is shown in Figure 2.16. A similar prototype (Phenix) has been operated in France.

A much larger commercial-sized [1200-^MW(e)j fast reactor, Superphenix, has been built in France and was commissioned in 1986. European utilities, design and construction companies, and research and development organizations have

collaborated on an advanced design known as the European fast reactor (EFR) with an electrical output of 1450 ^MW(e). The layout of the proposed EFR is il­lustrated in Figure 2.17. The fuel is in the form of pellets of mixed plutonium

and uranium oxides (20-25% Pu O) clad in austenitic or nimonic alloy steel tubes as illustrated in Figure 2.18. Each fuel element consists of 331 pins, each

8.2 in diameter with an active core length of about 1 m. The core power density is about 5 times that in a PWR and 1000 times that in a Magnox reactor.

Sodium-cooled fast reactors have been operated in the United Kingdom, the United States, France, the former Soviet Union, and Japan. In recent years, the commercial development of fast reactors has slowed down. Problems have been encountered in the steam generators in fast reactors, where it has not al­ways been possible to meet the requirement for complete watertightness of the tubes. However, the sodium-cooled fast reactor has some inherent safety fea­tures that may make it very attractive, despite its very high power density. We shall discuss these in Chapters 5 and 6.

Steam

Generator Cell DHR

Emergency Filtration System

Vault Cooling Filters

Electrical Switchgear

Sodium Storage/ Cover Gas Buffer Vessels

Concrete Vault Seismic Raft Base Mat

^^PLES ^AND PROB^LEMS

1 Power increase following increase in reactivity Example: A sudden increase in reactivity of a water reactor 1% beyond prompt criti­cality occurs. The neutron lifetime is 10-4 s. What is the increase in reactor power after 1/100 s? What processes are available to terminate the transient?

Solution: The reactor power increased by a factor of 1.01lf00 = 2.7 in 0.01 s. If such a re­activity increase is to be terminated before melting of the fuel occurs, then steam bub­bles must appear within a few hundredths of a second to expel the moderator and terminate the fission reaction.

Problem. What increase in reactivity would be required to increase the power of a water reactor by a factor of 2 in 0.01 s, assuming a neutron lifetime of 10-4s?

2 Decay heat removal

Example: A 4000-MW(t) P’^TC has been taken out of service. Use the data given in Table 2.2 to estimate the rate of decay heat generation after 1000 h and 1 year from shutdown.

Solution: From Table 2.2 we see that after 1000 h the decay heat rate is 0.11% of the full-power rate. Thus the decay heat generation rate after 1000 h is

4000 X 0.11
100

Similarly, after 1 year, 0.023% of full power is emitted as decay heat, giving the follow­ing value for decay heat generation:

4000 X 0.023
100

Problem: Assume that the shut-down reactor in the example is cooled by residual heat removal (RHR) water at 20°C. Calculate the RHR water flows required after 1000 h and 1 year if the rise in water temperature is to be restricted to 20°C.

3 Fuel investment in thermal reactors

Example: Using the data in Table 2.3, estimate the investment of enriched fuel that would be required for a 10-GW(e) program of AGRs and P^WRs, respectively.

Solution. With a figure of 11 MW(t)/tonne and a thermodynamic efficiency of 40%, the fuel required for the AGR program would be

lOx 103 1lx0.4

Similarly, assuming a thermodynamic efficiency of 32%, the fuel required for the P’^TC program would be

10 X 103

38.8 x 0.32

Problem: Assuming that the alternative programs were for 1000-MW(e) reactors, use the data from Table 2.3 to estimate the core volumes required for the AGR and PWR re­actor choices, respectively. Calculate the diameters of equivalent spheres required to contain these respective volumes.

BmuOG^^HY

Dent, K. H., et al. 0982). “Status of Gas Cooled Reactors in the UK." In Gas-Cooled Reactors Today, Proceedings of a Conference, Bristol, September 20-24, 1982, vol. 3, 247-58. British Nuclear Energy Society, London, 830 pp.

Duderstadt, J. J. 0979). Nuclear Power. Marcel Dekker, New York.

Haywood, R. W. 0975). Analysis of Engineering Cycles, Pergamon, Elmsford, N. Y.

Hirsch, P. B. 0990). The Fast-Neutron Breeder Fission Reactor. The Royal Society,

London.

International Atomic Energy Agency 0986). Nuclear Power Reactors in the World, ^^A, Vienna, April 1994.

International Atomic Energy Agency 0986). The Accident at Chernobyl Nuclear Power Plant and Its Consequences. Information for the IAEA Experts Meeting, August 25—29, 1986. Compiled by the USSR State Committee on the Utilization of Atomic Energy.

Knief, R. A. 0992). Nuclear Energy Technology: Theory and Practice of Commercial Nuclear Power, 2d ed. Hemisphere/Taylor and Francis, Washington, D. C., 770 pp.

McIntyre, H. C. 0975). “Natural Uranium Heavy Water Reactors.” Sci, Am, 233 (4):

17-27.

National Nuclear Corporation 0986). The Rusian Graphite Moderated Channel Tube Reactor. Report of a Critical Assessment following a Visit to Leningrad RBMK Station, March 1976 (republished May 1986).

Patterson, W. C. 0983). Nuclear Power, 2d ed. Penguin, Harmondsworth, U. K., 256 pp.

Weisman, J. 0977). Elements of Nuclear Reactor. Elsevier, New York.

Winterton, R. H.S. 0981). The Thermal Design of Nuclear Reactors, Pergamon, Elmsford, N. Y.

3

The various operational states for a liquid metal-cooled fast reactor (LMFBR)

can be listed as follows:

1. Normal operation and operational transients. The sodium in the circuit is al­ways kept in a molten state by heating the whole circuit with electrical resis­tance heaters wound on all the pipework. This maintains the sodium at a temperature of at least 100°C (the melting point of sodium is 98°C). The large pool of molten sodium responds rather slowly to heat input. Thus, the coolant takes some time tq reach operating temperature.

2. Upsets. Various categories of upset situations have been postulated for an LMFBR. Many are similar to those for water — and gas-cooled reactors, includ­ing loss of load, turbine trip, loss of feedwater, and loss of a single main cir­culating pump.

3. Emergency conditions. In an LMFBR, emergency conditions will occur if the upsets described above cannot be contained within normal operational pro­cedure. These include the following:

a. Loss of electric power (and resultant coast-down of the pumps). Loss of power supply to the primary coolant pumps causes them to coast down to zero speed. Under these circumstances, the reactor is immediately tripped and power may he reinstated to the circulators from emergency supplies (diesel-driven generators) that operate secondary electric motors (“pony” motors). However, the sodium pool itself represents a major heat sink. For instance, with the decay heat in the reactor alone, the sodium pool would take about 24 h to reach the boiling point if there were no heat removal at all. Moreover, the reactor has decay heat removal heat exchangers that are connected to the primary circuit and can remove the decay heat by natural circulation alone, without any electric power input to the reactor. The final heat sink from these removal systems is the atmosphere via air-cooled heat exchangers. Even if single-phase natural circulation is not established immediately after a reactor trip, sodium boiling in the core is an acceptable means of removing decay heat and the generation of two-phase flow within the core enhances the natural circulation to the extent that single-phase natural circulation is rapidly established.

b. Inadvertent increase in neutron population in the core. The rate of the fission reaction in the LMFBR can be increased by inadvertent removal of a control rod, movements of the fuel (e. g., by the fuel elements becoming bowed, as happened in the U. S. experimental breeder reactor EBR I incident described in Chapter 5), or sodium boiling in the core. Sodium boiling in the inner region of the core causes an increase in the rate of fission (neutron population), since sodium absorbs neutrons, and if it is partially vaporized, the absorption is reduced. However, if the boiling occurs in the outer region of the core, the reduced local density causes increased leakage of neutrons from the core and gives rise to a reduction in the fission reaction (neutron population). Thus, the effect of sodium boiling is usually negative for small reactors such as the prototype fast reactor (PFR) and positive for larger reactors, where any boiling is likely to be away from the boundary of the core. Great care must be taken to design LMFBRs to avoid failure in the control rod insertion mechanism, and systems are being designed to be capable of self-actuated shutdown, directly triggered by high temperatures in the core and requiring no out-of­reactor mechanisms.

c. Local damage within a fuel subassembly. The reactor core consists of hundreds of separate groups of fuel elements, which can be inserted or removed independently from the core. A typical subassembly consists of 300 pins 6 in diameter and I m long. Since an accident in the Enrico Fermi reactor (described in Chapter 5), considerable attention has been focused on the possibility of blockages occurring within individual subassemblies or groups of subassemblies. If the sodium flow is blocked, local melting of the cladding and possibly the oxide fuel could occur. The oxide fuel reacts with the sodium, limiting its useful lifetime, but the failure of a subassembly can usually be detected by specially provided instrumentation. Failure to detect the fault may lead to escalation of the upset into a fault condition (see below), with debris blocking an increasing area of the core, reducing the flow, and preventing cooling. Reduction of flow gives local sodium boiling, and this increases the reactivity in the region, making the problem worse.

d. Loss of heat removal from secondary sodium or steam systems. Here the system responds in the manner described for the loss-of-flow upset. The reactor is tripped and natural circulation cooling is set up, with heat released by the decay heat removal heat exchangers. The circulators may still operate under these circumstances; provision is made for driving them automatically via the pony motors.

To summarize, the primary objective in the design and operation of an LMFBR is to bring it, in response to the various operating states, to <1 condition

such that it can be cooled by either (1) primary circuit cooling by the interme­diate sodium circuit to the steam generators or (2) primary circuit cooling via the separate liquid metal coolant circuit to air-cooled heat exchangers. In the former case, the primary circuit uses forced circulation, while the secondary in­termediate circuit can rely on natural convection. Emergency boiler-feed is sup­plied to the steam generators, and the steam produced vented from the circuit. In the latter case, natural circulation in the primary circuit is sufficient to cool the core, and indeed if all the heat exchangers are operational, natural convec­tion is sufficient in the secondary circuit. However, if this is not the case, a pow­ered fan is necessary to force air across the heat exchanger.

For the fast reactor much attention has been given to the case that is beyond the design basis accident, namely, conditions under which quantities of molten fuel are produced. In this case it is postulated that the energy present in the molten fuel could be rapidly converted to a pressure shock wave and cause a vapor explosion. We shall consider this extremely unlikely event in Chapter 6.

REFERENCE і

Snell, V. G., et al. 0990): ‘‘CANDU Safety under Severe Accidents: An Overview.” Nuclear Safety 31 (January-March): 20—35.

7.1 INTRODUCTION

In Chapters 4-6 we have discussed hypothetical and actual accident conditions in reactors. Now we return to the discussion of the next phase of normal oper­ation, namely, the removal of the used fuel from the reactor and its subsequent processing.

In a nuclear reactor, the fissile material is gradually used up and converted to energy and fission products. During the nuclear reaction there are changes in the microstructure of the fuel due to the release of fission products, which ei­ther combine with the fuel or are released inside the fuel can. These changes have two effects: (1) a gradual deformation of the fuel and in some cases the can and (2) the release of fission products (such as xenon and iodine), which are themselves strong absorbers of neutrons, leading to a reduction in neutron population and a less efficient nuclear reaction. For these reasons, the fuel ele­ment must be removed from the reactor after a period of time and before all the fissile material is used up. Typically this period will be between 3 and 5 years for thermal reactors and 1 year to 18 months for fast reactors. For thermal reac­tors, 60 to 75% of the original fissile material is used up at the time of fuel re­moval. For the fast reactor, the utilization is much less, of the order of 25%. The fraction utilized is often referred to as the burn-up.

The fuel removed from a nuclear reactor contains three kinds of valuable material:

1. The unused proportion of the fissile material that was originally introduced with the fresh fuel.

2. New fissile material that has been bred as a result of the nuclear reactions, in particular, the reaction between neutrons and 238U to form 239Pu. The pluto­nium produced can be used as a fissile material in both thermal and fast re-

actors. Note that bred material also participates in the fission reaction while the fuel is still in the reactor; in a thermal reactor system, 25% of the heat production might arise from fission of material bred in situ.

3. Much of the original 23HU, the nonfissile isotope of uranium, still remains. This material is valuable as a fertile material for use, particularly in the blan­kets of fast reactors, where it is converted to 239Pu.

Of course, these valuable materials are mixed with a range of highly ra­dioactive fission products that form the waste from the nuclear cycle. Basically, there are two choices facing the nuclear power operator:

1. To discharge the fuel and store it safely without making any attempt to sep­arate the useful fissile-fertile materials from the fission products in the fore­seeable future. If such storage is regarded as permanent, this approach is colloquially referred to as the throwaway cycle in the sense that valuable re­sources are being disposed of. Such a cycle could be practical only if it was felt that the world’s uranium resources were adequate to operate thermal re­actors for a sufficiently long period. As discussed in Chapter 1, this would be a highly inefficient use of these resources.

2. To discharge the fuel, store it for a relatively short time (typically 1-5 years) to allow the more active fission products to decay and the decay heat to drop to manageable levels, and then to process the fuel chemically to sepa­rate the valuable fissile and fertile materials from the fission products, which can then be stored in a safe form.

If a program of fast reactor operation is envisaged, the second option is manda­tory; otherwise, far too much of the fissile material in the cycle will be wasted. Reprocessing the fuel is more expensive in the short term than simply storing it, and the decision about whether to reprocess in the case of thermal reactor fuels is closely related to the overall utilization strategy for nuclear energy in individ­ual countries. Where a program of fast reactors is envisaged, reprocessing of thermal reactor fuel is necessary in order to produce the initial inventory of plu­tonium for such a program. Typically, it would take about 15 years’ worth of spent fuel from a thermal reactor to produce the initial inventory for a fast re­actor of similar size.

In this chapter we discuss the removal of spent fuel elements from reac­tors, their transport to a long-term storage location or a reprocessing plant, and the problems of the reprocessing plant itself. The questions of long-term storage of nuclear waste products will be discussed in Chapter 8. This chapter concentrates on the thermal aspects of these operations, in line with the rest of this book.

Higher operating temperatures at lower pressures can be obtained by using molten salts as coolants. Molten metal hydroxides such as sodium hydroxide (caustic soda) have been suggested. The melting point of such substances tends to be rather high, though by the use of mixtures, as of sodium and potassium hydroxides, lower melting points (typically 190°C) can be obtained. The main problem with such systems is corrosion, and this has prevented their serious application.

In the early days of the development of nuclear power, many reactor systems were suggested in which the fissile material (e. g., in the form of uranium tetra — fluoride, UF4) was dissolved in a mixture of fused salts. When this mixture is passed through a vessel containing a moderator such as graphite, a fission re­action takes place, heating the uranium-containing fused salt. The fused salt is pumped from the reaction zone to a heat exchanger, where the heat is trans­ferred to another heat transfer fluid and ultimately to a power generation sys­tem. A reactor of this type, in which the fuel is actually dissolved in the coolant, is termed a homogeneous reactor and has the advantage that the fuel can be re­processed continuously. However, the corrosion and other problems associated with reactors of this type effectively rnle them out, though small prototypes have been operated.