Gaseous coolants have the great advantage of having a well-defined phase state. Unlike liquid coolants, they are not subject to a change of phase, with the resulting complicated two-phase flow problems during abnormal operating conditions. However, they have the disadvantages of a low heat capacity and low heat transfer coefficients, the latter necessitating heat transfer enhancement or low operating temperatures. A wide variety of gases have been considered for nuclear reactor cooling, but only those that have been used or have had se­rious evaluation will be discussed here.

On September 1, 1972, a routine start-up operation was proceeding on the Mill­stone 1 reactor in Connecticut. This reactor was a 660-MW(e) B^^. When the reactor had achieved less than 0.1% of full power, the operator noted that the water purification system was malfunctioning. He switched to a second water purification system and continued the start-up. About half an hour later the sec­ond system also failed and the operator began to shut down the reactor. When it became obvious that salt from seawater was penetrating the primary coolant circuit, the reactor was tripped rapidly. Upon investigation it was found that tubes in the condenser (which was cooled by seawater) had corroded, allowing a massive amount of seawater to enter the primary circuit. One consequence of the saltwater ingress was the failure of the instruments that measured power in the reactor; the failure was due to stress corrosion cracking of the stainless steel sheaths of the instruments, which are sensitive to chloride attack.

The reactor was successfully repaired and resumed operation. Although this

accident caused no injuries and no radioactivity was released, it demonstrates the relative vulnerability of direct-cycle systems such as the B^^ in comparison with indirect-cycle systems such as P^WR CANDU, or AGR. In a B^^ the pri­mary circuit coolant passes directly to the turbine and is condensed in the con­densers before returning to the reactor. If the condensers are cooled by seawater, ingress into the primary circuit is always a potential problem. One way to overcome this is to isolate the condensers in the event of leakage of sea­water, but this leads to loss of the main heat sink and a need to provide alter­native cooling or means of energy release.

In the case of the P^^ and B^WR the refueling is off-load. It takes place ap­proximately once a year over a period of 4—6 weeks. Other maintenance work on the plant is scheduled to be done at the same time, which means that high load factors are still achievable with these reactors.

To carry out refueling in a P^^ or B^WR the system is partially drained to bring the liquid level to below the level of the flange that connects the main part of the vessel to the top part (referred to as the top head). All the control rods are fully inserted into the core and unlatched from their mechanisms (which pass through the head). The bolts attaching the top head to the vessel are then loos­ened, the cavity in which the reactor sits is flooded with water, and the head is removed. The upper structures in the reactor vessel are removed to expose the fuel, and handling operations are carried out under a significant depth of water in the reactor cavity (typically 5-10 m). This water is also circulated through a heat exchanger to provide liquid cooling for decay heat removal. Approxi­mately one-third of the total number of fuel elements are removed in any one operation, namely, about 5^0-0 elements out of the total inventory of 200.

As shown in Figure 7.4, in the case of the P^WR the fuel is passed into a transfer canal, in which it is transferred horizontally out of the reactor building and into a water-filled fuel storage pond.

The refueling route for the B^^ is similar to that illustrated in Figure 7.4, but with the additional complication that it is necessary to remove all of the devices above the core used to separate the steam from the steam-water mixture leav­ing the core (see Figure 4.27 for an illustration of the reactor structure). The B^^ fuel elements are somewhat smaller than those in the P^WR and therefore a correspondingly larger number of fuel movements must be made.

Once every 3 years it is common practice to remove all the fuel and the lower core structures and to carry out a thorough inspection of the pressure vessel from the inside surfaces. This provides a guarantee of the integrity of this vessel, which is essential to the safety of the system. The internal structures and fuel are then recharged into the vessel and the reactor restarted. Typically, this triennial inspection process might take up to 3 months.

Sources and Resources

1.1 INTRODUCTION

This book is written from an engineer’s viewpoint, particularly that of a thermal engineer, that is, a design or research engineer concerned with heat production and utilization. We believe that the most important problems in the utilization of nuclear power concern the handling of thermal energy generated in the various processes. This includes handling under the normal operating and processing conditions and dealing with heat removal problems under the unlikely condi­tions of an accident. The problem of handling thermal energy associated with nuclear power does not stop when the fuel is removed from the power station; small amounts of heat are generated in the spent fuel before it is processed and in the waste products. The consequences of this are also the concern of the thermal engineer.

The approach that we shall take, therefore, is one that is not normally fol­lowed in general books on nuclear energy. We will follow the history of nuclear materials from their cosmic origins, through their terrestrial life span up to the time when they are used in nuclear reactors, and beyond. Although we will need to explain some elementary aspects of physics, the emphasis will be on what happens to the thermal energy.

We begin with the history of uranium in the earth, the decay of its isotopes, and the effect this decay has had on the earth as we know it. Comparisons are made with the earth’s other main energy source: the sun. Energy from the sun is derived either directly or through storage media such as fossil fuels, hydro­electric power, and winds.

The rate at which energy may be extracted from nuclear materials can be en­hanced by the self-sustaining process of nuclear fission. Nuclear fission does not normally occur in nature, but recent studies have revealed that nature an­ticipated Enrico Fermi by about 2 billion years in creating a natural nuclear fis­sion reactor by a series of extraordinary and improbable events. We shall use this example in introducing nuclear fission.

In the final part of this chapter, we compare the relative magnitudes of thermal energy resources of the various types: fossil fuel, nuclear, solar, and so forth.

Since the first air-cooled nuclear reactor built under the squash court of the Uni­versity of Chicago in December 1942, an amazing variety of nuclear reactors have been devised and many of them have been built. In all cases a coolant cir­cuit was included; the main components of such circuits and the circuits ap­plied in the most commonly used nuclear power reactors are described in Chapter 2. Of course, all reactor cooling circuits must include the reactor core itself, a means of circulating the coolant through the core, and a means of ex­tracting the heat from the coolant in order to maintain continuous cooling of the reactor and at the same time (in power reactors) generate useful power. In power reactors the means of extracting the heat from the coolant is almost uni­versally a heat exchanger, which produces high-pressure steam that can be used in a steam turbine to generate power. It is convenient to divide the various types of reactor circuits into three groups:

1. Loop-type circuits. The core itself is contained within a reactor vessel, and the primary coolant circulator and the steam generator are coupled to the reactor vessel by suitable pipe systems.

2. Integral-type circuits. The core, primary coolant circulator, and steam generator are contained within a single vessel, feedwater is fed to this vessel, and steam is taken from it to the turbine.

3. Pool-type circuits. The core and the primary coolant circulators are im­mersed in a pool of coolant. This arrangement is feasible only for unpressur­ized coolants such as sodium. The steam generator is usually outside the reactor containment vessel. This type of circuit is intermediate between the loop-type and integral-type circuits.

5.3.3 The EBR-1 Meltdown Accident

The U. S. Experimental Breeder Reactor I (EBR-1) had the distinction of being the first reactor to generate electricity. Construction of the reactor began in 1948, and electric power production started in December 1951. The reactor was designed for a thermal output of 1 MW(t) and an electrical power output of 200 kW(e). Of course, the power production was more for demonstration than for economic viability.

The core of the reactor is illustrated schematically in Figure 5.25a. During its lifetime, the reactor was operated with four different core configurations, all with fuel in metallic form. The first three cores were of highly enriched ura­nium, consisting mainly of U-235. The second core had a uranium-zirconium alloy fuel containing 2% zirconium. The fuel pins were 1.25 cm in diameter, and 217 pins in a triangular array were mounted in a central hexagon 19 cm across, forming the core of the reactor. The small size of this core illustrates the great compactness of liquid metal-cooled fast reactors. Around the central U-235 re­gion there was a blanket region containing natural uranium rods, as shown in Figure 5.25a. The coolant for the reactor was a sodium-potassium mixture (NaK) that is liquid at room temperature (see Chapter 3).

With the second core, power oscillations were observed at very low core flows. In an experiment to examine this effect beginning on November 29, 1955, with the core flow totally stopped and certain safety interlocks cut out, power was rapidly raised in order to determine the magnitude of a previously observed increase in reactivity with temperature. It had been intended to termi­nate the experiment with the fuel temperature at 500°C, but through the com­bination of this temperature effect and an operator error, the temperature rose to more than 720°C. At this temperature the uranium metal fuel and the stain­less steel can begin to interact, leading to the melting of about 40% of the core, but without explosion, plant damage, or radiation hazard.

As explained in Chapter 4, bringing the pins closer together in a fast reactor causes an increase in reactivity or neutron population. The mechanism by which the EBR-1 core meltdown occurred was related to this. It was possible for the rods to bow as illustrated in Figure 5.25b, and this gave an increase in reactivity that was self-propagating as the increased temperatures increased the amount of bowing. This accounted for the temperature effect that was being investigated at the time and that was subsequently explained theoretically. The core of EBR-1 was later removed and replaced by another core designed to eliminate the bow-

(4) External Air cooled control rods annulus

Graphite

7j across inside of flats

Upper shields and seal plales

(b)

Figure 5.25: The EBR-1 meltdown incident.

ing effect by the use of spacer ribs. The expansion of the ribbing with increasing temperature causes the core to expand, giving a negative rather than the previ­ously observed positive temperature coefficient of reactivity.

The EBR-1 reactor, which was finally shut down in December 1963, gave in­formation of great value related to the design of fast reactors. Now all fast reac­tor cores are designed with significant amounts of restraint so that they always have a negative temperature coefficient of reactivity. In fact, it may be possible in the future to design fast reactor cores that are inherently safe in that they ex­pand to switch off the nuclear reaction even if the control rods fail to actuate. This is one of the features of fast reactors that make them in some respects even safer than thermal reactors.

As mentioned in Chapter 7, the nitric acid stream containing the fission products after solvent extraction in the reprocessing plant is concentrated by evaporation and then held in storage tanks. A photograph of one of these tanks under construc­tion is shown in Figure 8.7. Nearly all the high-level waste from the nuclear work in the United kingdom, accumulated over the past 25 years, is stored in 15 such tanks at Sellafield in Cumbria, which contain a total of about 100m3 of liquid.

The stainless steel tanks are contained in concrete vaults, which are them­selves lined with stainless steel to provide further containment in the improba­ble event that the primary container should fail. The space between the tanks and the vaults is monitored, and provision is made for transferring the contents to spare tanks should the need arise. Heat is removed by several independent sets of cooling coils. Reinforced concrete, typically 2 m thick, in which the tanks are sited, protect the operators from direct radiation. Provided cooling is main­tained, there are essentially no radiological hazards. The possibility and conse­quences of an accidental loss of coolant were considered at the Public Enquiry on Windscale in 1978. In the extremely unlikely event of a total loss of coolant (estimated to have a probability of occurrence of 1 in 1 million for each year of operation), it would take hours for the contents to boil and days for them to evaporate, allowing ample time to take remedial action. During the period in which the fission products are generating significant quantities of heat, keeping them in a liquid form facilitates cooling. However, for long-term storage it is considered preferable to convert the waste into solid form. and a number of

processes have been considered for this.

Work on solidification of nuclear waste started in the 1950s, and by the mid — 1960s incorporation of wastes into glass (vitrification) was established on a lab­oratory scale. The method has been used on an industrial scale in France for a number of years, and the French AVM process (illustrated in Figure 8.8) has been adopted in other countries, including the United Kingdom. Among the al­ternative processes being investigated is the microwave vitrification process il­lustrated in Figure 8.9. A range of glass compositions have been developed that enable the constituents of the waste to he incorporated. The glasses have been shown to survive the effects of heating and radiation from the wastes without significant deterioration. They would dissolve veiy slowly over many thousands of years in freely flowing water. Dissolution in the sort of repositories likely to he used, where access to water is severely restricted, would he very much slower. Other solidification techniques include incorporation into various ce­ramics and forms of crvstalline rocks.

The vitrified waste is typically cast into stainless steel canisters and these can­isters dry-stored in a manner illustrated in Figure 8.5. The vitrified waste canis­ters will be stored in these natural convection air-cooled stores up to 50 years before final disposal. A typical glass block might be 30 cm in diameter and 1 m long, weighing about 0.2 tons. About 20% of the weight of such a block would be the fission products from the reprocessing plant, the rest being added mate­rials to help form the glass. As in the case of the spent fuel, the heat release is dominated by the caesium-strontium decay with a half-life of 30 years. It is gen­erally considered that surface temperatures for the block in the long-term store should fall below 100°C, and at this surface temperature a heat rejection rate of about 1 kW is achievable by conduction into the surrounding rocks. To avoid interactions between blocks within the rock matrix, a spacing of about 10 m in all directions is required. This could be achieved by tunneling to the required 1 km depth and then constructing a gallery from which holes, say, 200 m in depth, are drilled. The blocks could then be dropped in and the required 10 m

spacing achieved by infilling before dropping in the next block. The holes themselves would also be spaced out on an array of 10 m square.

For the British program, it is estimated that some 10,000 blocks will have been produced before the end of the century. This would imply the use of an array of around 50 x 50 blocks arranged, say, in a cube.

Again, the problem of leaching of fission products from the block and their transfer through the strata must be considered, and the thermal circulation and thermal buoyancy effects mentioned are very important in the medium term. Enough is now known about these systems to be sure that safe disposal of nu­clear waste is possible.

The Magnox reactor is illustrated schematically in Figure 2.4. The coolant is car­bon dioxide at a pressure of 20 bars (300 psia). The coolant is circulated through a core that consists of the moderator structure, which is built from

graphite bricks containing holes through which the coolant flows and in which the fuel elements are placed. Fuel elements consist of natural uranium bars clad in cans of a magnesium alloy known by the trade name Magnox (hence the name of the reactor). The alloy does not significantly absorb neutrons, so nat­ural uranium, rather than enriched uranium, can be used as a fuel. A typical Magnox core would be 14 m in diameter and 8 m high. The coolant gas leaves the core at 400 eC, flows to the steam generator, and from there flows back through the gas circulator to the reactor. In the earlier designs of Magnox reactors, the pressure vessel containing the core was made of steel. In later designs it was combined with the shielding in the form of a prestressed concrete pressure ves­sel, which also contained the heat exchangers (in the earlier designs these were external to the pressure vessel and the shielding as shown in Figure 2.4). Magnox reactors were constructed in the United ^ngdom, France, Italy, and Japan and have operated very successfully since their construction, which in some cases was around 35 years ago. The steam cycle efficiency of Magnox reactors is about

A Magnox fuel ^element

31%; this means that 69% of the nuclear heat is rejected to atmosphere via the cooling towers (Section 1.1.3).

A Magnox fuel element is shown in Figure 2.4b. The outside of the Magnox can is machined in a complex pattern of fins (“herringbone” pattern), which has been shown by detailed heat transfer experiments to be the optimal form. The swirl of the gas in the channel and the fins on the surface are an aid to heat trans­fer. The advantages and disadvantages of various coolants will be discussed in Chapter 3, where we shall also discuss some basic principles of heat transfer.

Although the Magnox reactor has been remarkably successful and reliable, it has disadvantages compared with some other reactor types. The principal one is its relatively low power output per unit volume of core. This leads to a large size for the core, a large investment in fuel, and high capital costs. Table 2.3 compares various reactors in terms of the average power generation rate per unit volume of the core (called the average volumetric power density). It also shows the rate of power generation per tonne of fuel (the averagefuel rating) and the power generation per unit length of fuel (the average linearfuel rat- infi). Compared with other reactors, the Magnox has a very low volumetric power density and a very low average fuel rating per unit mass of fuel. Both of these factors lead to high costs due to the high fuel inventory and large cores.

Before the accident at Three Mile Island, most attention was focused on the posmlated large-break LOCA. However, the Three Mile Island incident sharply focused attention on the fact that a small break (typically up to sizes where the reactor remains pressurized despite the break, say, up to 12-cm-diameter holes) in the primary circuit was, in fact, much more likely. At Three Mile Island this small break was due to a stuck-open power-operated relief valve. It could, however, also have occurred as a result of the break in one of the large number of small pipes attached to the primary circuit. Figure 4.20 shows a histogram of the number of pipes attached to the circuit as a function of pipe size and cross­sectional area. Also shown is the percentage area relative to the main coolant pipes. Figure 4.20 is for a German P’^TC. design, but the result is likely to be much the same for other designs.

The most important difference between small-break and large-break acci­dents is that in the former the reactor depressurizes relatively slowly. Reactor pressure as a function of time after the break is shown for various break sizes in Figure 4.21. Since the core may remain at a high pressure in a small-break LOCA, it is not possible to activate the accumulator or low-pressure injection system until late in the accident.

A typical sequence for a small-break LOCA is illustrated in Figures 4.22 to

.Intermediate

breaks yi Large breaks

5 20

Fi^^e 4.20: Connection pipe diameter/cross section/percentage spectrum of a P’^TC. Solid lines: primary loop system; dashed lines: pressurizer.

Fi^^e 4.21: Primary pressure versus time for small-break LOCAs in P’^TC. (O) Pri­mary temperature 175°C; (□) reflood tank empty. (Two HPI pumps; reflood tanks 4 x 286 m3; no LPI pumps.)

4.26. As with the large-break LOCA, the most serious effects are found when the break is in the reactor inlet pipe (the cold legs).

Following the initiation of the break, the pressure falls and the reactor trips. As the pressure falls below about 100 bars, the high-pressure injection system comes on. The pressure continues to fall to around 70 bars, when the hottest liquid in the circuit starts to vaporize and produce steam. First, the water in the pressurizer vaporizes. As the saturation condition is reached throughout the hotter parts of the primary circuit, steam bubbles form and because the pumps are stopped, will settle out in the upper part of the reactor as shown in Figure 4.22 There has been considerable controversy about whether to leave the cir­culating pumps operating or to stop them during a LOCA. If left on, they may assist in circulating liquid through the core, promoting its cooling. On the other hand, they may aid the loss of fluid by pumping it out through the breach. Cur­rent rules for operating P^WR indicate that the pumps should be stopped, this being considered to give the balance of advantage in general.

As a result of depressurization, steam forms and collects in the upper head of

the reactor but cannot escape via the breach because of the arrangement of the pipework. However, the loss of water is quite rapid, though it removes a rela­tively small amount of energy from the system. The water drains down to the level of the water inlet-outlet pipes on the reactor pressure vessel (the “noz­zles”) in about 250 s, during which the pressure may be maintained at a high level and still inhibit the actuation of the accumulators and LPIS.

During this phase, the steam generators are voided (i. e., their original water content is lost through the break) on the primary side, the steam from the reac­tor having access to the steam generators, as illustrated in Figure 4.23. Obvi­ously, the steam generators represent a potential heat rejection source, with the steam from the reactor core being condensed in the steam generator tubes and flowing back down the tubes and into the core Figure 4.23). However, this ben­eficial action is conditional on the secondary side being at a sufficiently low pressure (and corresponding low saturation temperature) to allow heat to be extracted via the secondary circuit. Assume, for instance, that the secondary — side pressure remains at its normal operating value of 70 bars (1000 psia) while

the steam from the primary system is arriving at the steam generators also at 70 bars. In this situation the saturation temperatures are identical and no conden­sation can take place. Therefore, it is imperative in this type of accident that sec­ondary-side cooling or depressurization is carried out.

Because of the continuing loss of water from the system, the core begins to dry out from the top downward (so-called core uncovery), as illustrated in Fig­ure 4.24. The system is still pressurized due to the formation of steam, which cannot escape through the cold leg break, being blocked by the water in the vessel and the pump. The pump has a U-bend under it (the pump loop sea[), and this remains full of water and blocks the steam from flowing from the ves­sel through the steam generator and pump to the break. The level in the pump side of the loop seal near the break (i. e., left-hand side of the loop seal at the left in Figure 4.25) is roughly equal to the level of water in the vessel since the two levels are connected by the (voided) steam generator and pump. Only when the levels reach the bottom of the U-bend can the steam pass the loop seal. At this point, steam from the core passes through the pump and out along the cold leg to the break. This results in a rapid depressurization. The water in

the remaining part of the core partially vaporizes, and the mixture of water and steam bubbles formed rewets the upper part of the core. As depressurization proceeds, the core may be dried out again (as in a large-break LOCA). How­ever, the depressurization permits actuation of the accumulator and LPIS sys­tems, and these rapidly reflood the core and bring it to a cold condition. In the longer term (usually longer than 350 s) heat is extracted in the way illustrated for the large LOCA in Figure 4.17.

Figure 4.27 shows the variation of water level and fuel clad temperature for small-break LOCAs with two different equivalent break diameters. The level of the two-phase mixture in the vessel is shown. Normally, the cooling is good for regions of the core that are in contact with this mixture, the core being over­heated above this level. The mixture level is much higher than the level of the liquid without steam bubbles would be. The latter level is referred to as the col­lapsed liquid level, and the phenomenon of increase of level due to the pres­ence of the bubbles is termed level swell. A similar phenomenon occurs in dispensing glasses of beer.

4.3.4 Alternative ECCSs

The descriptions above apply mainly to a typical U. S. P^WR where the emer­gency core cooling water is injected into the coolant inlet pipes (the cold legs) only. Alternative ECCSs have been used, the most important being that used in the German P^WR where the emergency core cooling water is injected into both the cold legs and the hot legs (the coolant outlet pipes from the reactor vessel). Such combined injection is claimed to offer advantages in more rapid quench­ing of the core and in lower peak cladding temperatures during a large-break LOCA. In the case of the small-break LOCA, it is claimed that faster depressur­ization occurs, allowing early actuation of the accumulator and LPIS systems.

As we saw in the previous section, there are a number of circumstances in which beds of fuel debris may be formed, initially submerged in a pool of coolant. If such beds can be effectively cooled, remelting is avoided and dam­age to the vessel or the cavity contained in the bed may be prevented. In recent years, and particularly since the accident at Three Mile Island, much attention has been given to the coolability of such beds.

The cooling of beds of debris is a highly complex process and is strongly af­fected by such variables as the bed particle size, the means of access of the coolant to the bed, the bed depth, and the system pressure. Some mechanisms for debris bed cooling, illustrated in Figure 6.4, are as follows:

1. Once-through flow through the bed. Here it is assumed that the liquid is able to reach the bottom of the bed and is then induced to flow into the bed under the action of natural or forced circulation. Natural circulation would be caused by the difference in density of the coolant inside the bed and outside the bed. This is the same kind of circulation that occurs in some forms of steam-generating boilers. Alternatively, the debris bed may be in a region of the reactor over which a pressure drop occurs in the circulation liquid, and this pressure drop would force liquid through the bed. As illustrated in Fig­ure 6.4a, the first phase is for the heat generated in the bed (from decay heat of the fission products trapped in the bed) to heat the liquid to its boiling point. Then, as the flow passes through the bed, the liquid is evaporated and ultimately converted totally to vapor. From this point on, the temperature rises rapidly with distance up the bed, and if the circulation is too low or the bed too deep, the particles may reach a temperature at which they begin to fuse together. This clearly represents a limit to this form of cooling.

2. Cooling of closed deep beds. Here, as illustrated in Figure 6.4b, the liquid can only enter from the top of the bed. The liquid trickles into the bed, cooling it and generating vapor, which must escape in the direction opposite to that of the liquid flowing in. This causes a flooding phenomenon of the type we discussed in Chapter 2, with the vapor resisting and limiting the entry of liq­uid at the top of the bed. This may mean that only the upper part of the bed is cooled and the lower part may become overheated. This limitation is more severe the smaller the particle size in the bed. Again, drying out and fusing of the lower part of the bed is the limit on cooling in this situation.

3. Shallow-bed cooling. If there is a shallow bed of particulate material on the bottom of the containment and this is covered by a liquid layer, then “chim­neys” may be formed in the layer (Figure 6.4c) through which the vapor may escape, the liquid passing into the bed by capillary action through the par­ticulate layer between the chimneys. This is an efficient way of cooling but can only be applied over a limited range of conditions.

Experiments and calculations show that in case 2, for a 1-m-deep bed, a heat dissipation rate of 750 kW/m3 may be achieved if the particles are 4 mm in di­ameter in a pool of water at 1 bar (atmospheric pressure). However, the maxi­mum dissipation rate before dryout and fusion of a bed composed of particles of 0.1 mm diameter would be only 20 kW/m3. Thus, the effectiveness of the de­bris bed cooling can be estimated accurately only if the particle size of the bed is known. Although a better understanding of the mechanisms of debris bed cooling is now beginnng to emerge, the main difficulty of predicting the parti­cle size that might result in different phases of the accident is still to be re­solved. A typical debris bed might have a power generation rate (from fission product decay) of 1000 kW/m3 some 3 h after initiation of the accident—about the time at which one might expect meltdown in a P’^TC. This Dower could be

(a)

dissipated in a bed 0.5 m thick provided the particle size was greater than 2 mm. These calculations are for a P^WR but a similar picture is obtained for the fast reactor, since its increased fuel rating (and hence fission product decay heating) is offset by the increase in latent heat of vaporization of sodium com­pared with that of water.