The chief disadvantage of UO; is its low coefficient of thermal conductivity which is also markedly de­pendent upon temperature, exhibiting a five-fold de­crease over the temperature range between 0°C and the melting point. This low conductivity is very im­portant in that it leads to the formation of high temperature gradients between the centre and the sur­face of the pellets. The first consequence of these gradients is to produce non-uniform thermal expansion which, in turn, causes cracking of the pellet as shown in Fig 1.36. In CAGR fuel, the use of hollow pellets reduces these gradients (and the fuel centre tempera­ture) below the values for solid fuel, nonetheless, cracking still occurs in the more highly rated pellets and the outer annulus of the pellet, adjacent to the clad, generally breaks up into a number of blocks. The radial cracks which separate these blocks have important consequences for pellet clad interaction dur­ing power transients, as explained later; a rough guide ta their number is that one crack is produced for every 2 MW/t(U) of rating.

In the early stages of irradiation, following power raising and. pellet fragmentation, the most important

process taking place in the fuel is densification, the rate of which is considerably accelerated by neutron irradiation. Depending upon the gas content of the finished pellet, a fraction of the porosity remaining after hiah temperature sintering will still be remov­able and it is this which is eliminated during irra­diation. In general, this process is complete after a burn-up of 2-3 GWd/t, the fractional change in vol­ume iV, increasing according to an equation of the form aV = aVo [1 — exp (-1 / I0)j where AVo is the ‘removable’ porosity (generally about 70% of the total which itself amounts to around 2% of the overall fuel volume), Ї is the irradiation and IQ is constant (Hargreaves and Collins, 1976 [5]). The main effect of densification on the fuel is that it decreases the length of the pellet stack and may allow the for­mation of large unsupported lengths of clad which can then collapse under the influence of the coolant pressure. In the case of CAGR fuel this is prevented by the use of a number of anti-stacking pellets which have a circumferential groove on their outer surface. These are distributed evenly along the pellet stack and the clad collapsed into the grooves by subjecting the pin to a high external hydrostatic pressure prior to loading in the element. This breaks up the pellet stack into a number of sub-stacks, so that any gaps which are formed by densification or ratcheting are spread throughout the pellet stack and prevented from ac­cumulating to form a single large gap where the clad could collapse. In PWR fuel this problem is avoided by loading the pins with high density fuel and/or by pre-pressurising the pins with helium (to about 10 bar) to reduce the driving force for collapse.

In highly rated fuel a second important process which occurs during these early stages is fuel grain growth leading to pellet restructuring. Whilst the outer rim of the fuel pellet remains in much the same condition as its pre-irradiation state, at higher tempera­tures (1000-1600°C) equiaxed grain growth occurs. As the grain boundaries migrate they sweep up the intragranular pores allowing them to coalesce and sin­ter by grain boundary diffusion. In solid fuel pellets, where the centre temperatures can be higher (I800°C), the grains become columnar, radiating out from the pellet centre. In this region diffusion is rapid and pores can sinter quickly.

As irradiation proceeds, fuel swelling occurs due to the production of solid and gaseous fission products. In the former case swelling occurs simply because the products of each fission event will almost invariably have a lower density than the atom from which they were formed; this type of swelling is inexorable and proceeds at a rate which is rather less than 1 volume r° per 10 GWd/t(U). In CAGR pins, where the clad and fuel are often in close contact, this swelling is reflected in both pellet and pin dimensions. The for­mation of these elements also has an effect upon the fuel chemistry. In UCb each fission event removes a uranium or plutonium atom from the lattice leaving two excess oxygen atoms. Some of the fission atoms will, of course, react with these to form oxides, but this is not universally the case: noble metals and noble gases (about which, more later) are also formed. Hence the effect of irradiation is such as to gradually make the fuel more hyperstoichiometric. The fuel stoichiometry will also vary across the pellet section with the temperature; overall the effect will be to raise the oxygen potential within the pin, allowing the possibility of internal dad corrosion.

The high pressure high temperature water in the RCS will undergo radiolysis (principally due to 7-rays) in passing through the core, as discussed by Cohen (1969) U8]. A number of elements, activated species and compounds can be formed such as H + , OH-, H02-, eaq, H202, H*, OH’ and the ultimate stable species of interest are hydrogen (H2) and oxygen (02) sith hydrogen peroxide (H202) being formed under lower temperature conditions.

A simple reaction scheme would be as follows,

H* + H* — H2

2H20 + 2eaq — — H2 + 20H “

он + н2о2 H;G + HOr

но2- + H02- — H20; + 02

If allowed to reach equilibrium, therefore, the RCS would carry a significant level of dissolved oxygen with consequent implications for corrosion of RCS materials.

In — addition to formation reactions and the dis­sociation of hydrogen peroxide at high temperatures, oxygen and hydrogen peroxide are removed by the recombination reactions:

2 + 4 H* — 2H20 H202 + H’ — H20 + OH*

The rate constants of these reactions increase with temperature and therefore the hydrogen concentration required to maintain very low’ oxygen levels decreases with temperature.

In order to suppress the radiolysis of water and maintain the dissolved oxygen concentration at an acceptably low level, the RCS is operated under a hydrogen overpressure. An RCS dissolved hydrogen concentration of 25-50 cm3 (STP)/kg is maintained by means of hydrogen addition at the volume control tank in the Chemical and Volume Control System (CVCS). The formation of molecular oxygen is there­fore inhibited and the decomposition of water pre­vented, primarily by scavenging of hydroxyl radicals (which are the precursors of molecular oxygen) by the reaction scheme as follows:

The equilibrium concentrations of the species 02 and H02~ are shown in Fig 1.53 (taken from Solomon 1977 [29]) as a function of initial hydrogen concen­tration. The concentrations of oxygen (02) and the species H02" (a precursor of oxygen) are seen to be very low for the dissolved hydrogen concentration range 25-50 cm3 (STP)/kg. The residual RCS oxygen level under these conditions is in the range 1-5 ppb (parts per billion or 1 in 109), which implies very low corrosion rates for circuit materials possibly augmented by decreased corrosion product release rates.

There are two significant implications of the primary coolant containing 25-50 cm3 (STP)/kg of hydrogen, and its maintenance by the hydrogen purge at the volume control tank of the CVCS. The first relates to the conventional safety of handling gaseous hydrogen,

noting that gaseous and liquid samples withdrawn from the system will contain hydrogen and should be treated accordingly in any sampling/analysis procedure.

The second concerns the effect of the concentration (or partial pressure) of hydrogen on the thermody­namic stability of the oxides present on metal sur­faces, and this is discussed later.

Concrete is strong in compression but weak in tension. For the vessels to be capable of withstanding the gas pressure forces they have to be prestressed to ensure that the bulk of the concrete is under compression

under all possible working and shutdown conditions. This is achieved by steel cables (tendons) which pass through ducts in the vessel walls (Fig 2,12). The size and arrangement of the tendons in the vessel walls is unique to each design of reactor. The tendons are tensioned by means of hydraulic jacks and the loads are transferred to the concrete by means of anchor­ages. Freyssinet type anchorages were used at both

Oldbury and Wylfa. In this type of anchorage the strands of the tendon are wedged between male and female cones (see Fig 2.13).

Unlike most prestressed concrete structures, the tendons are not grouted-in but installed and stressed after the concrete has been cast, i. e., the vessels are ‘post-stressed’. Since the tendons are ungrouted it is possible to check their loading throughout the reactor

Fig. 2.13 ‘Freyssinet’ multi-strand anchorages

life and to re-tension if the load falls below a speci­fied value. Inspection of the tendons for corrosion throughout reactor life is also possible.

There is also reinforcing steel in the vessel, par­ticularly in areas where there may be tensile stresses in the concrete and in the buttresses, which take the anchorage loads.

The vessel design has to meet three main strength requirements:

• Its failure with increasing pressure has to be pro­gressive, not sudden.

• The ratio of the ultimate failure pressure to design pressure must be a specified minimum value (2.5 when the ultimate load behaviour is predictable, as in spheres and cylinders, and 3.0 when the failure mode is not known with any certainty). [16]

The ultimate failure mode of concrete pressure ves­sels is complex and not easily determined by calcula­tion. For each design of pressure vessel, the ultimate failure mode and load has been determined from models; the models also show that the failure me­chanism is progressive. Once the failure mode of a vessel is known, the ultimate load factor can be cal­culated.

Liner

To prevent leakage of the carbon dioxide coolant gas through the concrete a thin steel liner is at­tached to the inside of the vessel. The liner is lugged to the concrete to ensure that it follows the adjacent concrete movements when the vessel is prestressed, when the concrete shrinks and creeps during its life and when there are differentia] temperatures between the concrete and liner. The lugging also prevents the liner buckling under the applied strains. A number of factors such as liner thickness, erection tolerances, and adjacent plate thickness and yield strengths have to be taken into account in the design of the lugging.

The liner also acts as shuttering when the concrete is cast to form the vessel.

The outputs of mean current ionisation chambers are measured by DC amplifiers. A typical logarithmic power measuring channel is shown in Fig 2.50.

An essential feature of nuclear reactor instrumen­tation is that the pow’er must be measured at all times including shutdown. Since the shutdown power is in

Txbie 16

Mam characteristics oj a pulse counter channel

Head amplifier

Input impedance

Maximum input voliage

Tv pica I po lari’mg voltage tor counter

M a x t :r. m : г j :oie:H protect і tm

(_ urreni gain input to 5l) Р output

Bandvv idih at — 3dB

Output voltage when powered via SI.) m of cable Output impedance

Period and doubling time IDT) amplifier and display unit

Analogue

Zero error

Full scale deflection error Period amplifier time constants Internal meter scale External meter drive into 500 ti

Recorder output

Response time: to 20 s DT indication for a 10 s DT ramp input applied at

со via the ramp test input jack ^ 2 s

the region of six to seven decades less than the full power, this requirement necessitates the provision of an amplifier with a range covering many decades. It therefore becomes convenient to use a logarithmic amplifier, the output of which is proportional to the logarithmic of the input current.

The required logarithmic characteristic between in­put current and output voltage can be obtained by using a semiconductor diode as the input resistor of a negative feedback DC amplifier (Fig 2.51); thereby obtaining the advantages of virtual earth operation, e. g., short response time and relative insensitivity to input circuit leakage resistance to earth.

Typically the range of measurement is nine decades 10- 12 to 10“3A.

An important parameter to be measured during the start-up of a reactor is the doubling time. For a con­stant doubling time the power rises at an exponential rate, i. e.,

I = I0exp(t/T)

where 1 = ion chamber current at time t,

I0 = initial ion chamber current,

T = reactor period.

I ‘ 2.51 Basic logarithmic DC amplifier

Піе reactor period is simply related to the doubling ume Ти (the time for the power to double itself,

also rc! erred to as DT) since:

II, — 2 and e.p{Td T) = 2 Ти — ln2 = 0.693T.

It the exponentially rising current (I) from an ionisa — ;lon chamber is fed into a logarithmic amplifier the output voltage (V) is given by:

V = Klnl = K1 n(It>expft T)J = K[ln l0 + (t/T)l

= Klnl0 + К 17T

where К is an amplifier constant.

Thus the output of the logarithmic amplifier rises linearly with time. If differentiated with respect to time this will give a voltage proportional to FT, i. e., inverse of the reactor doubling time. A typical range is 10 to — 10 s, the scale being as in Fig 2.51. This signal can be used for control purposes and to ini­tiate a reactor trip if it becomes dangerously short. Further details are given in Table 2.7.

Тлві. е 2.7

Parameters of a typical logarithmic DC channel

Mam amplifier Input range Output

Period output

Chamber polarising supply Power supply

Fail alarm

Fail safe trip unit Trips on voltage

The channel is provided with circuits which detect any inconsistency between the different modes and there are comprehensive test and calibration facilities. Polarising voltage and tell-tale circuits are also pro­vided in a similar way to those in other channels. The unit can provide indication of:

• Neutron flux on a logarithmic scale.

• Doubling time.

• Linear flux.

• Power deviation.

Further details are given in Table 2.8.

Campbelling channels are being installed in the Heysham 2 AGR but have not been installed on other CEGB reactors.

Part of the gas circuit in the reactor is formed by a gas baffle, which has to maintain its position relative to the core and the standpipes in the vessel roof under all operating conditions so that the passage of fuel stringers and control rods into and out of the core is not impeded. The Heysham 2 gas baffle (Fig 2.85), in addition, supports the diagrid and part of the boiler weight.

When fulfilling its function, the gas baffle has to withstand the pressure difference between gas circu­lator outlet pressure and boiler gas inlet pressure. The baffle operates with gas at core inlet temperature on the inside and over part of the outside, and is insu­lated from the core outlet temperature over the re­mainder of the structure by a thermal shield.

The baffle is a continuous welded structure of high quality mild steel consisting of three main sections; dome, cylinder and skirt. The Heysham 2 design has a cylindrical section of 13.85 m diameter and which is mainly 35 mm thick.

Reactor safety considerations require that the baf­fle must have a high integrity throughout its opera­tional life. Its integrity must be such that gross dis­ruptive failure, which could cause damage to other re­actor components, or result in excessive fuel tempera­tures due to gas bypassing of the core, can be dis­counted. The gas baffle must also continue to support the components which are attached to it and provide access to and from the core for the fuel stringers and control rods.

The integrity requirements are achieved by high quality design and fabrication and full volumetric inspection of welds and the components (plate, forg­ings and cruciforms) from which the gas baffle is made.

Reactor depressurisation is indicated by a post-trip reactor gas pressure of less than 28 bar. The basic X and Y systems sequences are followed as for the pressurised trip, except that emergency boiler feed is immediately introduced into the main boilers, and the gas circulator variable frequency converters are signalled to increase the circulator speed at approx­imately constant power, to provide sufficient gas flow

to remove the reactor heat at reduced pressure. The operators are required to take action to make addi­tional water available to the EBF system.

A reactor pressure support system is provided in order to enhance heat transfer for some very low probability’ depressurisation faults, which could in the limiting case, result in only a single quadrant being available for reactor cooling. This system requires to be introduced into service by the operators and maintains the reactor pressure above 2 bar. Also for any depressurisation faults, the operators must ini­tiate the air ingress prevention system to maintain a CO; atmosphere in the vessel for at least 72 hours post-trip, to prevent oxidation of the fuel and core whilst temperatures remain high.

Reactor trip on high pressure

The most likely cause of this trip is a major boiler tube failure. In order to limit the amount of water entering the reactor, the boilers are isolated on the steam side by signalling the boiler steam stop valves and the start-up line isolating valves to close imme­diately the reactor trips.

When fission heat production ceases on reactor shut­down or trip, the main feedwater pumps continue to supply the steam generators with a relatively small feedwater flow to remove reactor decay heat. Even if the primary coolant pumps are not being driven, natural buoyancy forces ensure that primary coolant continues to circulate between the core and the steam generator U-tubes. The steam generator U-tubes are between 10 m and 15 m higher than the core. Feed control is achieved by the use of small bypass lines around the main feedwater control valves, together with speed control of just one of the six main feed — water pumps. Each one of the pumps is rated for 25% reactor duty and can supply ample feed flow for post-trip heat removal purposes.

If the main feedwater pumps are not available, for example, following loss of 11 kV power supplies, the much smaller auxiliary feedwater pumps can sus­tain steam generator cooling. Two pairs of auxiliary feedwater pumps are provided; one pair electrically — driven, the other pair steam-turbine-driven. The elec­tric pumps deliver feed to each steam generator via separate nozzles; the steam-turbine-driven pumps de­liver to the main feed lines. The steam generated in the cooling process is dumped on the main condensers if normal power supplies are available or, alternative­ly, bled-off to atmosphere via power operated relief valves.

Cooling in this manner allows the reactor to be kept in a stable and safe condition for an indefinite period, albeit in a hot and pressurised state. Reducing the pressure and temperature of the secondary side steam system is achieved by increasing the feed flow and the dump rate. This is the first step in taking the reactor itself to a cold state to facilitate plant repairs, maintenance or refuelling.

However, as the temperature is reduced the speci­fic volume of steam increases and heat removal by dumping steam becomes less effective. Below about 177°C, reactor cooling is achieved by using water- to-water heat exchangers in the residual heat removal system. In this system, two pumps each take coolant from one hot leg (core outlet), pass it through a heat exchanger at a rate of 800 m3/h and return it to two cold legs. The pumps and heat exchangers are located in the auxiliary building basement and are separated from the primary circuit by multiple isolation valves to prevent inadvertent loss of coolant outside the re­actor containment.

The reactor decay and stored heat, removed via the residual heat removal system, is dissipated to the environment via an intermediate closed loop demin­eralised water system (the component cooling water system). This system provides a service to other reactor and process auxiliaries. The component cooling water system rejects heat via one of two paths. The first is to an auxiliary sea water circuit, known as the essential service water system. Alternatively, if sea water supplies fail, heat is rejected to the atmosphere by air-cooled heat exchanger units, termed the reserve ultimate heat sink. Both the essential service water and reserve ultimate heat sink systems are multiple redundant, and one half of either of these systems is adequate to maintain reactor cooling and ensure safety.

The above description of the plant and systems used to achieve post-trip heat removal indicates that the emphasis is on AC-powered, motor-driven main and auxiliary feed pumps, together with general support service pumps and fans. An alternative capability is provided by steam-driven auxiliary feed pumps and power operated relief valves. However, even this al­ternative capability has a requirement for DC sup­plies. The design of the electrical system recognises this and delivers the required AC and DC supplies in a manner which supports the independence, re­dundancy and functional diversity required between the two sets of equipment.

Figure 2.139 provides a basic outline of the elec­trical design. The preferred power supplies to both the main and essential electrical systems are provided from the grid supply operating in parallel with the turbine-generators. In normal operation, this arrange­ment ensures a secure 11 kV power supply to the station based on multiple redundant paths. It allows maintenance to be carried out and has no requirement for switching during or following a reactor trip. Al­though the essential electrical system is usually sup­plied by the 11 kV main power system, it has an independent integral power supply capability based on four-way redundant 7 MW diesel generators and two- way redundant small battery charging diesels. The four diesel generators support the essential loads in situa­tions where the preferred main power system supplies fail. The additional battery charging diesels provide

Post-trip heat removal and electrical auxiliary supplies

assurance that steam-driven plant and essential DC equipment can be kept operating indefinitely in sit­uations where all AC power supplies fail. This fea­ture also assists in ensuring effective realignment of AC power supplies following an extended loss of AC pow er.

In the context of post-trip heat removal, the im­portant items supplied at 11 kV by the main power system are the reactor coolant pumps, the main feed pumps and the main circulating water pumps. This voltage level is suitable for the motor ratings se­lected, and the redundancy of supply paths to the 11 kV boards and the delivered reliability ensure good availability. The design aim is to achieve station ‘run through’ and the reactor coolant pumps are closely coupled to the generators to simplify the electrical supply paths in grid failure situations. However, fail­ure of all 11 kV power supplies following a trip is
acceptable by virtue of the electrical supplies provided at 3.3 kV, 415 V AC and ПО V DC by the essential system. Failure of 11 kV supplies to the main power system also disables the 3.3 kV-powered condensate extraction pumps.

Essentia! electrical supplies are provided at 3.3 kV via the diesel generator backed, four-way redundant essential electrical system to the motor-driven aux­iliary feed pumps, the residual heat removal pumps, the component cooling water pumps and the essential service water pumps. The power supplies to the pumps and fans associated with the reserve ultimate heat sink are provided at 415 V from the essential electrical system. These supply arrangements provide a high degree of confidence that an AC-powered route for post-trip heat removal can be established and main­tained. In the unlikely event that all AC power sup­plies fail, the no-break essential electrical system AC supplies, which normally provide control and support functions to the AC power systems, ensure the con­tinued availability of a post-trip heat removal route, using the steam-driven auxiliary feed pumps and at­mosphere dump valves. The availability of battery charging power supplies via independent diesel gen­erators ensures long term availability of DC power supplies beyond normal battery discha’rge times.

To ensure that the plant and equipment is operated in a safe, correct and consistent manner, plant operating instructions are written to cover the various modes of operation. The operating instructions also cover op­eration under fault conditions so that any corrective action which is taken will ensure that it is done in the safest way to minimise the risk of any further damage.

The site licence calls for a set of operating rules, which when written will be assessed by the Nuclear Installations Inspectorate (Nil) for acceptance. Upon acceptance they are issued back to the licensee as part of the approval licence. The operating rules are a set of instructions primarily concerned with reactor safety giving the minimum amount of protective de­vices that are to be in service at any time the reactor is operating, the minimum of back-up services to be maintained (e. g., CO: stocks, essential electrical sup­plies), the maximum value of operating limits (e. g., moisture in reactor vessel, maximum temperatures), and the way in which derived figures are procured in a statistical calculation. The operating rules are an instruction to the licensee and usually these are con­veyed to the operator by way of the operating in­structions. The plant operating instructions should be written to include the requirements of the operating rules and are therefore mandatory. The instructions tor the various operations should cover the following:

• State of plant before the operation commences.

• Conditions of the reactor at commencement (e. g., CO: density, fuel and core temperatures, etc.).

• The method by which the operation is to be carried out and the limits that apply during each phase of

the operation.

• Operation under steady conditions.

The moderator coefficient of react їх і t> is, as noted aboe, highly dependent on the boron concentration.

* 0rder to maintain a negative coefficient, the boron concentration must be less than about 1000 ppm. This! ч brought about by attaching burnable poisons to the initial charge of fuel. Burnable poisons are not required for subsequent charges of fuel although this may become necessary і I the fuel burn-up is increased over that currently envisaged. At the start of a fuel cycle the coefficient is about — IS mV °С for hot, full power conditions (at zero power the coefficient is vcry close to zero as the absence of xenon poisoning 7 compensated by a higher than normal boron lex cl, typically 1300 ppm compared with about 800 ppm at full power). At the end of a fuel cycle when the boron concentration is zero, the coefficient is about -70 mV °С.

2.3 Rod worth changes

As noted above, the absorption cross-sections of most of the materials in a reactor core vary as I v. Con­sequently, their absorption cross-sections decrease with increasing moderator temperature, varying as the in­verse of the square root of the absolute temperature. The control rods used to shutdown the reactor con­tain boron inserts and, whilst boron is a I/v absorber, the absorption cross-section is so large that all thermal neutrons entering the rod are absorbed. Hence the absorption of the rod is effectively independent of temperature, but the absorption in the rest of the core decreases with increasing temperature. The re­activity worth of the control rod is proportional to the ratio of absorptions in the rod to absorptions in the remainder of the core and, therefore, increases with temperature.

This effect has important consequences in estimat­ing the shutdown reactivity of a core. The worth of sector rods used for reactor control is virtually in­dependent of temperature as these rods do not contain boron.

To describe the kinetic effects of a change in neutron power, consider for example a rod run-in fault. De­viation of reactivity from the value zero causes a reduction in neutron power which in turn causes a reduction in thermal power via delays due to the thermal inertias in the system. The reduction in fuel temperature, with its negative temperature coefficient ot reactivity, gives a positive change in reactivity which is in the opposite direction to the initial disturbance, and il no other factors were acting then the reactor parameters would stabilise at new values appropriate to the new conditions, i. e., reduced power and tem­perature. Any other negative temperature coefficients ol reactivity, such as the fuel outer sleeve in an AGR at star; of life, assist in this process (moderator tem­perature coetliciem of reactivity in a magnox reactor

negative at start of life, but since there are cur­rently none in this state it is not worth considering such a case).

^ hen the reduction in fuel channel gas outlet temperature is sensed by the gas outlet temperature auto control system (where fitted and in use), the controller will withdraw the regulating rods to restore the temperature to the demanded value. In the case of a single rod fault this automatic action, with no operator intervention, may be adequate to recover the situation initially, then the reactor control engineer can take action to restore the longer term situation. For example, if a rod has run-in because of an electrical fault, the fault is cleared, the rod in ques­tion is restored to its normal position, and the auto control system responds so that the regulating rods are returned automatically to their normal positions. Prompt operator intervention is helpful in the ini­tial stages of the fault, for example, he can assist the auto control system by pulling trim rods or other rods which are on manual control such as bulk rods. Prompt operator intervention may be important to counteract the action of an auto control system which is too fast-acting. For example, at Berkeley (magnox) the auto control system restores temperatures so quick­ly that the reactor may trip on excess temperature because the temperature is restored faster than the following rate of the auto-reset temperature trip am­plifiers in the safety lines. In this case the auto con­trol system is disengaged as soon as the temperature starts to rise, the auto control system having arrested the initial fall, and temperature restoration is then carried out at a rate determined by the reactor con­trol engineer. Where no auto control system is fitted, or if it is not in use, prompt action by the reactor control engineer is of course necessary, and the action which he takes will be similar to that of an auto control system, although he has only one pair of hands whereas the auto system has several!

If temperatures can be restored within a few min­utes, the disadvantageous effect of the positive tem­perature coefficient of reactivity of the moderator in a magnox reactor is barely noticeable because it has a long time constant of 10-20 minutes (in an AGR the bulk moderator temperature is held largely con­stant by the re-entrant gas flow). If the moderator temperature does fall because of the rod run-in, ad­ditional corrective action on regulating rods, etc., will be necessary to offset this negative reactivity, and if reactivity from rods is limited because of high fuel irradiation, a situation which exists on several magnox stations, there is a strong incentive for prompt action by the reactor control engineer to restore tempera­tures and power before the moderator temperature is significantly affected.

If, however, there is a more generous amount of reactivity available from rods, so that a reduction in moderator temperature can be tolerated at least in the locality of the failed rod, it may be preferable to al­low temperatures in that locality to remain depressed until either the fault is rectified and the failed rod restored, or to restore temperatures in the locality more slowly. The latter is easier from the reactor con­trol engineer’s point of view and less likely to result in problems which may arise in taking quicker cor­rective action.