CLEMENTINE, a very small reactor with a power of up to 25 kW, is noted as the one reactor which was cooled by mercury. It was fueled by plutonium clad in steel with a slug of natural uranium at each end. Control of the system was achieved by moving uranium reflector and boron-10 safety rods. Figure 4.43 shows a cross section of the reactor (48).

Irradiation defects eventually caused the uranium plugs to swell and burst the cladding, allowing the plutonium and mercury to mix. This mixture is

a pyrophoric solution that vastly complicated the eventual disassembly of the reactor system. In addition, the uranium safety block swelled and stuck immovably.

One interesting safety highlight during the operation of the reactor con­cerned the possibility of damaging the system by propelling some object down the neutron window (seen to the right of the core in Fig. 4.43). Just in case, as a safety measure, the guards in the building were disarmed rather than stand the chance of one of them firing his revolver down into the core.

Safety related incidents are likely to continue; it is up to the safety engi­neer to learn from the errors so far committed, to collect the data from each occurrence, and to create more reliable systems.

Local failure might include local subchannel blockages, defected fuel, or the failure of a pin that inadvertently included more highly enriched fuel than the design called for. Section 4.4 has already discussed such accidents and their possible consequences, including the possibility of a propagation of that failure to other pins and other assemblies.

The previous discussion showed that a propagation of failure from a case that resulted only in the release of some fission gases was very unlikely and even when molten fuel was involved, current analysis seemed to show that failure propagation did not occur. The most probable local defect was a subchannel blockage due to a failure of the wire wrap or grid supports or possibly from crud deposition. Such a case is the subject of much analysis and experiment aimed at determining the possible course of events and at showing that such a failure would be localized within a single assembly.

Propagation of a failure in which only gas release occurs is not considered a credible CDA initiator. A failure following a very low probability local blockage is also not expected to give rise to more than localized core damage, but this has not yet been conclusively demonstrated. No such failure propagation has ever been observed; nevertheless the accident should be retained as a low ranking candidate CDA initiator for further consideration.

Such disturbances may be separated into visual and political considera­tions.

6.4.1.1 Visual

Individuals living in the locality, preservation groups, and the power industry itself are all concerned in lessening the impact of the power plant on the countryside. This can be done by siting in relation to geographical features, architectural design, and landscaping.

The power plant may be sited advantageously to diminish the height of buildings, to hide transformer yards, to remove power transmission lines from skylines, and to use existing roads and facilities where possible. Archi­tectural design enters into the picture by the design, placing, and coloring of buildings that blend rather than strike the eye. The British Steam Gen­erating Heavy Water Reactor (7c), for example, was built in the lee of a small hill, half below grade; the building was colored green and oriented to diminish its effect on the area, all in order to improve its visual impact. Landscaping of every industrial installation is of course commonplace although less effective with very large power plant buildings.

Nuclear power plants are clearly much simpler to make visually attractive than fossil-fueled power plants because of the lack of a smoke stack and the lack of vast fuel storage areas around the site. Both types of system have the same problems of power transmission and their associated high voltage lines.

The fast reactor is a smaller reactor for a given power output, but as far as associated plant is concerned, it is essentially the same size as the thermal light water systems and has the same siting visual effects as other nuclear plants.

The surface temperature drop to the coolant is calculated from an empi­rical heat-transfer coefficient that depends on the channel geometry, the fuel element and its cladding, and the coolant velocity and thermodynamic properties. There are a number of such correlations for different coolants and different ranges of operation.

Water. The Dittus-Boelter correlation (7a) is

Na = 0.023(Re)»-8(Rr)»-33 (1.36)

Sodium. The main correlations all follow a general equation of the form (7b)

Na = A + 0.025(Re)os(PI)°-8 (1.37)

and in both cases the heat-transfer coefficient h can be calculated from the Nusselt number Na, given by

Na = hdJkA (1.38)

These correlations usually result in the heat transfer being proportional to flow to the 0.8 power because of the Reynolds number dependence.

Internal heat-transfer coefficients are calculated from the conductivities combined as in Eq. (1.25). The main uncertainties occur in not knowing the condition of the fuel or the fuel/cladding gap.

Alternatively heat-transfer coefficients can be calculated from the steady — state temperature distributions for a given heat input [Eqs. (1.29)—(1.32)] if they are known from experimental measurements.

Probability analysis is best used where failure statistics are good, for example, in control systems that are made up largely of conventional components. To date, probability analysis is not used regularly in accident analysis.

However, pseudo-probability analysis, otherwise known as engineering judgment, has been supplemented in some cases by statistical analysis and by partial probability analysis. For example, failure of scram systems has been variously estimated at between 10-4 and 10-e per reactor year despite the difficulty of quantizing possible short-circuit paths. On a somewhat firmer basis (17), from an analysis of safety and shim rod hang-ups in the SM-1A reactor at Fort Greeley in Alaska during a 72-month period, the probability of failure to scram on a two out of three system is calculated to be 4.1-Ю-4 per reactor year. Such an analysis applies to a particular reactor system but in general can be applied to any plant protective system composed of similar parts.

On a somewhat more speculative basis and applying now directly to the sodium-cooled fast reactor, one analyst has applied the figure of 10~8 per reactor year to the probability that a small local failure in a fuel subassembly would propagate to a neighboring assembly. The error on this probability figure would be as much as two orders of magnitude either way!

In general then, the fast reactor is in the same position as the thermal systems: there is a desire to use probability theory to aid in the assessment of reactor safety, and a good deal of work is being done to apply the theory to protective systems and systems that use components for which failure rates are reasonably well known (18-20). In addition, further effort is being applied to overall containment analysis, where quantitative engineering judgment is used in place of failure rate data (21, 22).

When failure rates for reactor components become known (see Section 3.1), then probability analysis of system safety will undoubtedly be an established assessment method. Such a method will clearly be used initially to improve the reliability of reactor subsystems, such as the protective system or emergency core-cooling system (19), but ultimately general safety accident analysis should benefit by an application of these methods.

Each type of scram has a combined delay which depends on the delays in obtaining signals from the system and delays in acting upon these signals.

(a) System physical transport delays are usually small except for the transmission of temperature in the reactor, plenums, and heat exchangers. The transport of fission products from failed fuel to fission product and delayed neutron detectors elsewhere in the system also incurs a delay.

(b) Instrument delays vary with the instruments in question.

(1) Reactivity instrumentation, flux, and period meters incur a delay due to noise filtering; the latter incur the more serious delays as they need more filtering.

(2) Thermocouples have varying delays depending on whether the couple is sheathed or not. A |-in. ungrounded thin sheath thermocouple in a LMFBR might have a 1000 msec delay with a further 10 msec for its amplifier.

(3) Flux/flow ratio measurement delays depend on the fact that separate measurements are divided. Electronic dividers may have no delay, whereas electromechanical servos might add 100 msec and thermal dividers could add 500 msec.

(4) Tachometer measurements are rapid; a dc tachometer filter delay would be about 5-10 msec whereas an ac induction tachometer can have filter delays from 5 to 50 msec.

(5) Flowmeters can have delays of 10 msec or less.

(6) Pressure meters incur a delay through the diaphragm and isola­tion device which may be between 250 and 500 msec, depending on the capillary tubing used. There may be an added 100 msec sensor delay.

(7) Under — or overvoltage relays would have to be protected from spurious operation due to surges and false alarms by time lags which could amount to 500 msec.

(c) Logic delays following the reception of the signal would typically be 20 msec.

(d) Once the signal has been received and analyzed and dispatched to the shut-down mechanism, a further rod release delay including the scram breaker dropout would add a further 120 msec.

2.6.3.1 Total Scram Delays

From these various component delays we obtain the total typical delays in a LMFBR system shown in the accompanying tabulation.

The temperature signals are noticeably slower and have additional phys­ical delays which are not included in the tabulated figures. These transport delays range from 500 msec in the IHX measurement to 1000 msec at the outlet plenum or even longer at low flow levels.

Scrams using such combined delays as these are used in the safety evalua­tion representations.

Coolants other than liquid metals are possible and previous sections of this book have dealt with reactor systems using supercritical steam and helium as coolants.

Although the U. S., British, French, and Soviet national fast breeder programs are based on sodium, gaseous coolants have not been ruled out. Indeed the gas-cooled system shows considerable breeding potential. Figure

4.2 shows comparative breeding ratios for different fuels and coolants.

Steam gives rise to a poorer breeding ratio, because it degrades the spectrum while low density helium, although a moderator, actually improves the spectrum in comparison to sodium.

In order to obtain adequate heat removal, a nonmetallic coolant must be used at high pressure: over 3000 psi for supercritical steam and of the order

+ See Dalle-Donne (4a) and Hummel and Okrent (4b).

of 1000 psi for the helium gas-cooled system. This high pressure then gives rise to problems associated with depressurization of the system.

Gas-cooled fast reactors are susceptible to water-flooding accidents, either from an external source or from the high-pressure steam generator. The resultant reactivity increase can be prohibited by the addition of resonance absorbers to the system to absorb the thermalized neutrons, but in practice this would probably be unneccessary because flooding-induced reactivity changes would be slow transients and could be engineered out of the system.

Despite the economic potential of gas-cooled fast reactors, the remainder of this volume will refer only to sodium-cooled systems, unless otherwise noted, because sodium is the chosen coolant for the first generation of fast — breeder power reactor plants around the world.

In any safety text a summary of previous accidents in similar systems is most important, because it is only by our experience that we learn. An excellent survey of all reactor accidents up to 1964 is contained in the work of Thompson and Beckerley, referred to in the Preface. The following section merely highlights those main operating experiences and occurrences which have taken place in fast reactors. No systematic classification is intended; rather it is hoped to show a general picture of the multitude of kinds of problem that arise and that have to be accounted for in a com­prehensive safety evaluation.

4.6.1 Survey

No really large fast reactors have yet been built; the largest which are presently under construction are the Prototype Fast Reactor (PFR) in Britain, Phenix in France, and BN 350 and BN 600 in the USSR, all of which are above 550 MWt. In the USA, the Fast Flux Test Facility (FFTF) will include a 400 MWt reactor. Table 4.7 shows a listing of the main para­meters of the most important reactors already built or in the planning stage.

Apart from CLEMENTINE which was cooled by mercury, all fast — reactor systems to date have been sodium — or NaK-cooled. There have been no gas-cooled versions.

Some of these systems have suffered operational teething problems before and after start-up, and others have experienced accidents and incidents dur­ing operation. These accidents have never resulted in any risk to the public, and only two have resulted in considerable material damage and delays. The following sections outline some of the major problems.

Following the dose calculations for particular operating and accident conditions, the containment design will be chosen to ensure that the Federal Regulations embodied in 10 CFR 20 and 10 CFR 100 are met and, indeed, are bettered.

The annual dose from a PWR from its design basis is 5 mrem/yr although it would be expected to be considerably better than that and values in the region of 0.0063 mrem/yr are likely to be experienced at the site boundary (3b, 6). These values should be compared to the 500 mrem/yr allowed by 10 CFR 20, and they should also be compared to the natural background of 100-600 mrem/yr and the 75-100 mrem/yr of man-made background (see Tables 5.3 and 5.4).

To put these values in perspective it is worth noting that for an increase in altitude of 600 ft there would be an increase in cosmic radiation of approx­imately 5 mrem/yr. Thus the federal regulations allow a release correspond­ing to 25,000 ft,+ the PWR design basis allows 600 ft, but actual emissions are in the range of an equivalent 12-in. increase in altitude!

A fast reactor has no difficulty in bettering the LWR release design figures. Indeed current designs allow for essentially zero release plants during normal operation with on-site storage of fission product wastes.

Recent work is directed toward deriving a model for energy transfer to the sodium by considering high energy fuel particles being ejected from the core. These are presumed to transfer energy to the sodium through which they pass, leaving a trail of vaporized sodium along their flight paths (see Fig. 5.13).

Such a model would include the three conservation equations for mass, momentum, and energy and an equation of state for the vaporization of the sodium.

Conservation of mass

dmjdt = —dmv/dt (5.20)

Conservation of momentum

(mh + m{ + |ttjv + mg) dvb/dt = A(PV — Pg) — (mL + mt + mv + mg)g

(5.21)

Conservation of energy

— (»V — »l) dmv/dt = (mv duv/dt) + (mt дщ/dt) + vbA(Pv — Pg) (5.22) Equation of state

Hl = ub{T) (5.23)

In these equations m is mass, и is internal energy, P is pressure, v is velocity, A is the cross-sectional area of motion, and the subscripts L, f, v and g refer to the liquid sodium, fuel, sodium vapor, and cover gas (see Fig. 5.14).

Fig. 5.14. Sodium hammer model, including the sodium vapor driving force.

The analysis requires a fuel model in order to compute the heat transferred from the fuel (m{du{jdt). It assumes that incompressible sodium is being driven, by the pressure of the sodium vapor produced, against the cover gas. Such a model also needs an allowance for direct momentum transfer, since the previous driving force considered in Section 5.5.2.1 is also a driving force here. It should take account of the fact that fuel particle populations of differing sizes may be shot into the sodium in different times and the

final driving force is an integrated effect of sodium vapor produced by a large number of individual energy transfer events.

Such a model could be included into a larger hydrodynamic model and could be expanded to include the effect of vessel internals. Table 5.13 shows some sample results from such a model, results which show a very moderate sodium hammer behavior, for in one case it did not even reach the vessel plug before the sodium vapor condensed. Later improved versions of this analysis included waves of particles reaching the sodium thus resulting in much larger slug energies up to 20 or 30 MW-sec (27b).