In principle, it is important to maintain a subcritical system during the entire period in which the system is subjected to thermal failure, so that there is no excursion and no explosion, and to continue to maintain that subcriticality during the decay period until the damaged core can be removed.

Even if there had been a core disruption, the debris would need to be maintained in a subcritical condition to avoid further, and possibly worse, excursions. This would need to be done either until debris could be removed after the decay heat had subsided so that the fuel could freeze.

Figure 5.15 illustrates the system used in the Fermi Reactor {29). It comprises:

(a) a zirconium clad conical flow guide designed to disperse any molten fuel from the core into a distributed and subcritical configuration;

(b) a series of zirconium clad plates in a melt-down section through which

the fuel could possibly melt (this melt-down system would provide a time delay to reduce the decay heat level before the fuel reached its final resting place);

(c) a further internal cone in the melt-down section to maintain sub­criticality by dispersion;

(d) the vessel and the guard vessel, both of which provide delay time while the fuel penetrates them, and provide a coolant system boundary for some sodium cooling while the fuel is melting downward; and

(e) a graphite crucible outside the vessel designed to catch and retain any molten fuel which reaches it (at this point, the molten mass would be cooled by sodium in the flooded vault).

In fact, as Section 4.6 relates, the zirconium cladding on the first cone worked loose and blocked the coolant channels, thus causing a small melt­down of two assemblies. The molten fuel from these assemblies froze in the lower section of the core.

It is worth learning the lesson that safety features in the system should also be evaluated for their possible adverse effect on safety in other respects.

Figure 5.16 illustrates the system used in the Dounreay Fast Reactor (30). It comprises:

(a) annular fuel elements with an inner cladding (vanadium) having a lower melting temperature than the outer cladding (niobium) (intended to direct any fuel release following fuel element failure down the inside of the element itself);

(b) downflow to help molten fuel leave the core and make the system subcritical rather than to retain it in the system (downflow forces comple­ment gravity);

(c) an invessel cone to disperse the fuel in the same way as in Fermi;

(d) an outside vessel—a graphite, steel lined cone—again to disperse the melt into a subcritical configuration and into its final catchpots; and

(e) twenty-four steel lined drains or catchpots to take any molten fuel away into the bedrock on which the reactor stands.

This system has never been used, and the reactor has never experienced more than minor fuel failures, in which no molten fuel was involved.

The modern designer faces problems very similar to these, with the added complication that the fast reactors under consideration are now much larger. Dounreay is 72 MWt, the Fermi plant is 200 MWt, while present day systems in design are from 800 to 2500 MWt. With the larger plants, even in decay mode, more heat is produced than in Fermi at full power. This demonstrates the magnitude of the task of providing for cooling of a damaged core even when shut down.

Scaling up the simple solutions becomes prohibitively expensive and uncertain. It is difficult to justify such addition to the design to cope with an accident when confronted with a growing amount of evidence saying that such an accident will never occur.

Some recent work indicates that it may be possible to allow the molten fuel to take its own course in penetrating the vessel and falling onto the vault floor. Melt-through of the vessel would take about 5-10 min if the whole core were involved, although fuel cooled by a pool of sodium above might never penetrate the vessel if only one or two assemblies were involved.

Even if the fuel emerged from the vessel, it could be contained in from 5 to 20 ft of concrete depending on whether cooling was applied through a cooling pipe system close to the concrete bed or not. Other factors which are critical to such a calculation are the delay before the fuel arrives at the concrete, what heat is being produced by decay, and, moreover, in what form the debris arrives at the concrete. Exactly what occurs is dependent on the type of concrete in use (31a, b) and it is important to have compatible
materials and concrete which does not dissociate or produce gaseous sub­stances such as carbon dioxide which could over-pressurize the system. No design solution has yet been found or, moreover, has been shown to be needed.

Further study at ANL is showing that the heat flux from debris is such that, with some in vessel cooling, the best and most promising position in which to retain debris is above the core support plate. This has the ad­vantage of being simpler to design to and somewhat more believeable than core catchers used to date.