The sequence of events outlined in Section 6.2 applies to a PWR; the situation with regard to other reactor types can he summarized as follows:

Boiling-Water Reactor (B^R). The situation is veiy similar to that in the P^^ regarding sequences of core meltdown, fuel-water interaction, and ulti­mate disposition of the molten fuel pool.

CANDU. The melting sequence is not considered to be very likely because of the large pool of moderator heavy water through which the individual fuel channels pass. Analysis of the heat transfer events following a loss-of-coolant accident and failure of the emergency core cooling system has indicated that significant fuel melting would not occur and provided the means of extracting heat from the moderator were still intact, the accident would be controlled. However, should a single-pressure tube fail and the moderator become pres­surized as a result of the release of high-pressure steam into it, the moderator could be expelled and its cooling effectiveness for the other channels removed. That this is at least a remote possibility is indicated by the accident at Lucens, described in Chapter 5. If the moderator was expelled, fuel melting would pro­ceed in the same way as for the other water reactors; again, this event could be contained provided there were no steam explosions or other events that dis­rupted the containment.

Magnox Reactor. The inherent basic safety features of the Magnox reactor (the fact that the graphite itself may absorb a great deal of heat and that decay heat removal can be maintained even if the reactor is depressurized) have led to the view that a full core meltdown is not credible. However, some studies have been done on the effects of meltdown of single channels, specifically those with the highest rating. That such single-channel events are credible is borne out by the accidents in this type of reactor discussed in Chapter 5. In the Magnox reactor systems, such events can lead to small releases of activity since the reactors do not have the hermetic containment that is provided for water re­actors.

Advanced Gas-Cooled Reactors (AGRs). Full meltdown accidents are not considered credible for this type of reactor for much the same reasons as men­tioned above for the Magnox reactors. Furthermore, with AGRs, much higher fuel temperatures can be sustained before fuel damage since the fuel is in the oxide form and clad in stainless steel (in a Magnox reactor the fuel is in the form of uranium metal clad in magnesium alloy). Tests in the Windscale proto­type AGR showed that the fuel temperature can approach to within 50°C of the melting point of steel without clad meltdown and significant fuel damage. How­ever, single-channel fuel melting due to local blockage effects, or due to the dropping of a fuel stringer during the refueling operation, is still considered possible and is taken account of in the design. As explained in Chapter 4, the rise of temperature following a loss-of-coolant accident in an AGR is very slow indeed compared with that in a P’^TC or a B’^TC. This means that there is time to take alternative actions, even if off-site power is lost and the local power sup­plies feeding the emergency circulators fail to operate immediately. It is inter­esting to compare the situations in an AGR and P’^TC; in the AGR the consequences of a fuel meltdown would be more serious since it does not have hermetic containment; on the other hand, the probability of a meltdown is even smaller than in the case of the P’^TC.

Liquid Metal-Cooled Fast Reactors. The very high fuel ratings in fast reac­tors have led to much interest in the possibility of core meltdown and its con­sequences. One accident scenario is that of failure of all the primary sodium coolant pumps and complete failure of the reactor shutdown system. As the sodium reaches its boiling point in the channels of maximum rating, sodium boiling and voiding occur, and this has a net positive reactivity effect on the re­actor, which accelerates the heating. Melting of the fuel and cladding occurs in about one second after sodium voids are formed in a particular fuel assembly. In the area of that assembly there is a complex mixture of liquid fuel, sodium vapor, liquid steel, fuel fragments, fission gas, and steel vapor. If the fuel chan­nel walls melt, adjacent channels may also be damaged and melted.

Calculations of the consequences of these events are highly complex be­cause of the coupling between the nuclear reactions, the heat transfer processes, and the fluid flow processes. Two different outcomes are possible, depending on such things as the reactor design and reactor state at the begin­ning of the accident:

1. If, during the meltdown, a large fraction of the original fuel has managed to remain within the active core region, an extremely large increase in reactiv­ity occurs and the fuel is actually blown apart and dispersed by the fission product gases in the interstices of the fuel pellets. The dispersal of the fuel terminates the nuclear reaction, though the resultant shock wave may dam­age the reactor structure and breach the containment.

2. If the fuel inventory has been reduced to about half the original amount by gradual leakage, or if large quantities of blanket materials have diluted the fuel, a severe power excursion will not occur. The molten fuel will fall ro the bottom of the reactor and the sequence of events will be similar to that de­scribed above for the P’^TC, including the possibility of a vapor explosion due to interaction between the molten fuel and the liquid sodium still in the vessel. The possibility of this form of accident has drawn great attention to the reliability of shutdown systems in fast reactors; one possible design ap­proach is to arrange the core structures so that an excessive increase in core temperature causing its thermal expansion will trigger an automatic shut­down of the reactor. Combined with the fact that the decay heat can be re­moved by natural circulation to air-cooled heat exchangers and the enormous heat capacity of the sodium coolant, this inherent shutdown sys­tem would give the fast reactor system a “walkaway" safety capability that is not available in other reactors, which depend on the operation of active sys­tems demanding operator actions and/or totally reliable power supplies.

Clearly the attention given to core meltdown accidents varies from reactor to re­actor and depends on the assigned credibility for such accidents. In general, the objective is to bring down the likelihood of an accident and in particular its public consequences to a minimal level.