Evaluation of the bases for estimating alpha-mode failure probabilities

Much research has been devoted to understanding fuel-coolant interactions (FCIs) that might occur during core-melt accidents in water reactors. One major objective of this research is the evaluation of the probability of containment failure due to an in-vessel steam explosion (alpha-mode failure). Such a failure could conceivably contribute significantly to the risk associated with these accidents. Similarly, demonstrating that such a failure is extremely unlikely could significantly reduce the uncertainty concerning estimates of risk. Most analyses of alpha-mode failure decompose the failure into several stages, all of which are assumed to be necessary for containment failure to occur. The following is the description generally employed (throughout this paper, words in single quotes indicate qualitative terms which must eventually be quantified): A 'large' amount of melt accumulates in the core region, a 'large' fraction of which pours 'rapidly' into the lower plenum, which contains a 'large' quantity of water. The melt mixes 'efficiently' with the water to form a premixture. The premixture is 'triggered' and an 'energetic' explosion ensues which accelerates a water-fuel-structure 'slug' up the core barrel, creating a missile out of the upper head with a 'velocity' sufficient to propel it through the containment dome. Hypotheses and models have been developed which claim that one or more of these stages is impossible. This paper will examine the bases for those claims and the extent to which they agree with experimental observations and calculations, and with each other. 37 refs., 1 tab