Impact of source term research on regulatory processes and NPP design

Expectations of secondary containment have changed considerably over the past few years as new challenges have appeared. Far from eliminating the detailed analysis of accidents at the extremes of severity and probability scales, the use of PRA (probabilistic risk assessment) has focussed attention on the continuity of the risk spectrum. Containment itself has evolved from a passive system designed to respond to a single DBA (design basis accident) into an active system expected to survive a spectrum of far more severe challenges. To expect a single structure to contain radioactivity as well as act for the multiple barriers it may have to replace, is clearly 'optimistic'. Hence we are now faced with the options of doubling up the containment wall - a 2-barriers containment - or of recognizing that in certain circumstances a delayed and controlled leak may be preferable to the risk of catastrophic failure. If the resulting point on the probability and consequences curve is constant with a risk that falls as the size of the accident increases, then one can at least argue one is doing one's best. Licensors find a risk continuum difficult to work with for a variety of reasons, of which uncertainty is just one. As a result the nuclear community is trying to evolve a small number of bounding scenarios (or a small fault tree) leading to containment failure. Current research aims at reducing the very large uncertainties in estimated source term and their probabilities -handling of uncertainty is very important as is shown in the NUREG 1150. In fact, release of fission products involves a long sequence of processes each of which carries uncertainty. Risk uncertainty will be even greater since that includes the uncertainties attached to the various probabilities used. It is not surprising that the uncertainty attached to a source term can be very large when one realizes that it includes all the uncertainties involved in the modelling of the accident. (author). 16 refs