Probabilistic models for fuel rod puncturing in an LWR core during a LOCA

The number of fuel rods which puncture during an LWR loss-of-coolant accident (LOCA) must be estimated as part of the plant radioactivity release analysis. Due to the great number of fuel rods in the core and the great number of contributing parameters, many of them associated with wide uncertainty and/or truly random variability limits, probabilistic methods are well applicable. A succession of computer models developed for this purpose is described together with applications to a WWER-440 PWR. Deterministic models are shown to be seriously inadequate and even misleading under certain circumstances. A simple analytical probabilistic model appears to be suitable for many applications. Monte Carlo techniques allow the development of such sophisticated models that errors in the input data presently available for practical problems probably become dominant in the residual uncertainty of the core-wide fuel rod puncture analysis.