Impact of consistency between modeling detail and physical uncertainties on severe accident predictions

Abstract

The complexity of phenomena occurring during severe accidents in nuclear reactors, combined with a limited amount of experimental evidence, do not allow for formulating detailed reliable models or making highly accurate predictions of accident progression. Thus, the modeling consistency and a proper understanding of the uncertainties associated with the results of any computer simulations, including those caused by the imperfection and inherent limitations of the available experimental data used in model validation, are critical for improving accident mitigation capabilities and for enhancing the safety of current and future generations of nuclear reactors.The objective of this paper is to give an overview of selected issues illustrating the importance of:(a) identifying the dominant phenomena governing the progression of core meltdown accidents, and (b) formulating models which are consistent with our understanding of the underlying physics and chemistry and of the increasing level of randomness as the accident progresses. The results used as examples, in particular those pertaining to the research performed by the authors and their collaborators, have been obtained over past several years and documented in a several reports (in particular, in the USNRC NUREG series), but have never been included in copy-righted publications. The sources of any other experimental data used in the discussion of specific coupled experimental/modeling issues (typically, also reports of various agencies or labs) are clearly identified in the text. Since some of the examples include recent unpublished results of computer simulations performed using updated versions of the models which have already been published in other journals, only brief information about such models is presently shown, assuming that the reader can find details in the corresponding references.