In order to address the risks associated with the operation of ageing pressure boundary components, many assessments incorporate probabilistic analysis methodologies for alleviating excessive conservatism of deterministic methodologies. In general, deterministic techniques utilize conservative upper bound values for all critical parameters. Equally, defense-in-depth assessments for the nuclear industry employ probabilistic methods in order to estimate potential risks associated with unanticipated events to demonstrate adequate margins associated with the licensed activity.
Probabilistic approaches typically invoke the Monte-Carlo (MC) approach where a set of critical input variables, assumed independent, are randomly distributed and inserted in deterministic computer models. Estimates of results from probabilistic structural integrity assessments are then compared against assessment criteria, at times, based on the assumption that these results follow normal distributions. However, this assumption is not always valid, as normality depends both on the initially assumed distributions of the input variables and linearity, or lack thereof, of the deterministic model. In particular, the characteristic of a system function (either a linear or a non-linear system function) and the sampling region of input parameters affect the level of normality of the MC simulation results.
As a proof of principle, a specific case study is presented. A system function is chosen based on the steady-state thermal creep of Zr-2.5Nb Pressure Tube (PT), instead of a full deterministic computational model, to show whether it can give rise to MC results that deviate from normality. The consequence of the deviation from normality when compared against assessment criteria is briefly discussed. It is noted that this study does not deal with analysis of Probabilistic Safety Assessments, also known as PSAs.
Introduction of a video assignment: advantages and disadvantages from the students’ perspective