T.Dvoršek, L.Cizelj

A methodology aiming at assessing the efficiency of a given set of possible maintenance approaches has already been proposed by the authors. It pointed out better performance of the degradation specific over generic approaches in (1) lower probability of single and multiple steam generator tube rupture (SGTR), (2) lower estimated accidental leak rates and (3) less tubes plugged.

A sensitivity analysis was also performed pointing out the relative contributions of uncertain input parameters to the tube rupture probabilities. The dominant contribution was assigned to the uncertainties inherent to the regression models used to correlate the defect size and tube burst pressure.

The uncertainties, which can be estimated from the in-service inspections, are further analysed in this paper. This gives an estimate of time dependence of the estimated single SGTR probabilities, taking into account (a) the initiation rate of newly detected defects, (b) growth rate of old defects and (c) different available maintenance options (e.g., traditional 40% wall thickness and alternate bobbin coil voltage repair criteria). A special emphasis is given to the statistical interpretation of inspection results and related uncertainties, which can affect the estimated rupture probabilities.

A realistic numerical example is provided to illustrate the effects of above assumptions. The data used is obtained from a series of inspection results from Krško NPP with 2 Westinghouse D-4 steam generators. The results obtained are considered useful in planning the remaining safe lifetime of affected steam generators.