Probabilistic optimisation of the maintenance of steel components in French nuclear power facilities

Abstract

Certain structural steel components located in the secondary part of PWR (Pressurized Water Reactors) nuclear power plants are subject to an atmospheric corrosion phenomenon in a confined environment affecting their bottom by the external surface. For one of them (a vessel), several campaigns to measure the thickness of the bottom of the vessel are available over time over a population of measurement points.

It is possible to assess, from these measurements, a distribution of thickness loss kinetics and, at each measurement point, a distribution of residual thickness over time, and a risk of under-thickness. By combining these individual risks, we obtain an overall risk of under-thickness over time for the entire population of measurement points. By imposing a limit on this risk, we deduce a number of points to repair (patch).

It is this predictive probabilistic approach to maintenance that is detailed in this article, both methodologically and industrially. First, statistical tests [1] confirm that corrosion thinning is significant. As no relevant physical model is available for the corrosion rate, it is directly estimated from the data. Normal distributions (adjusted on the available measurements) can be accepted for thickness loss kinetics. A best-estimate reference distribution is considered. The mean corrosion rates for different campaigns are consistant. It is found that at the time limit considered, the overall under-thickness amounts to around 50% if no repair is planned before. To reduce this risk to an acceptable level estimated at 10% (resp. 5%), 8 (resp. 11) points have to be repaired before this time limit.

A structural reliability approach [2] has therefore demonstrated its ability to solve an industrial problem, even based on the elementary ﾓR-Sﾔ case. On a scientific point of view, this study confirmed the need to perform overall risk assessments, based on system reliability (parallel systems), and consequently the relevance of the recent corresponding evolution of the uncertainty treatment software OpenTURNS [3].

References

1. Saporta, G., 2006. Probabilit�s, Analyse des Donn�es et Statistique, 2nd edition, Technip, 2006 2. Lemaire, M., 2009. Structural Reliability. Ed. by Jacky Mazars. London: ISTE Ltd., Wiley. 3. Baudin, M., Dutfoy, A., Iooss, B. and Popelin, A.-L., 2017. ﾓOpenTURNS: An industrial software for uncertainty quantification in simulation,ﾔ In: Handbook of uncertainty quantification, R. Ghanem, D. Higdon and H. Owhadi (Eds), Springer. HAL version - Springer link