Quantifying Uncertainty in Flow Accelerated Corrosion (FAC) Predictions
9th International Conference on CANDU® Maintenance - 2011 December 04-06


Presented at:
9th International Conference on CANDU® Maintenance
2011 December 04-06
Location:
Toronto, Canada
Session Title:
Fitness for Service Session C3

Authors:
Mikko Jyrkama (University of Waterloo)
Mahesh Pandey (University of Waterloo)
  

Abstract

The fundamental challenge with the FAC analysis of nuclear piping is accessibility, leading to the inability to make precise measurements at fixed locations.  The lack of fixed location referencing introduces considerable uncertainty into the predictions, particularly for methods that rely on point-to-point comparison (e.g., of the minimum point between consecutive outages).  The uncertainty is further compounded by the contact and coverage errors associated with the NDE inspection tools.

To minimize the uncertainty in prediction, a new method for estimating the FAC thinning rate and pipe lifetime is developed in this paper using a probabilistic sample based approach.  Although the methodology was developed in the context of feeder assessment, involving the analysis of a large dataset of over 1500 separate feeder wall thickness scans from seven different operating nuclear reactors, it is equally applicable to other piping subject to the FAC process where the precise alignment of consecutive wall thickness measurements is not possible.

The developed methodology focuses on a small circular area or patch in the vicinity of the minimum point, rather than only the minimum point itself.  Because the FAC process is relatively stationary over time, the analysis is insensitive to the patch size, and therefore provides a highly robust way to estimate wall thinning rates.  The only significant uncertainty is due to the inspection probe coverage and contact error, which is minimized by the sample based approach and a surface interpolation and fitting scheme.

The new method can readily be used to estimate pipe lifetimes in a probabilistic setting.  This directly supports risk-informed decision making at the station by quantifying the present and incremental risk in the piping system over time.

Individual Conference-Paper Copies (Electronic Where Available):

  • For CNS members, the first 5 copies per calendar year are free, and additional copies are $10 each.
  • For non-members, the price is $25 for the first Conference-paper copy in a request, and $10 each for additional copies of papers in the same conference and in the same request.
  • Contact the CNS office to order reprints.