Probabilistic Flaw Growth Rate Estimates Using Multiple Inline Inspection Tool Run Data Analyses

Abstract

Many pipeline operators perform run comparisons (or “run-corns”) during assessment of their inline inspection data. Analysts match corrosion pit depths reported in back-to-back tool runs and estimate corrosion growth rates, remaining lives, and appropriate reinspection intervals. This approach is generally deemed more accurate than using an assumed constant growth rate for all reported features. However, the accuracy of each estimate is subject to the depth sizing error inherent in the measurements. The larger a dataset, the higher the likelihood of many calculated corrosion growth rates being misleading. This is due to there being a higher likelihood of extreme depth sizing errors. A first run undercall followed by a second run overcall compounds the depth sizing error and may result in an overly conservative decision. A first run overcall followed by a second run undercall and may result in a missed defect. Both scenarios can lead to a misuse of resources. This work is a statistical analysis of flaw growth that considers and accommodates for depth sizing errors. The result is that the growth rate for each ILI reported flaw is considered probabilistically. Advanced data ingestion algorithms allow analysts to align multiple ILI tool run data sets quickly and conveniently, opening the door to advanced data analytics, including work with non-linear flaw growth rates. This work looks at corrosion and SCC growth rates (approximated as linear) and fatigue crack growth rates that follow established Paris Law behavior (non-linear).