Det Norske Veritas (U.S.A.), Inc. (DNV GL) prepared this paper in order to study the repeatability of inspection results between subsequent in-line inspections. DNV GL has access to a significant amount of data that spans many different pipeline operators, ILI vendors, inspection years, and inspection technologies. DNV GL is well suited to complete this study as a result of our access to these various data sets.
Over 55,000 one-to-one metal loss defect comparisons were assembled from ILI-to-ILI analyses. Reported metal loss defect depths, lengths, and widths spanning from 2003 through 2015 from 13 pipeline operators and 36 pipeline segments were compiled to meet the objectives of this paper. Inspection technologies include axial magnetic flux leakage (MFL), ultrasonic wall thickness (UTWT), spiral MFL, and circumferential MFL ILI.
From analyses of these data, the following conclusions were generated:
• Effect of ILI vendor: ILI repeatability is generally improved when the same ILI vendor is used (when compared to using two different ILI vendors in subsequent inspections), but this is not always true.
• Reported metal loss depths: ILI repeatability decreases with increasing metal loss depth.
• Pipe geometry and type: ILI repeatability is better in larger diameter pipelines and with increasing wall thickness.
• POF classification: ILI repeatability is better for pitting, general corrosion, and axial grooving defects as compared to the other POF classifications.
Based on these insights, the authors make the following recommendations:
• Pipeline operators should consider using the same ILI vendor and tool if the goal is to identify change and/or corrosion growth in the pipeline segment. A raw signal review is encouraged in order to verify the presence, or lack thereof, changes in metal loss morphologies. The raw data review is especially important when comparing inspections from two different ILI vendors.
• If the goal is to identify corrosion growth, and a pipeline operator uses different ILI vendors, it is recommended that a statistical review of one-to-one matched metal loss features take place to identify candidate locations that are more likely to be growing. The candidate locations should have a raw signal review in order to verify whether or not growth is taking place.