This paper is based on some of the findings resulting from the PRCI project NDE-4E which examined the performance of various crack ILI technologies based on an industry data mining exercise. One of the project deliverables was an extensive database of crack in-line inspection (ILI) and excavation information collected from operators. This was used to characterize the field performance of ILI technologies with respect to detection, identification and sizing of crack features.
Thousands of feature measurements were collected from several operating companies. The data were validated to ensure completeness, consistency and accuracy, and stored in a database. The main results to date are as follows:
• Most profiles have shallow areas along the feature length — they are not semi-elliptical or parabolic in shape as usually assumed in burst pressure computations. Shallow regions increase the defect length, but do not increase the defect cross-section proportionally.
• If it is assumed that the ILI tool requires a minimum depth before a portion of a feature is reported, it follows that the length reported by the ILI tool would differ from that observed in the field. This is due to the fact that the ILI tool would be reporting the length of the crack at some threshold depth and the field would be reporting the feature length at the surface.
• Using depth profiles, an effective detection depth was calculated, that is the average depth at which the ILI length aligns with the profile. This was determined to be an average value of 0.88 mm for cracks and 0.83 mm for crack-fields which exceeds published ILI tool specification.
• Shallow parts of a crack profile have significant impact on the field-measured length, but do not significantly impact the burst pressure.
• For 83% of the data, the remaining strength factor (RSF) calculated using ILI data is within 10% of the RSF value calculated using field profiles.
• Feature depth recorded in depth categories by the ILI tools show sensitivity to their proximity to long seam welds for crack-like features.
• Overall, Burst pressure is more relevant to pipeline integrity than either length or depth individually and burst pressure calculated from ILI measurements is conservatively biased compared to burst pressure calculated from profile data.