In-line inspection (ILI) systems continue to improve in the detection and characterization of cracks in pipelines, and are relied on substantially by pipeline operators to support Integrity Management Programs for continual assessment of conditions on operating pipelines that are susceptible to cracking as an integrity threat. Recent experience for some forms of cracking have shown that integration of data from multiple ILI systems can improve detection and characterization (depth sizing, crack orientation, and crack feature profile) performance. This paper will describe the approach taken by a liquids pipeline operator to integrate data from multiple ILI systems, namely Ultrasonic axial (UC) and circumferential (UCc) crack detection and Magnetic Flux Leakage (MFL) technologies, to improve detection and characterization of cracks and crack fields on a 42 miles long, 12-inch OD liquid pipeline with a 38-year operating history. ILI data has indicated a large number of crack features, including 4000+ crack features reported by UC, 1000+ crack features by UCc, and 2500+ metal loss features reported by MFL. Initial excavations demonstrated a unique pattern of blended circumferential-, oblique- and axial-orientated cracks along the entire extent of the 42-mile pipeline, requiring advanced methods of data integration and analysis. Applying individual technologies and their analysis approaches showed limitations in performance for identification and characterization of these blended features. The outcome of the study was the development of a feature classification approach to classify the cracks with respect to their orientation, and rank them based on the depth sizing by using multiple datasets.
Several sections of the 42-mile pipeline were cut-out and subjected to detailed examination using multiple non-destructive examination (NDE) methods and destructive testing to confirm the crack depths and profiles. These data were used as the basis for confirming the ILI tool performance and providing confirmation on the improvements made to crack detection and sizing through the data integration process.