There are many threats that can impair pipeline integrity and can lead to significant damage and subsequent failure if not detected. Current in-line inspection (ILI) technology has been developing in recent years and now it is possible to reliably detect and size anomalies using multiple sensors technologies or multiple tools. However, it is not always easy to characterise anomalies and discriminate those that are defects of significant interest from those that are relatively benign.

In addition, with the innovations in ILI technology mean there are significantly more data streams to review. Reviewing ILI data often involves expending significant effort over long periods of time while coming up with the same result—in other words, poring over the data until one is “blue in the face” yet still not being able to discriminate pipeline defects accurately. What becomes important, then, is the “art of looking”, the ability to see beyond the easily observable, to bring things out of the side lines and into sharper focus.

What the industry requires, then, is a different way of “looking.” Algorithmic processes can be defined to look for well-defined objects such as welds in ILI data; however, it is often necessary to use a heuristic or not fully algorithmic process to look for less well-defined anomalies such as dents with gouges. 

  Multiple ILI datasets can be leveraged using innovative techniques, resulting in a process to discriminate critical damage features from other anomaly types. Once discrimination is complete, anomalies can then be assigned to “buckets” based upon characterised features and prioritised for further investigation based upon three criteria: their effect on integrity, their environment and the impact of further investigation. This paper will discuss obstacles related to analysing some of the more complex defects such as pin-hole metal loss, crack-like features and also mechanical damage