The ability to characterize metal loss and gouging associated with dents and the identification of corrosion type near the longitudinal seam are two of the remaining obstacles with in-line inspection (ILI) integrity assessment of metal loss defects. The difficulty with denting is that secondary features of corrosion and gouging present very different safety and serviceability scenarios; corrosion in a dent is often not very severe while metal loss caused by gouging can be quite severe. Selective seam weld corrosion (SSWC) along older low frequency electric resistance welding (ERW) seams also presents two different integrity scenarios; the ILI tool must differentiate the more serious SSWC condition from the less severe conventional corrosion which just happens to be near a low frequency ERW seam. Both of these cases involve identification difficulties that require improved classification of the anomalies by ILI to enhance pipeline safety.
In this paper, two new classifiers are presented for magnetic flux leakage (MFL) tools since this rugged technology is commonly used by pipeline operators for integrity assessments. The new classifier that distinguishes dents with gouges from dents with corrosion or smooth dents uses a high and low magnetization level approach combined with a new method for analyzing the signals. In this classifier, detection of any gouge signal is paramount; the conservatism of the classifier ensures reliable identification of gouges can be achieved. In addition to the high and low field data, the classifier uses the number of distinct metal loss signatures at the dent, the estimated maximum metal loss depth, and the location of metal loss signatures relative to dent profile (e.g. Apex, Shoulder).
The new classifier that distinguishes SSWC from corrosion near the longitudinal weld uses two orientations of the magnetic field, the traditional axial field and a helical magnetic field. In this classifier, detection of any long narrow metal loss is paramount; the conservatism of the classifier ensures that high identification of SSWC can be achieved. The relative amplitude of the corrosion signal for the two magnetization directions is an important characteristic, along with length and width measures of the corrosion features.
These models were developed using ILI data from pipeline anomalies identified during actual inspections. Inspection measurements from excavations as well as pipe removed from service for lab analysis and pressure testing were used to confirm the results.