72 Application of Advanced Data Analytics

Abstract

In recent years there has been greater and greater desire to reduce pipeline maintenance costs through improvements to the effectiveness and efficiency of integrity management programs. Extracting as much information out of data obtained through each In-line Inspection (ILI) to better understand sizing tolerance performance and accuracy is an area of key interest as it has direct impact on repair decisions.  Today, higher data volumes captured through multiple sensing techniques are being collected by ILI tools than ever before and the interpreted results are established as one of the most effective assessment techniques for managing pipeline safety.

In addition to this improved condition assessment data, validation programs are generating exponentially increasing volumes of now highly reliable and accurate truth data.  This essential combination of high-quality inspection and validation data provides the opportunity to re-think how we establish performance specifications for the different ILI technologies. Traditional methods of utilizing relatively low volumes of isolated artificial defects that attempt to represent defects expected to be found can be replaced with vast ranges of actual real-world defects.  Extensive “Big Data” libraries of high-resolution field measurements married to raw signal data captured by the tools provides new and exciting opportunities for innovative improvements that can be made to the detection, characterization and sizing of pipeline anomalies.

This paper will present a machine learned technique applied to vast quantities of dig and tool data to improve metal loss (corrosion) sizing tolerance performance.  This method goes beyond the common industry practice to divide metal loss into a small number of categories based on arbitrary discrete defect length and width to include a much wider range of factors that truly affect Magnetic Flux Leakage (MFL) metal loss sizing to predict an individual defect-by-defect tolerance.  This provides a more precise prediction of metal loss tolerances that reduces the general conservatism that has been built into existing tolerance specifications over the years while maintaining the necessary conservatism when needed to ensure pipeline safety. This paper will also present examples of the specific benefits operators may realize utilizing these newly predicted tolerance specifications.

Read/Download PaperGradeAvailable after the event