Performance assessment of multi-MFL inspection using feature-based POD

Abstract

Oil and gas pipelines, which transport large quantities of oil products and natural gas, are subject to pipeline failures caused by corrosion. Magnetic flux leakage (MFL) is one of the most popular non-destructive testing (NDT) techniques for the detection of pipeline corrosion. Since individual MFL is insensitive to the corrosion components that are parallel with its magnetic field, two types of MFL tools with perpendicular magnetic fields are usually employed in one inspection to detect all corrosion defects. This study applies probability of detection (POD) to quantitatively assess the detection capabilities of two individual MFL tools and their combination. Due to the characteristics of MFL inspection, this study proposes the construction of the POD model as a function of two geometric features, namely the volume and the orientation, which have a significant influence on the MFL signal response. Detection results from two MFL tools are integrated using logical OR operation to study the POD of their combination. With the proposed POD model, the minimum criteria that ensure a corrosion defect will be reliably detected by MFL tools are studied in this paper. The validity of the proposed POD model is justified on the data collected from an in-service pipeline.