Understanding the Impact of Correlated Corrosion Inline Inspection Uncertainties on Reliability

Abstract In an industry where evolving technology and data-driven decision-making are critical in effectively managing the integrity of pipeline assets, this paper addresses the need for validation of measurement uncertainties of pipeline inspection technologies. The study delves into the less explored domain of error correlation in Magnetic Flux Leakage Axial (MFL-A) Inline Inspection (ILI) tools. It identifies the extent at which various sources of random and systematic errors are correlated at different distance spans. Factors such as tool speed, sensor lift off, debris, material variations, and magnetic field inconsistencies influence measurement random error while the feature morphology and orientation can influence systematic errors in the sizing algorithms. The study utilizes Non-Destructive Examination (NDE) ultrasonic corrosion defect data to empirically conduct an in-depth analysis at various spatial levels — cluster, pipe joint, and pipeline — to understand the extent of correlation on measurement uncertainties. This includes a statistical assessment of over sizing and under sizing correlations in depth measurements. Our study is applicable to advanced probabilistic reliability models, such as performing reliability assessment with the Effective Area model, where the co-relations of uncertainties within a cluster could impact the results. The results of this study will help engineers better understand what level of correlations to assign for probabilistic burst pressure assessments. Our study seeks to determine whether depth measurement errors are consistently present along the entire pipeline, within short spans, or in specific clusters, thereby influencing the correlation of uncertainties and affecting the overall reliability model. If strong correlations are found across small inspection spans but are not consistently represented across the entire inspection, then API 1163 level 2 and level 3 validations require field data to be sufficiently spread across the inspection to be representative of the entire inspection. This finding would highlight the value of using field data spread across the pipeline, and that an operator’s confidence on tool performance can be impacted by both sample size and field measurement span. This study aims to help risk-informed decision making with a more accurate representation of depth measurement uncertainty.