Statistical Analysis of Dig Operations Leading to Productive Repairs

Abstract

Inline inspection data from several runs spanning many years is available for individual pipeline segments, but compilation of this data into a comprehensive picture of pipeline integrity necessarily relies on computational tools. A critical advantage of modern data storage, analysis, and visualization techniques is the relative ease of performing statistical assessments of integrity operations. Data from a single user of OneBridge Solution’s software may comprise over 1,000 in-line inspections (ILI) runs, hundreds of pipe segments, several million aligned anomalies, and thousands of repair records. Automated alignment of ILI data allows a single physical anomaly to be reliably tracked through many years of growth and repeated measurement and then correlated to repair records. We present a study of cases where ILI anomaly measurements warranted a dig operation in which repair actions were either performed or found to be unnecessary. The fraction of dig operations leading to a productive repair varies with the condition triggering the dig and discretionary choices about dig condition parameters. The analysis is done through the exploration of different methods of corrosion growth forecasting in use by operators and how they compare. The measures that have been taken into consideration for the purpose of this study include half-life vs. pit-to-pit where the effectiveness of identifying and mitigating fast-growing anomalies is compared across models. Further exploration of how forecasting and building a dig program based on pit-to-pit alignments and a comprehensive growth model through the advances in data science and machine learning can bring efficiency improvements and an overall reduction in risk. We analyze the relationship between these parameters, ILI measurements, repair-to-dig ratios and the impact on operational spend. We examine whether a reduction in overall inspection frequency and expenses is possible through advanced growth modeling. Ultimately this would provide a more accurate view of long-term operating costs and allow for operators to consider scenarios relating to repair and replacement of assets.