Why AI-Driven Analytics Are Essential for Next-Generation Pipeline Condition Assessments

Condition assessments of metallic pipelines deploying “next-generation” free-swimming ultrasonic inspection platforms provide high-resolution, multi-layered information on metal wall corrosion, out-of-roundness, and lining condition. While these data layers inform efficient pipeline management decisions, they also generate immense amounts of data to be analysed and interpreted. With the rapid advancements in computing hardware, artificial intelligence (AI) in the form of machine learning has been widely applied in data analytics and image recognition problems in recent decades. We show the promising results of using support vector machine (SVM), and then convolutional neural network (CNN) on the problem of analyzing pipeline condition assessment data for wall-loss defects. Additionally, we discuss the challenges that pipes of various materials and manufacture dates pose to AI-driven analytics, and how these challenges are being overcome. Finally, we will look to the future and explore AI analytics’ potential to monitor change in pipeline conditions over time. This will power predictive analytics allowing pipeline owners to adjust inspection intervals, or in the case of some larger utilities, enable the ability to predict where they need to perform in-field work to optimize capital expenditures and resource planning.