May 28.

Webinar: Machine Learning for Health Monitoring of Uninspected Pipelines

This webinar is in collaboration with NACE Malaysia:

For decades, the pipeline industry has been collecting in-line inspection (ILI) data for pipelines all around the world, in addition to an abundance of historical data on design, construction, operations and environment. The availability of these datasets has led the industry naturally towards machine learning as a supporting technique for various integrity management activities. By observing trends from the past, we can better understand our current and future assets.

One particularly promising application of machine learning is condition prediction in pipelines that cannot be inspected using ILI. This is true for approximately 40% of the world’s pipelines.

Previous work by the authors has described how the current condition of an uninspected pipeline may be estimated using machine learning, with a specific focus on the threats of internal and external corrosion. Using machine-learning models trained on thousands of pipelines from around the world, the condition of an uninspected pipeline can be predicted with an accuracy that was previously unattainable with traditional techniques. Further work has focused on the prediction of future condition. By estimating deterioration rates for an uninspected pipeline, we can simulate its future state and begin to make more proactive decisions on repair, mitigation and future monitoring.

Herein, the authors build upon all previous work by proposing a complete health-monitoring framework for uninspected pipelines, again with corroded pipelines as the primary focus. The framework is exemplified and validated with case studies on real pipelines, thereby demonstrating the power of machine learning for integrity management decision support. It is furthermore proposed how the techniques could be extended to other threats, such as cracking and geometric defects.

Learn more and join this webinar, which will be presented on May 28, 2021 at 3 pm (MYT).

Date
May 28, 20212021-05-28T08:00:00
Location
online
URL
to the event website