About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
High Performance Steels
|
| Presentation Title |
Machine Learning Guided Insights Into Hydrogen Embrittlement of Pipeline Steels |
| Author(s) |
Zach Bare, Lucas Teeter, Shree Ram Acharya, Kyle Rozman, Michael Gao, Richard Oleksak |
| On-Site Speaker (Planned) |
Zach Bare |
| Abstract Scope |
Hydrogen is an important fuel in the future energy landscape. Steel pipelines are among the most cost-effective options for transporting hydrogen; however, safety and reliability are concerns due to the risk of hydrogen embrittlement (HE). An improved understanding of both the experimental methods of testing for HE, and the steel properties affecting HE, will help assess hydrogen compatibility of existing pipelines and guide development of new HE-resistant steels. Herein, a comprehensive database of tensile testing of API 5L grade pipeline steels in hydrogen was compiled. Machine learning models were developed to predict HE response and to assess the importance of both the testing methodology and steel properties impacting it. Thermodynamic descriptors derived from CALPHAD simulations were further incorporated to enhance model performance and provide additional insight into important steel properties. The resulting models demonstrate strong predictive performance and offer valuable insights into the factors affecting HE in low-alloy steels. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Iron and Steel, Machine Learning, |