About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Accelerated Qualification Methods for Nuclear Reactor Structural Materials
|
| Presentation Title |
Machine Learning-based Correlation of Charpy Impact Properties for Sub and Standard sized Specimens |
| Author(s) |
John Merickel, Yugandhar Sreenivasulu, Isshu Lee, Yalei Tang, Rongjie Song, Aleksandar Vakanski, Fei Xu |
| On-Site Speaker (Planned) |
John Merickel |
| Abstract Scope |
Mechanical testing of sub-sized specimens is crucial for the nuclear industry, allowing for maximum utility of the limited space inside reactors for irradiation testing. However, smaller specimens can exhibit different material behavior across sample dimensions; a phenomenon known as the “specimen size effect”. In this study, more than 4,000 Charpy impact data points were collected from over 250 different Charpy curves, covering 54 parameters such as material type, composition, manufacturing, irradiation conditions, specimen dimension, and mechanical properties through a comprehensive literature review. This study utilizes literature data from SA533B and SA508 steel alloys to correlate sub- and standard-sized specimens’ Charpy impact properties using machine learning (ML) based models. This study also determines key factors influencing Charpy impact properties and compares the effectiveness of ML models with existing analytical methods in addressing the specimen size effect. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Mechanical Properties, Nuclear Materials, Machine Learning |