About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Artificial Intelligence Applications in Integrated Computational Materials Engineering (AI-ICME)
|
| Presentation Title |
Harnessing AI for Prediction of Abnormal Grain Growth Using 3D Experimental Data |
| Author(s) |
Michael R. Tonks, Zhihui Tian, Woohyun Eum, Vishal Yadav, Amanda Krause, Joel Harley |
| On-Site Speaker (Planned) |
Michael R. Tonks |
| Abstract Scope |
Traditional grain growth simulation methods fail to fully explain the 4D grain growth phenomena revealed by non-destructive 3D microscopy. Our physics-regularized interpretable machine learning microstructure evolution (PRIMME) framework offers a novel solution by seamlessly integrating physical principles with AI-driven prediction capabilities. This presentation details how we have adapted PRIMME to accommodate 3D experimental datasets from X-ray diffraction microscopy (3D-XRM). We show how both misorientation and inclination dependent anisotropy can be learned by PRIMME. We also illustrate the usage of a recursive neural network to reduce the amount of training data from five sequential microstructures to just two. Finally, we show changes to PRIMME to reduce the memory usage for 3D simulations. This work illustrates the potential of AI and ICME integration for advancing quantitative understanding of microstructural evolution, with direct applications to process optimization and materials design for controlled grain structures. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
ICME, Computational Materials Science & Engineering, Machine Learning |