About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Material Responses Investigated Through Novel In-Situ Experiments and Modeling
|
| Presentation Title |
Grain-Scale Plastic Deformation Transmission Prediction in Ti-7Al During Creep Using High-Energy Diffraction Microscopy and Graph Neural Networks |
| Author(s) |
Yuefeng Jin, Xiongye Xiao, Peiyupei Zhang, Wenxi Li, Amlan Das, Katherine Shanks, Paul Bogdan, Ashley Bucsek |
| On-Site Speaker (Planned) |
Yuefeng Jin |
| Abstract Scope |
Plasticity is often modeled as continuous in space and time; however, experimental observations reveal that it can occur as localized grain-scale bursts that propagate through the grain network. Such events can precede macroscopic yield and are critical for understanding room-temperature creep and dwell fatigue in α-phase titanium-aluminum alloys. In this study, we use far-field high-energy diffraction microscopy (ff-HEDM) to characterize grain-scale plastic deformation in a Ti-7Al alloy during creep loading. To predict plastic deformation transmission paths, we use a graph neural network (GNN), where grains are nodes and grain contacts form edges labeled as transmission or no-transmission. Although individual material properties show expected trends, no single property alone provides reliable prediction. GNNs integrate neighboring grain features and demonstrate robust predictive capability. Topological descriptors further enhance prediction, and subgraph-based modeling highlights the importance of local structure. Feature importance analysis identifies which material and topological attributes most strongly influence plastic deformation transmission. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Titanium, Characterization, Machine Learning |