About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Verification, Calibration, and Validation Approaches for Mechanical Modeling of Metals
|
| Presentation Title |
Quantifying the Effects of Missing Grains and Stitching in ff-HEDM Measurements Using Crystal Plasticity and Machine Learning |
| Author(s) |
Huy Tran, K.M.Zaheen Nasir, Seunghee Oh, Xuan Zhang, Mark Messner, Tianchen (Gary) Hu, Jun-Sang Park |
| On-Site Speaker (Planned) |
Huy Tran |
| Abstract Scope |
Far-field high-energy diffraction microscopy measurements at the Advanced Photon Source provide grain-resolved datasets for mechanical model validation, but limited beam size, incomplete grain reconstruction, and stitching across small scanning windows can introduce uncertainty in the measured microstructure and response. In this work, we quantify the effects of missing grains and stitching on the macroscopic stress-strain response, local strain and stress fields, and rare-event grain identification. A framework combining crystal plasticity modeling, a novel stitching algorithm, and a machine-learning rare-event indicator based on graph segmentation using Nye tensor features is developed. Synthetic microstructures with varied grain sizes, shapes, and spatial distributions are generated, and virtual sampling is used to mimic ff-HEDM measurements. The resulting effects on reconstructed response and rare-event localization are assessed. In parallel, dedicated ff-HEDM experiments with varying scanning windows are performed, and comparisons of the experimental trends and their implications for model verification and validation will be discussed. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Characterization, Modeling and Simulation, Mechanical Properties |