About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Novel Strategies for Rapid Acquisition and Processing of Large Datasets From Advanced Characterization Techniques
|
| Presentation Title |
Segmentation Methods for Tracking Dislocation Dynamics |
| Author(s) |
Johann Haack, Adam Cretton, Antonella Gayoso Padula, Axel Henningsson, Grethe Winther, Henning Friis Poulsen |
| On-Site Speaker (Planned) |
Johann Haack |
| Abstract Scope |
Recent advances in Dark-Field X-ray Microscopy (DFXM), particularly the new beamline at ID03, have significantly increased data acquisition rates and enabled in situ studies of dislocation patterning under realistic deformation conditions, placing new emphasis on efficient and robust data analysis. DFXM is a synchrotron-based technique that provides non-destructive, 3D-resolved maps of local orientation, elastic strain, and dislocation density within deeply embedded volumes of crystalline materials. A key challenge in analysing these datasets lies in segmenting domains, whose boundary misorientations are often well below 0.1°, rendering traditional threshold-based methods ineffective. To overcome this issue, we use an iterative region-growing approach, clustering the centroids of randomly seeded cells to generate a probability map for each voxel in the 2D and 3D data. We demonstrate its application to plastically deformed aluminium single crystals and highlight its utility for tracking domain evolution across load steps, enabling dynamic studies of dislocation boundaries in bulk samples. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Modeling and Simulation, Characterization, |