Filtern
Dokumenttyp
- Zeitschriftenartikel (1)
- Vortrag (1)
Sprache
- Englisch (2) (entfernen)
Referierte Publikation
- nein (2) (entfernen)
Schlagworte
- Synchrotron X-ray tomography (2) (entfernen)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (1)
The 3D microstructure of an Al alloy matrix composite with two ceramic reinforcements was investigated by synchrotron X-ray tomography. A deep learning algorithm was used for the segmentation of four different phases. We show that convolutional networks with the U-Net architecture are able to solve complex segmentation tasks with small amount of training data.
The 3-dimensional spatial distribution of liquid water in different gas diffusion layer (GDL) materials was analyzed using synchrotron X-ray tomography. The capability of the method was demonstrated by virtually separating the GDL components in order to facilitate individual analysis of fiber material, liquid water and gas filled pore spaces. The influence of hydrophobic surface treatment on the water distribution in the GDL was illustrated by analyzing three GDL materials with different degrees of hydrophobicity. In the least hydrophobic sample, liquid water tends to form larger clusters which stretch out about several hundred µm inside the porous GDL. In contrast, only small water clusters were found in the strongly hydrophobic material with high Polytetrafluoroethylene (PTFE)-content as the liquid is partially pressed out of the GDL. Additionally, the influence of fiber orientation on the water distribution in the felt material was demonstrated.