Beschreibung
The rising use of recycled aluminum content in alloy production leads to an accumulation of certain tramp elements like Fe, Si, Cu, Mn and thus to an increasing content of brittle intermetallic phases (IMPs). The Fe-rich IMPs can exert either beneficial or detrimental effects depending on their size, type and morphology and the intended application. For wrought alloys, fine and homogenously distributed Fe-rich IMPs show favourable effects on the mechanical properties, due to an enhanced strain-hardening potential and overall smaller grain size. The phase fragmentation during the rolling process is strongly influenced by the IMPs morphology in the as-cast state. Various morphologies have been reported in literature for the Fe-rich IMPs. The β AlFeSi phase often form thin plates, whereas the α Al(Fe,Mn)Si phases often exhibits blocky structure or complex branched morphologies, also known as “Chinese script”. Detecting the latter, often-disconnected structures in the 2D cross section as entity is not feasible with conventional segmentation models leading to a misleading impression of their actual size and fragmentation behaviour. Thus, a hybrid segmentation approach is established, in which a specially trained AI model is applied in a first step to segment Chinese script and blocky phases, followed by conventional thresholding to identify fine platelets. This allows for a high-throughput segmentation of BSE-SEM mappings without the need for run-time-intensive measurements or time-consuming and complex sample preparation.| Zeitraum | 14 Apr. 2026 → 16 Apr. 2026 |
|---|---|
| Ereignistitel | 5th. Material Sciene Colloquium |
| Veranstaltungstyp | Konferenz |
| Ort | Lech am Arlberg, ÖsterreichAuf Karte anzeigen |
| Bekanntheitsgrad | International |