About this Abstract |
Meeting |
MS&T23: Materials Science & Technology
|
Symposium
|
Advances in Ferrous Metallurgy
|
Presentation Title |
Segmentation of Microscopy Images of Lower Bainite (LB) and Tempered Martensite (TM) High Strength Steels |
Author(s) |
Xiaohan Bie, Manoj Arthanari, Evelin Barbosa de Melo, Jun Song, Steve Yue |
On-Site Speaker (Planned) |
Xiaohan Bie |
Abstract Scope |
High strength steels (HSS) are crucial structural metals with superb strength; however, they are very susceptible to hydrogen embrittlement (HE), a phenomenon that H ingress in steel makes it go through ductile to brittle transition. Two common microstructure types within HSS are LB and TM. Experiments found that LB is more resistant to HE than TM, attributed to carbide precipitates’ different distribution and morphology. However, characterizations of microstructures in LB and TM are mainly qualitative. Employing semantic segmentations in deep learning, we quantitatively analyzed different metrics of carbide precipitates in LB and TM HSS of similar strength. The carbide volume percentages in LB and TM HSS are found rather similar. However, carbides are more evenly distributed in TM, while marginally better aligned in LB. The deep learning model achieved 97.38% accuracy, providing a time-efficient workflow for analyzing carbides in HSS, and a potential route towards quantitative microstructural analysis in metals. |