| About this Abstract |
| Meeting |
2020 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing: Materials Design and Alloy Development II
|
| Presentation Title |
A-57: Design of Easy-to-Use Structural Alloy Feedstocks for Additive Manufacturing Using Machine Learning Methods |
| Author(s) |
Akansha Singh, Ben Rafferty, Jeremy J. Iten, Jacob Nuechterlein, Branden B. Kappes, Sridhar Seetharaman, Aaron Stebner |
| On-Site Speaker (Planned) |
Akansha Singh |
| Abstract Scope |
Metal additive manufacturing is a multi-process and machine dependent complex procedure and requires a specialist to manufacture a part. We are trying to design an alloy feedstock which should be easy to print and consistent regardless of the machine, and also whose manufacturing requires minimal trained members and uses a minimum set of processing tools. For this, we identified several descriptors such as mushy zone width, thermal expansion, miscibility gap, etc., to define printability. We performed several CALPHAD based computer simulations on iron-based alloys and gathered a large dataset for alloy compositions, and their thermodynamic properties. In order to relate alloy compositions, stable phases and their properties to the descriptors of printability, we employed machine learning methods and estimated the associated error in the predictions. Using our developed model, we would be able to design the alloy feedstock with highly printable, functional, and reproducible printed properties. |
| Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |