Informing Data-Driven Recyclable Aluminum Design with Phase Constraints
The design of new aluminum alloys for recycling constitutes a large, multi-dimensional composition space. Within a Bayesian optimization blending model, phase or composition constraints are necessary to narrow the design space within and focus on regions that may contain viable alloys. This thesis f...
| Auteur principal: | |
|---|---|
| Autres auteurs: | |
| Format: | Thèse |
| Publié: |
Massachusetts Institute of Technology
2024
|
| Accès en ligne: | https://hdl.handle.net/1721.1/153790 https://orcid.org/0000-0001-8244-7575 |
| Résumé: | The design of new aluminum alloys for recycling constitutes a large, multi-dimensional composition space. Within a Bayesian optimization blending model, phase or composition constraints are necessary to narrow the design space within and focus on regions that may contain viable alloys. This thesis focuses on the 6xxx-series system through a broad dataset comprising all known Al-Mg-Si composition and thermodynamic calculations of individual compositions. Explicit matrix phase constraints are not superior to basic Al composition constraints within lean alloys. To suppress detrimental 𝛽-AlFeSi constituent phases and promote strengthening 𝑀𝑔₂𝑆𝑖 intermetallic phases, ratios of Fe/Mn=1 and Mg/Si=1.73 are sufficient. Equilibrium and Scheil calculations of commercial compositions show non-linear regimes where the Fe/Mn=1 rule may be too conservative and point towards ways to expand the amount of Fe that recycled aluminum can accommodate. |
|---|