About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Advanced Materials for Reusable Rocket Engines
|
| Presentation Title |
Property-driven design of compositionally graded alloys using a generative AI framework |
| Author(s) |
Jixuan Dong, Bo Ni, Hasan Al Jame, S. Mohadeseh Taheri-Mousavi |
| On-Site Speaker (Planned) |
Jixuan Dong |
| Abstract Scope |
Additive manufacturing (AM) enables spatially tailoring the composition of alloys thus not trading exceptional properties for global combined property optimization as in monolithic components. However, the design of gradient systems remains costly and time-consuming partially due to the need for comprehensive property data across composition transitions. This study proposes a generative AI-based framework that lays the foundation for efficient gradient property predictions. CALPHAD-based integrated computational materials engineering (ICME) simulations were employed to compute thermophysical and mechanical properties for monolithic alloys and binary gradients linking Inconel 718, Monel K-500, and Incoloy 909 for reusable propulsion systems application. Using GPT generative AI model, we evaluated the potential for developing foundation models for forward prediction and inverse design of composition/processing-property by combining the monolithic alloy data. We will discuss the challenges and opportunities in using this framework for efficient design of gradient AM alloys. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, ICME, Other |