| Author(s) |
Jian Cao, Changjie Sun, Ping Guo, Anindya Bhaduri, Rujing Zha, Wei Chen, David Dunand, Ian McCue, Horacio Espinosa, Malachi Landis, John Reidy, Wenpan Li, Stefan Knapik, Nicholas Berger, Rowan Rolark, Hongshun Chen, Garrett Mathesen, Tiffany Wu, Changjie Sun, Kenhee Ryou, Lele Luan, Kareem Aggour, Scott Oppenheimer, Yigitcan Comlek, Bharath Pidaparthi, Liping Wang, Ahmed Elghandour, Michael Kinstrey, Aaditya Chandrasekhar, Zihan Chen, Guanzhong Hu, Ryan Zhou |
| Abstract Scope |
Metallic materials are critical for enhancing the energy efficiency and performance of engineering systems. To shorten the material development cycle, we developed a high-throughput system for alloy fabrication and testing of temperature-dependent properties, including initial yield stress, elastic modulus, hardening, creep, and oxidation. Specifically, we created a novel parallelized bulge test instrument using an array of mini dimples with varying compositions fabricated via Directed Energy Deposition. Inverse methods for extracting material properties have been developed. A physics-coupled machine learning model, PIRATE (Physics-Informed Research Assistant for Theory Extraction), predicts composition-property relationships in the compositional space between experimentally sampled points. Furthermore, we developed a differentiable finite element solver, JAX-FEM, and integrated it into a gradient-based optimization framework, which optimizes local material composition while considering temperature-dependent mechanical properties and creep deflection. This integrated framework of material design, fabrication, evaluation and part optimization lays the foundation for component-material co-design and accelerate the design-to-deployment cycle. |