| Abstract Scope |
This talk will summarize the MIT/CMU/Lehigh University team's effort under the DARPA METALS program. This project is focused on optimal design of functionally graded turbine rotors for reusable oxygen-rich turbopumps, where there are risks of thermal fatigue, burst, and metal fires. I will describe new test methods for high-throughput characterization of metal flammability and thermal fatigue resistance. I will also describe a generative AI design framework which uses these experimental results as inputs to determine optimal composition gradients that maximize the competing objectives of fatigue life, burst speed, and oxygen compatibility. The outputs from the genAI system highlight the opportunities in using such frameworks to determine high-performance designs in a non-linear performance landscape where conventional optimization methodologies struggle. |