Machine Learning Accelerated Physics-Based Modeling for Defect Mitigation in Metal Additive Manufacturing

We propose the “Machine Learning Accelerated Physics-Based Modeling for Defect Mitigation in Metal Additive Manufacturing (AM)” project to mitigate AM defects by developing a low-cost, interpretable surrogate model for defect formation during the AM process. The breakthrough over the traditional experimental survey approach of the AM design space is reduced time and costs to determine build parameters that will produce defect free components. After completing this project, the Agency and LaRC AM-focused projects will gain the capability to build segregation defect-free parts for high temperature aerospace usage. We anticipate that sensitivity in the computational fluid dynamic (CFD) models might delay our project and will mitigate this risk by using the surrogate modeling to do additional calibration and supplement the CFD simulations. Once successful, the outcomes of this work will be infused to our current project, our NASA partners, and interested aerospace original equipment manufacturers.

The target state by the end of the first IRAD period of performance is having the software fully procured, installed, and tested and a first version of the surrogate model developed. Once working, the software will be used to do initial parameter studies to compare to experimental observations from GRC. The work is expected to be ready for project incorporation after the 2nd year of support, at which point it will be transitioned into the TTT project both LaRC PIs are funded on. After transitioning to other funding sources, the framework developed in this project will be extended to predict other defects, including lack-of-fusion and keyhole porosity, that are critical for qualifying and certifying AM parts. By not funding this project, the proposed capabilities for preventing defects will not be developed, which is important since defects are a critical aspect preventing wide aerospace usage of AM parts. Not funding this would also reflect a missed opportunity for developing LaRC-GRC collaboration, cross-center networking, and future cross-center projects and collaborations. This work developing high-fidelity models combined with machine learning to predict and mitigate defects will enable AM to realize long-sought goals of increased efficiency and sustainability relative to conventional processes.

This is early/mid-stage (TRL 3) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.