Image Courtesy: Shutterstock
Researchers at Washington State University have used artificial intelligence to identify a faster and potentially cheaper way to 3D print a high-performance NASA-developed metal alloy. The system found successful printing configurations from a search space containing more than 100 million possibilities while requiring just 40 physical experiments.
The breakthrough could make GRCop-42, a copper, chromium, and niobium alloy designed for extreme aerospace environments, compatible with more widely available commercial 3D printers. The AI-guided approach may also offer a new way to tackle scientific problems where testing every possible option would be prohibitively expensive. The findings were reported by Washington State University researchers.
GRCop-42 was developed by NASA for applications requiring both high thermal conductivity and the ability to withstand extreme temperatures. The alloy is used in aerospace systems, including liquid rocket engine combustion chambers, but its demanding manufacturing requirements have limited its accessibility.
Traditional approaches require researchers to manually test different combinations of printing parameters, including laser power and other machine settings. With more than 100 million potential configurations, a comprehensive trial-and-error approach would be impractical.
The WSU team instead used data from 37 previously unsuccessful printing attempts to train an AI model. The system estimated which untested configurations were most likely to succeed and recommended small groups of experiments that could either produce a viable result or provide valuable information to improve future predictions.
“We were essentially applying the AI so that we efficiently choose candidates from this very large search space,” said Azza Fadhel, a PhD student and the study’s first author.
After three months of testing, the researchers identified six successful configurations while conducting only 40 experiments in total. Most notably, they successfully printed GRCop-42 using a 500-watt laser, a significantly lower power level than previously required for the challenging material.
“Ninety percent of commercial printers cannot print this metal alloy,” said Jana Doppa, who led the research. Finding workable parameters, he said, could help “democratize” access to the material by allowing it to be printed on commercial equipment.
The researchers believe their approach could eventually be applied to other metal alloys and scientific fields, including drug discovery, where researchers face enormous experimental search spaces and high costs.
