AI helped find optimal parameters for 3D printing NASA alloy for rocket engines

Edited by: Svitlana Velhush

Researchers at Washington State University used artificial intelligence to find a way to print NASA's high-performance alloy GRCop-42 on a 3D printer more cheaply and easily. This material made of copper, chromium, and niobium is designed for extreme conditions: it conducts heat excellently and retains strength at high temperatures, which is why it is used in combustion chambers of liquid rocket engines.

Printing GRCop-42 usually requires a powerful laser and is expensive. Most commercial printers cannot handle it. Manually going through all possible settings — more than 100 million combinations — is practically impossible: each experiment consumes material, time, and money.

The team started with data on 37 failed printing attempts. Based on this, the AI developed a model that estimated the probability of success for untested settings. The system proposed small groups of new configurations, balancing between promising options and those that helped better understand the search space.

In three months and with only 40 experiments, they managed to find six working settings. One of them allowed, for the first time, printing the alloy at a record low laser power — 500 watts. This opens the way to using more affordable equipment.

"Ninety percent of commercial printers cannot print this alloy," notes Professor Jana Doppa, who led the study. "The found parameters allow working on ordinary machines and essentially democratize printing of this material."

The first author of the paper, graduate student Azza Fadhel, adds that previously some configurations simply melted the workpiece. Now AI helps avoid such costly mistakes.

The success of the project was published in the Proceedings of the AAAI Conference on Artificial Intelligence. The method can be adapted for other alloys and tasks where success is rare and the number of options is huge. But how exactly AI coped with such a huge space of options showed that the approach works even with real costs for materials and equipment.

In the future, this will reduce energy consumption, decrease printer wear, and make the alloy accessible to universities, small laboratories, and companies without specialized high-power equipment.

24 Views

Sources

  • AI searched 100 million possibilities and found a cheaper way to 3D-print a NASA rocket alloy

Read more articles on this topic:

US PCE Holds at 3.7%, Above Forecast and Slightly Lifts Fed Rate-Hike Expectations 📊 July PCE inflation rose 3.7% year over year, above the 3.6% forecast, while the monthly increase came in at 0.2% versus expectations of 0.1%. 🔍 Core PCE rose 3.3% YoY and 0.2% MoM, both in

Image
Reply
Did you find an error or inaccuracy?We will consider your comments as soon as possible.