AI Just Made It Possible to 3D Print a NASA Rocket Alloy on Standard Machines

A Rare Alloy, Now Within Reach

GRCop-42 is one of those materials that sounds like it belongs in a rocket engine, because it does. Developed by NASA, the copper-chromium-niobium alloy stays strong at extreme temperatures and carries heat fast. That makes it ideal for liquid rocket combustion chambers and other aerospace parts. The catch is that it is a pain to 3D print. Until now, it has needed specialized high-power laser equipment that most universities, small labs, and mid-sized shops do not own.

A team at Washington State University thinks that is about to change. Led by computer science professor Jana Doppa, the researchers used an AI-driven experimental design process to find settings that let GRCop-42 print on lower-power commercial machines. The work was published in the Proceedings of the AAAI Conference on Artificial Intelligence and won the conference's Innovative Deployed Application Award.

The problem is scale. The team estimated there were more than 100 million possible combinations of laser power, scan speed, hatch spacing, and other parameters. Testing them all would cost too much and take too long. A single failed print can cost hundreds of dollars, and post-print analysis can stretch across days. The WSU group started with data from 37 previous unsuccessful attempts, then built a model that predicted which untested settings were most likely to work.

How the AI Chose the Settings

The system did not just chase the most promising options. It also picked configurations in uncertain parts of the search space, so each experiment improved the model even when it failed. That balance between exploitation and exploration is what let the team move quickly. Within 40 experiments and three months, they found six working configurations across different power levels, including a successful print at 500 watts.

Ninety percent of commercial printers cannot print this metal alloy, Doppa said in a WSU release. Given that we were able to find these feasible process parameters, it allows us to use those commercial printers, and we are essentially democratizing the printing of this alloy.

PhD student Azza Fadhel, the paper's first author, took a practical view of the failures. They would give me back the results, and I liked all of them, even if they failed, because every result improved our AI model. That attitude makes sense for a problem where most configurations melt, crack, or distort.

Beyond Rocket Science

Lowering the power requirement matters beyond the lab. High-power laser systems are expensive to buy and run. They also wear faster and use more energy. If GRCop-42 can be printed on standard equipment, more organizations can experiment with it. That could speed up development of heat exchangers, thermal management parts, and aerospace components without forcing every user to buy a top-tier metal printer.

The researchers say the same AI framework could apply to other alloys and additive manufacturing systems. It could even move beyond 3D printing into scientific discovery problems where experiments are expensive and successes are rare. Drug discovery is one example Doppa mentioned.

There is still work to do before GRCop-42 becomes a routine material on a desktop metal printer. The WSU study found feasible settings, not a fully qualified production process. Aerospace parts still need certification, repeatability testing, and post-processing development. But the result shifts the conversation. A material once locked behind high-power hardware now looks like it could be reachable for a much wider range of printers, with AI doing the heavy lifting of figuring out how.