AI Finds Low-Power Way to 3D Print NASA's GRCop-42 Alloy

What makes GRCop-42 so hard to print

GRCop-42 is a copper-chromium-niobium alloy NASA developed for rocket and engine parts that see extreme heat. It handles high temperatures well, which is why it appears in combustion chambers and nozzles. But that same strength creates a manufacturing headache.

Copper reflects infrared laser light rather than absorbing it. It also pulls heat away from the melt pool faster than most metals. Traditional laser systems need high power to compensate, typically running at 900 watts or more. That puts the process out of reach for smaller shops running mid-range equipment.

AI searched 100 million configurations

A team from Washington State University and the University of Minnesota tried a different approach. They used a method called Bayesian Experimental Design for Additive Manufacturing, or BEAM. Rather than brute-force testing every combination, the AI narrowed the search space based on each result.

Starting with a five-axis powder-fed directed energy deposition system and a 1000 watt fiber laser, the team tested powder feed rate, gas flow, substrate thickness, scan speed, and layer height. The AI explored fixed power levels of 950, 700, 600, and 500 watts. Within 40 experiments it found at least one viable configuration at each level, including three successful setups at 600 watts.

A 500-watt result that holds up

The lowest-power configuration used a 0.075 RPM powder feed rate, 7 L/min gas flow, 250 mm/min scan speed, and a 0.3 mm layer height. Successful builds generally paired high scan speeds with thin layers. After machining a cylinder printed at 700 watts, the team found a solid interior with minimal defects and a clean transition between the Inconel 718 substrate and the GRCop-42 build.

The study does not qualify parts for flight. That requires a separate qualification process. But widening the process window down to 500 watts matters. Lower-power DED systems cost less and are more common in small and medium manufacturing facilities. If GRCop-42 can run on those machines, adoption should increase.

Where this fits in the broader picture

GRCop-42 has already shown promise on conventional powder-bed fusion machines. Auburn University printed it on an EOS M290 in 2025. Nikon SLM Solutions developed LPBF parameters that reached 99.97 percent density. The WSU work adds DED to that list and shows that AI-guided experimentation can compress months of trial and error into a few days of targeted runs.

The team plans to extend the approach to other alloys and compositions. For now, the 500-watt proof of concept is a concrete step toward making NASA-grade copper alloys practical outside specialized facilities.