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

Researchers at Washington State University (WSU) have leveraged artificial intelligence to overcome a significant bottleneck in the additive manufacturing of GRCop-42, a high-performance copper-chromium-niobium alloy developed by NASA. Prized in the aerospace sector for its exceptional thermal conductivity and creep resistance at temperatures exceeding 700°C, GRCop-42 is ideal for liquid rocket engine combustion chambers. However, its high thermal conductivity and reflectivity typically require extremely high laser energy densities, restricting its use to specialized, high-wattage industrial 3D printers that are inaccessible to approximately 90% of the commercial market. To democratize access, the WSU team deployed a Bayesian optimization framework to navigate a massive search space of over 100 million parameter combinations. By training their machine learning model on data from only 37 prior failed experiments, they successfully identified viable configurations after just 40 physical trials. The most significant outcome was the discovery of settings that achieved high-density, defect-free prints using a laser power of only 500 watts—a record low for this alloy. This reduction enables processing on standard, lower-cost commercial 3D printers. For engineering professionals, this breakthrough represents more than just a reduction in capital expenditure. Lowering the power threshold decreases energy consumption, minimizes residual thermal stress within printed components, and reduces equipment wear. Furthermore, the study provides a scalable methodology for using AI to rapidly optimize the processing parameters of other refractory or notoriously 'unprintable' alloys, potentially accelerating the development of advanced thermal management systems across industries such as automotive and high-power electronics.