MIT Engineers Use Machine Learning to Develop Ultra-Strong Aluminum Alloy for 3D Printing

Creating alloys means combining two or more metals to achieve new and improved properties, such as greater strength or resistance to corrosion. Traditionally, researchers test countless combinations to find the perfect formula, a process that can take months or even years. Simulations can speed things up, but a team of MIT engineers has found an even faster route. In a recent study, they used machine learning to identify a lightweight, 3D-printable aluminum alloy that is five times stronger than conventionally manufactured aluminum. Instead of running over a million simulations, their model narrowed the field to just 40, leading them straight to the ideal mix.

When the team printed the aluminum alloy, it performed as predicted. It was as strong as the most robust aluminum alloys produced today through traditional casting methods. This new material could be used to create strong, lightweight, and heat-resistant components, like fan blades in jet engines. “If we can use lighter, high-strength material, this would save a considerable amount of energy for the transportation industry,” Mohadeseh Taheri-Mousavi, who led the work as a postdoc at MIT and is now an assistant professor at Carnegie Mellon University, said.

The alloy design concept (Credit: Taheri-Mousavi et al.)

The Role of Machine Learning

The research came out of a course that Taheri-Mousavi took at MIT in 2020, where the students had to use computational simulations to design high-performance aluminum alloys. One of the keys to achieving a strong alloy is having its microscopic constituents, its “precipitates,” be as small and as densely packed as possible. So, the class methodically combined aluminum with various types and concentrations of elements to simulate and predict the resulting alloy’s strength. However, the tests failed to deliver a stronger result, leading Taheri-Mousavi to wonder if machine learning could do better.