AI Is Optimizing LPBF for Titanium Alloys

Manufacturing parts using titanium alloys is complex and expensive, requires a lot of patience and users make only slow progress. Even modern technologies such as metal 3D printing can only help to a limited extent. Currently, trial and error is the main approach utilized in order to gradually identify the optimum manufacturing conditions. However, titanium parts are of great value in the aerospace, defense and maritime sectors. It is therefore important to speed up the production of these parts in order to meet demand more quickly and reduce costs. That is why researchers at the Johns Hopkins Applied Physics Laboratory (APL) and the Whiting School of Engineering have been searching for a solution. As such, they have developed a new technology that enables fast, stable and precise processing of titanium parts for additive manufacturing using AI.

In recent APL research, AI has been used in many different ways in various fields and examined for its potential, but also to a certain extent for its risks and side effects. The study “Machine learning enabled discovery of new L-PBF processing domains for Ti-6Al-4V,” which was published in Additive Manufacturing in December 2024, covers the opportunities for AI in process control and optimization. As the title of the study suggests, the research team focused on the titanium alloy Ti-6AI-4V, which is valued in numerous industries for its high strength and low weight. The aim of the research was to create optimal conditions for faster processing of the alloy and thus obtain precise, strong end parts.

Brendan Croom, a senior materials scientist at Johns Hopkins APL, in the lab (photo credits: Johns Hopkins APL/Ed Whitman)

AI Goes Beyond the Processing Limits of Titanium Alloys

As with all materials, processing conditions have an influence on the material properties. Laser power, scanning speed, etc. can determine how the material hardens and whether it is ultimately solid, flexible or brittle. The correct configuration of the process parameters is therefore largely responsible for the quality of the end parts. This is only possible as well through constant trial and error and adjustment.