A team from Carnegie Mellon University presents an AI agent-based workflow to predict the behavior of alloys in metal 3D printing by laser powder bed fusion (LPBF). The system integrates thermodynamic models and defect maps to evaluate the printability of compositions, optimizing materials and process parameters with less costly experimentation.
Integration of Thermodynamic Models and Defect Maps ?”¬
The methodology combines Thermo-Calc software, which predicts phases and thermodynamic properties of an alloy, with models that simulate lack-of-fusion defect formation based on laser parameters. The AI agent correlates this data to predict whether a given composition, under certain conditions, will produce a dense and defect-free part. This allows virtually screening hundreds of combinations.
Farewell to the print and pray method ??
This could mark the end of the golden age of glorified trial and error, where designing a useful alloy required the faith of a monk and the budget of a small nation. Now, instead of crossing your fingers and hoping it doesn't turn out like a colander, a digital agent will coldly tell you that your master composition is, in reality, a disaster waiting to happen. An advance for science, but a blow to the romance of the workshop.