A data-driven artificial intelligence approach will allow additively manufactured (AM) part fabrication to be more easily moved from machine to machine without the need to repeat full qualification. This presentation will review a multi-disciplinary effort focused on streamlining qualification procedures for metallic additively manufactured ground vehicle components. The goal is to characterize feedstock, microstructures, defect contents, and mechanical properties of test components produced on various AM machines. These data together with pedigreed manufacturing process parameters, in-situ process monitoring and machine learning are used to define knowledge that can readily be transferred when production is moved from one machine or AM process to another. The project is focused on 316L stainless steel, and to-date sample plates and tensile samples have been fabricated using both laser powder bed fusion and wire-fed laser-directed energy deposition. This presentation will discuss similarities and differences in the relationships between these processes and from machine to machine. An overview of the proposed machine learning approach will also be provided. Through this data-driven approach, the amount of work that needs to be repeated for qualification on the new machine or a different AM process will be reduced, allowing for wider adoption of AM.
- Define how combining experimental data with machine learning can uncover correlations in the processing-structure-properties relationship.
- List similarities and differences for alloy 316L stainless steel components produced via laser powder bed fusion, and by wire-fed laser directed energy deposition.
- Describe how a data-driven machine learning approach can be used to streamline additive manufacturing qualification.