Continuous Learning for Additive Manufacturing Processes Through Advanced Data Analytics

Additive manufacturing (AM) is a promising manufacturing technique for end-use parts that can solve challenges for American manufacturers in many industries, e.g. aerospace, defense, automotive, energy, and healthcare. However, despite the potential that AM offers, the rate of AM adoption in industry is very slow. This is because AM suffers from low repeatability and quality consistency issues, which lead to high cost and time requirements in AM qualification. Conventional manufacturing is mature enough such that there are commercially available engineering software and data analytics tools (e.g. injection molding simulation, casting simulation, finite element analysis) that elucidate Parameter-Structure-Property (PSP) relationships. Unfortunately, in the AM industry, these engineering software tools are barely nascent and not fully validated for AM, and appropriate data analytics tools for understanding PSP relationships do not exist. Furthermore, the database schema that holds and structures the data upon which AM engineering software tools draw is not standardized for AM data. Senvol proposes to develop a robust AM data schema that will be able to structure multiple types of AM data, including in-situ monitoring data, microstructure data and non-destructive testing (NDT) data; and to further develop a data analytics software tool that will analyze AM’s PSP relationships.