Additive manufacturing (AM), more commonly known as 3D printing, has existed for years but has not yet been adopted within common markets due to the lack of quality, versatility, and trustworthiness of the products produced by the process. The ultimate goal of AM processing quality assurance is a digital thread that tracks a part through each stage of its design, build, and post-production performance. Many post-production problems stem from preventable build-time errors that go undetected. Specifically, there is a major issue in Laser Powder Bed Fusion (LPB-F) prints of relatively high failure rates and unexpected flaws in prints. This unreliability prevents AM from being adopted in manufacturing processes. Researchers currently perform analyses on build-time sensor data to mitigate these challenges on an ad hoc basis.
This project focuses on streamlining and consolidating techniques for analysis to allow researchers to gain insights efficiently and effectively. The resulting product is a reference tool suite to illustrate how AM build data from retrofitted sensors can be parsed, analyzed, explained, and visualized in an understandable and easily exchangeable file format. The suite is based in Jupyter notebooks, a markup format that displays Python source code, output, and plain text in a cohesive fashion. The product’s interface will focus on the build data produced by the sensors on the EOS M-290 ; however, the resulting algorithms, sensor data analysis, defect consideration, and graphical user interfaces will be modular and extensible to other applications. The finished product will be utilized for other machines and sensor types, and adaptable to a wide variety of analyses to make the digital thread possible.