Automated defect detection for X-ray computed tomography – cost-effective inspection for additive manufacturing

EU funded research project develops alternatives to manual CT data analysis

The most time-consuming factor during non-destructive testing with X-ray computed tomography (CT) is the manual data analysis. But with automated defect detection in CT data, there is much potential for reducing inspection costs – thus making additive manufacturing (AM) of components more economical.

A Bremen-based cooperative research project is looking for ways to use machine learning to find defects automatically in CT data. In this project, Testia GmbH works together with the Fibre Institute Bremen and Kolbes Messtechnik. The collaboration is funded by the European Regional Development Fund.

Since May 2019 the partners work together in the Project LURAFO2008A „Automatisierte Fehlerdetektion und Fehlerbewertung von Röntgen-Computertomographie (CT) Daten hochkomplexer 3D Metall- und Faserverbundbauteilen“ (Automated defect detection and evaluation in CT data of highly complex 3D metal and fiber-composite parts), which is planned to run until fall 2021.

Research goals of the project:

  • Analysis and qualification of short fiber injection molding processes and 3D long fiber printing processes, with high-definition 3D X-ray microscopy (XRM) and CT.
  • Based upon the above: Development of automated defect detection in CT data
  • Researching artificial intelligence methods for automated defect detection
  • Development of in-line capable process assurance concepts for 3D long fiber printing

Testia GmbH is mainly taking part in the second and third area.

Short description of the research project for automated defect detection

The project aims to analyze and develop in-line capable process assurance concepts and methods of automated defect detection for 3D manufacturing. These methods will be based on X-ray computed tomography and partly on machine learning. The highly complex components are difficult to approach with conventional non-destructive testing and analysis, while at the same time demanding high standards of in-line and in-service quality assurance. This does not only apply to material surface quality, but also structural integrity. To ensure consequent adaption for structures in the aerospace sector, it is necessary to develop and validate automated inspection and analysis methods that are applicable for the respective level of complexity.