Quality control in metal additive manufacturing
Metal additive manufacturing (or 3D printing of metal) is a promising technique in production environments. However, many companies struggle with quality control. High-end products such as turbine blades, impellers, functional parts for aeronautics, etc., have a dense metallic structure. For these applications, customers demand consistent and certified quality levels. Lower quality parts often contain porosities and other solidification defects. These are caused by suboptimal process conditions, e.g., when new parts are produced for the first time, or when, in a serial production, process conditions start drifting. To identify defects, to meet quality standards and to deliver certificates, a monitoring system which relies on in-process data is necessary.
Flanders Make participates in CoE RAISE
As part of the European Centre of Excellence in Exascale Computing "Research on AI- and Simulation-Based Engineering at Exascale" (CoE RAISE), we develop AI techniques for keyhole and lack-of-fusion anomaly detection. These AI and HPC methods are refined with a strong focus on processing big data in efficient workflows. In a first phase, data is generated by monitoring the melt pool during printing with a high-speed camera. During the print, a variability of speed and power is applied to characterize the process, and to increase the rate of anomalies created with respect to normal operation. The additive manufactured samples are analysed by micro–CT X-ray scanning, acting as ground truth for the training of the deep learning models. In combination with the log of the printing parameters and the high-speed melt pool monitoring video it leads a robust model, able to detect anomalies that occur during printing.
Interested in this technology?
For this project, we are looking for companies that will benefit from the results of the use case described above. Are you interested in the outcome of this project? Then please fill in the form below. We will contact you as soon as possible.