In-process Defect Monitoring and Correction in Additive Manufacturing of Aluminum Alloys

Metal additive manufacturing (AM) has become increasingly popular to fabricate complex, light-weight, and high- efficiency components for use in the aerospace industry; however, there are inherent limitations in existing AM processes that have delayed widespread implementation for aviation applications. Porosity is just one example of the key characteristics that can impact the mechanical strength of an AM part. This research focuses on a real-time feedback system to detect and correct defects during the powder bed fusion process of aluminum alloys. In this study, AlSi10Mg coupons were built using various AM parameters. The build process was continuously monitored via a high-frequency in-situ infrared camera which had been integrated into a commercial metal powder bed fusion machine. Porosity information (pore location and size) of the as-built AM coupons were characterized using x-ray computed tomography. The monitoring results were post processed and correlated with porosity location, indicating a strong relationship between abnormal sensing signal and pore formation. This demonstrates that the real-time abnormal sensing signal can be a good indicator for identifying pore formation during the AM process. Additionally, Sentient Science Corporation (Sentient) used its advanced modeling technique to simulate the AM build process regarding the melt pool geometry, porosity, and microstructure. Prediction of porosity level at different AM parameters aligned well with the experimental results. Advanced modeling results showed that careful selection of AM settings is required to correct in-process defects. Repair parameters must be tailored to achieve satisfactory correction of individual defects. Combining the in-situ defect monitoring and advanced simulation capabilities enables the creation of a closed-loop feedback control system that provides automatic defect detection and correction action in powder bed additive manufacturing process.