Additive Manufacturing of Metals: Equipment, Instrumentation and In-Situ Process Monitoring: Process Monitoring and Modeling Methods
Program Organizers: Ulf Ackelid, Freemelt AB; Ola Harrysson, North Carolina State University; Joy Gockel, Colorado School Of Mines; Sneha Prabha Narra, Carnegie Mellon University
Tuesday 2:00 PM
October 19, 2021
Room: A121
Location: Greater Columbus Convention Center
Session Chair: Sneha Prabha Narra, Carnegie Mellon University
2:00 PM Invited
Combined In-situ Monitoring of Meltpool, Powder Layer, and Part Topography for Laser Powder Bed Fusion (LPBF) Based Metal Additive Manufacturing: Xiayun Zhao1; 1University of Pittsburgh
Comprehensive in-situ process monitoring is desired for metal additive manufacturing quality assurance and control, since it can provide critical insights of diverse process signatures and material behaviors. We have designed, developed, and deployed an in-house system for monitoring a commercial LPBF machine (EOS M290) to quantify ultrafast meltpool dynamics and layerwise powder bed and printed part surface topography. The synergistic in-situ monitoring approach has a great potential to diagnose online the process anomalies and part defects more thoroughly. In this talk, I will introduce the newly developed monitoring technology, which features a reinvented coaxial meltpool monitoring system in conjunction with an off-axis camera and an improved fringe projection based topography measurement method. Our cost-effective systems and LPBF-specific methods can be readily applicable to monitor other machines. I will also present our current experimental results and findings and conclude with a perspective of huge heterogeneous monitoring data analytics and computing.
2:40 PM
Melt Pool Level Flaw Detection in Laser Hot Wire Additive Manufacturing Using a Trained Convolutional Long Short Term Memory Autoencoder: Brandon Abranovic1; Sulagna Sarkar1; Jack Beuth1; 1Carnegie Mellon University
This work focuses on deep learning enabled process monitoring for large-scale laser hot wire additive manufacturing using video data, which was collected using camera mounted on the robot arm pointed at the melt pool. Initial work consists of the unsupervised training of a convolutional long short term memory autoencoder to reconstruct footage from anomaly free single beads. The trained architecture was used to reconstruct footage where anomalies including arcing and wire stubbing occurred. Wire stubbing is a condition where un-melted wire impacts the solid bottom of the melt pool, leading to jittering of the wire. The model’s ability to faithfully reconstruct the video was quantified by computing a regularity score between the raw frames and model outputs, with low regularity scores being indicative of an anomaly. Preliminary results have demonstrated the model’s robustness in detecting the anomaly classes of interest to this study.
3:00 PM
Materials Characterization of Anomalies Identified Through In-situ Process Monitoring Data Analytics: Jonathan Ciero1; Dylan Christman1; Kyle Ryan2; Shuchi “SK” Khurana3; Thomas Spears2; Joy Gockel1; 1Wright State University; 2Open Additive, LLC; 3Addiguru
Laser powder bed fusion (LPBF) is a rapidly growing metal additive manufacturing (AM) technology for the fabrication of complex end-use parts. However, LPBF has difficulties assuring quality parts with minimal defects during the build process without additional post-processing that adds to the total cost and overall part manufacturing time. The use of in-situ process monitoring can assure build quality and can support qualification and certification of AM parts. Using three different in-situ sensors, data was collected over multiple builds of Alloy 718 with strategically generated process anomalies using processing parameters and problematic geometries. Materials characterization methods were employed to quantify defects in anomalous regions as indicated by the in-situ process monitoring data analytics. Anomalies detected during the AM builds are associated with specific material defects to further guide the development of robust process monitoring techniques to detect quality issues that could lead to part failure.
3:20 PM
Physics Guided Machine Learning DED Melt Pool Width Prediction: Brett Diehl1; Clara Mock2; Lester Hitch2; Brandon McWilliams2; Berend Rinderspracher2; 1Oak Ridge Associated Universities; 2CCDC Army Research Laboratory
Directed energy deposition (DED) is of interest to the aerospace and defense industries for the production of novel and complex geometries, as well as repair applications. However, variability during the build process can result in deviations in final component geometry, structure, and mechanical properties which adds to the complexity of process planning and slows down adoption of this technology. Neural nets were trained to predict melt pool width, given input drivers such as build height, laser power, laser speed, and thin wall length. Physical constraints on the relationships between the 1st and 2nd derivatives of input drivers and melt pool width were enforced using custom loss functions, yielding physics-guided neural networks (PGNNs). PGNNs predicted the melt pool width with a higher performance (R-square = 0.991) than traditional neural networks (R-square = 0.884). Physics-based loss functions performed superior to traditional methods of regularization and were a superior method of training on unbalanced datasets versus sample/class weighting. This work demonstrates the benefits of enforcing physical constraints on machine learning predictions of additive manufacturing processes using finite estimates of mathematical expressions of physical laws.