Researchers at Nanyang Technological University Singapore (NTU Singapore) have developed a new, fast and inexpensive imaging method to assess the quality of 3D printed metal parts.
Capable of analyzing the microstructure and material quality of parts, the imaging method could be a game-changer for industries such as aerospace where rapid, low-cost assessment of critical parts such as turbines and blades may be important for maintenance, repair and overhaul (MRO).
“Thanks to our inexpensive and fast imaging method, we can easily distinguish good 3D-printed metal parts from defective ones,” said Matteo Seita, assistant professor at NTU. “Currently, it is impossible to tell the difference without evaluating in detail the microstructure of the material. No two 3D printed metal parts are equal, even if they were produced using the same technique and have the same geometry.
“Conceptually, it’s akin to how two otherwise identical wooden artifacts can each possess a different grain structure.”
Imaging of microscopic crystals
Most 3D printed metal alloys are made up of a multitude of microscopic crystals that differ in shape, size, and atomic lattice orientation. Scientists are able to determine the properties of an alloy, such as its strength and toughness, by mapping these microscopic crystals. The researchers liken this to looking at the grain of wood, where wood is strongest when the grain is continuous in the same direction.
However, until now, analyzing this microstructure in 3D-printed metal alloys has been a laborious and time-consuming process, usually involving scanning with electron microscopes, which can cost between $73,000 and $1.5 million.
Seita and his team developed their new imaging method to provide the same quality of information in just minutes, and at a much lower cost. They used a system consisting of an optical camera, a flashlight and a laptop that runs proprietary machine learning software developed by the team, which cost around $18,400 in total.
The method involves treating the surface of the 3D printed parts with chemicals to reveal its microstructure before the camera is used to take multiple optical images as the flashlight illuminates the metal in different directions. The scientists’ proprietary software then analyzes the patterns produced by the light reflected off the surface of the part and deduces their orientation within 15 minutes.
Improve the quality assessment of 3D printed parts
The NTU researchers believe their imaging method could simplify the certification and quality assessment of 3D-printed metal alloy parts, especially those produced via laser-based techniques.
The microstructure of 3D printed metal can vary due to laser intensity and speed, cooling times, and the type and brand of metal powders used, which means that the same 3D printed design in two Different machines or production houses can result in parts of varying quality.
The NTU team’s software uses a neural network that mimics the way the brain forms association and processes thought combined with machine learning to predict crystal orientation in metal parts, based on differences in the scattering of light on the surface.
The software successfully created a complete “crystal orientation map” capable of providing information about the shape, size and orientation of the crystal atomic lattice in a 3D printed part.
According to the NTU team, the method could benefit critical parts and industries in particular with quick and relatively inexpensive quality assessment. The team is in discussions with NTUitive, NTU’s innovation and enterprise company, to explore the possibility of starting a spin-off company or licensing the patented imaging software to industry players.
Further information on the study can be found in the document entitled: “A Machine Learning Approach to Map Crystal Orientation Using Light Microscopy,” published in the journal NPJ Computation Materials. The study is co-authored by Mr. Wittwer and Mr. Seita.
Advances in Quality Assurance of Metal Parts
According to the World Economic Forum, simplifying the qualification process for 3D printing will be the key to its industrialization. Gregor Reischle and Christophe Blanc of TÜV SÜD recently echoed this point, stating that quality assurance could provide a faster route to serial additive manufacturing.
In the past, machine learning has been deployed to detect defects in 3D printed parts and improve the quality of individual metal layers by reducing “spatter” during the metal 3D printing process. X-ray imaging has previously been used to mitigate defects in 3D printed metal parts, while individual powders are continuously modified to improve the quality and surface finish of metal end parts.
There have also been more recent developments in this area, with Pennsylvania State University receiving a grant worth $180,000 from scientific technology company 3M to explore quality control methods for the metal 3D printing, and engineers at Lawrence Livermore National Laboratory developing a way to optimize properties. of parts produced by Liquid Metal Jet (LMJ) 3D printing.
Subscribe to the 3D Printing Industry Bulletin for the latest additive manufacturing news. You can also stay connected by following us on Twitter and love us on Facebook.
Looking for a career in additive manufacturing? To visit 3D printing works for a selection of roles in the industry.
Subscribe to our YouTube channel for the latest 3D printing video shorts, reviews and webinar replays.
Featured image shows analyzing unique crystal patterns on the surface of a 3D printed metal can pave the way for certification and quality assessment of parts made by additive manufacturing. Photo via NTU Singapore.
