CT scan of a fiber-reinforced composite showing the direction of fibers. Arrows show the different direction of fibers in different regions of the material.Aerospace and automotive applications increasingly rely on lightweight composite parts made from polymers reinforced with carbon or glass fibers. With the rise of additive manufacturing methods like 3D printing, it is becoming possible to design novel filaments and control their orientation within composites to tailor performance to specific applications. But the ease of manufacturing advanced composite parts via 3D printing has a downside. The ubiquity of 3D printers makes it much easier for counterfeiters to copy parts.
“A lot of intellectual property is invested in developing composites material and 3D printing conditions,” points out Nikhil Gupta of New York University. “Reverse engineering using imaging methods can allow this IP to be stolen. Machine learning can even automate this process… [to] produce high quality replicas by unauthorized personnel.”
Gupta and his team show that novel and innovative composite materials can be reverse engineered using three-dimensional imaging and reconstruction techniques combined with powerful machine learning [Yanamandra et al., Composites Science & Technology 198 (2020) 108318, https://doi.org/10.1016/j.compscitech.2020.108318 ]. Not only can the size and shape of a part be reconstructed in this way, but also the tool path – or unique instructions – that the 3D printer needs to manufacture it.
The researchers took a set of scanning electron microscopy (SEM) to measure dimensions and micro-CT (µCT) scans to identify the orientation of fibers within a composite. A sub-set of these images was then used to train the machine learning algorithm. The accuracy of the algorithm in identifying the 3D printing tool path was then verified with a second sub-set of images. The team used over 78,000 images to train the algorithm and a further 5000 µCT images to test the accuracy of its predictions.
“Compared with the manual methods of reverse engineering, the machine learning algorithm is a powerful method where a few CT scan images are able to provide the part design and 3D printing tool path,” explains Gupta. “Our study shows that products developed with investment of significant amount of time and money can be easily reverse engineered.”
The work raises concerns over the security of intellectual property in 3D-printed composite parts. Modern capabilities in designing and producing high-performance fiber-reinforced composites can be harnessed to produce almost-indistinguishable high-quality replica parts quickly and cheaply.
“Machine learning methods are being used in the design of complex parts but, as our study shows, they can be a double-edged sword, also making reverse engineering easier,” says Gupta. “Awareness of security issues will encourage researchers to develop new tool paths and composite materials that consider security as a design constraint.”