AI Advances Reliability of Aerospace-Grade Composites
Researchers at the University of British Columbia’s Okanagan campus have developed a data-driven method to better understand and optimize advanced fabric composites, materials widely used in aerospace, automotive, construction, and high-performance sports equipment. By combining materials informatics with machine learning, the team has devised a faster, more accurate way to assess the internal structure of these complex materials and predict their mechanical performance.
Fabric composites differ from simpler, one-dimensional composites in that they are woven from interlaced fibers, creating a structure that is both lightweight and exceptionally strong. Their geometry can be tailored to deliver specific performance characteristics, but this same complexity makes them difficult to analyze. Dr. Abas Milani, Professor in UBC Okanagan’s School of Engineering and founding Director of the Materials and Manufacturing Research Institute, explains the engineering challenge: “For example, if we want the wings of an aircraft to resist specific high shear forces, building a composite material with a particular microstructure will help us achieve that.”
Traditionally, engineers have relied on experimental testing or numerical modeling to characterize these materials. While effective, these methods are resource-intensive, requiring costly equipment, high-performance computing, and significant time. They also often assume idealized geometries. In reality, manufacturing introduces imperfections such as waviness, voids, and fiber misalignment, all of which can influence performance. “Experimental or numerical modelling techniques are effective tools, but they are time-consuming and require expensive devices or high-power computers,” says doctoral student and study co-author Tina Olfatbakhsh. “They also often assume the material geometry to be perfect, although, in the actual manufacturing process, textile composites can have many different internal complexities like waviness, voids and even fibre misalignment. This complicates matters significantly.”
The UBC team’s approach begins with high-resolution, non-destructive imaging of composite specimens using X-ray computed tomography. This technique captures the intricate three-dimensional architecture of the fibers in situ, revealing the true microstructural features of the material. These images are then processed using machine learning algorithms trained to correlate structural details with measured mechanical properties. The result is a predictive model that can estimate performance directly from imaging data, bypassing the need for exhaustive physical testing.
This methodology not only accelerates analysis but also feeds into a growing materials database. By cataloging microstructural images alongside performance metrics, the database becomes a shared resource for researchers and manufacturers worldwide. “This database is a good opportunity to exchange knowledge with scientists around the world to prevent doing repetitive tests and analysis,” Olfatbakhsh explains. “Now, whenever they need a specific performance, they know which material arrangement to choose using this database.”
The work is closely tied to the Composite Research Network’s (CRN) Okanagan Node, where academic and industry partners collaborate to advance composite manufacturing. Dr. Milani, principal researcher at CRN’s Okanagan Node, emphasizes the importance of keeping testing capabilities aligned with material innovation: “As manufacturers develop more innovative composite materials that are formulated at the micro-scale, our testing needs to keep pace so we can ensure the integrity and strength of these new microstructures.”
Olfatbakhsh, who manages the CRN Okanagan Node laboratory, notes that the new method is not only accurate and effective but also compatible with existing manufacturing workflows. “By streamlining the analysis using machine learning techniques, we are making great strides towards a framework for smart, data-driven design and optimization of woven fabric composites,” she says. “Our findings are a promising step forward for the smart design of next-generation tactile composites, especially in prominent industries like aerospace and transportation.”
The research, published in *Composites Science and Technology* and supported by the Natural Sciences and Engineering Research Council of Canada, represents a convergence of advanced imaging, computational modeling, and collaborative data sharing. It reflects a broader trend in materials science toward integrating artificial intelligence with experimental techniques to accelerate the development of high-performance, application-specific materials.
