Determination of fiber orientation in large-volume, short-fiber-reinforced plastic components by computed tomography using tracer fibers

• Applicants: Professor Dr. Marc Daniel Leonhard von Kreutzbruck; Professor Dr.-Ing. Sven Simon
• Subject Area: Plastics Engineering ; Lightweight Construction, Textile Technology ; Measurement Systems
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 556215542

The strength and stiffness and therefore the performance of fiber-reinforced plastics depend significantly on the fiber morphology (fiber orientation, length and volume). An established method for determining fiber morphology is computed tomography (CT), which enables fast, high-resolution and non-destructive 3D analysis of fiber morphology throughout the entire volume of the test specimen. This represents a major advantage over destructive, two-dimensional and time-consuming micrograph analysis, which can only be used in 2D layers. In addition to these many advantages of computed tomography compared to the alternative method, CT has the disadvantage that typically only relatively small components in the cm range can be analyzed non-destructively, as the resolution of the fibers in CT depends on the component size and the fibers in larger components (above the cm range) can no longer be resolved. A further increase in resolution using X-ray detectors with higher resolution reaches its limits, as the pixel area becomes smaller by a factor of four for a doubling of resolution and the noise increases accordingly. Furthermore, the possibilities of increasing resolution beyond this with detectors with higher resolution are also technologically limited. In contrast, this project aims to improve the maximum component size at which the fibers can still be resolved or detected by at least an order of magnitude (factor of 10). For this purpose, a small proportion of strongly X-ray attenuating tracer fibers, which behave mechanically and rheologically almost identically to the reinforcing fibers, will be added to the reinforcing fibers in the production of some components as an example. These tracer fibers attenuate X-rays more intensively than conventional reinforcing fibers, which means that, unlike conventional reinforcing fibers, they can also be detected and analyzed in larger test volumes. Due to the rheologically and mechanically similar behavior of reinforcing fibers and tracer fibers, the orientation of the detected tracer fibers can be used to infer the orientation of the surrounding reinforcing fibers. The resolution at which the orientation of the tracer fiber can still be detected is to be significantly increased with the methods of artificial intelligence using neural networks. Neural networks are particularly suitable for segmenting tracer fibers in the presence of high noise, as it is often the case in computer tomography. This increase in resolution based on artificial intelligence, in conjunction with the concept of tracer fibers, enables the above-mentioned increase in the maximum component size by an order of magnitude at which the fiber orientation of the entire component can still be detected. This makes it possible to produce better components by adapting the design and optimizing the manufacturing process.

• DFG Programme: Research Grants