Presenting at ECCM22 in Oslo — Statistically Equivalent 3D CFRP Microstructure Generation
Published:
Presentation at ECCM22, Oslo
Will be presenting part of my PhD research at the 22nd European Conference on Composite Materials (ECCM22) in Oslo, organized by ESCM.
About the Work
In this work, we developed an open-source and scalable algorithm for generating statistically equivalent 3D fiber composite microstructures. Rather than directly reconstructing a specific X-ray μCT dataset, the framework generates new synthetic fiber networks that preserve the key statistical features of the real microstructure — including measured fiber misalignment profiles, fiber size variability, and geometrical contact constraints.
The method combines a statistically optimized modelling layer with an in-house contact solver inspired by plant-like growth mechanisms. This allows the generated microstructures to remain statistically equivalent to real high-resolution X-ray μCT scans, while also being physically feasible and ready for CAD export.
In the attached video, the algorithm generates almost 2,400 fibers with realistic misalignment profiles, and demonstrates capability for generating microstructures at the scale of a 12K tow.
Beyond unidirectional CFRP, the algorithm can be applied to different fiber-based architectures — relevant for applications where microstructure matters, such as resin impregnation, structural FEA, compaction studies, fiber-to-fiber interactions, and electrical conductivity simulations.
Presentation Details
- Title: Statistically Equivalent Microstructure Generation of As-Manufactured 3D Unidirectional CFRP: From High-resolution X-Ray Micro-Computed Tomography to High-fidelity Fiber Network Architectures
- Session: SS17 — Image-based analysis of composite features
- Time: Monday, 17:45
- Location: Oslo, Norway