Search

A novel pipeline for the generation of synthetic tomograms of woven composite materials, to be used for training of machine learning based segmentation algorithms is presented. The pipeline is completely based on open source software and heavily utilizes the graphical processing unit for fast data generation. The proposed method generates a surface mesh of the woven geometry, scans it, reconstructs the scan, and generates a voxel labeling of the generated tomogram. It is demonstrated that the method can generate images that show good agreement with experimentally produced x-ray computed tomography images of a 3D-woven carbon fiber reinforced polymer composite.

For fibre-reinforced composites, most of their mechanical properties is tied to the fibre scale. Thus, imaging-based characterisation demands resolving fibres to characterise these materials accurately. However, high resolutions limit the field of view and lead to lengthy acquisition times. Emerging non-destructive imaging technologies and algorithms now accurately provide fibre orientations without detecting individual fibres. Studies show that voxel sizes up to fifteen times the fibre diameter are feasible, still allowing accurate tensile modulus predictions. Our presented software incorporates sub-voxel fibre orientation distributions using ultra-low-resolution three-dimensional X-ray tomography data in a numerical model, providing an effective method for characterising these materials.

Three dimensional (3D) fibre-reinforced composites have shown weight efficient strength and stiffness characteristics as well as promising energy absorption capabilities. In the considered class of 3D-reinforcement, vertical and horizontal weft yarns interlace warp yarns. The through-thickness reinforcements suppress delamination and allow for stable and progressive damage growth in a quasi-ductile manner. With the ultimate goal of developing a homogenised computational model to predict how the material will deform and eventually fail under loading, this work proposes candidates for failure initiation criteria. The criteria are evaluated numerically for tensile, compressive and shear tests. The extension of the LaRC05 stress based failure criteria to this class of 3D-woven composites is one possibility. This however, presents a number of challenges which are discussed. These challenges are related to the relative high stiffness in all directions, which produce excessively high shear components when projected onto potential off-axis failure planes. To circumvent these challenges, strain based criteria inspired by LaRC05 are formulated. Results show that strain based failure predictions for the simulated load cases are qualitatively more reasonable., Part of proceedings ISBN 9781510896932QC 20201106