Automated CT-Based Segmentation and Modelling of Textile Composite Materials

Sammanfattning

Textile-reinforced carbon fibre composites are increasingly considered for aerospace applicationsdue to their high specific stiffness and strength together with their ability to be tailored towardsspecific loading conditions. However, the mechanical behaviour of such materials is strongly influencedby the as-manufactured textile architecture, making accurate geometrical representationsimportant for reliable prediction of homogenised material properties. X-ray computed tomography(XCT) combined with computational homogenisation provides a potential route for generatingrepresentative volume elements directly from physical composite samples.The present work investigates the applicability of an existing XCT-to-FE workflow, previouslydeveloped for 3D layer-to-layer angle interlock composites, to a 2D five-harness satin (5HS) wovencarbon fibre reinforced polymer composite architecture representative of aerospace applications.The workflow combines XCT imaging, machine learning-based image segmentation, and voxelbasedfinite element homogenisation in order to estimate elastic material properties from XCTreconstructions of as-manufactured composite samples.Both virtual and physical investigations were carried out. A virtual 5HS composite samplegenerated in TexGen was utilised to evaluate XCT scan parameters and quantify segmentationperformance against known ground truth data. In addition, physical XCT scans were performedon both resin-injected and dry fibre samples in order to assess the applicability of the workflowto manufactured materials. Experimental tensile testing combined with digital image correlationwas further conducted in the principal material directions to provide reference elastic propertiesfor comparison with the numerical predictions.The results showed that the pre-trained segmentation model was capable of identifying the generaltextile architecture, but that additional transfer learning using synthetic 5HS training data wasrequired to obtain segmentation quality suitable for reliable voxel-based finite element modelling.Following transfer learning, a volume-wide pixelwise agreement value of 94% was obtained for theevaluated virtual sample. The study further showed that attenuation contrast significantly influencesthe segmentation performance, with the dry fibre samples visually exhibiting substantiallyimproved material constituent separation compared with the resin-injected sample.Computational homogenisation of the segmented XCT reconstruction produced elastic stiffnesspredictions showing reasonable agreement with the experimentally measured tensile stiffnessvalues, with a deviation of less than 12%. Although discrepancies between experimental and numericalresults remained, the work demonstrates that the XCT-to-FE workflow can be extendedto previously unseen 2D woven textile architectures through the introduction of additional targetarchitecture-specific training data.Overall, the results demonstrate the potential of combining XCT imaging, machine learningbasedsegmentation, and computational homogenisation for non-destructive characterisation ofas-manufactured textile-reinforced composite samples