Image-driven analysis of tensile mechanical properties of 2.5D woven composites

2.5D woven composites have the complex meso-structures owing to the mutual compaction between fiber bundles, where the fiber bundles have the significant variations in local fiber distribution and fiber volume fraction along the fiber bundle paths. These complex microstructures directly affect the mechanical performances of the composites. An image-driven surrogate modeling approach is developed to construct meso-scale finite element (FE) models analyzing transverse tensile failure of fiber bundles. The transverse stress distributions in fiber bundles are rapidly predicted by the surrogate model, which are further used to correct the matrix stress concentration factor. The internal fiber bundle states of the 2.5D woven composites are studied by Micro-CT characterization, which can be recognized as the input to rapidly determine the transverse mechanical properties of the local fiber bundles. A refined FE model of the 2.5D woven composites combining the local variation mechanical properties of the fiber bundles can successfully reproduce the progressive damage evolution, identifying the main damage localized in bundle interlacing and bending regions, which are also validated by the tensile experiment.