Deep learning based tensile-shear damage evolution mechanism of quasi-isotropic satin weave C/SiC composites

4D X-ray CT in-situ tensile testing, along with deep learning technology, was used to characterize the damage and failure process of quasi-isotropic lay-up satin C/SiC composites under tensile loading, and to reveal the damage evolution mechanism under the coupled action of (0°/90°) lay-up tension and (±45°) lay-up shear. Using the deep learning image segmentation method, damages were extracted for quantitative analysis based on intelligent detection of matrix cracks, delamination of the material under loading. Furthermore, damage and failure mechanisms were investigated by examining the fracture morphology. It is found that ±45° oblique cracks account for the major part of matrix cracks. Oblique cracks were mainly induced by small cracks at the initial loading stage. Although transverse cracks were less than oblique ones, their lengths and crack opening distance developed rapidly. Delamination was induced by the deflection of matrix cracks along the interfaces between adjacent layers. For a (0°/90°) satin lay-up, transverse split took place in the tissue point region of 90° fibre tows, and accompanied by bending in the floating length region of 90° fibre tows. Fractures occurred in the tissue point region of 0° fibre tows, and accompanied by longitudinal split in the floating length region of 0° tows. For a (±45°) satin lay-up, oblique split and relatively sliding occurred in −45° (or +45°) tows. While fractures accompanied by fibre tow bridging and bending took place in the +45° (or −45°) tows of the same lay-up.