Damage evolution analysis of C/SiC composite specimens with holes based on synchrotron radiation CT and deep learning

The damage evolution behavior of center-notched C/SiC composite components is investigated under uniaxial tensile loading by conducting in situ mechanical experiments based on synchrotron radiation X-ray CT observations with deep learning algorithms，3D digital image-based finite element method（3D IB-FEM），and digital volume correlation（DVC） method. The intelligent identification of different types of internal damage， including fiber bundle cracking， matrix cracking， and delamination， in center-notched C/SiC composite specimens under uniaxial tensile loading is achieved. Furthermore， the relationship between damage evolution and strain concentration is established. Deep learning-based damage identification and quantitative analysis indicate that both the center notch and initial pores influence the locations of damage initiation. The results of the three-dimensional digital image-based finite element analysis reveal the effect of initial pore geometry on crack initiation， while DVC results demonstrate the correlation between widespread strain concentration in fiber bundle regions and delamination damage， as well as final fracture. The three damage forms—fiber bundle cracking， matrix cracking， and delamination—are interrelated to some extent. As the load increases， adjacent delamination areas connect to form matrix cracks， which may further propagate into fiber bundle cracks. Under tensile loading， the primary failure modes of fiber bundles include fiber bundle fracture， fiber bundle splitting， and fiber bundle sliding. The presence of a center notch does not change the failure modes of the fiber bundles.