Deep learning-based Meso-scale reconstruction and bending failure mechanism of spreading fabric/felt needle C/SiC laminated composites

Needled carbon/silicon carbide (C/SiC) composites are widely used in the field of aerospace. However, reducing porosity and enhancing mechanical properties remain critical challenges. To address these challenges, this paper investigates the bending damage behavior and failure mechanisms of spreading fabric/felt needled C/SiC laminated composites (SFNPCS). Three kinds of SFNPCS, SFNP-15Gs (needle depth: 15 mm, hook type: G), SFNP-15Fs (needle depth: 15 mm, hook type: F), and SFNP-11Gs (needle depth: 11 mm, hook type: G), were innovatively designed and prepared. Then, a deep learning–based yarn segmentation method was employed to achieve a reconstruction of the meso-scale model of SFNPCS. Finally, the bending properties and progressive damage behavior of SFNPCS were investigated through a combined experimental and numerical approach. The results demonstrated that the densification efficiencies of SFNP-15Gs, SFNP-15Fs, and SFNP-11Gs are 572.58%, 566.87%, and 552.80%, respectively, with SFNP-15Gs achieving 1.5% and 3.6% higher efficiency than SFNP-15Fs and SFNP-11Gs due to its G-type needles, increased fiber coverage, and deeper needling. SFNP-15Gs also exhibits the highest bending strength (162.63 MPa), 6.20% and 18.06% higher than SFNP-15Fs and SFNP-11Gs, respectively. The primary failure mechanisms of SFNPCS include matrix fracture, fiber pull-out, interface debonding, and varying degrees of delamination. • Innovative design and preparation of three kinds of spreading fabric/felt needled C/SiC composites (SFNPCS). • A high-fidelity meso-scale model of SFNPCS was reconstructed using a deep learning-based yarn segmentation strategy. • The bending mechanical performance and failure mechanisms of SFNPCS was revealed by the combination of experiment and numerical method.