Micro-CT Based Statistical Geometry Modeling and Numerical Verification of 2.5D Sic_f/Sic Composite

This work presents a statistical approach to establish the refined meso-scale model for angle-interlock (2.5D) woven SiC_f/SiC composites considering various pore defects. This method firstly extracted the regions of interest to obtain the original data of yarn and pores from X-rays computed tomography (Micro-CT). Then, the geometric parameters of the collected yarns and pores were statistically analyzed based on the scanning tomograms. Consequently, a non-uniform pore distribution model (N-model) was established by introducing the virtual small pores and giant pores construction. Furthermore, for the validation of N-model, the traditional ideal model without pore defects (I-model) and uniform small pore distributing model (U-model) were reconstructed. Also, the tensile tests were carried out. The elastic modulus of composite obtained from the N-model, U-model, and I-model are 44,354.58 MPa, 42,932.23 MPa and 40,477.25 MPa respectively, while that of experimental result exhibit 41,659.85 MPa, the prediction accuracy rate of the N-model could reach up to 95%. The results showed that the proposed N-model is capable of accurately predicting the mechanical behaviors, full-field stress distribution and damage of 2.5-dimensional (2.5D) woven SiC_f/SiC composites, validated by the experimental tests.