Numerical simulation of torsional deformation for 3D woven preforms based on the virtual yarn modeling strategy

SiC fiber-reinforced ceramic matrix composites have been widely applied in aerospace fields due to their excellent high-temperature performance. During the forming process of complex curved-surface components, torsional deformation of preforms is inevitable. Current modeling approaches are generally limited to unit-cell scale and lack simulation methods that simultaneously consider computational accuracy and efficiency. In this study, we propose the concept of virtual yarns, which decouples the characteristics of low bending stiffness and high tensile stiffness of yarns through shell/truss hybrid elements. The bending stiffness is calibrated based on cantilever beam experiments. A full-scale torsion model of SiC preform was established to predict torque-torsion angle curves and simulate torsional deformation of SiC fiber 3D woven preforms. The accuracy of the model was verified by scanning the torsional deformation of the yarns through Micro-CT. The results demonstrate that the proposed virtual yarn model exhibits mechanical responses consistent with experimental torsion tests on SiC preforms, and can simulate torsional deformation with high fidelity, achieving similarity coefficients exceeding 94.28%. Compared with traditional fiber-level modeling, the computational efficiency of virtual yarn model is significantly improved, providing a new idea for the engineering simulation of large-scale components.