High-fidelity modeling of 3D woven composites considering inhomogeneous intra-yarn fiber volume fractions

The accuracy of strength prediction of 3D woven composites is strongly dependent on the quality of the utilized mesoscale geometry model. This paper aims at presenting a novel high-fidelity modeling approach for 3D woven composites considering both realistic mesoscale architectures and inhomogeneous intra-yarn fiber volume fractions (iy-FVF) along the yarn path. The unit cell model is numerically reconstructed from simulated results of digital element analysis. An Alpha-shapes algorithm was utilized to extract the shape of yarn cross-sections, and a digital-element-chain tracing method was established to avoid artificial inaccurate torsion deformation in the yarn geometry. The inhomogeneity of iy-FVF along the yarn path was specially characterized, and its effect on the mechanical performance of the composite was analyzed based on the progressive damage model. Furthermore, the proposed numerical method was employed to investigate the influence of the weft tow size on the tensile properties of the composite. It is shown that the predicted results of the high-fidelity model agree well with experiments. Iy-FVF varies significantly in warp yarns of the investigated 3D woven composite. Considering the inhomogeneous iy-FVF, predicted warp stiffness and strength increase about 10%, leading to higher predictive accuracy.

Automated summary

This paper proposes a high-fidelity modeling approach for accurately predicting the strength of 3D woven composites. The approach considers realistic mesoscale architectures and inhomogeneous intra-yarn fiber volume fractions along the yarn path. By numerically reconstructing the unit cell model and utilizing algorithms, the modeling accuracy is improved. The results show that considering inhomogeneous intra-yarn fiber volume fractions can improve the predictive accuracy.