Fast prediction of 2D C/SiC compression performance based on self-consistent clustering analysis

In this paper, the self-consistent clustering analysis (SCA) method was used to investigate the progressive damage behavior of 2D C/SiC under uniaxial compression load. The SCA method clusters the grid elements by strain concentration tensor, which greatly reduces the degree of freedom of the model and improves the computational efficiency of the model without significantly reducing the computational accuracy. The whole method consists of two stages: offline and online. In the offline stage, the k-means algorithm was used to decompose and cluster the high-fidelity composite unit cells and calculate the interaction tensor between different clusters, and finally the reduced-order model was generated. At the online stage, the mechanical response was obtained by solving the discrete Lippmann-Schwinger equations based on the reduced-order model. The SCA method was applied to predict the compressive strength of 2D C/SiC. When the total number of clusters is 64, compared with the experiment, the calculation accuracy of the compressive strength solution is reduced by 1% compared with the traditional finite element method, but the overall calculation efficiency is improved by 34 times. When the clustering time spent in the offline stage is not considered, that is, the meso-structure of the material is known in advance to solve its mechanical behavior, the time of one-time online calculation is only 6s, and the calculation efficiency is 10⁴ times higher than that of the traditional finite element method. It has broad application prospects in the fields of rapid design of structural performance and rapid prediction of structural state.