Greedy-based approach for generating anisotropic random fiber distributions of unidirectional composites and transverse mechanical properties prediction

A novel adaptive greedy-based generation (GBG) algorithm is proposed to generate 2D random distribution of fibers for unidirectional composites. Inspired by greedy algorithm, the fitness function is introduced to better overcome the jamming limit of the traditional hard-core method than most published algorithms and generate representative volume element (RVE) with high fiber volume fraction adaptively. The good agreement of sta-tistical analysis with experiments suggests that the generated anisotropic RVEs is statistically equivalent to real composites. The influence of anisotropy on mechanical properties and progressive failure process is further revealed based on finite element method (FEM) using a progressive damage model. The results show that the predicted transverse elastic and strength properties are both anisotropic, which is not consistent with the traditional transverse isotropic hypothesis. The proposed algorithm can generate RVEs with high fiber volume fraction (up to 80%) and provide anisotropic RVEs statistically equivalent to real composites, and thus can serve as a promising method to guide the design and analysis of composites.

Automated summary

A novel adaptive greedy-based generation algorithm is proposed for generating 2D random distribution of fibers in unidirectional composites. The algorithm overcomes the jamming limit of traditional methods and generates representative volume elements with high fiber volume fraction adaptively. The generated anisotropic volume elements are statistically equivalent to real composites, as confirmed by statistical analysis and experiments. The algorithm can guide the design and analysis of composites by providing anisotropic volume elements with high fiber volume fraction.