Development of RVE-embedded solid elements model for predicting effective elastic constants of discontinuous fiber reinforced composites

• An RVE model with embedded solid elements is developed.
• A modified RSA algorithm based on the Boolean operation is proposed.
• The developed model can simulate the deformation of fiber more realistically.
• The strict mesh conformity at the interface is avoided.
• The overall calculation processes are accomplished automatically.

In this work, an RVE model with embedded solid elements (RVE-ESE) is developed to predict the effective elastic constants of discontinuous fiber reinforced composites (DFRC). Unlike the traditional finite element (FE) modeling, the matrix and fibers can be meshed separately and independently due to the implementation of the embedded element technique, which enables the ease of RVE meshing without strictly satisfying the mesh conformity at the interfaces. Since the structured meshes can be used for the matrix, it becomes quite easy to impose the periodic boundary conditions on the RVE. In addition, a modified random sequential adsorption (RSA) algorithm based on the Boolean operation is proposed to generate the RVE by using the Python script in the ABAQUS/CAE. This algorithm can determine whether a new generated fiber overlaps with the pre-existing ones without looping calculation. Some important issues related to the modeling accuracy, such as the RVE size, the fiber separation effect and the mesh sensitivity of both matrix and fibers, are also investigated. It is found that the proposed model agrees very well with the traditional FE direct meshing approach when the element sizes of the matrix and fibers in the RVE-ESE model are sufficiently fine. Compared with commercial software Digimat, the current model yields relatively higher prediction on the elastic moduli of the composites with randomly distributed fibers. However, for the composites with aligned fibers, Digimat shows lower result for the transverse Young's modulus while a bit higher prediction for the longitudinal one.