An automated framework for homogenizing thermo-mechanical properties of composites

Abstract

Automated finite element (FE) homogenization of composite RVEs requires consistent treatment of periodic meshing, boundary constraints, analysis generation, and postprocessing, and coordinating these tasks can become labor-intensive for repeated RVE calculations. This work presents Meso-FEH v1.0, an open-source Python framework for automated mechanical and thermal homogenization of periodic composite RVEs. The framework integrates periodic meshing, periodic boundary condition (PBC) construction, analysis generation, job execution, and postprocessing within a unified Abaqus-based workflow. Conformal meshes on opposing RVE boundaries are generated using scripted copy mesh pattern operations, while PBCs can be constructed either through direct Abaqus input-file generation or through the Abaqus/CAE model database. The framework is verified using randomly distributed spherical SiC particle-reinforced aluminum RVEs. Periodic mesh correspondence and PBC enforcement are confirmed through coordinate and displacement checks. For the tested models, the input-file-based PBC implementation is approximately 40 times faster than the CAE-based implementation for the elastic, thermal-expansion, and thermal-conduction analyses and approximately 254 times faster for the elasto-plastic analysis. Mesh- and RVE-size sensitivity analyses support the adopted discretization and RVE dimensions. The predicted elastic properties, coefficients of thermal expansion, thermal conductivities, and uniaxial elasto-plastic responses show generally good agreement with analytical models, available experimental measurements, and Digimat-FE results. Meso-FEH provides an efficient and reproducible platform for automated Abaqus-based RVE homogenization and high-throughput generation of effective composite properties.