Data and Manuscripts for Voxel-Based Finite Element Analysis of Spherical and Aligned Ellipsoidal Al-SiC Particulate Composites

SUMMARY This repository archives a high-fidelity, open-access benchmarking dataset detailing the 3D micromechanical elastic behaviour of multi-phase Al-SiC particulate composites. The included working manuscripts detail the formulation and analysis of representative volume elements (RVEs) containing mono-dispersed spherical inclusions and aligned ellipsoidal inclusions (aspect ratio ≈ 2:1) at a high volume fraction of ~40%. The numerical models feature a 6:1 phase contrast ratio (Matrix: E=70 GPa, ν=0.33; Inclusions: E=410 GPa, ν=0.17) discretised on a structured voxel grid of 1 million hexahedral elements (3 million degrees of freedom). All models utilize 1/8 symmetry boundary conditions to capture geometry-induced elastic anisotropy (showing a ~15% stiffness increase along the major axis). The simulations and underlying geometries were generated using an optimized, proprietary parallel Fortran 90/OpenMP solver pipeline. This dataset is explicitly published to serve as high-resolution training, verification, and benchmark data for Machine Learning (3D CNN), deep learning property prediction, and custom FE solver verification. CALL FOR COLLABORATION The optimized mesh generation code can output alternative particulate arrangements on demand in 1–2 minutes and solve in 6-8 minutes. We actively welcome data-driven machine learning and mechanics research groups to reach out via the contact information in the manuscripts to collaborate on custom training sets (e.g., bi-modal distributions, varying aspect ratios, or custom volume fractions).