CT Scan Material Field Extraction, Voronoi Meshing and Harmonic Basis Function FEM of 3D Woven Composites

Fibre-reinforced composites (FRCs) are ubiquitous in mass-critical applications like aerospace and wind turbines due to exceptional specific strength and stiffness. However, FRCs exhibit highly anisotropic properties, with strength and stiffness maximized along the reinforcing fibre directions. 3D woven architectures in offer improved forming and increased toughness over traditional multi-layer FRCs, though challenges remain in their modelling, manufacture and validation. This thesis presents a number of novel contributions for the industrial production of these composites.

The Computed Tomography Wavelet Volume Orientation (CT-WVO) method presented extracts orthotropic material orientations from CT scans efficiently. Tests on synthetic datasets demonstrated high accuracy; for cross-ply simulations, 83\% of vectors were recovered within 1.5 degrees error, outperforming traditional tow-tracking analyses in speed while operating on first principles without black-box weights.

A Voronoi-based meshing algorithm segments heterogeneous domains into self-similar regions, producing quality meshes devoid of sliver triangles. Evaluated via intra-element variance as a proxy for discontinuity isolation, the algorithm performed 4–6 times better than voxel meshes with comparable degrees of freedom. This facilitates accurate analysis of complex woven architectures where interlacing tows follow non-straight paths and cross-sections.

Lastly, a novel finite element formulation extends Harmonic Basis Function (HBF) elements with polynomial refinement to ensure continuity within heterogeneously refined meshes. The pipeline was validated against experimental measurements using seven 3D-woven specimens produced per ASTM D7264B four-point bending standard. Bulk flexural moduli estimated via the proposed pipeline converged within 5\% of experimental values in the linear elastic regime, verifying the model's predictive capability.

These findings contribute to efficient design and quality assurance of 3D woven composites, with applications extending to various orthotropic structures beyond fibre-reinforced materials. The algorithms can replace traditional methods piecewise, or as a vertically integrated FEA pipeline.