University of British Columbia
Data-driven processing–property–defect relationships in the forming of flax fibre composite reinforcements
Abstract
dc:descriptionNatural Fibre-Reinforced Composites (NFRCs) offer lightweight and sustainable alternatives for industrial applications including automotive. However, similar to synthetic composite options, forming-induced defects such as wrinkles can reduce NFRCs’ performance. Predicting these defects is particularly challenging for natural fibre reinforcements due to the inherent variability in fibre geometry and properties. This study applies a data-driven approach to predict wrinkle formation in flax and glass composite reinforcements. The fabrics’ shear, bending, tensile, and friction behaviours were experimentally characterised to capture forming-relevant mechanical properties. The fabrics were subsequently formed in multi-layer configurations over tetrahedron and square tools, followed by 3D scanning to quantify wrinkle distributions. Unsupervised data-driven methods were used to identify processing–property–defect relationships. Structural response metrics (surface distortion, volume distortion, wrinkle index, compliance) and image-texture descriptors (contrast, dissimilarity, energy, homogeneity) were calculated from post-forming surface maps to quantify wrinkle severity. These descriptors, together with fabric design parameters (fabric weave, orientation, layers, grammage, and thickness), were used to train Linear Regression (LR), Partial Least Squares (PLS), and Artificial Neural Network (ANN) models. Results showed that flax twill reinforcements exhibited the best forming performance, followed by the biaxial fabrics, while the glass fabric exhibited the poorest performance under both the tetrahedron and the square forming tools. For the square geometry, a 45° orientation maximised in-plane shear accommodation and minimised wrinkle formation across all woven reinforcements. This effect was most pronounced for the flax twill with 367 gsm areal weight, which demonstrated the best overall formability due to its balance of bending and shear stiffness. ANN models were able to capture nonlinear relationships between input material properties and wrinkle severity descriptors, achieving R² ~ 0.90 for SD, VD, and dissimilarity, compared to R² ~ 0.70 for LR and PLS models. The dissimilarity index showed the most consistent predictive performance and was able to capture both defect severity and fabric architecture effects, with lower values for glass twill reinforcements and higher values for the flax non-crimp fabric reinforcement. Overall, integrating structural and image-texture descriptors with process parameters and mechanical properties provides a promising machine-learning framework for rapidly predicting NFRC forming quality.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy - PhD
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Mechanical Engineering
- Grantor dc:publisher
- University of British Columbia
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Margoto, Olivia Helena
- Contributors dc:contributor
-
- Milani, Abbas S.
- Abdin, Yasmine
Rights
dc:rights- Statement dc:rights
-
- Attribution-NonCommercial-NoDerivatives 4.0 International
- Language dc:language
- eng
Identifiers
dc:identifier.*- Handle dc:identifier
- http://hdl.handle.net/2429/95645