Quantifying Intra-Tow Fiber Volume Fraction in GFRP: A Comparison of 3D Non-Destructive X-ray Computed Tomography and Destructive Optical Microscopy

Accurate fiber volume fraction characterisation is crucial for modelling the permeability of the glass fiber reinforced fabrics used in manufacturing of wind turbine blades. In this paper, we analyse a specific sample of UD glass fiber with different non-destructive and destructive imaging modalities, and compare their ability to characterize the intra-tow Fiber Volume Fraction (FVF) accurately. The images are segmented using a thresholding method, and the FVFs are estimated for an area common to all the imaging modalities. Following this approach, the local FVF is also analysed for a larger Field of View (FOV) for three selected imaging modalities, capturing nearly an entire glass fiber bundle. The estimated local FVFs from the low-resolution X-ray CT do not match the Synchrotron X-ray CT local FVF distribution, and subsequently, a generative adversarial network model is applied to enhance the information in the low-resolution X-ray CT. The resulting methodology of mapping the low-resolution images to high-resolution images of glass fibers through neural networks is demonstrated to allow for a better estimate of the local FVFs with the faster imaging modality.