Abstract
The purpose of this work is to find an effective image segmentation method for lab-based micro-tomography (µ-CT) data of carbon fiber reinforced polymers (CFRP) with insufficient contrast-to-noise ratio. The segmentation is the first step in creating a realistic geometry (based on µ-CT) for finite element modelling of textile composites on meso-scale. Noise in X-ray imaging data of carbon/polymer composites forms a challenge for this segmentation due to the very low X-ray contrast between fiber and polymer and unclear fiber gradients. To the best of our knowledge, segmentation of µ-CT images of carbon/polymer textile composites with low resolution data (voxel size close to the fiber diameter) remains poorly documented. In this paper, we propose and evaluate different approaches for solving the segmentation problem: variational on the one hand and deep-learning-based on the other. In the author’s view, both strategies present a novel and reliable ground for the segmentation of µ-CT data of CFRP woven composites. The predictions of both approaches were evaluated against a manual segmentation of the volume, constituting our “ground truth”, which provides quantitative data on the segmentation accuracy. The highest segmentation accuracy (about 4.7% in terms of voxel-wise Dice similarity) was achieved using the deep learning approach with U-Net neural network.
Highlights
The development of realistic models of woven composite material for meso-scale simulations has grown significantly over the last decades
Existing approaches for generation of the material finite element model can be grouped roughly in three categories: (1) creation of idealized geometry based on statistical data; (2) extraction of geometry from the result of a numerical simulation; (3) development of a model based on real micro-tomography (μ-CT) images
It is not prohibitive to use such segmentation for the further processing, because the sharp edge regions can be naturally reconstructed on the meshing stage in case of the correctness of tows geometry
Summary
The development of realistic models of woven composite material for meso-scale simulations has grown significantly over the last decades. Existing approaches for generation of the material finite element model can be grouped roughly in three categories: (1) creation of idealized geometry based on statistical data (e.g., analytical representation of a tow surface); (2) extraction of geometry from the result of a numerical simulation (e.g., simulation of fabric compaction); (3) development of a model based on real micro-tomography (μ-CT) images. The resulting volume of idealized tows is often less than the real one. Increasing the fibers volume fraction within a tow is used to preserve correct overall material properties, which in turn can have an undesirable effect on the simulation result [3]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have