Automated reconstruction and conformal discretization of 3D woven composite CT scans with local fiber volume fraction control

Abstract

An ad-hoc methodology for voxel-based geometry reconstruction is proposed. The reconstruction procedure outlined is fully automated and allows transforming explicit voxel RVE geometries into implicit smoothed geometries through a level set-based processing. During this post-processing, the yarns interpenetrations are suppressed, a thin gap is inserted between the contacting yarns and the fiber volume fraction is controlled locally by modifications of the yarns cross sections shapes. A conforming tetrahedral mesh can be subsequently generated based on the implicit geometrical description for finite elements simulations. Puck failure indices are used here to compare the damage initiation locations in the microstructure. Results show that damage initiates early and is rather randomly distributed in voxel-based discretizations, whereas it occurs more in the yarns contacting zones and in the higher fiber fraction volume regions for the smoothed RVE.

A review of applications of CT imaging on fiber reinforced composites

“…XCT provides an opportunity for a detailed examination of the inner structure of fiber reinforced composites, providing details of the composite microstructure, defect population, and damage distribution. [122][123][124] 3D images obtained by micro-CT can be used for a realistic modeling of composite materials. A range of researchers have predicted the properties of fiber reinforced composites based on the numerical models from XCT.…”

“…A direct method based on X-ray tomography imaging has been widely used to determine FE models based on the real geometry of 3D reinforcements. [122][123]129 The separations of the yarns in different directions were usually specialized based on the homogeneity parameter. XCTbased meso-FE simulation of the transverse compaction of a 3D orthogonal reinforcement had been introduced using a hyperplasic model developed for fiber bundles.…”

“…3D solid tow generated by analogous algorithms represented geometrical models for more detailed failure analyses. Also, a voxel-based geometry reconstruction based on 3D woven composite CT scans had been proposed by B. Wintiba et al 122 It transforms explicit voxel RVE geometries into implicit smoothed geometries through a level set-based processing in Figure 13. During the post-processing, the yarns interpenetrations are suppressed, a thin gap is inserted between the contacting yarns and the fiber volume fraction is controlled locally by modifications of the yarns cross-section shapes.…”

“…Voxel-based models. (a) Voxel geometry Represent Volume Elements, (b) Post-processed Represent Volume Elements 122.…”

Geometrical and deep learning approaches for instance segmentation of CFRP fiber bundles in textile composites

“…The structure tensor approach is a frequently used method for local fiber orientation analysis. The method results in images that are then used for an intensity-based segmentation [12][13][14]. However, a non-gradient based method to analyze the fiber texture is employed in work [15], where the Grey Level Co-occurrence Matrix was used to analyze the texture of µCT images of composites.…”