Digital volume correlation for meso/micro in-situ damage analysis in carbon fiber reinforced composites

Recently, in-situ X-ray Computed Tomography (CT) has shown its potential for 3D damage analysis in composite materials. However, the characterization of damage in X-ray tomograms is not always straightforward, and it requires post-processing of the 3D images. In this study, we explore the potential of Digital Volume Correlation (DVC) for the detection and characterization of damage in fiber-reinforced composites, where fibers provide the required 3D speckle pattern. Preliminary analysis via digital deformation of 3D images is performed to verify the applicability of DVC for quantification of deformation and damage in a carbon/epoxy laminate and to estimate the measurement errors. Then, real-deformation images, acquired with synchrotron CT during in-situ tensile loading of the laminate, are analyzed with DVC to detect different damage mechanisms. A rough analysis is performed at the mesoscale using subset-based DVC, followed by a more detailed investigation at the microscale via finite-element-based DVC. Damage appears in the DVC strain fields as local strain magnification. Crack opening displacement can be estimated reliably via the jumps in displacement fields. DVC proves to be a promising tool for damage characterization in X-ray tomograms of fiber-reinforced composites, especially when simple methods such as grayscale thresholding are not adequate.

Automated summary

This study explores the potential of Digital Volume Correlation (DVC) in detecting and characterizing damage in fiber-reinforced composites using in-situ X-ray Computed Tomography. Through preliminary analysis of digital deformation images and real-deformation images acquired during in-situ tensile loading, DVC proves to be a promising tool for quantification of deformation and damage at both mesoscale and microscale levels. Damage mechanisms in fiber-reinforced composites are successfully detected and characterized using DVC, showing the limitations of traditional methods like grayscale thresholding.