A supervised iterative approach to 3D microstructure reconstruction from acquired tomographic data of heterogeneous fibrous systems.

Graphical abstract Highlights • A new technique to reconstruct fiber architecture from tomography is presented. • Determines 3D fiber architecture from supervised reconstructed stacks of 2D data. • Approach is suitable for complex morphologies and high volume fractions of fibers. • Technique enables fiber tracking to incorporate with in situ mechanical loading. Abstract The recent ubiquitous utilization of short fiber reinforced composites (SFRCs) in applications that require complex shape conforming, light-weight materials with good strength properties, calls for in-depth studies to understand the underpinning physics behind SFRCs response to load and consequently damage. Since the accuracy of such studies is contingent on successful sub-volume characterization, the intricate sub-volume architecture of SFRCs requires conscientious methodology that addresses complex morphologies like fiber cross-overs, which is by no means a trivial endeavor. This paper proposes a novel framework that hinges on the synergy between robust 2D segmentation and 3D volume reconstruction techniques to faithfully reconstruct the fiber architecture of 3D X-ray tomograms of SFRCs. The implications of this framework not only include a platform that fully characterizes the complex sub-volume, but also provides a convenient means of incorporating tracking algorithms necessary for the in-situ characterization of the reconstructed fibers, if desired. [ABSTRACT FROM AUTHOR]