This paper presents a comprehensive study on the statistical characteristics of angle, tortuosity, curvature and wave magnitude to deepen the understanding of realistic fiber misalignments within unidirectional composites. A feasible fiber path reconstruction procedure has been optimized, which can be applicable to other types of composites. The high-resolution micrographs are acquired through X-ray computed tomography. The individual fiber segmentation is implemented using a U-Net deep learning method, and the fiber trajectories are reconstructed with the aid of a tracing algorithm. The stepped fiber trajectories are slightly smoothed and a polynomial fitting formula is adopted to quantitatively describe the fiber paths. The statistical characteristics corresponding to the differential tortuosity, misalignment angle, spatial curvature and wave magnitudes are comprehensively analyzed, with emphasis on their fitting distributions and scanning length effects. The collected data indicate that the statistical distributions of differential tortuosity and angle, curvature, and wave magnitude can be well fitted by normal, lognormal and Weibull equations, respectively. Particularly, the differential tortuosity and wave magnitude are overall features of individual fiber trajectory, which are highly correlated with the scanning length. In contrast, the angle and curvature are local features, so a smaller scanning length could obtain convergent results.
Read full abstract