Search for a command to run...
• Micro-computed tomography of Syntactic foams made from high-density polyethylene (HDPE) reinforced with hollow glass microspheres (HGM) • Automated K-means clustering algorithm for segmentation, was employed to characterize the internal microstructure of additively manufactured HDPE syntactic foams at various infill densities. • The automated segmentation method successfully differentiates the matrix material from the reinforcing particles and voids, allowing for a precise evaluation of the phases. Syntactic foams made from a high-density polyethylene (HDPE) matrix reinforced with hollow glass microspheres (HGMs) are used in weight-sensitive applications given their desirable mechanical performance and low density. However, accurately measuring the internal microstructure features, especially void and particle volume fractions, is compromised by the inadequacy in standard image segmentation methods. In this research, microcomputed tomography (µCT) imaging, in conjunction with a K-means clustering algorithm for segmentation, was employed to characterize the internal microstructure of additively manufactured HDPE syntactic foams at various infill densities, allowing for the quantification of void and particle volume fractions. The segmentation method successfully differentiates the matrix material from the reinforcing particles and voids, allowing for a precise evaluation of the phases. The results reveal a steady increase in Φ void from 3.28 % up to 20.98 % upon a reduction in infill density, with Composition A having 55.8 vol% HGMs of 0.14 g/cm 3 particle density, exhibiting lower void content compared to Composition B, which contains 37.6 vol% HGMs with a density of 0.32 g/cm 3 . Validation against manual segmentation using LabKit yielded an average F1 score of 0.926, demonstrating high segmentation accuracy for identifying voids. Additionally, the analysis distinguishes between internal HGM voids and raster-induced matrix voids, a crucial distinction in understanding porosity formation during additive manufacturing in syntactic foams. The resolution limitation of µCT imaging, 4.8 µm, poses challenges in accurately resolving the thin walls of HGMs of wall thickness 0.37 µm, leading to underestimation of the particle volume fraction. The results provide a quantitative framework for evaluating the microstructural characteristics, contributing to the optimization of syntactic foams for advanced engineering applications.