Heterogeneity detection of void structure in discrete particle assemblies: Case-based analysis on asphalt mixtures utilizing digital image processing

Discrete Particle Assemblies (DPAs), such as Hot Mix Asphalt (HMA), display notable performance variability, often driven by internal void structure heterogeneity. This study aims to develop a practical and applicable framework for quantifying and analyzing structural heterogeneity in HMA voids. To this end, high-resolution 3D digital image processing was employed to extract and reconstruct void structures from a randomized asphalt mixture database. Subsequently, void morphology, spatial distribution, and volumetric classification are quantified using statistical, fractal, and multifractal descriptors. Parameters such as void ratio, Homogeneity Index, and Shape Index are used to characterize global and local heterogeneity, while multifractal spectrum metrics provide deeper insight into geometric complexity. Finally, the correlation between particle interference and structural heterogeneity was analyzed. The results confirm that the void structure exhibits significant heterogeneity. Specifically, the coefficient of variation for Statistical Descriptors can reach about 30 %, and the proportion of outliers can be as high as 7 %. For Fractal Geometry-Based Descriptors, these values are even larger. Furthermore, particle interference was found to greatly increase void heterogeneity by altering void shape. Therefore, controlling particle skeleton features can serve as an effective way to regulate void structure heterogeneity, providing theoretical support for improving the engineering reliability of granular materials.