Characterizing aggregate shape evolution during mechanical abrasion using digital image analysis and multivariate analysis of geometric shape descriptors

Abstract Aggregate shape is a key factor influencing the performance of aggregates used in pavement, concrete, and geotechnical applications. However, the progressive evolution of aggregate morphology under mechanical abrasion and the ability of commonly used shape descriptors to capture these changes remain insufficiently understood. This study investigates abrasion-induced shape evolution using digital image analysis and multivariate statistical techniques. A total of 39,600 crushed rock particles, ranging from 5 to 30 mm, were subjected to controlled milling at eleven abrasion levels from 0 to 2000 revolutions, and fourteen geometric shape descriptors were quantified. The results showed a systematic transformation from angular and irregular particles to smoother, more rounded, and compact morphologies with increasing abrasion. Correlation analysis revealed redundancy among several descriptors, indicating opportunities for descriptor reduction. Effect-size analysis identified convexity as the most sensitive descriptor, followed by circularity, form factor, sphericity, and shape factor. Multivariate analysis revealed a dominant morphology-evolution pathway and a distinct transition zone between 300 and 1000 revolutions, where the most significant shape transformations occurred. Beyond this range, changes became progressively less pronounced, indicating shape stabilization. The findings provide a practical basis for selecting effective shape descriptors for aggregate morphology characterization and quality assessment.