Investigation of particle breakage along phase transition points using the particle partition potential (P3) and the loading intensity (LI) factor
摘要
This study investigates the impacts of shearing and volume change on the evolution of particle size distribution in silica (Ottawa) and calcareous (Fiji Pink) sands, during the direct shear test. Results reveal Fiji sand's higher shear resistance and extensive grain crushing compared to Ottawa sand, due to differences in mineralogy and grain shape. The findings provide insights into granular soil behavior under increasing stress denoted as phase transition points that naturally occur with increasing shear. It was observed in the particle size distribution (PSD) of both sands that grain crushing initially increases the shear fraction of fines but then gradually reduces the percentage of larger grain diameters. The analysis of PSD evolution during shear was conducted using Hardin's Br, Lade's B10, and Marsal's Bm breakage index frameworks. The micro- and macro-mechanical aspects of direct shear show that to model shear effects on grain breakage, the effects of dilation (− dεv/dεh), internal grain friction (μ), and vertical effective stress (σ′v) conditions must be considered. The combination of these effects leads to the proposition of the particle partition potential (P3) parameter, which represents the average stress conditions acting on sand during shear. P3 can be computed from stress, strain, and volumetric change data only, but it nevertheless shows a strong linear correlation with Hardin’s Br parameter, which is measured by comparing pre- and post-test PSDs. Particle breakage was also related to the loading intensity (LI) parameter which combines the magnitude of force chains formed within particles and the duration of loading. P3 is linearly correlated with both P3 and Br, thus pointing to the effectiveness of LI in quantifying particle breakage across different sands. Finally, three additional datasets from the literature were used to calculate LI and P3 from stress–strain data and compute Br. The computed and measured Br values were well correlated with R2 = 0.91.