Figure 10. Comparison of the solid fraction from experiments and DEM simulations.

Videos taken during the shear tests (available in the supplementary information) reveal drastically different macroscopic flow behavior for the monodispersed and binary mixtures. For the monodispersed smaller sized beads, a bulk motion of the system, with coordinated movement of particles in a wave-like fashion is observed. For the monodispersed larger size beads, a start-stop and rotational motion of individual particles is observed within the bulk. However, these individual movements of large particles are not observed in the video with 0.50 large glass bead mass fraction, suggesting that at lower concentrations, small particles hinder the ability for the larger particles to rotate. The 0.75 system exhibited the highest degree of segregation of all the mixtures tested. For the 0.75 system, in regions of low levels of particle segregation, the macroscopic flow behavior of the binary mixture resembles that of the 0.5 system. However, in regions of high levels of particle segregation, with a higher concentration of large glass beads near the surface, rotation of individual large glass beads is observed. Time-lapse images of the video for the 0.75 system are extracted and shown in Figure 11. In the images, a few particles are outlined to indicate their movement with time. From these images, it is evident that the highlighted large glass bead is rotating. As the larger glass bead rotates, multiple small glass beads on the surface of the larger particle can be identified. This behavior indicates that the addition of small glass beads, even at low concentrations, can adhere to the larger glass beads.

While spinning movement is observed for the large particles, the small particles move as a bulk in a wave-like motion. Furthermore, for the 50:50 binary mixture, the larger particles show no evidence of spinning movement, and the bulk flow is similar to that of the small particles. Therefore, an explanation for the unique shear stress trend is proposed based on this observed macroscopic flow behavior. The hypothesis is that large particles can spin, rather than move transversally, minimizing the bed dilatation during shearing. And, the addition of smaller, more cohesive particles and the adhesion between large and small particles minimizes or prevents the spinning of the large particles. This, in turn, results in a larger shear stress of the binary blend required to induce flow.