We experimentally study particle scale dynamics during segregation of a bidisperse mixture under oscillatory shear. Large and small particles show an underlying asymmetry that is dependent on the local particle concentration, with small particles segregating faster in regions of many large particles and large particles segregating slower in regions of many small particles. We quantify the asymmetry on bulk and particle scales, and capture it theoretically. This gives new physical insight into segregation and reveals a similarity with sedimentation, traffic flow, and particle diffusion.

We experimentally investigated how a binary granular mixture made up of spherical glass beads ͑size ratio of 2͒ behaved when flowing down a chute. Initially, the mixture was normally graded, with all the small particles on top of the coarse grains. Segregation led to a grading inversion, in which the smallest particles percolated to the bottom of the flow, while the largest rose toward the top. Because of diffusive remixing, there was no sharp separation between the small-particle and large-particle layers, but a continuous transition. Processing images taken at the sidewall, we were able to measure the evolution of the concentration and velocity profiles. These experimental profiles were used to test a recent theory developed by Gray and Chugunov ͓J. Fluid Mech. 569, 365 ͑2006͔͒, who derived a nonlinear advection diffusion equation that describes segregation and remixing in dense granular flows of binary mixtures. We found that this theory was able to provide a consistent description of the segregation/remixing process under steady uniform flow conditions. To obtain the correct depth-averaged concentration far downstream, it was very important to use an accurate approximation to the downstream velocity profile through the avalanche depth. The S-shaped concentration profile in the far field provided a useful way of determining what we refer to as the Péclet number for segregation, a dimensionless number that quantifies how large the segregation is compared to diffusive remixing. While the theory was able to closely match the final fully developed concentration profile far downstream, there were some discrepancies in the inversion region ͑i.e., the region in which the mixing occurs͒. The reasons for this are not clear. The difficulty to set up the experiment with both well controlled initial conditions and a steady uniform bulk flow field is one of the most plausible explanations. Another interesting lead is that the flux of segregating particles, which was assumed to be a quadratic function of the concentration in small beads, takes a more complicated form.

Optical measurement techniques such as particle image velocimetry (PIV) and laser Doppler velocimetry (LDV) are now routinely used in experimental fluid mechanics to investigate pure fluids or dilute suspensions. For highly concentrated particle suspensions, material turbidity has long been a substantial impediment to these techniques, which explains why they have been scarcely used so far. A renewed interest has emerged with the development of specific methods combining the use of iso-index suspensions and imaging techniques. This review paper gives a broad overview of recent advances in visualization techniques suited to concentrated particle suspensions. In particular, we show how classic methods such as PIV, LDV, particle tracking velocimetry, and laser induced fluorescence can be adapted to deal with concentrated particle suspensions.

We experimentally investigated the spreading of fluid avalanches (i.e., fixed volumes of fluid) down an inclined flume. Emphasis was given to the velocity field within the head. Using specific imaging techniques, we were able to measure velocity profiles within the flowing fluid far from the sidewalls. We studied the behavior of Newtonian and viscoplastic fluids for various flume inclinations and initial masses. For the Newtonian fluids tested (glycerol and Triton X100), we compared the measured velocity field with that predicted by lubrication theory. Provided that the flow Reynolds number Re was sufficiently low (typically Re < 1), there was excellent agreement between theory and experiment except for the very thin region just behind the contact line. For higher Reynolds numbers (typically Re ∼ 10), the discrepancy between theory and experiment was more marked (relative errors up to 17% for the body). As viscoplastic materials, we used Carbopol ultrez 10. For the body, agreement between theoretical and measured velocity profiles was fairly satisfactory whereas it was very poor for the tip region as the curvature of the free surface became more pronounced: the velocities were not only much lower than those predicted by lubrication theory, but there was also evidence of slipping in the flow part adjacent to the contact line.

No abstract