Numerical Study of Quiescent and Sheared Sedimenting Suspensions

Citation

Chan, Hong-Man (1988) Numerical Study of Quiescent and Sheared Sedimenting Suspensions. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/c07k-ws42. https://resolver.caltech.edu/CaltechETD:etd-11062007-132325

Abstract

In this thesis we reported the results of our investigation into the behavior of particles settling in a quiescent and sheared suspensions using a molecular-dynamics-type simulator. The effects of varying the concentration over the range of 0.0025 to 0.05, and the shear rate between 0.0 and 20.0 on the average particle velocity, the diffusion coefficient, the variance of the particle velocity, the average particle-distribution function, the particle-distribution-correlation time, and the average velocity and variance autocorrelation times are reported. We found that the first correction to the sedimentation velocity for concentration (c) depended upon c rather than c^1/3, and that no structure with a length scale of O(ac^(-1)/3) developed in the particle-distribution function. Shearing the suspension caused a nonuniform particle-angular distribution to form, and the particle correlation times and the diffusion coefficients to decrease. Using the diffusion coefficient from the simulations, we modelled the spreading of the interface between the supension and the clear fluid region. We found that the spreading of the interface could cause misinterpretation of the experimental measurements of the sedimentation velocity, and that this might explain the discrepancies among the experimental data.

In building the simulator, we needed to determine what effects N, the number of particles used in the simulator, have on the simulation results. Neither the sedimentation velocity, nor the average particle-distribution function changed when N varied from 27 to 125. However, we found that both the variance of the particle velocity and the diffusion coefficent grew with N. We have shown that for an homogeneous suspension, these statistics will converge, although the number of particles needed may still be very large, especially in the low concentration range which we are interested in.

Thesis Files