Fluid Mechanics of Solid-Liquid Suspensions – An Experimental and Numerical Study

Doktorsavhandling, 2005

Multiphase flows are common in many applications in chemical, mineral, biochemical and several other process industries. Examples of equipment in which multiphase flows are encountered are stirred vessels and multiphase jets, in which suspensions are agitated by an impeller in the former case and distributed through a nozzle in the latter. To understand the hydrodynamics of multiphase flows, measurements and theoretical models are needed to form a base on which conclusions about several aspects of the properties of such flows can be drawn. However, performing measurements and numerical simulations on multiphase systems is a delicate procedure and entail several difficulties that must be overcome before reliable data can be acquired and conclusions can be drawn. This thesis concerns measurements as well as numerical simulations of multiphase flows in a stirred vessel and a confined jet. The measurements are performed using Laser Doppler Velocimetry (LDV) and Particle Image Velocimetry (PIV), and the numerical simulations are performed using the granular Eulerian/Eulerian multiphase methodology. To be able to access and perform measurements of each phase in multiphase systems, a method using refractive index matching is developed. In the measurements performed using PIV, image analysis is used to separate the two phases while, in the case of LDV, features inherently present in the LDV technique are used to accomplish the separation. In the experimental studies, the maximum volume fractions of solids at which measurements can be performed differs between the setups and measurement techniques. In the case of a stirred vessel, the maximum mean solid content is 1.5% when using PIV while 9% when measuring the continuous phase using LDV. In the confined jet the maximum volume fraction is 1.9% when using PIV. The method of using image analysis in combination with refractive index matching, used in the experiments performed with PIV, works well but is sensitive to the quality of mixture composition and surface quality/material homogeneity of the secondary phase. The phase separation is also accompanied by several parameters and settings that have to be determined with frame-to-frame examination. The technique used in the LDV measurements similarly works well and allows for measurements at higher concentrations at the cost of measurement time. From the measurements of stirred vessels it is found that the mean velocities of both phases decrease while the RMS velocities increase with increasing solid content. It is also found that the RMS velocities increase with increasing particle size. The numerical evaluations of the stirred vessel, performed using the Sliding Mesh methodology, reveal good general agreement with experimental data, but discrepancies, increasing in magnitude with increasing solid content, are found regarding various aspects of the flow field. Examples are the impeller jet, where the predicted jet is wider than the measured counterpart and the wall region where the predicted axial velocities are generally lower in magnitude compared to experimental data. The predictions of the mean axial slip velocities are in accordance with measurements but the ones acquired from simulations are generally larger in magnitude than their measured counterparts. The experimental and numerical evaluations of the confined jet show that the differences in axial velocity between the two phases are small and that the axial RMS velocities generally increase with increasing volume fraction. Additionally, the RMS velocities for the dispersed phase are generally larger than those of the continuous. The numerical simulations captures the hydrodynamics fairly well, but the continuous phase centreline velocities are underpredicted close to the inlet and overpredicted further away, except at high concentrations where they are underpredicted throughout the region. No conclusion can be made regarding the accuracy of the two drag models used, as each performs better than the other at different volume fractions and locations.