The effect of the particle size distribution on non-Newtonian turbulent slurry flow in pipes

Abstract

The handling of solid-liquid suspensions is an important concern within the chemical and processing industries and many theoretical models have been proposed to try and explain and predict turbulent flow behaviour. However, the prediction of turbulent flow from only the viscous properties of non-Newtonian suspensions has over the years been questioned by researchers. This thesis considers theoretical models well established in the literature and the Slatter model, which uses both the rheology of the suspension and the particle size distribution of the solids. These models are used to analyze the experimental data and the effect that particle size and the particle size distribution has on turbulent flow behaviour. The literature concerning the rheological fundamentals relevant to fluid flow in pipes has been examined. The Newtonian turbulent flow model as well as the non-Newtonian models of Dodge & Metzner, Torrance, Kemblowski & Kolodziejski, Wilson & Thomas and Slatter have been reviewed. Test work was conducted at the University of Cape Town's Hydrotransport Research Laboratory using a pumped recirculating pipe test rig. The test apparatus has been fully described and calibration and test procedures to enable collecting of accurate pipeline data have been presented. Three slurries were used in test work namely kaolin clay, mixture I (kaolin clay and rock flour) and mixture 2 (kaolin clay, rock flour and sand) with ad,s particle size ranging from 24/Lm to 170/Lm. The yield pseudoplastic model has been used to model and predict the laminar flow of the suspensions that were tested and the meth9J adopted by Neill (1988) has been used to determine the rheological constants. The pipeline test results have been presented as pseudoshear diagrams together with the theoretical model lines providing a visual appraisal of the performance of each model. The Slatter model predicts the test data best with the other theoretical models that were considered tending to under predict the head loss. The reason the Slatter model performs better than the other theoretical models is because this model can account for the wall roughness and particle roughness effect. Evidence to support this statement has been presented. This thesis highlights the fact that the particle size distribution is a vitally important property of the suspension and that it does influence turbulent flow behaviour. It shows that turbulence modelling using the particle roughness effect (eg Slatter, 1994) is valid and can be adopted for non-Newtonian slurries. It is concluded that the particle size distribution must be used to determine the particle roughness effect and this effect must be incorporated in the turbulent flow analysis of non-Newtonian slurries.