Flow of non-newtonian fluids in open channels

Abstract

Flume design for homogeneous non-Newtonian fluids is problematic and not much research has been conducted in this field. This application is industrially important in mining where slurries have to be transported to processing or disposal sites at higher concentrations because water is becoming a scarce and expensive commodity. This thesis addresses the problem of flume design and develops predictive models for the laminar, transitional and turbulent flow behaviour of non-Newtonian fluids in rectangular open channels. The relevant literature pertaining to Newtonian and non-Newtonian pipe and open channel flow is reviewed and research aspects are identified. A unique test facility was designed, constructed and commissioned for this project. The facility includes a 5 m-long by 75 mm-wide rectangular tilting flume, as well as a 10 m by 300 mmwide rectangular tilting flume that can be partitioned to form a 150 mm wide flume. The flumes are in series with an in-line tube viscometer which has tubes of diameter 13, 28 and 80 mm. The experimental investigation covers a wide range of widths (75 mm-300 mm), slopes (1º-5º), flow rates (0.05 l/s-45 l/s), relative densities (1.0067-1.165), volumetric concentrations (0%-10%), and yield stresses (0-21.3 Pa). The fluids tested are kaolin and bentonite slurries and CMC and Carbopol polymer solutions. The resulting database of empirical flow behaviour enabled the identification of the important flow behaviour characteristics. Existing models are compared and evaluated using the experimental database compiled for this thesis and it is concluded that no model exists to predict the database compiled for the various materials from laminar flow through the transition region into turbulence. For the correlation of laminar flow data, a Reynolds number was developed from the Reynolds number proposed for pipe flow by Slatter (1994). Using this Reynolds number, all the laminar flow data available was collapsed onto the 16/Re line on a standard Moody diagram. Criteria were developed to predict the onset of transition and the onset of ‘full turbulence’. These criteria are functions of the Froude and Reynolds number as well as the viscous characteristics of the fluids. These models performed better than the methods proposed by Naik (1983) and Coussot (1994), which were based on the Hanks criterion. A turbulent flow model was developed based on the turbulent model presented by Slatter (1994) for pipe flow. Flow predictions using this model were more accurate than those presented by Torrance (1963), Naik (1983), Wilson and Thomas (1985), and Slatter (1994). The new models were tested with the database compiled for this thesis as well as with two published data sets, one by Naik (1983) and the other by Coussot (1994). The new flow models predicted all the available data within acceptable limits, providing a basis for design. A new and experimentally validated design protocol is presented for the design of rectangular non-Newtonian open channel flow in laminar, transitional and turbulent flow.