Centrifugal pump derating non-Newtonian slurries: analysis of the viscosity to be used in the hydraulic institute method

Abstract

Centrifugal pumps are the most commonly used pumps in slurry transport systems. The design of pumping systems dealing with liquids more viscous than water requires a reliable method of pump performance prediction for the pump selection. For Newtonian fluids, the Hydraulic Institute method is well established, but there is no generally accepted method for non-Newtonian fluids. Many authors have fallen back on using the Hydraulic Institute method for non-Newtonian fluids. This requires a constant viscosity while non-Newtonian fluid viscosity varies with the shear rate. The question arises: What viscosity should be used in this method for non-Newtonian fluids? Two approaches have been developed: the use of a Bingham plastic viscosity made by Walker and Goulas (1984) and the use of the apparent viscosity calculated using an “equivalent hydraulic pipe” diameter, designed by Pullum et al. (2007). Previous results obtained from these two approaches are not in agreement. Therefore, the aim of this study is to explore a suitable procedure to determine a representative non-Newtonian viscosity to be used in the Hydraulic Institute method to predict the pump performance. To achieve this goal, a set of data was experimentally obtained and the existing data were reused. Test work was conducted using the pump test rig in the Flow Process Research Centre at the Cape Peninsula University of Technology. A Warman 4/3 pump was tested, using four concentrations of kaolin suspension and three concentrations of CMC solution. Five pump speeds were chosen to run these tests: 1200, 1400, 1600, 1800, and 2000 rpm. An additional data set obtained by testing two submersible centrifugal pumps with eight concentrations of sludge, in Stockholm, Sweden, was also analysed. These sets of data were analysed firstly according to the Walker and Goulas (1984) approach and secondly according to the Pullum et al. (2007) approach. The use of the apparent viscosity led to the better pump head prediction. The results of this prediction were close to those obtained in the Pullum et al. (2007) work, and even better in some cases. On the other hand, the use of the Bingham plastic viscosity showed better pump efficiency prediction, although the Walker and Goulas (1984) efficiency prediction range was achieved only for one pump out of five. The apparent viscosity reflected the non-Newtonian behaviour but it could not represent alone the non-Newtonian viscosity because of the poor efficiency predictions and the sensitivity of the Pullum et al. (2007) approach to a change in viscosity. From the results of this work, it is advisable that the pump performance prediction be done using both apparent and Bingham plastic viscosity, the apparent viscosity for the head prediction and the Bingham plastic viscosity for the efficiency prediction.