The effect of shear on dewatering of flocculated suspensions

© 2005 Dr. Brendan Robert GladmanThe ability to separate a suspension into its respective solid and liquid constituents is an important requirement in the chemical, wastewater and mineral industries. Typically, separation occurs in open, large diameter tanks known variously as thickeners, settlers or clarifiers. The design and operation of these devices have been based, until recently, on kinematic models and macroscopic mass balances. The problem with these approaches is that consolidation in the bed is not described accurately and consequently, the area required for thickening is often grossly overestimated. Recently, Buscall and White [24] proposed a 1−D phenomenological theory of dewatering that encompasses both sedimentation and consolidation, providing a more solid grounding for understanding, simulating and optimising dewatering in a range of devices, including thickeners. This theory identifies two important rheological parameters; a concentration dependent yield stress, Py (φ) and hindered settling function, R(φ). Despite representing a significant improvement over a kinematical approach, Buscall and White’s dewatering theory involves a number of simplifications so that in practise, simulations often underestimate dewatering in full sized thickeners [97, 153]. One aspect of thickening that is poorly understood is the effect of raking. At the base of the thickener, a rake transports the thickened sediment to the outlet. An additional effect from raking is to increase the average solid concentration in the underflow [33, 46]. Raking introduces normal and shear stresses, which cannot be described within a one-dimensional framework. Therefore, observed differences between predicted and measured thickener underflow concentrations are attributed to the action of the rake. The aim of this thesis is to develop a better understanding of how shear stresses effect compressional dewatering in both pilot and full scale thickening operations. Before attempting to quantify the effect of shear on dewatering, it was considered necessary to first establish that the 1-D model was capable of predicting dewatering in the absence of shear. Up until now, no known studies have been undertaken to validate the model under controlled conditions. To approximate one-dimensional flow with no shear, a tall narrow column with no moving parts was used. Two solid fluxes and several bed heights were studied, and the outputs from the column were compared with the 1-D model predictions. The results show that under ideal conditions, the model predicted the underflow solid concentration to within 10 %. The effect of shear on dewatering was investigated using a Couette shear device. Couette geometry was chosen to provide uniform shear. Since in Couette flow, no normal stresses act in the direction of rotation, the mechanism behind dewatering can investigated. These experiments showed that shear caused dewaterability to improve up to a critical shear rate, beyond which dewaterability was adversely affected. The relationship between this critical shear rate and flocculation conditions was investigated by using different flocculant dosages. The shear modified Py (φ,γ) and R(φ,γ) can be input to the 1−D model, thereby incorporating shear indirectly. As a result, the model predicted an order of magnitude increase in solids flux. The above procedure was used to characterise the dewaterability of a real thickener feed as a function of shear rate. The optimum shear rate was determined by finding the minimum R(φ,γ). Then, Py(φ) and R(φ) were input into the thickener model. The predicted underflow concentration could then be compared against plant data. Even when shear is taken into account, the model still under predicts the performance of the thickener. To understand this result, the pilot column work was revisited since the control over experimental conditions was far greater. To introduce shear, concentric cylinders were installed in the column and rotated at a fixed speed. Thus, the effect of shear and bed height on underflow density were determined at different rates of shear. This showed that the underflow concentration increased with bed height; a result not expected based on the model prediction. The effects of shear on underflow density were secondary to bed height. The bed height dependence can only be explained if dewatering is not steady but changes over time. For a four metre bed height the residence time is eight times longer than a one metre bed. Improvements in dewatering could be related to time dependent restructuring of aggregates which would result in an associated change in R(φ). By fluidizing suspensions for times corresponding to the residence times in the tall column, R(φ) and Py(φ) could be determined, as functions of volume fraction and time. Aggregate properties including structure and density were measured before and after fluidization using focussed beam reflectance measurement (FBRM) and floc density analysis (FDA)