Optimisation of thickener performance: incorporation of shear effects

© 2017 Dr Adam CrustDewatering processes such as sedimentation and consolidation of colloidal suspensions are important to a wide range of industries, including waste water and mining. Within the mining industry, massive quantities of tailings comprising mineral suspensions are processed daily. The ability to effectively dewater tailings plays a major role in the successful operation of the process. Modern dewatering theory developed by Buscall and White (1987) can predict the dewatering behaviour of such suspensions. Combining modern dewatering theory with a phenomenological model can be utilised in the design, operation and understanding of a dewatering process. A lab scale batch settling test can be accurately predicted through the combination of modern dewatering theory and developed phenomenological models. For a mineral suspension settling within a laboratory batch settling test, the combination of modern dewatering theory and phenomenological model can accurately predict the result of the settling test. However, using the same approach, the ability to accurately predict the full scale performance of gravity thickeners (Usher and Scales 2005, Usher et al. 2009, Zhang et al. 2013, Grassia et al. 2014) eludes. Instead, current full scale gravity thickener models result in an under estimation of thickener solids flux by a factor up 100 (Usher 2002). Shear effects, introduced by raking within a thickener for example, are known to affect the settling of suspensions; however, these have not been completely accounted for within current phenomenological models. Shear alters the structure of a flocculated aggregate and subsequently alters the dewatering behaviour of the overall suspension. This effect, known as aggregate densification, can be attributed to the enhancement in dewatering observed when a suspension experiences shear. Aggregate densification theory has previously been partially included into thickener models (Usher et al. 2009, Zhang et al. 2013, Grassia et al. 2014); however these models assume time independent densification and do not account for the dynamic nature of the process. The dependencies on densification parameters have not been critically studied. This thesis investigates the role of shear effects on suspension dewatering. The key aim of this research is to quantify the effect of shear on dewatering and in particular, the enhancement of dewatering in a thickener induced through shear. The research objective is to develop a process, in which simple batch settling tests can be performed, the results analysed and utilised to predict the full scale thickener performance that correctly takes dynamic shear effects into account. Aggregate densification theory was applied to sheared laboratory batch settling scenarios to quantify the effect of shear on material property characterisation. Sheared laboratory batch settling tests were performed using a polymer flocculated calcite as a representative mineral suspension and further analysed using novel analysis methods based on modifications to current techniques such as predictive modelling and densification analysis developed by van Deventer (2012). Batch settling tests were performed to investigate the relationship between aggregate densification parameters and both network stress and the magnitude of shear. Manipulating the shear magnitude showed no significant trend in the extent of densification however trends were observed in the rate of densification. The rate of aggregate densification increased with shear up until a critical shear value. Above this critical shear value, the rate of densification was observed to be constant. It is believed that this critical shear value corresponds to the minimum requirement for particles to collide such that below this value, there is insignificant kinetic energy to cause any noteworthy deformation to the aggregates. The effect of network strength on aggregate densification parameters was also investigated. It was observed that a flocculated suspension with sufficiently high solids concentration such that the aggregates interact and become networked, densify a further 5% compared to shearing un-networked aggregates. In terms of thickener performance, raking the networked solids resulted in an order of magnitude increase in thickener throughput. Raking within a thickener is currently performed at the base of the thickener in order to aid transport of the suspension. This increase in thickener throughput due to raking the networked suspension is therefore already implemented however unaccounted for within current thickener models. The implementation of these findings into thickener models has the potential to account for the current underestimation. To completely exploit the phenomenon of aggregate densification, the results of the work suggest that the shear within the system needs to be at or above a critical value and that a network bed must be present, maintained, and raked. Current one-dimensional (pseudo two dimensional) steady state thickener models were modified so as to account for the time dependent nature of aggregate densification. This exercise involved combining the theories of aggregate densification; sedimentation and suspension bed consolidation. The model uses material dewatering properties obtained from laboratory batch settling tests and thickener operational parameters as inputs to produce steady state solids flux predictions for a range of underflow solids concentrations. Modelling scenarios involved manipulating thickener operational and aggregate densification parameters. The results indicate that an order of magnitude increase in thickener performance can be achieved when the shear rate is above the critical value determined during sheared laboratory batch settling tests. Additionally, model applications to real systems have been discussed and methods of utilisation to achieve process optimisation have been demonstrated. As a result of this experimental and modelling work, a new method for full scale thickener modelling from simple laboratory batch settling tests has been developed. This method incorporates the rate dependence of aggregate densification and hence provides increased accuracy in the estimation of thickener performance compared to previous models. It is important to note that this newly developed model has a few assumptions and limitations. Assumptions and limitations include line settling, negligible wall effects, all aggregates are equal, no solids exit the overflow, steady state operation, straight walled, equal distribution of shear and no aggregate breakage. Although some assumptions are inherent to the model being one dimensional it is suggested that further work goes into addressing such issues as particle and aggregate size polydispersity and aggregate breakage in order to further increase the accuracy of the model. Additionally, the development of a transient model in which shear history is included is recommended