Nupehewa, J, Palmer, J, Suvio, P, Koponen, V & Safonov, D 2023, 'Developing predictive empirical filtration models for advanced tailings handling', in GW Wilson, NA Beier, DC Sego, AB Fourie & D Reid (eds), Paste 2023: Proceedings of the 25th International Conference on Paste, Thickened and Filtered Tailings, University of Alberta, Edmonton, and Australian Centre for Geomechanics, Perth, pp. 324-338, https://doi.org/10.36487/ACG_repo/2355_25
(https://papers.acg.uwa.edu.au/p/2355_25_Nupehewa/)
Abstract: There has been a significant improvement in tailings dewatering techniques over recent years. However, the effects of tailings properties on the filtration process have not been vastly investigated. The different properties of tailings cause significant effects on the feasibility of filtration. Modifying or optimising filtration equipment for different tailings mixtures often requires testing procedures that involve high expenses. Therefore, an increasing demand has arisen to improve the prediction of tailings properties for efficient thickening and filtration using the available mineralogy and particle size data. Developing a suitable solution to do so is the focus of this paper. Experimental studying of tailings and their fractions helps to understand the empirical relationship of the physical properties of tailings, such as particle size distribution (PSD) and air permeability, to the filterability properties like average cake porosity and cake resistance. Also, it is vital to study how the changes of fine particle concentration affect these parameters of the tailings mixture. An algorithm developed based on the filterability of different particle size fractions of chosen minerals should be able to predict the filtration rate for a user-defined tailings blend. This type of model will be useful in evaluating the performance and economics of different tailings treatment models and studying the feasibility to produce tailings disposal solutions. It was discovered that separating fine fractions from tailings, significantly improves the filterability of the remaining portion. This opens up several further possibilities for advanced tailings handling systems. One possibility is to perform cost-efficient tailings filtration for the coarser fractions of the tailings and keep the fine fractions as slurry, which then mix with the filtration cake to form paste for surface disposal or backfill. This approach potentially allows mining companies to achieve paste rheology at lower opex and capex compared to the conventional paste thickener technology.
During the study a set of laboratory experiments were conducted to fraction the tailings and determine the empirical relationships between the physical and filterability properties of each fraction and their different mixtures. Development of a filterability parameter prediction model with the PSD data and known parameters of original tailings fractions allows the possibility of predicting new, untested materials. The accuracy of the predictions depends on the degree of similarity between the new material and the original tailings material used for empirical study in this stage. By inputting the filterability parameters, suspension properties and operation conditions, the developed filtration models are able to predict the filtration process parameters such as total filtration time, final filtration volume and final cake thickness, etc. The outcome of a validated predictive filtration model can be utilised to trade-off between filtered, thickened, paste and combined treatment of tailings by modifying the tailings feed.

Keywords: tailings filtration, tailings dewatering, particle size distribution, tailings fractioning, filterability properties, predictive empirical filtration models, advanced tailings handling