Laboratory Methods for Characterising the Floatability of Ore

Flotation circuits are often modelled mathematically so the process can be analysed and optimised quickly and inexpensively. In these models, the propensity of mineral particles in a slurry to be collected by attachment to air bubbles is termed floatability. In Floatability Component Models, the ore is represented by a number of discrete floatability components, each with a mean flotation rate constant. Multiple Stream Batch Testing (MSBT) is widely used to generate the data for modelling flotation circuits. The method is empirical, requiring a plant survey and laboratory batch tests to be carried out on the streams of the industrial plant. To design new plants, there is a need to characterise the floatability of ores before they are processed in a flotation plant. Towards this end, two laboratory batch flotation tests were developed in this study: one involving a series of batch tests performed at different air flow rates (the Changing Air Flow Rate technique, CAFR), and one involving the reflotation of batch test concentrates (the Reflotation of Batch Concentrates technique, RFBC). Both the CAFR and RFBC techniques were evaluated theoretically and experimentally. In each case, the technique results were modelled and the derived ore floatability parameters were compared with those obtained using the MSBT method (which was used as a benchmark). For the theoretical analysis, a measured floatability distribution obtained from the literature was used to generate synthetic data for the CAFR, RFBC, and MSBT techniques. A flotation model was built from each of the three data sets. The parameters derived from the RFBC and CAFR models were unique and statistically stable and the same as those derived from the MSBT method. When tested experimentally at three mine sites, the parameters derived from the RFBC and CAFR models were unique and statistically stable. Upon comparison with the MSBT method, the results were inconsistent as the parameters were not the same in all cases (within experimental error). Further analysis indicated, for the RFBC technique, that this was due to ageing during the batch tests. For the CAFR technique, this was due to some doubt as to whether the shape of the floatability distribution was conserved with a change in air rate. Although the parameters derived from the different methods were statistically different, they were of similar magnitude. A sensitivity analysis was undertaken to determine if the parameters derived from the CAFR models were sufficiently close to the MSBT parameters to predict the same simulation results. (A sensitivity analysis was not conducted with the RFBC parameters due to problems with the experimental data.) The results showed that the model predictions were sensitive to arbitrary changes in the input parameters. However, the parameters derived from the CAFR models predicted the same performance as the original parameters derived from the MSBT models. This indicated that the two methods produced similar overall ore floatabilities. Therefore, it was concluded that the CAFR parameters were accurate enough to produce the same model predictions as the MSBT parameters. The CAFR technique produces flotation rate constants at the laboratory scale. At this scale, the technique can be used as a comparison tool to evaluate the floatabilities of different ores, to compare the effect of different reagent suites on a given ore, or to evaluate the impact of changing the grind of a given ore. However, the rate constants need to be scaled to produce a model that can predict full scale plant performance. A method of scaling the rate constants from laboratory to industrial sized plant cells needs to be developed to improve the accuracy of the simulations conducted using parameters derived at the laboratory scale. The aim of developing a method to characterise ore floatability offline is to apply the technique in the laboratory and then be able to scale the resulting parameters to predict performance in either brownfield or greenfield operations. This research is a first step in the complex process of achieving this goal