A methodology for gangue management in the flotation of a PGM-bearing ore through laboratory tests, mineralogical analysis and circuit modelling

Abstract

With the ever-increasing demand for sustainability in the mineral processing industry, many operations are focusing on minimising their energy usage and maximising their smelter capacity. One approach to achieving this, is by ensuring concentrates fed to smelters do not contain excess unwanted gangue minerals inadvertently recovered during flotation. In the case of the South African platinum group mineral (PGM)-bearing UG2 chromitite ores, the flotation feed mostly consists of hydrophilic gangue minerals, chromite, pyroxene and plagioclase, with talc being the only floatable gangue mineral at concentrations of approximately 1 wt%. Therefore, it is to be expected that the mechanism of recovery for the two most common gangue minerals, chromite and pyroxene is by entrainment only, whereas talc may report to the concentrate by true flotation in the absence of sufficient quantities of depressant. However, evidence from a Western Limb UG2 concentrator indicated that large quantities of pyroxene were being recovered that could not be accounted for by entrainment alone.To further explore this observation, a continuous laboratory hybrid flotation cell was used in a series of experiments to generate deep froths that could mimic plant-scale entrainment functions. Using quantitative mineral analysis, material flows from the hybrid cell could be quantified in terms of their mineralogy and particle characteristics. Subsequent decoupling of these into entrained and floatable fractions showed that about half of the gangue in the concentrate reported there by true flotation and half by entrainment. The large amount of pyroxene reporting by true flotation was found to be due to finely disseminated talc throughout the pyroxene particles. A subsequent full-scale survey of the actual flotation plant and resulting mass balance and model of the flotation plant, resulted in simulations that could be used, in conjunction with the laboratory-based studies, to predict the effects of process changes. These simulations predicted a four-fold improvement in grade, together with a similar reduction in mass pull. Implementation on site resulted in improvements that were very close to those predicted by the simulations. This study serves to illustrate an important methodology that can be utilised to reduce mass pull, improve concentrate grade and reduce environmental footprint.