Predictive modelling of mineralogical and textural properties of tailings using geochemical data

Tailings deposits pose a significant threat to the environment and/or contain residual metal contents that may be of economic interest. The mineralogical and textural properties of the tailings dictate both processing behaviour and potential for acid mine drainage. Automated Scanning Electron Microscope-based image analyses enable valuable quantitative mineralogical and textural data to be obtained. Such methods are, however, time and cost intensive. This study investigates the extent to which geochemical data can be related to mineralogical and textural data of tailings materials. Models based on the dominant components (PCs) from principal component analysis were tested for mineral abundance, particle and mineral grain size distributions (PSD, GSD), and degree of liberation. Sedimentary-style deposition of the tailings is represented by the main trend between PCs 1 and 2. This being the dominant process in the tailings, PC 1 is a robust estimator for mineral contents, PSD and mineral GSDs. Degree of liberation is less well constrained by the PCs. This study clearly demonstrates that prediction of mineralogical and textural parameters of tailings with geochemical data is possible. Considering that the dominant processes should be the same in similarly deposited tailings deposits, these findings are widely applicable.