The rate variable batch test (RVBT) - A research method of characterising ore floatability

As part of the AMIRA P9 project, a methodology has been developed by which a comprehensive flotation model can be built from survey data around an existing plant, combined with laboratory batch flotation tests carried out on samples of a number of the circuit streams. The output of the model, in terms of ore floatability, is a discrete number of floatability classes for each mineral of interest, each with its own first order rate constant, and each comprising a certain mass fraction of the feed. Currently, this methodology must be applied to an operating plant during ore treatment. The ambit of the modelling methodology would be greatly extended if the floatability of an ore could be characterised before it is processed in a flotation plant. Developing a laboratory procedure to predict the fl oatability parameters will allow for better planning and forecasting of the fl otation performance of a given ore. The Rate Variable Batch Test (RVBT) method is such a procedure, and is the subject of this paper. The RVBT method involves floating a number of identical samples of an ore in a laboratory batch cell at different input air fl ow rates. It was developed and tested at three porphyry copper ore mine sites in conjunction with surveys of the operating plants to enable the floatability parameters obtained from the RVBT procedure to be compared with those derived from the current best practice method. The results indicated that the models developed from the RVBT data were able to represent the experimental data well and the derived kinetic parameters were statistically reliable; even though these parameters were not identical to those derived from the operating plant models. Nevertheless, a sensitivity analysis showed that the model predictions (mass, recovery, grade) were not overly sensitive to changes in the derived input parameters, i.e., for the same ore, the RVBT parameters predicted final grades and recoveries comparable to those predicted from a model developed using the current method