Insights from dynamic models for an online flotation pulp sensor

Abstract:

It is well known that pulp-phase dynamics strongly impact flotation performance, but until recently online measurements of pulp behaviour were not possible. Stone Three have developed such an online pulp sensor (OPS) that performs in-situ machine vision to generate pulp measurements. A comparison of steady-state data from this sensor and the Anglo Platinum Bubble Sizer has been published, showing that the online pulp sensor produces measurements that are responsive to relevant process changes. These results highlighted that the exact calibration of the OPS requires more work. In this paper, the analysis is extended to empirical dynamic modelling of the OPS parameters, such as bubble size and superficial velocity, as functions of inputs including airflow, pulp level and reagent flows. 

The dynamic models provide good predictions of the bubble parameters and give insight into the effects of the input parameters on the intrinsic variables affecting the flotation performance. The dependence of concentrate grade and recovery on pulp parameters has been modelled for plants with online analysis. Where assays are used for grade and recovery analysis, online information on these performance indicators are not readily available. The models suggested in this work may prove valuable to link online pulp parameters to key performance indicators of the cell or bank, and these measurements may then be used for advisory systems and eventually closed-loop control.