In order to develop mathematical models a small industrial scale (700mm lip length, 900mm diameter) wet drum magnetic separator was installed in a purpose built experimental rig. Initial tests indicated that with this rig the process outcomes of interest, magnetics recovery and concentrate solids content, could be determined with very good reproducibility. A campaign of factorial experiments, done to determine the major response trends, were followed by a campaign of incremental experiments done to examine the response to particular variables over a wider range. A substantial data set of 191 trials has been generated which was used for model development and validation.
Observations carried out as an adjunct to this work, as well as magnetic theory, suggests that the capture of magnetic particles in the wet drum magnetic separator is by a flocculation process. Thus a model based on this concept and containing adjustable parameters was developed. This model was then fitted to a randomly chosen 80% of the data, and validated by application to the remaining 20%.
It was further proposed that the concentrate in the separator is dewatered by a drainage process, which is defined by an initial moisture and a drainage rate. Again a model based on this concept, also including a model of residence time in the drainage zone of the separator, and containing adjustable parameters was developed. This model was then fitted to a randomly chosen 80% of the data, and validated by application to the remaining 20%.
Both models were found to be good fits to the data, and most of the error in the models could be explained in terms of experimental error in determining each process outcome. The applications of both models in simulation either of an individual magnetic separator or of a circuit containing primary and secondary separators with classification of primary tails was also demonstrated.