Abstract
This work assessed the potential of a single stage Reflux Classifier to upgrade itabirite iron ore to high grade at satisfactory recovery. Part I reported on the detailed findings of the experimental program and the physical transport of the particles through the system. A key purpose of the present paper was to deduce the underlying partition surface from relatively basic feed information on the Fe assays obtained as a function of the particle size. Conversion of the feed data into a simple binary description based on the density of hematite and density of silicates was used. This approach then provided a basis for applying the partition surface to a given feed to predict the separation performance. Data from the experiments were compared to values predicted from the partition surface. A least squares objective function was used to deduce the parameters governing the partition surface. Across 12 of the experiments, the key exponent, n, governing the partition surface was found to be 0.26+/-0.02. This result was in very good agreement with the value of 0.28 determined previously. The second key parameter, the Ep, was also determined for each of the experiments. The lowest Ep, found to be 365 kg/m3 for a low slimes viscosity, was also in good agreement with the result reported previously for a deslimed feed. This work provides confidence in the application of the partition surface to predict similar dense mineral separations, and stronger insights into the mechanisms responsible for the separation.