Mathematical models have been developed which can predict mass recovery in high-intensity crossbelt and induced magnetic roll (IMR) separators. These models are based on fundamental principles which are used to determine a critical susceptibility for recovery at a crossbelt pole or IMR splitter. The models have been designed for incorporation into a computer simulator being developed at the Julius Kruttschnitt Mineral Research Centre for the mineral sands industry.
A standard feed characterisation technique has been devised to determine the feed specific parameters for the models. This is used to quantify the distribution of magnetic susceptibility in the feed to these separators and makes the models general or non site-specific in nature.
Fieldwork was conducted around seven different industrial applications of these separators in mineral sands dry plants in Australia. At these sites the effect of machine design and operating variables on separator performance was studied. The data were used in the development of the models and for obtaining values of machine specific model parameters using least-squares fitting techniques.
The crossbelt model contains one fitted parameter, referred to as the gap constant, for each recovery stage. This constant is a function of crossbelt pole geometry and pole air gap. Machine design and operating variables such as feed rate, feed temperature, main belt speed and flux density are explicitly incorporated in the model. The efficiency of recovery of magnetic particles at each pole is calculated using a probability expression based on the number of particle layers on the main belt. The model was fitted to all machine survey data and in all cases was able to reproduce the observed mass recovery data within the uncertainty associated with the measured variables.
An expression defined as the lambda function was developed to define the magnetic force index at all points in the field of an IMR separator and is the key feature of the IMR model. The function contains three parameters which are dependent on pole geometry, rotor laminae dimensions and air gap. Variables such as roll speed, splitter position, feed temperature, feed rate and flux density are explicitly incorporated in the model. The parameters in the lambda function were fitted using machine survey data and the calculated size-by-size mass recoveries were compared with observed values. The model was able to reproduce these data over the operating range of splitter positions but could not do the same over a wide range of feed rate values. This is thought to be due to the effect of interparticle forces not considered in the model. In practice it is not a significant shortcoming of the model since feed rate is rarely used as a control variable in these separators.