The behaviour of electrostatic separators in the mineral sands industry has long been considered a "black art". Although a great deal is understood about the physical processes involved in the separation, little is known about how these processes interact in a dry plant environment.
This thesis has developed mathematical models of high tension roll (HTR) and electrostatic plate separators (ESP) based on extensive sampling of plant equipment combined with laboratory-scale testwork and particle charge measurements. Variables investigated ranged from atmospheric conditions to machine settings and feed mineral behaviour.
Machine separation behaviour has been quantified by the mass distribution of mineral along the splitter axis, termed "Product Spatial Distributions" (PSDs). The random nature of the separation process demands that a probability
distribution of mineral behaviour be included to account for variations in particle surface condition, shape, orientation, etc.
Each feed mineral is characterized by an average charge relaxation time reflecting its relative surface conductivity. This is combined with a variance term to generate the PSD for each size fraction of mineral. The separation forces are calculated from the machine settings which, combined with the mineral characterizations, determine how each component will behave in the separator.
The models predict the mass recovery of each mineral component to all products by the superimposition of splitter(s) onto the PSDs. These component mass flows are combined to determine the product size and assay distributions. The models can be combined to simulate the performance of any number of separation stages from a single machine to an entire dry plant electrostatic circuit.
The models are fully compatible in their input/output formats to allow the integration of the two types of machines in the simulated circuit. These models have been incorporated into a commercial simulator software package - JKSimSand.
Although the models are intended primarily for the optimization of machine performance and circuit design, they may also be used for the training of operating personnel and to evaluate machine designs. Further work to extend usage into process control applications and expert systems is also possible.