It is well known that classification performance influences grinding efficiency. Typically classification performance is assessed using the overall stream particle size distributions. However, most ores contain both liberated particles of specific densities, and composite particles with a range of intermediate densities. The dense liberated particles are most likely to continue to report to the underflow long after they have been reduced to the required size. This unnecessary recycle to the mill can lead to over- grinding. If the dense particles reporting to the underflow are the intended target of recovery, they may be too small to be recovered easily, while if they are gangue material they may report to the concentrate due to entrainment. Thus in these multi- component feeds classification is influenced not only by particle size, but also by particle density and shape, and an overall classification curve gives a misleading picture of classification performance.
Current hydrocyclone models described in the literature fail to take into account the range of particle densities present in a realistic feed. Ideally, the models should predict the partition curves for each significant density class in the feed, given the cyclone geometry and operating conditions. Additionally, the models should take into consideration the potential for particle interaction, expected to vary with cyclone operating conditions.
This thesis describes an experimental investigation to test two hypotheses: 1) different density components do not behave independently during classification, even when they are fully liberated, and 2) the operating conditions of the hydrocyclone influences the degree of component interaction. The investigation comprised pilot-scale testwork using selected mixtures of two components having different particle densities. The results suggested that both hypotheses are true.
A model for individual component behaviour was developed based on the experimental data. The equations developed for cut size show that the behaviour of both components are driven by both per cent solids and the interaction between the proportion of the light component and the per cent solids present in the feed. The difference in the behaviour lies in the additional terms present in the equation for the low density component d50c, representing flow rate and a size distribution term for the low density component.
Based on the trends seen in the experimental work and the literature dealing with particle behaviour inside a hydrocyclone it is proposed that the mechanism by which the observed component interaction occurs is particle displacement in which there is a shift in the d50c of the light component becoming larger with the addition of dense material of the same size distribution due to the hydrocyclone classifying based on the immersed mass of the particles. In order to investigate the particle displacement mechanism, an experimental method, Positron Emission Particle Tracking (PEPT), was investigated that had the potential to determine the movement and position of a particle inside a hydrocyclone containing slurry. PEPT has previously been used to determine particle behaviour in tumbling mills containing slurry but has never been used to track the behaviour inside a hydrocyclone operating with slurry. The work provided qualitative information about the movement of a particle in slurry inside a hydrocyclone.