Modelling and Control of Classifying Cyclones

Hydrocyclones are one of the key technologies for the classification of fine coal. However, there have been limited advancements in modelling and simulation of hydrocyclones in the last 20 years. This project was concerned with the potential of using Coal Grain Analysis (CGA) and vibration monitoring to improve understanding and performance monitoring of hydrocyclones.

One of the drawbacks of using hydrocyclones for size separation is that they separate on a combination of drag and particle density rather than just size. As a result, higher density particles will have a lower cut point than lower density particles, retaining low value clay materials and discarding higher value coal. The degree to which this cut point changes with density is not well understood; currently for modelling theory, there are two schools of thought with respect to the impact of particle density, and they use either 0.5 or 1 as 'n', the exponent for the density correction term. However, conventional float sink, and sizing technologies have limitations for determining this exponent at the small particle sizes involved.

CGA is an optical technique by which the size, shape and maceral composition of particles can be estimated under a microscope. From the maceral breakdown of each particle, the ash and density of the overall particle can be estimated. By taking samples of the cyclone feed, underflow and overflow, partition curves for individual density fractions can be determined to allow the cut point shift with density to reviewed.

This work was carried out in a pilot plant setting with a single coal feed type collected from the desliming cyclone feed pipework of a western NSW coal washery. This site was selected because CGA had been carried out successfully in the past on a similar size fraction, and the material was found to have a reasonable size and density distribution.

The project concluded:

• CGA can be used to develop sizing and density curves at a fraction of the cost and effort that conventional float sink testing would involve.
• Preliminary evidence and model verification suggests that the current cut point vs density relationship used to simulate cyclones overstates the impact of density.
• Partition curves for individual density fractions were steeper than the overall partition curve.
• The particle aspect ratio appears to influence cyclone performance and CGA can be used to evaluate the impacts of this.
• Vibration monitoring appears to be able to detect roping cyclones, but in-plant trials are required to confirm the difference in vibration is not lost in general plant vibration.
• CGA analysis is not currently at a mature enough stage for routine analysis of in plant cyclone performance, but reductions in turn around time and resolving aggregate issues may change this.