The classifying cyclone, which was first patented in 1891, has played a significant role in the coal industry during the past 30 to 40 years. It is well regarded due to the relatively low capital, operating costs and its versatility.
Even with the large amount of models available, the way in which a cyclone operates is still not completely understood. Studies in the area of coal modeling are limited as the majority of the models were developed for mineral applications. This thesis investigates some of the common models developed over the years.
The thesis experimental work was undertaken at Collinsville Coal, a Thiess operation. Currently, there is no full-time metallurgist on site to conduct the required surveys and as a result the company is experiencing some operating problems with the primary classifying cyclones. Previous test work, mainly plant audit information, does not provide much insight into the physical and operating parameters of the cyclones.
It was proposed that a constant feed from one stockpile be run through the plant producing coking coal while the primary classifying cyclones were tested. In order to obtain the best results the plant feed and operating parameters were maintained at constant levels wherever possible.
The sampling campaign was focused on testing a number of variables that were easily adjusted such as vortex finder and spigot diameter. The pressure was also used as a variable by adjusting the number of operating cyclones; larger numbers of operating cyclones gives a lower pressure.
For the analysis of the data it was proposed to use at least two modeling methods for each parameter of cyclone efficiency including the cut-point, sharpness of separation and solid and water recovery.
The fitted models (compared to the predictive models) produced results that were closer to the actual values for the test work.
A larger vortex finder diameter increases the solids recovery to underflow and the cut-size. The water-split decreased as the diameter increased. All of these theories follow the literature. The test work on the vortex finder wall thickness was inconclusive due to inconstancies in the data. A larger spigot diameter indicated an increase in recovery of solids to the underflow, cut–size and water split. The recovery of solids to the underflow is the only parameter that followed the literature. A rise in the pressure showed an increase in the recovery of solids for the underflow, which is supported by the literature. The cut-size and water recovery were inconclusive relative to the pressure changes.
Based on the test results, it is recommended that Collinsville should operate with a spigot size between 3.4 to 4.4 cm. Less than 3.4cm produces roping while exceeding 4.4cm produces a large diameter of spray around the cyclone. An 8 cm vortex finder performed more effectively than the 10 cm for a larger range of conditions. The thickness of the vortex finder wall had very little effect in the cyclone operation.