Coal Preparation » Gravity Separation
Vast quantities of fine coal, worth hundreds of millions of dollars, are disposed as tailings from many coal preparation plants. Often this coal has not responded well to flotation and may have suffered oxidation. Or, the cost of installing flotation is prohibitive. Gravity separation exploits differences in particle density hence any surface oxidation of the particles is of no concern. Further, gravity separation is potentially more economic than flotation, but the applicable size range of conventional gravity based technologies does not extend effectively to below 0.10 mm. The Graviton can in principle lower this limit down to 0.010 mm, thus revolutionising fine coal processing.
The objective of this study was to develop and test a pilot‐scale version of a continuous steady state separator consisting of REFLUX™ Classifiers mounted within a high speed centrifuge. This device is known as the REFLUX™ Graviton. Previous laboratory‐scale work (ACARP Project C19039) showed that the benefits of the inclined channels in the REFLUX™ Classifier multiply with the benefits of operating with high G‐forces, thus generating a capacity advantage within the inclined channels of more than 1000 fold over conventional settling.
The first stage of the project was conducted using fine silica powder in order to investigate the effect of parameters such as feed and fluidisation rates, without any complications due to variations in particle density. Sharp separations with Ep values of less than 10 µm and separation sizes d50 in the range 10 µm to 20 µm were routinely obtained across a wide range of operating conditions, with feed flowrates over 100 L/min (6 m3/h) and feed pulp densities up to 30 wt.% solids. Overall, the silica work has shown that this system delivers strong systematic separations at relatively fine sizes.
The silica program was followed by tests on hydrocyclone overflow coal normally discarded to tailings. About 40 wt.% of this feed consisted of +0.020 mm well‐liberated coal with less than 10 wt.% ash, however the remaining 60 wt.% of ‐0.020 mm slimes was over 60 wt.% ash. Hence the separations focussed on simple desliming in the first instance. As a hydraulic device, the separations involved a strong dependence on both particle size and density. Experiments were performed at flow rates up to 150 L/min (9 m3/h) and feed pulp densities up to 9 wt.% solids. Here the size partition curves are really composite curves from each particle density, hence the composite Ep values were much larger than for silica. Despite this, strong evidence of desliming of the feed was achieved, though not to the extent required for exploitation. A single stage deslime reduced the ash from 42 wt.% to 31 wt.% at a yield of 47 wt.%. Therefore two stage investigations were conducted. One pair achieved a final product ash of around 24 wt.% ash, with an overall two‐stage yield of 39 wt.% and combustible recovery of 50 wt.%.
Further work was undertaken to achieve a stronger gravity separation, the goal being to remove the higher density mineral matter in the first stage. A high feed rate was applied in the first stage to promote shear induced lift in the inclined channels to help convey the coal to the overflow, leaving the larger and denser mineral matter behind to discharge as underflow. The overflow was then reprocessed in order to achieve more effective desliming in the absence of the coarser mineral matter. The second stage underflow product was then further processed in a third stage to remove the remaining slimes. It is suspected that the feed rate applied to the first stage gravity separation may have been too high, limiting its effectiveness. The net result was therefore just a repeat of the earlier desliming work, coupled with an initial yield loss from the first stage gravity separation. More work is recommended to examine this question.
This project has demonstrated at pilot scale the ability of the REFLUX™ Graviton to separate fine particles over a wide range of conditions. Clearly the silica based suspension separated with high efficiency. However, with the coal, the subtle variation in density of the feed produced inefficiencies. For example, the very lowest density coal particles tended to report with the slimes to the overflow, which is why those particles existed in the cyclone overflow from the plant in the first place. The rest of the fine coal and higher density particles reported to the underflow of the Graviton as product. While this effect was small, its impact on the overall product ash was significant. We are convinced the Graviton offers significant potential, but this depends on identifying the right problem. For example, flotation concentrate could be processed at a relatively low rate, achieving efficient upgrading through desliming, with significant reduction in the product ash.
The Graviton system was derived from a modified fine coal centrifuge, and hence there were some compromises in the design that was used. When the overflow emerges from the Graviton, and flows towards the floor of the unit, splashing is known to occur. This splashing causes some material to by‐pass and report with the underflow via the underflow launder.
Significant effort went into minimizing this effect, however, some by‐pass continued to contaminate the product. It is estimated that for feed rates of 100 L/min this by‐pass causes up to 5% of the ultrafine particles to report to the underflow.
Clearly, further design work is required to improve this system and prevent this by‐pass. Redesign of the fluidisation chamber could also reduce the direct entrainment of ultrafine particles to the underflow. Other work should focus on scale‐up. It has been shown that the Graviton can beneficiate particles down to 0.010 mm in size at flow rates of up to 150 L/min (9 m3/h) when using two small modules. A full‐scale centrifuge with larger G forces due to a higher speed and larger diameter, 8 modules per layer, 3 layers of modules, and wider 2 mm channels, would permit a 100 fold scale‐up to 900 m3/h. It is unclear whether all of these changes could be introduced. However, any increase in the feed rate would require redesign of the feed entrance.