Gravity Separation and Desliming of Fine Coal

Experiments were performed to test whether two Reflux Classifier units in series are able to effectively clean a feed containing -2.0mm coal. The first unit, called the "gravity unit", used narrow 6mm wide channels to perform a density separation. The overflow from this unit contained coal contaminated with fine high-ash slimes. This was fed into the second unit, called the "deslime unit", which used wider 12mm channels to perform a size classification to remove these slimes. Experiments were performed at laboratory scale using a pair of units with cross-sectional area of 100mm× 80mm and also at pilot scale at a coal preparation plant using a pair of Reflux Classifiers with 600mm× 600mm cross-sectional area.

In general high quality gravity separations were achieved at both laboratory and pilot scale down to particle sizes of 0.100mm, with some density separation evident down to sizes as low as 0.038mm. Low set point densities were needed in order to ensure fine mineral matter reported to the first stage reject. This approach improved the quality of the subsequent desliming. The performance of the laboratory and pilot scale deslime units were similar, indicating that full-scale desliming units can be confidently designed based on laboratory trials. The cut size (S50) varied linearly from 0.05 to 0.35mm with increases in the overflow channel velocity from 10 to 60m3/(m2h). The Ep values increased from 0.02 to 0.08mm (Whitten factor α from 2 to 8) over the same range, increasing scatter in the results as the overflow channel velocity increased. The linear dependence of the S50 on velocity in the Reflux Classifier was consistent with the theory, and with the significant throughput advantage of the technology. The gravity-deslime combination was able to reliably produce clean coal products, provided the gravity separation was run at a low set point.

It has previously been established that gravity separation in the Reflux Classifier proceeds at a high throughput. In contrast desliming is governed by the volumetric throughput, hence low throughputs are essential when targeting relatively fine desliming. In the laboratory tests, the highest volumetric feed rate used was around 50m³/(m²h) while at pilot scale the work was limited to less than 40m³/(m²h). The implication with respect to the effect of the solids throughput on desliming at different sizes is discussed.