Dry Classification of Fine Coal

The classification and beneficiation of coal with spirals or cyclones inherently involves high water consumption. The issue of water supply, particularly in arid areas, is further complicated by problems associated with waste tailings generation and disposal. The development of alternative dry processing methods would alleviate these water consumption and tailings disposal issues. This study examines the application of dry beneficiation technology to separation of coal and mineral matter samples using a novel gas-solid fluidised bed.

The innovative device, known as the Reflux Classifier (RC), combines a conventional fluidised bed with a set of parallel plates. An internal particle recycling mechanism develops from the continual upwards fluidisation of particles into the inclined channels formed between the plates, and the downwards segregation of the particles via the inclined surfaces. The self-recycling effect helps prevent misplacement of particles during the classification process and consequently improves the separation quality. The focus of the study is on evaluating the potential of the device for pneumatic size classification and gravity beneficiation of nominally -2mm coal particles. The study also examined particles up to 4 mm in size.
 
Size classification in the RC relies on the complex particle elutriation mechanism where high gas flow rates initially launch particles into the inclined channels. Here, the coarser particles segregate from the gas flow and return to the bed, whereas the finer particles are entrained by the gas and removed from the system. Hence, relatively higher flow rates are capable of suspending larger particles which, in turn, can be transported from the apparatus.

Using compressed air as the fluidisation gas, a series of flow rates were studied to examine the size classification of -2 +0mm coal samples in the RC. The underflow and overflow samples were collected and the partition data calculated, in turn the equilibrium size separation cut point (S50) was determined. A data set is shown in Figure 1. The RC generated separation cut points ranging from 0.06 - 1.1mm for the 0.903 – 2.917 m/s flow rates investigated. The imperfection (I = (S75-S25)/2S50) of each classification experiment was also determined using the partition data. This study has shown that a typical separation cut point of 0.25mm was achieved with an imperfection value of I = 0.12. Generally, the imperfection term increases as the gas flow rate increases. The equilibrium separation cut point was used to establish a relationship between the S50 and the velocity.

Figure 1: The effect of flow rate on the S50 for experiments with a 6 minute run time. The graph depicts an exponential increase in the cut point, S50, as the flow rate is increased.

This study also investigated the gravity beneficiation of nominally -2 +0.25mm coal and mineral matter samples in the RC. Air was used as the fluidisation gas but it was discovered that a dense medium was required in the apparatus to promote a density separation rather than a size classification. Illminite was originally used as the dense medium but was later substituted by magnetite due to the likelihood that magnetite would be preferred in the coal industry. A range of variables were studied which included the dense medium quantity, size range and the variation in the medium inventory throughout the experiment, the solids inventory, and the particle size range analysed, the gas velocity, the particle residence time in the vessel, and the RC channel geometry. Several different multistage process systems were examined to identify any improvement in quality of the separation by reprocessing either the product or reject streams, or by simulating a continuous process. Results from all the methods are shown in Figure 2.

The amount of magnetite required was related to the physical dimensions of the RC. Under excessive medium conditions, the RC had a tendency to entrain solids with the overflow by plug flow, whereas insufficient media promoted size separation. The ideal dense medium particle size range was -0.212 +0.150 mm. A medium that was too fine was impossible to retain in the system and suffered from its cohesive nature. The use of an excessively coarse medium was problematic to recover from either the overflow or underflow and also prevented fine particles from escaping the bed. Each experiment processed typically 200 g of feed. The feed particle size range for this study was varied to determine the optimum size range the RC could process. Although it was discovered that a range from -2 +0.5mm provided the best result, separations of particles covering the range -8 +0.25 mm were achievable.

Multistage process systems were examined to improve the results from a single stage split operation. Here either the products continued to the cleaner or the reject would be directed to a rougher, for reprocessing after the first separation was completed. While these multistage operations had several process advantages, the best performing arrangement was the system that reprocessed the product in a series of cleaners. This system was called the Product Fractionation path way. In this process, the continual removal of the reject material from the product stream reduced the likelihood of misplacement of mineral matter and ensured a higher quality product. Figure 2, which compares yield-ash results with the washability curve, shows that this method was significantly more effective than all of the other methods investigated.

Figure 2: Comparison of all the different methods investigated with reference to the Float/Sink washability Curve. The particle size range was -2+0.25 mm. The Product Fractionation method shows the best separation of all the different methods.

In general, the dry beneficiation results are poorer than the separations possible using a water based separation. However, the higher mineral matter content obtained for a given yield may be offset against the reduction in the moisture content.