Winnowing as a dry separation method for fine coal
Abstract
Dry coal processing is fast becoming favoured as a fine coal beneficiation technique. The decline in suitable good-quality coal, as well as the ongoing decrease in usable process water, makes dry beneficiation of the fine coal fraction even more crucial. Dry coal beneficiation is relatively young with most of the methods still in the developmental phase. Additionally, many of the dry processing options available are better suited for coarser (+6 mm) and also easy-to-beneficiate feed. Air winnowing for fine particles is established in both agricultural and pharmaceutical industries, which indicates that the method can be used to separate smaller coal particles on a density basis and therefore, may prove effective in the coal industry as well. A proof of concept study was conducted using computer-based simulations and physical experiments. The modelling and simulation of the system were used to assist with the design and optimization of separation unit, while the physical experiments served as a validation of the method and findings. The initial winnowing unit, consisted of a closed box through which a horizontal air stream was developed. Coal particles were dropped into the air stream and displaced horizontally across a distance (x), depending on the particle size and density, thus actuating a separation based on these two parameters. The modelling was done using basic equations of motion in an iterative process to obtain an initial estimate of the particle displacement. The results obtained indicate that a separation chamber with a height of 0.6 m and a length of 1 m would be sufficient to separate any -6 mm particle, using the airflow velocity range of 10 m·s⁻¹ to 30 m·s⁻¹. At this stage, the width of the chamber is not important due to the assumption that there is no sideways movement. The simulation in Star CCM+ confirmed the findings of the model and provided a virtual representation of the separation using streamlines. This aided in the determination of both the airflow pattern and the particle displacement tracking. The airflow pattern indicates that there is a possibility of backflow developing in the bottom of the separation chamber, with low-velocity eddies forming over time. This could possibly influence the separation and may need further investigation. The tracer test was conducted using cube-shaped particles and from the results, a baseline for the separation was established in the form of three prediction matrices. The matrices can be used to determine the displacement of any -6 mm particle provided that the density and airflow velocity is known. The matrices were used to test the capability of the unit in terms of size and density separation. The results obtained show that tracer particles can be separated into three distinct size classes with some overlap observed. The density separation proves sufficient, with efficiency values (EPM) of 0.16 and 0.17 at density cut-points of 1500 kg·m⁻³ and 1700 kg·m⁻³, respectively. After the tracer test confirmed that it was possible to separate particles by size and predict the density cut-point with some degree of accuracy, a coal test was conducted as a final validation of the method. Low-density coal (average density ± 1400 kg·m⁻³) from Mozambique was used for this experiment since the aim is to achieve good separation at a relatively low-density cut. The coal sample was prepared using the RhoVol analyser (developed by DebTec) and the experiment was conducted to determine the capability of the winnowing unit. The coal tests prove that size separation can be achieved at an airflow velocity of 21 m·s⁻¹ and the density cut-point can also be predicted by using prediction matrices. The results show efficiency values (EPM) of 0.11 and 0.09 at density cut-points of 1400 kg·m⁻³ and 1500 kg·m⁻³ respectively. The data obtained from the study indicates that air winnowing can be used to separate fine coal particles based on both size and density. However, in order to achieve an efficient density separation, a narrow size range distribution of 2:1 is required (Patil & Parekh, 2011). The prediction matrices were proved to be accurate to some degree and overall the method delivers promising results. With some improvements to the model, simulation, method and design, the separation efficiency can also be improved. The next phase of testing will include upscaling the process to a demonstration plant, optimizing the current unit and testing the method on different qualities of coal. This will result in the culmination of the research conducted on the method of winnowing thus far.
Collections
- Engineering [1403]