MSc (Eng), Faculty of Engineering and the Built Environment, University of the Witwatersrand, 2011 / Most of the remaining coalfields in South Africa are found in arid areas where process water is scarce and given the need to fully exploit all the coal reserves in the country, this presents a great challenge to the coal processing industry. Hence, the need to consider the implementation of dry coal beneficiation methods as the industry cannot continue relying on the conventional wet processing methods such as heavy medium separation. Dry coal beneficiation with an air dense-medium fluidized bed is one of the dry coal processing methods that have proved to be an efficient separation method with separation efficiencies comparable those of the wet heavy medium separation process. Although the applications of the fluidized bed dry coal separator have been done successfully on an industrial scale, the process has been characterized by relatively poor (Ecart Probable Moyen), Ep values owing to complex hydrodynamics of these systems. Hence, the main objectives of this study is to develop a sound understanding of the key process parameters which govern the kinetics of coal and shale separation in an air fluidized bed focusing on the effect of the particle size, shape and density on the performance of the fluidized separator as well as developing a simple rise/settling velocity empirical model which can be used to predict the quality of separation. As part of this study, a (40 x 40x 60) cm air fluidized bed was designed and constructed for the laboratory tests. A relatively uniform and stable average bed density of 1.64 with STDEV < 0.01 g/cm3 was achieved using a mixture of silica and magnetite as the fluidizing media. Different particle size ranges which varied from (+9.5 -16mm), (+16 -22mm), (+22 -31.5mm) and (+37 -53mm) were used for the detailed separation tests. In order to investigate the effect of the particle shape, only three different particle shapes were used namely blockish (+16 -22mm Blk), flat (+16 -22mm FB) and sharp pointed prism particles (+16 – 22mm SR).Different techniques were developed for measuring the rise and settling velocities of the particles in the bed. The Klima and Luckie partition model (1989) was used to analyze the partition data for the different particles and high R2 values ranging from (0.9210 - 0.9992) were recorded. Average Ep iii values as low as 0.05 were recorded for the separation of (+37 -53mm) and (+22 -31.5mm) particles under steady state conditions with minimum fluctuation of the cut density. On the other hand, the separation of the (+16 -22mm) and (+9.5 – 16mm) particles was characterized by relatively high average Ep values of 0.07 and 0.11 respectively. However the continuous fluctuation or shift of the cut density for the (+9.5 -16mm) made it difficult to efficiently separate the particles. Although, particle shape is a difficult parameter to control, the different separation trends that were observed for the (+16 -22mm) particles of different shapes indicate that particle shape has got a significant effect on the separation performance of the particles in the air fluidized bed. A simple empirical model which can be used to predict the rise/settling velocities or respective positions of the different particles in the air fluidized bed was developed based on the Stokes’ law. The proposed empirical model fitted the rise/settling data for the different particle size ranges very well with R2 values varying from 0.8672 to 0.9935. Validation of the empirical model indicate that the model can be used to accurately predict the rise/settling velocities or respective positions for all the other particles sizes ranges except for the (+9.5 – 16mm) particles where a relatively high average % error of (21.37%) was recorded. The (+37 -53mm) and (+22 -31.5mm) particles separated faster and more efficiently than the (+16 -22mm) and (+9.5 -16mm) particles. However, the separation efficiency of the particles can be further improved by using deeper beds (bed height > 40cm) with relatively uniform and stable bed densities. Prescreening of the coal particles into relatively narrow ranges is important in the optimization of dry coal beneficiation using an air fluidized bed since different optimum operating conditions are required for the efficient separation of the different particle size ranges and shapes. The accuracy and the practical applicability of the proposed empirical model can be further improved by carrying out some detailed rise/settling tests using more accurate and precise equipment such as the gamma camera to track the motion of the particles in the fluidized bed as well as measuring the actual bed viscosity and incorporate it in the model.