Abstract
The main objectives of this study were to quantify the classification efficiency of a binary mixture of two different particle types and to demonstrate that CPFD can be used to simulate the main features of the classification process. A lab-scale cylindrical fluidized bed (8.4 cm inner diameter, 150 cm height), equipped with pressure sensors and a video camera for recordings, was applied in the experiments. The particles used in the study were ceramic beads (median diameter 70 µm, skeletal density 3830 kg/m3) and steel shot (290 µm, 7790 kg/m3). Ambient air was used as the fluidization medium.
The minimum fluidization velocities of pure ceramic beads and steel shot were found to be 0.015 m/s and 0.240 m/s, respectively. In the experiments with a binary mixture of the two materials, the fluidization column was filled by alternating layers of ceramic beads and steel shot. In principle, segregation of the two particle types can be obtained by applying a suitable gas velocity within the interval defined by the two minimum fluidization velocities, resulting in a top layer of mainly lighter and smaller particles (flotsam) and a bottom layer of mainly heavier and larger particles (jetsam). In the experiments, the air velocity was gradually increased until the entire bed was fluidized. A gradual rearrangement of the multi-layer structure into a two-layer structure was observed. The rearrangement started at the top and then progressed downwards until most of the steel particles were collected at the bottom of the bed and practically all the ceramic beads were gathered at the top. The velocity at which the flotsam and jetsam layers were clearly segregated was found to be 0.180 m/s. The jetsam layer contained less than 0.3 wt% ceramic beads indicating that an almost pure steel shot fraction could be produced through fluidized bed classification. The flotsam layer was, however, less pure, with a steel shot content up to 24 wt%, suggesting that this layer may need a second classification stage for improved purity.
Computational particle-fluid dynamics (CPFD) simulations of the same setup were performed using the commercial software Barracuda. The simulated minimum fluidization velocity of steel shot, applying the Ergun drag model, perfectly matched the experimental value. For the ceramic beads, however, the simulations, applying the Wen-Yu drag model, gave a value lower than the experimental value. Still, the simulations were able to capture the general behavior of the particles in the bed observed in the classification experiments, i.e. the rearrangement of the layers, even if a higher gas velocity was required for complete classification of the particles. The formation of air pockets was also observed in the simulations, as in some of the experiments. The results suggest that Barracuda CPFD simulations can be a useful tool in design and evaluation of fluidized bed classifiers.
The minimum fluidization velocities of pure ceramic beads and steel shot were found to be 0.015 m/s and 0.240 m/s, respectively. In the experiments with a binary mixture of the two materials, the fluidization column was filled by alternating layers of ceramic beads and steel shot. In principle, segregation of the two particle types can be obtained by applying a suitable gas velocity within the interval defined by the two minimum fluidization velocities, resulting in a top layer of mainly lighter and smaller particles (flotsam) and a bottom layer of mainly heavier and larger particles (jetsam). In the experiments, the air velocity was gradually increased until the entire bed was fluidized. A gradual rearrangement of the multi-layer structure into a two-layer structure was observed. The rearrangement started at the top and then progressed downwards until most of the steel particles were collected at the bottom of the bed and practically all the ceramic beads were gathered at the top. The velocity at which the flotsam and jetsam layers were clearly segregated was found to be 0.180 m/s. The jetsam layer contained less than 0.3 wt% ceramic beads indicating that an almost pure steel shot fraction could be produced through fluidized bed classification. The flotsam layer was, however, less pure, with a steel shot content up to 24 wt%, suggesting that this layer may need a second classification stage for improved purity.
Computational particle-fluid dynamics (CPFD) simulations of the same setup were performed using the commercial software Barracuda. The simulated minimum fluidization velocity of steel shot, applying the Ergun drag model, perfectly matched the experimental value. For the ceramic beads, however, the simulations, applying the Wen-Yu drag model, gave a value lower than the experimental value. Still, the simulations were able to capture the general behavior of the particles in the bed observed in the classification experiments, i.e. the rearrangement of the layers, even if a higher gas velocity was required for complete classification of the particles. The formation of air pockets was also observed in the simulations, as in some of the experiments. The results suggest that Barracuda CPFD simulations can be a useful tool in design and evaluation of fluidized bed classifiers.
