Modeling of three phase inverse fluidized bed using artificial neural network

Abstract

Fluidization of a three-phase system can be achieved either with co-current up flow of gas and liquid or down flow of liquid and up flow of gas. Three Phase Inverse Fluidised Bed (TPIFB) falls in the second category. Because of high gas hold up and residence time, this type of fluidized bed has more mass transfer coefficient. In present work, experiments were conducted using polyethylene hollow spheres coated with benzoic acid and using water as solvent with air as the fluidizing medium. The data thus generated was used for developing models using Artificial Neural Networks (ANN). It has been observed that the ANN model developed has excellent accuracy level of more than 90%.