Designing an Efficient Artificial Intelligent Approach for Estimation of Hydrodynamic Characteristics of Tapered Fluidized Bed from Its Design and Operating Parameters

A tapered fluidized bed with variable fluid velocity throughout the bed length is a special type of fluidized system. Accurate estimation of hydrodynamic characteristics of the tapered fluidized bed is required for adjusting the operational conditions, optimum design, and process control of this system. In this way, minimum fluidization velocity (U-mf), minimum velocity of full fluidization (U-mff), and maximum pressure drop (Delta P-max) are the main hydrodynamic characteristics of the tapered fluidized bed. In this study, an artificial neural network (ANN) paradigm was developed for prediction of these parameters. Parameters of the ANN model was adjusted through minimization of the absolute average relative deviation (AARD) and mean square error (MSE) via a back-propagation algorithm. Finally, the proposed model predicted the experimental data of U-mf, U-mff, and Delta P-max, with AARD of 1.1%, 1.36%, and 0.89%, respectively, while the best obtained results by five different empirical correlations were 4.12%, 9.4%, and 5.14% for U-mf, U-mff, and Delta P-max.