Minimum Fluidization Velocities of Binary Solid Mixtures: Empirical Correlation and Genetic Algorithm‐Artificial Neural Network Modeling

• Published in: Chemical engineering & technology 2022-01, Vol.45 (1), p.73-82
• Main Authors: Let, Sudipta, Bar, Nirjhar, Basu, Ranjan Kumar, Das, Sudip Kumar
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Binary mixtures; Empirical analysis; Experimentation; Fluidized beds; Genetic algorithms; Levenberg‐Marquardt algorithm; Liquid phases; Liquid‐solid fluidized beds; Minimum fluidization velocities; Modelling; Neural network modeling; Neural networks; Solid phases
• ISSN: 0930-7516; 1521-4125
• DOI: 10.1002/ceat.202100170

Experimental investigation of the fluidization behavior in single and binary solid‐liquid fluidized beds of nonspherical particles as solid phase and water as liquid phase was performed in a Perspex column. Different particle sizes were used to prepare single and binary mixtures with different weight ratios for fluidization. Minimum fluidization velocity increased with increasing average particle size and decreasing sphericity for the binary mixture. An empirical correlation was developed to predict the minimum fluidization velocity. Genetic algorithm‐artificial neural network (GA‐ANN) modeling was applied to predict the minimum fluidization velocity for single and binary solid‐liquid fluidized beds. The application of GA‐ANN analysis leads to designing binary solid‐liquid fluidization systems without experimentation. Although less studied than solid‐gas fluidized beds, liquid‐solid fluidized beds have many applications. The fluidization of the latter was studied experimentally, and an empirical correlation to predict the minimum fluidization velocity was developed by genetic algorithm‐artificial neural network modeling, which will allow the design of solid‐liquid fluidization systems without experimentation.