Development of Novel Dimensionless Parameters for Accurate Estimation of Properties in Fluidized Beds

Articles in Press

Document Type : Research Article

Authors

1 LBMPT, Laboratory of Biomaterials and Transport Phenomena, University of Medea, Algeria.

2 College of Science and Technology, University of Medea

3 Medea University

Abstract

Fundamental hydrodynamic properties in the fluidized beds are the maximal pressure drop ΔPmax and the minimal fluidization velocity Umf. Accurately forecasting these variables is necessary to maximize the functionality and architecture of this system. This study introduces a fresh technique for multilayer modeling. Two dimensionless numbers were generated by utilizing the Ergun equation. The initial numerical value, π1, represents the energy dissipation from fluid friction within the fluidized bed. It demonstrates a clear relationship with the fluid velocity in the Ergun equation, denoted as "U." This characteristic is consistent with the properties of the dimensionless viscous (friction) number. In contrast, the second numerical value π2 represents the dissipation of kinetic energy within the bed and exhibits a direct proportionality to U2, akin to the dimensionless inertial number. Furthermore, the hydraulic diameter was created to differentiate between the conical and cylindrical beds. The ideal configuration for the multilayer perceptron (MLP) model was identified as a two-layer feedforward architecture comprising fourteen hidden neurons. The study determined that the precision of ΔPmax and Umf was 6.11% and 9.77% using the Average Absolute Relative Difference (AARD) metric. Additionally, the Mean Squared Error (MSE) metric yielded values of 0.0146 and 2.9 × 10-4 for ΔPmax and Umf, respectively. Furthermore, the Root Mean Squared Error (RMSE) metric resulted in values of 0.120 and 0.0172 for ΔPmax and Umf, respectively. The R-squared values for the respective variables were 0.935 and 0.989. The correlation coefficients (R values) were also found to be 0.967 and 0.994, respectively. The findings obtained from the simulation were compared to the experimental data, indicating a significant level of agreement and confirming the correctness and dependability of the MLP technique.

Keywords

Main Subjects

Iranian Journal of Chemistry and Chemical Engineering

Articles in Press, Accepted Manuscript
Available Online from 09 December 2023

Files

Share

How to cite

Statistics