Loading…
Correlating Interfacial Area and Volumetric Mass Transfer Coefficient in Bubble Column with the Help of Machine Learning Methods
Several empirical correlations are available to estimate the volumetric mass transfer coefficient and effective interfacial area for bubble column reactors. But these empirical correlations are applicable over the range of experimental conditions. By considering the broad range of parameters in a da...
Saved in:
Published in: | Industrial & engineering chemistry research 2023-02, Vol.62 (5), p.2104-2123 |
---|---|
Main Authors: | , , , , , , |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Several empirical correlations are available to estimate the volumetric mass transfer coefficient and effective interfacial area for bubble column reactors. But these empirical correlations are applicable over the range of experimental conditions. By considering the broad range of parameters in a database, data-driven machine-learning methods can be used to correlate the design parameters. In this work, a generalized machine learning-based methodology is presented to calculate the volumetric mass transfer coefficient and effective interfacial area with independent parameters. Machine learning methods such as support vector regression (SVR), random forest (RF), extra trees (EXT), and artificial neural networks (ANN) have been used. An extensive set of experimental data points (1245 data points for volumetric mass transfer coefficient and 526 for interfacial area) have been extracted from the literature. The predictors, column diameter, column height, sparger design, sparger location, percentage free area, superficial gas and liquid velocity, pressure, temperature, density of gas and liquid, viscosity of gas and liquid, and diffusion coefficient, have been used to calculate design parameters. The performance of the machine learning methods has been compared using statistical parameters: mean absolute percentage error (MAPE), mean square error (MSE), and determination coefficient/prediction accuracy (R-square). Statistical analysis shows that the predictive ability of machine learning methods is better than that of traditional regression methods. The extra tree predictions were accurate compared to other methods, and the statistical parameters by extra trees were found to be the best among the methods. For the volumetric mass transfer coefficient, R-square = 0.98, MSE = 2 × 10–4, and MAPE = 15.67, and for the effective interfacial area, R-square = 0.98 and MAPE = 12.04. |
---|---|
ISSN: | 0888-5885 1520-5045 |
DOI: | 10.1021/acs.iecr.2c02820 |