Prediction of oxygen mass transfer of plunging hollow jets using regression models

Jet aerator finds an essential application in the treatment of liquid wastes by increasing the microbial activity and decomposition of organic matter into liquid waste. Diverging plunging jet emitted from the conical end of hollow jet aerators enhances the surface disturbances to a large extent and signifies its usefulness in gas mass transfer and liquid mixing applications. The present study’s focus is on the application of soft computing approaches in estimating oxygen mass transfer by plunging hollow jets. The data observed experimentally from the oxygen transfer study conducted on hollow jet aerators having jet angles of 30°, 45° and 60° are used for modelling. The efficacy of Support Vector Machines (SVM) regression, M5 tree and multiple nonlinear regression is explored with both dimensional and non-dimensional data-sets. SVM regression approach (with RBF and PUK as kernel functions) is observed to be more accurate than M5 tree and multiple non-linear regression methods. The evaluation of the results using all the computing techniques tested indicates the improved performance using non-dimensional data-set as compared to dimensional data-set. Moreover, a parametric study based on SVM modelling is conducted to analyse the influence of jet parameters on the prediction of oxygen transfer.