Prediction of extraction efficiency in RDC column using artificial neural network

Abstract EN

An application of neural network technique was introduced in modeling extraction efficiency in RDC column, based on a data bank of around 352 data points collected in the open literature.

Three models were made, using back-propagation algorithm, the extraction efficiency was found to be a function of seven dimensionless groups: Weber number (we), (V d / Vc), (µc / µd), (Ds / Dt), (Dr /Dt), (Zc / Dt) and (Zt / Zc).

Statistical analysis showed that the proposed models have an average absolute error (AARE) and standard deviation (SD) of 12.23 % and 10.61 % for the first model, 5.35 % and 6.21 % for the second model, 8.34 % and 7.59 % for the third model.

The developed correlations also show better prediction over a wide range of operating conditions, physical properties and column geometry.