Determination of Surface Tension and Viscosity of Liquids by the Aid of the Capillary Rise Procedure Using Artificial Neural Network (ANN)

Abstract: The present investigation entails a procedure by which the surface tension and viscosity of liquids could be predicted. To this end, capillary experiments were performed for porous media by utilizing fifteen different liquids and powders. The time of capillary rise to a certain known height of each liquid in a particular powder was recorded. Two artificial neural networks (ANNs) were designed and used to separately predict the surface tension and the viscosity of each liquid respectively. The surface tension predictor network had six inputs, namely: particle size, bulk density, packing density and surface free energy of the powders as well as the density of the probe liquids together with the capillary rise time of the liquids in the corresponding powders. The viscosity predictor network had surface tension as an extra input. In order to correlate the surface tension and viscosity as predicted by the corresponding artificial neural network to their experimentally determined equivalents, two different statistical parameters namely the product moment correlation coefficient (r2) and the performance factor (PF/3) were used. It must be noted that for a perfect correlation r2 = 1 and PF/3 = 0. The results of the present work clearly showed that the artificial neural network approach is able to predict the surface tension (i.e. r2 = 0.95, PF/3 = 16) and viscosity (i.e. r2 = 0.998, PF/3 = 13) of the probe liquids with unsurpassed accuracy.