Machine learning approach for modelling and predicting interfacial tension and rheology of crude oil nanoemulsions stabilized by rhamnolipid biosurfactant

Abstract

Forecasting the properties of nanoemulsions without engaging in expensive and time-consuming experimental research can yield significant benefits across multiple applications. This study examines the capability of machine learning to precisely forecast the interfacial tension and viscosity of crude oil-water nanoemulsions stabilized by rhamnolipid biosurfactant. Four artificial neural network models were created and assessed for nanoemulsions composed of different concentrations of crude oil and biosurfactants. The performance evaluation of the artificial neural network models demonstrated mean squared error values below 2.26E-03 and coefficients of determination greater than 0.999, signifying exceptional predictive accuracy. The mean overall deviation for all models was determined to be around 0.004%, indicating a negligible divergence from experimental results. The findings indicate that the developed artificial neural network models can accurately and reliably predict interfacial tension and viscosity values, providing an efficient alternative to experimental methods, with potential applications in optimizing nanoemulsion formulations for industrial purposes.