Predicting Emulsion Viscosity Using Artificial Intelligence

Abstract

Abstract Water-in-oil (W/O) emulsions are present in some stages of exploration and operation of oil fields. Therefore, it is important to define the dynamic viscosity behavior of these emulsions for flow assurance purposes. This study applied an Artificial Intelligence Algorithm called Artificial Neural Network in order to predict the emulsion dynamic viscosity of thirty (30) oils from Brazilian basins. The Artificial Neural Network used as inputs rheological data considering different values of temperature, shear rate, oil °API and water fraction in order to predict the dynamic viscosity. All the algorithms and data handling were made using Python computational language. Nevertheless, the results obtained in this paper were statistically compared to the results obtained using other classical correlations presented in the literature. The model proposed could generate predictions with more precision when compared to the other correlations.