Prediction of Crude Oil Viscosity of Iranian Oil Field Using Artificial Intelligence

Crude oil viscosity, which is one of the most important PVT properties, is usually determined from laboratory PVT tests. In addition, to determine the viscosity, several correlations have been proposed for different regions. But, due to regional changes in crude oil compositions, none of the correlations could be applied as a universal correlation. On the other hand, laboratory determination of viscosity is very expensive and time consuming. In this study, by using Adaptive Neuro-Fuzzy Inference System (ANFIS), which is one the powerful techniques of artificial intelligence, an intelligent model was proposed to predict the viscosity for Iranian oil fields. A total of 113 data sets of different crude oils from Iranian reservoirs were used. Data sets include viscosity and conventional PVT properties. Among the data sets, 84 data sets were selected randomly for constructing the intelligent model, and the other included 29 data sets were used for model testing. The measured mean squared errors (MSEs) of predicted viscosity from the model in the test data was 0.006 and correlation coefficient (R2) between predicted values from the model and experimental values in the test data was 0.997 which shows a good agreement between the predicted and experimental data.

Prediction of Crude Oil Viscosity of Iranian Oil Field Using Artificial Intelligence Keywords:

viscosity , crude oil , PVT properties , artificial intelligence , ANFIS