Experimental Measurement and Accurate Prediction of Crude Oil Viscosity Utilizing Advanced Intelligent Approaches

Prediction of Viscosity of Blends of Heavy Oils with Diluents by Empirical Correlations and Artificial Neural Network

“…Over the years, publications have appeared devoted to evaluating the models available in the literature for predicting the viscosity of mixtures containing heavy oils and improving their accuracy. ,,,,− The models can be classified as empirical correlations and metaheuristic methods. The metaheuristic methods have found wide application in the viscosity modeling of individual petroleum fluids, − while the viscosity modeling of heavy oil blends using artificial intelligence methods has been reported in only two recent research reports. , Gao et al announced that the artificial neural network (ANN) predicted the viscosity of heavy oil diluted with light oil with an average absolute deviation percentage (% AAD) value of 8.4% versus 11.0% for the best empirical correlation (double log). Ssebadduka et al reported a % AAD value of 0.8272 for the ANN model versus a % AAD value of 3.86% for the best empirical correlation when the viscosity of blends of bitumen with light oil was investigated.…”