A Generalized Model for Estimating the Viscosity of Crude Oil

Currently, most crude oil viscosity correlations rely only on API gravity and temperature, resulting in significant limitations in accurately predicting crude oil viscosity. To address this issue, this paper systematically explores the effect of API gravity, and crude oil components (saturates, aromatics, resins, and asphaltenes content), on viscosity based on 251 crude oil samples through sensitivity analysis. To overcome the shortcoming of traditional models, this paper proposes an innovative generalized viscosity model that combines the Levenberg–Marquardt (LM) and universal global optimization (UGO) methods to fully consider the effects of API gravity and various crude oil components. To verify the effectiveness of the model, this paper divides the 251 crude oil samples into a training set (202 samples) and a test set (49 samples) and compares the prediction results of the new model with the traditional. The results show that the prediction accuracy of the new model on the training set and test set is significantly better than that of the traditional model, with the minimum average absolute relative deviation reaching 10.13% and 12.4%, respectively. This study not only improves the accuracy of crude oil viscosity prediction but also provides early warning of increased pipeline friction caused by abnormal viscosity, avoids suspension accidents, and ensures the safe operation of long-distance pipelines.