(60ap) Modeling Crude Oil Viscosity Based on Tuning Friction Theory Parameters Using Characterized Distillation
AIChE Spring Meeting and Global Congress on Process Safety
2020
2020 Virtual Spring Meeting and 16th GCPS
Spring Meeting Poster Session and Networking Reception
Spring Meeting & GCPS Poster Session
Wednesday, August 19, 2020 - 3:00pm to 4:00pm
A proper viscosity model to simulate reservoir and process fluid is necessary. The Friction Theory (FT) model has been known as one of the most successful viscosity models for the past several years. Since crude oil composition along with temperature and pressure changes during the process, the changes should be considered. For this reason, the FT model needs to use a characterization method to predict viscosity of crude oil. The characterization using Chi-Square method and SARA-based method has been previously used for the FT model. One of the main problems of these two characterization methods is uncertainty in determination of the required properties of the heaviest fraction to model viscosity of oils, especially heavy oils. This model needs many parameters to be tuned at the same time, and it can be one of the main reasons of deviation from the experimental viscosity because of lack of simultaneous convergence of all adjustable parameters. The objective of this work is to provide characterization based on distillation, assuming that it is capable of fixing these two problems.
Malthen fractions using Guassian extrapolation as pseudo-components and C5-asphalthene as a single component are characterized for each oil. The FT model applied to predict viscosity has five parameters, including a1, a2, b1, b2, and c2.The distillation cuts of crude oil were selected as a reference family, and the five parameters of the FT model were correlated to the boiling temperature. The correlation deviation for other chemical families was correlated to specific gravity. Finally, each of these five parameters was calculated using the difference between these two provided correlations. These parameters were fitted for asphalthene.
The model was tested on 11 light oils from the Middle East and 10 heavy oils from various geographical regions. The overal Average Absolute Deviation (AAD) values of predicted viscosity, which were derived from the experimental viscosity of 11 light oils and 10 heavy oils, were 38 and 45%, respectively. Since overall AAD of crude oils whose experimental density was available was not considerably different from density calculated using Peng-Robinson (PR) equation of state, it was concluded that the high overall AAD was not related to density. Since No good correlation between the parameters c2 to the boiling point, specific gravity, or any other proper physical property of the distillation cuts was found, it was assumed that the overall AAD may be related to the c2 parameter. Thus, the c2 parameter was tuned using a multiplier and a viscosity point at atmospheric pressure. Overall AAD for 11 crude oils and 10 heavy oils decreased by 4 and 6%, respectively.
Therefore, it can be concluded that predicting viscosity of oil (both light and heavy) using the FT model based on distillation and correlation between the fluid parameters /boiling temperature and specific gravity of distillation cuts is a faster and more reliable method than using the model based on Chi-Square method, SARA-test characterization and simultaneous tuning of multiple parameters.
Malthen fractions using Guassian extrapolation as pseudo-components and C5-asphalthene as a single component are characterized for each oil. The FT model applied to predict viscosity has five parameters, including a1, a2, b1, b2, and c2.The distillation cuts of crude oil were selected as a reference family, and the five parameters of the FT model were correlated to the boiling temperature. The correlation deviation for other chemical families was correlated to specific gravity. Finally, each of these five parameters was calculated using the difference between these two provided correlations. These parameters were fitted for asphalthene.
The model was tested on 11 light oils from the Middle East and 10 heavy oils from various geographical regions. The overal Average Absolute Deviation (AAD) values of predicted viscosity, which were derived from the experimental viscosity of 11 light oils and 10 heavy oils, were 38 and 45%, respectively. Since overall AAD of crude oils whose experimental density was available was not considerably different from density calculated using Peng-Robinson (PR) equation of state, it was concluded that the high overall AAD was not related to density. Since No good correlation between the parameters c2 to the boiling point, specific gravity, or any other proper physical property of the distillation cuts was found, it was assumed that the overall AAD may be related to the c2 parameter. Thus, the c2 parameter was tuned using a multiplier and a viscosity point at atmospheric pressure. Overall AAD for 11 crude oils and 10 heavy oils decreased by 4 and 6%, respectively.
Therefore, it can be concluded that predicting viscosity of oil (both light and heavy) using the FT model based on distillation and correlation between the fluid parameters /boiling temperature and specific gravity of distillation cuts is a faster and more reliable method than using the model based on Chi-Square method, SARA-test characterization and simultaneous tuning of multiple parameters.