Generalized Oil Viscosity Model For the Effects of Temperature, Pressure And Gas Composition

Abstract

A generalized viscosity correlation and its application are presented in this paper for heavy oils, bitumens and light oils from Saskatchewan. The correlation is shown to predict accurately the viscosity at any desired temperature and pressure, and also at any concentration of dissolved gases. Interestingly, the correlation predicted with the same accuracy in both kinematic and absolute viscosities. By adjusting the parameters of the shape factor, the correlation is shown also to predict accurately the viscosity of various Chinese waxy crude oils, containing up to 40% water, at different temperatures. The average absolute deviations obtained with the correlation are: 4.8% Saskatchewan light oils in kinematic viscosity, 4.6% and 2.2% for the viscosity of Chinese crude oils. The input parameter required for this correlation is just one viscosity measurement of the dead oil made at atmospheric pressure and at a convenient temperature such as 30 ºC where the composition of the oil sample is unaltered during the measurement. Introduction Numerous efforts have been directed towards the development of correlations capable of predicting adequately crude oil viscosity as a function of temperature, pressure, and/or composition(1–7). This is principally because viscosity values of crude oils and crude oils containing dissolved natural gases are required in various petroleum engineering calculations. Several empirical and semi-empirical viscosity-temperature correlations have been summarized by Reid et al.(8) The most convenient viscosity correlations in use are those based on one-parameter equation since they involve fewer numerical computations. Some of the one-parameter correlations, however, are only suitable for the specific oil type for which they were developed. New parameters and constants have to be developed for any slight changes in the physical oil properties such as density, API gravity, molecular weight, etc., thus involving extra numerical computations. Others require several experimental viscosityalues at varying conditions of temperature, pressure or composition to make predictions. The correlation we present here is a generalized one-parameter viscosity equation already published by Puttagunta et al.(9) The applicability and accuracy of the correlation have been demonstrated in several other publications(10–13). Therefore, the focus of this paper is not to present a new viscosity correlation but to evaluate the correlation on new experimental data recently obtained for a variety of Saskatchewan crudes. The Puttagunta et al. viscosity correlation is as follows: Equation (1) Available In Full Paper. where S is a parameter which depends on b, C is a constant, T is temperature in °C, P is gauge pressure (MPa), X is dissolved gas concentration, (mole %), and μ is viscosity in Pas. b is the viscosity characterization parameter obtained from one viscosity measurement at 30 °C and one atmosphere as follows: Equation (2) Available In Full Paper. In the correlation, ∑Cti represents the sum of the concentration terms for all i gases dissolved in the heavy oil or bitumen, and thus Ct is given as: Equation (3) Available In Full Paper.

Modelling of Bitumen-and-Solvent-Mixture Viscosity Data Using Thermodynamic Perturbation Theory

“…Our model reproduces the available experimental data with an acceptable accuracy as compared with the Miadonye et al (1997) model. However, it is worth noting that the Miadonye et al (1997) model has been tuned for different types of bitumen and the comparison presented here is to demonstrate the applicability of the developed model and not to criticize the Miadonye et al (1997) model.…”

“…The AARD with these experimental data was less than 7.4%. The model of Miadonye et al (1997) underestimated the experimental data of CH 4 -saturated bitumen, with an AARD at approximately 28%. As shown in Table 6, the best fit was observed at 300 K. Results showed that the deviation from the experimental data increased with increasing temperature.…”

“…The calculated error for this model in the prediction of the viscosity of compressed Cold Lake bitumen was 2.1%. To compare our proposed model with another model, we reworked the model of Miadonye et al (1997) and compared our results with this model. We have selected this model because this model can consider the effect of temperature, pressure, and composition of the dissolved solvent, similar to our proposed model.…”

“…The AARD from experimental data at 376 K was 10.5%. AARD between experimental data and the model of Miadonye et al (1997) for each temperature is also shown in Table 6.…”

“…Better accuracy could be observed at 330 K with an AARD equal to 6.1%. Because the model of Miadonye et al (1997) has not been tuned for N 2 , this model was not applied to regenerate viscosity data of N 2 -saturated bitumen.…”