Abstract

The corresponding states viscosity model is widely accepted in the petroleum industry for calculating oil and gas viscosities. In its classical form the model predicts the viscosities of complex mixtures using methane as a reference compound. As a result of the chosen reference compound the corresponding states model has good predictive capabilities for gases and light oil systems, whereas the predictive performance deteriorates for heavier oil systems. The present work discusses how far the performance of the corresponding states model can be stretched in its classical form with methane as thereference compound. It further outlines a procedure based on a viscosity correlation for stable oil mixtures that may take over when the methane reference temperature becomes too low to apply methane as reference compound. In the temperature region where the new correlation is to take over from the classical corresponding states there is a smooth transition between the viscosity results obtained with either method. Therefore the new correlation can be seen as an extrapolation of the corresponding states method to be applicable to heavy aromatic oil mixtures at relatively low T.

Introduction

Viscosity is a key transport property both for subsurface simulations, well design and pipeline and process simulations. A number of correlations for viscosity ofpetroleum systems exist, ranging from simple ones requiring information on bulk properties like API gravity and temperature such as the Beggs and Robinson correlation (Beggs and Robinson, 1975) to the morecomplex ones relying on the composition of the mixture in question. The most popular of the latter type is the method by Jossi et al. (1962), which was developed for reservoir fluids by Lohrenz et al. (1964), most oftenreferred to as the Lohrenz-Bray-Clark (LBC) correlation. The LBC correlation is not in general predictive butbeing a polynomial it can easily be tuned to viscosity data. As it further executes fast in computer code, it is an obvious choice in e.g. reservoir simulation models where CPU time constraints apply. If on the other hand no experimental data are available for tuning, there is a great risk of serious malpredictions. Another pitfall with viscosity correlations may occur when tuning the model entirely to stabilized oil viscosity data. In this situation live oil and gas viscosity predictions from the correlationcan become highly unrealistic. For these reasons, the more reliable, but also more complex, corresponding states model is an important tool for the petroleum industry. The corresponding states viscosity model has good predictive capabilities for gases and light oils whereas the predictive capabilities deteriorate for heavier oil systems. This is fundamentally because methane at the corresponding reference state would be a solid for whichno viscosity data are available. It follows that for a heavy oil system a heavier reference compound would be more appropriate. Change of reference compound however requires viscosity data in a wide temperature and pressure range for a heavier hydrocarbon as for example n-decane. Unfortunately most viscosity data for heavier compounds are measured at atmospheric pressure, and hence the pressure effects cannot be captured properly.

This content is only available via PDF.
You can access this article if you purchase or spend a download.