This paper gives a comparative study of empirical viscosity correlation expressions. The documented expressions include Lee et al.'s exponential expression, Dempsey-Standing's polynomial expression, Dranchuk-Islam- Bentsen's representation of Carr et al's data, Gurbanov and Dadash-Zade's expression, Lohrenz et al's expression, and a regression expression of Gonzalez and Lee's data. The first four expressions are two-step procedures, while the equation representing Gonzalez and Lee's data is a one-step procedure, and Lohrenz et al's expression is basically a semi-empirical composition dependent method. Some of these expressions are capable of estimating viscosities of impure natural gases which contain non-hydrocarbon components like N2 CO2and H2S. A detailed analysis of application region, accuracy, and computational superiority for these empirical expressions of viscosity correlations is presented and a two-step procedure with combined equations is recommended.
In the petroleum industry, natural gas viscosities at reservoir conditions or elevated pressures and temperatures are of particular importance for reservoir engineering calculations. The only accurate way to obtain the viscosity of a gas is to determine it experimentally, using apparatus such as a rolling-ball pressure viscometer or a capillary-tube viscometer. In the early petroleum industry, many experimental efforts were made and many data were obtained. In practical applications, however, petroleum engineers always rely on empirical correlations of natural gas viscosities. The reason is perhaps twofold: performing good experiments is quite tedious, and also compared with the inaccuracy of some of the other data that are used in the same equations with the gas viscosity, the viscosity might be the most accurate even though it comes from empirical data which were correlated graphically for engineering convenience. With the advent of computers, many semi-empirical equations for gas viscosities were developed to meet the industry's demand. Most of these equations stemmed directly from the regression of experimental charts that were already well correlated, while others were developed based on experiment, supported by molecular theories.
The viscosity of a pure gas depends on temperature and pressure, but for a natural gas, or a gas mixture, it is also a function of the composition of the gas. In addition, when a natural gas is not pure, or there are some non-hydrocarbon components (such as N2, CO2 and H2S) present in the gas, its viscosity under pressure and temperature has to be corrected.
Conventionally, a two-step procedure is usually used to estimate natural gas viscosities under pressure and temperature.. In the first step, µg the viscosity at low pressure or atmospheric pressure and the desired temperature T is estimated if the apparent molecular weight Mg or the gas gravity G and the desired temperature are known. When it is not a pure hydrocarbon gas, the viscosity µg has then to be corrected for all non-hydrocarbon components. In the second step, the viscosity of the natural gas under pressure and temperature is calculated through equations that predict the ratio of µ to µg based on the principle of corresponding states, that is, the ratio of µ to µg is expressed as a function of both the reduced pressure.
