449940 Developing a Dead Heavy Oil Viscosity Correlation for Lower Fars, Kuwait

Monday, November 14, 2016: 2:42 PM
Taylor (Hilton San Francisco Union Square)
Adel Elsharkawy, Petroleum Engineering, Kuwait University, Kuwait, Kuwait and Saad Alrashdan, Petroleum Engineering


Viscosity of the produced fluid is an important property for oil and gas industry. It is used in calculations of fluid flow in reservoirs, designing production tubing and flow lines. Accurate estimation of heavy oil viscosity is the most important, as it is one of the key parameters in evaluating a given prospect for the possibility of cold oil production and planning the type of thermal Enhanced Oil Recovery (EOR) methods such as hot water, steam cycling, steam flooding, steam assisted gravity drainage or in-situ combustion. Viscosity of crude oil is usually measured at reservoir temperature at various pressures using bottom-hole sample or surface recombined sample.

Occasionally, the sample is difficult to obtain or becomes unreliable. In this case calculation methods are used to estimate the crude oil viscosity as a function of easily measured production information such as API gravity or detailed compositional data. Many correlations have been presented to estimate crude oil viscosity for crude oils of a certain field or region. Application of these correlations to other parts or regions results in large error in estimated viscosity because each crude oil has its own physico-chemical characteristics that makes it different from other crudes.

 In this study, the accuracy of the well-known correlations are tested against heavy oil viscosity data collected from Lower Fars formation in Kuwait.  Later, a new correlation is presented to estimate the dead oil viscosity of Lower Fars.  Accuracy and reliability of the proposed model is compared to published correlations such as Beal (1946), GlasØ (1980), Kartoatmodjo and Schmidt (1994), and Al-Omair et al. (2014).The results showed that the new correlation is able to predict the dead oil viscosity with mean absolute relative error of 0.16, correlation coefficient equal of 96%, and standard deviation of 0.17.

Extended Abstract: File Not Uploaded