This paper presents an experimental approach to minimize uncertainties involved when modeling CO2-crude oil mixtures. Very little data on CO2-crude oil mixtures have been published in the literature, and most do not cover the range of temperatures necessary for a broader interpretation of data during equation of state (EOS) modeling. Also fluid viscosity is rarely measured. This work addresses both issues, and also develops new experimental procedures.

Phase behavior studies of CO2 mixtures with a live oil sample from a UAE field were carried out using three different temperatures (54°C, 64°C and 74°C) and maximum pressure of 4500 psi. CO2 was injected into a cell containing the crude oil at a suitable working pressure of 2500psi, and swelling tests were conducted. Fluid properties such as saturation pressures, crude oil densities, viscosities, relative oil volumes and swelling factors were determined. It was confirmed that the swelling factors increased with a rise in temperature at constant injection volumes. In addition to the phase behavior studies, a modified operating method was developed for the capillary tube viscometer which required approximately half as much sample. The procedure took into account the integrity of the homogenous mixture as well as the dead volumes of the equipment, and gas trapping effects during sample charging. Results obtained from this method were repeatable and consistent with the routine method, with an average deviation of 0.34%. At the final step of CO2 injection in crude oil, a viscosity reduction by a factor of 4.4 was recorded. The method should be beneficial for subsequent research projects involving viscosity measurements of CO2–crude systems.

A commercial software was used in the phase behavior modeling of the laboratory data. The software incorporates the Peng-Robinson equation of state (PR-EOS) for modeling the experimental data obtained for the range of temperatures. The reservoir fluid sample was first characterized into several components, and then the EOS parameters and viscosity model were tuned to match the experimental data. The outcome of the EOS modeling established that the critical properties, binary interaction coefficients, acentric factors and the selection of tuning weights were the most sensitive and influential parameters. An 11 component lumping scheme was found to accurately match experimental data at all three temperatures.

You can access this article if you purchase or spend a download.