Relative permeabilities (kr) are crucial flow functions governing the fluid distribution within and production from petroleum reservoirs under various oil-recovery methods. To obtain these important reservoir parameters, conventionally, it is required to take rock samples from the reservoir and perform appropriate laboratory measurements. Although kr is expressed as a function of fluid saturation, it is now well-known that kr values are affected by pore structure and distribution, absolute permeability, wettability, interfacial tension (IFT), and saturation history. These rock/fluid properties often change from one region of the reservoir to another, but it would be impossible to perform kr measurements for all regions of a reservoir. Generally, performing experiments on a core with higher permeability is faster and easier than a low-permeability rock. Therefore, assuming all other parameters such as wettability, IFT, and displacement direction are the same for two rocks with different permeabilities, the question becomes how do we estimate the kr of a rock with lower permeability from available (measured) kr of a higher-permeability rock? How do we account for wettability and IFT differences? A normalization technique has been proposed to remove the effect of irreducible water and trapped saturations, which would be different under different conditions. The relative permeabilities can then be denormalized and assigned to different regions (rock types) of the reservoir on the basis of their own irreducible water and trapped saturations.

The objective of this study is to introduce a methodology to predict the gas/oil kr for new rock/fluid conditions (such as permeability, wettability, and IFT) by use of existing gas/oil kr data measured at different conditions. By use of measured data from coreflood experiments, we show that by applying an appropriate normalization technique one can adequately predict kr of rocks with different permeability and wettability conditions in two-phase gas/oil flow. However, the results show that the effect of IFT change cannot be captured by normalization techniques. To improve the methodology, a new hypothesis is introduced and proposed here on the basis of dynamic trap saturation. Finally, by use of our experimental data, we evaluate the validity of the Coats (1980) IFT scaling method. We demonstrate the shortcomings of the method and offer an improvement to its prediction.

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