One major problem in the prediction of the performance of Water Alternating Gas (WAG) process is the uncertainty associated with the changes in three-phase relative permeability (kr) values of oil, gas and water in different cycles, which is known as cyclic hysteresis. In this work we have investigated the effect of cyclic injection on three phase kr by performing a series of coreflood experiments under both water-wet and mixed-wet conditions. WAG experiments started with water injection (I) followed by gas injection (D) and this cyclic injection of water and gas was repeated (IDIDID). Three-phase relative permeabilities were obtained analytically from the coreflood data using an extension of Buckley-Leveret formula to three-phase flow.
The results show the importance of properly accounting for irreversible kr hysteresis loops (especially for gas and oil) in the processes involving cyclic injection under three-phase flow conditions. Gas relative permeability (krg) dropped in successive cycles under both water-wet and mixed-wet conditions krg hysteresis was larger in the water-wet system compared to the mixed-wet case. The results also reveal cyclic hysteresis for oil relative permeability (kro), which tends to increase in successive gas injection periods. The improvement in kro was larger in the water-wet system. In both water-wet and mixed-wet systems, the largest krw hysteresis happens for the transition from two-phase (oil/water system) to three-phase system (from 1st water injection into 1st gas injection) and the subsequent WAG cycles does not show much hysteresis for krw in our experiments.
The paper also offers insights into and explanations for the observed cyclic hysteresis behaviour based on our understanding of the pore-scale and core-scale displacement mechanisms involved in WAG injection. The results highlight some serious shortcomings of the existing reservoir simulators for reliable simulation of oil recovery processes involving three-phase flow and flow reversal.