Physical model experiments of the gas-assisted gravity drainage process

Abstract

The displacement of oil by gas injection in oil reservoirs is an attractive method of improved oil recovery. Commercial gravity-stable gas injection projects have demonstrated excellent recoveries; however, their application has been limited to dipping reservoirs and pinnacle reefs. Horizontal gas floods and the water alternating gas (WAG) processes, practiced in horizontal type reservoirs, have yielded less than satisfactory recoveries of 5-10%. The Gas Assisted Gravity Drainage (GAGD) Process being developed at LSU extends the concept of gravity-stable gas floods to horizontal type reservoirs to improve volumetric sweep and oil recovery. This experimental study consists of a series of visual experiments to study the effects of operating parameters such as capillary number, the Bond number, gravity number and mobile water saturation on the GAGD process. The experiments were performed in a visual physical model packed with uniform glass beads of various sizes and by injecting gas at various pressures, rates and initial water saturations. The results have been correlated against dimensionless numbers characterizing the role of gravity and capillary forces. This has also enabled the comparison of the physical model results with those from core floods and field projects. The run time of the physical model experiments have been scaled to the required time in the field to obtain similar recoveries. Good correlations are obtained between the Bond and capillary numbers with cumulative oil recovery. Results indicate that these correlations are not only valid for immiscible GAGD floods but may be applicable for miscible GAGD floods. This enables us to predict oil recoveries from similar processes on commercial scale if sufficient rock and fluid data is available. Significantly better oil recovery is obtained during the early life of the project at constant pressure gas injection. Higher recoveries are obtained during gravity-dominated flow as opposed to capillary or viscous dominated. Experimental results show that the composition of the injected gas has little effect on oil recovery during immiscible gas injection. Recovery versus gravity number data from the physical model, core floods and commercial field projects, all fall close to a straight line on a semilog plot. This indicate that the physical model is capable of capturing the realistic mechanisms operating in the field projects and that these experimental runs may be reasonably extrapolated to field scale.