Experimental Studies on the Waterflood Residual Gas Saturation and Its Production by Blowdown

Abstract

Summary. Experiments performed on long sandstone cores determined the waterflood residual gas saturation and the mobilization of this trapped gas by reduction of pressure. The trapped gas apparently did not become mobile immediately as it expanded: the gas saturation had to increase appreciably to a critical value for gas remobilization. This new finding has unfavorable implications for possible processes to increase liquids recovery from gas-condensate reservoirs and for enhanced recovery from gas reservoirs invaded by water. Introduction Pressure maintenance processes offer the possibility of producing Pressure maintenance processes offer the possibility of producing a higher liquid recovery from a gas-condensate reservoir than natural depletion because less liquid phase condenses in the reservoir. Pressure maintenance is usually achieved by gas injection, but it Pressure maintenance is usually achieved by gas injection, but it has been suggested that water injection may have some advantages. Because of the very favorable mobility ratio for the displacement of gas by water, a waterflood should have a high volumetric sweep efficiency and thus produce a high recovery of gas and liquids, provided that the residual gas saturation is low. Waterflood residual gas saturations in the range 0. 18 to 0.50 have been observed, however, so the overall recovery would depend on the extent to which this trapped gas becomes mobile during a subsequent blowdown phase. Two-phase gas/water relative permeabilities are usually determined in displacement experiments at constant pressure. Primary-drainage curves are obtained by starting with a fully Primary-drainage curves are obtained by starting with a fully brine-saturate-core and increasing the fractional flow of gas. Subsequent decreases in the gas fractional flow yield the imbibition curve, and further increases provide the secondary-drainage curve. These conventional measurements reveal hysteresis in the gas relative permeability. As illustrated in Fig. 1, the imbibition curve is permeability. As illustrated in Fig. 1, the imbibition curve is different from, and much lower than, the primary-drainage curve because of gas trapping. (In practice, the imbibition curve is often calculated from the primary-drainage curve, and the residual gas saturation by use of the theoretical Land model.) The brine-phase relative permeability curves also show hysteresis, but the differences are generally small and sometimes negligible. It is usually assumed, following Killough, that the secondary-drainage curve retraces the imbibition curve, which is our experience for oil/water flow in cores similar to those used in these experiments, though some differences have been reported for gas flow in sandstone cores. If the conventionally determined secondary-drainage curves of the type shown in Fig. 1 were also applicable during the depressurization of a waterflooded gas-condensate reservoir, gas would become mobile immediately because of gas expansion, and a small increase in gas saturation from the residual value would produce a high fractional flow of gas. This paper reports results of experiments designed to check this effect and to determine the residual gas saturation. These experiments, performed on two long sandstone cores, are the first of a series designed to study the problems of waterflooding gas-condensate reservoirs and are also relevant to the process of enhanced gas recovery by depressurization of gas reservoirs that have been invaded by water. The initial experiments have used a simple gas (methane) rather than a gas condensate, though future work is planned to investigate the effects of a condensing liquid phase, to elucidate the mechanisms involved, and to extend the results to cores from other formations. Experimental Apparatus The experiments were performed in two high-pressure gas displacement rigs. A simplified line diagram of the more sophisticated rig is shown in Fig. 2. In the rest of this paper, this rig will be referred to as Rig 1. An epoxy-resin-coated sandstone core, fitted with pressure tappings, 1.68 m [5.5 ft] in length and 4.5 cm [1.8 in.] in diameter, was held in the rig's pressure vessel. Fluid displacement was achieved by maintaining a constant pressure on fluid stored in reservoirs at the inlet to the core and withdrawing fluid from the exit face with a two-cylinder pump. After leaving the core, samples of the produced fluids were collected in small (20-mL) pressure vessels, removed from the rig while their contents were pressure vessels, removed from the rig while their contents were maintained at rig pressure and temperature, and their compositions determined. When installed on the rig, the sample bottles were initially full of water, offering the possibility that small volumes of produced gas dissolved in this water and thus went undetected. The produced gas dissolved in this water and thus went undetected. The rig is equipped with nucleonic instruments that allow brine saturations to be measured at positions along the core through use of a gamma neutron reaction. 10 Experimental data are collected through a minicomputer system that also organizes the collection of fluid samples and the operation of the nucleonic measurement apparatus, as well as providing the operator with diagnostic output and assistance with operations through interactive procedures. Fig. 3 shows a simplified line diagram for the other rig, which will be referred to as Rig 2. It has core assembly arrangements similar to those of Rig 1, but uses a shorter core of 0.84-m [2.8-ft] length. Fluids are injected into the core with high-pressure liquid chromatography pumps and, in the case of a gas phase, from an intermediate free piston vessel maintained at a constant temperature. After the produced fluids leave the core, they pass through a backpressure valve, which maintains the required rig pressure, and having been flashed down to ambient conditions, pass to a measurement system. This rig is also equipped with minicomputer-controlled nucleonic instruments though, in this case, saturations could be determined only over the 30-cm [12-in.] length immediately upstream from the exit face of the 84-cm [2.8-ft] -long core. Experimental Method Core Properties. The cores were of a well-cemented aeolian sandstone from the Clashach quarry in Scotland. This sandstone is water-wet, has well-sorted grains, and a low clay content. SPERE P. 387