Summary A compositional steam injection simulator was used to study the effects of noncondensable gas injection on oil recovery by steam flooding. Steam flooding oil recoveries resulting from injection of various mixtures of gas and steam were investigated for both light- and heavy-oil reservoirs. The results indicate that noncondensable gas injection with steam can accelerate production significantly early in the life of a typical heavy-oil project, but the cumulative recovery over a 5-year steam injection period is about the same as that obtainable with steam injection alone. The early production increase was seen to result mainly from the additional sweep of the reservoir provided by the injected gas. In a typical light-oil reservoir, the injection of non-condensable gas was seen to accelerate the oil recovery as a result of increased volume of the displacing gas phase and lowering of the oil viscosity by gas dissolution in the oil. There also was a small (6 to 7%) increase in oil recovery over a 5-year steam injection period. This increase is attributable to enhanced steam distillation and viscosity reduction of the oil by oil-soluble gas. Introduction A number of new processes for generating steam, including downhole steam generators, involve injection of noncondensable gas with steam into the reservoir-notably, the downhole steam generator developed by Sandia Natl. Laboratories and Zimpro's surface-operated direct wet air oxidation process. Companies marketing steam generation systems involving simultaneous injection of steam and noncondensable gas (CO2 and N2) have suggested that improved oil production resulting from the presence of a noncondensing production resulting from the presence of a noncondensing gas phase in the reservoir might be possible. Many mechanistic theories have been advanced to support this improved recovery, and some limited laboratory experiments tend to corroborate these theories. However, how oil recovery is affected by a complex, three-phase (water/oil/gas), multicomponent environment generated by gas/steam injection has not been studied in detail. More recently, reservoir simulation studies have evaluated the effects of noncondensable gas injection with steam on heavy-oil recovery. Some of these have shown dramatically accelerated oil recovery when CO2 is injected with steam as compared with steam injection alone. Other studies, however, show that the acceleration is only marginal-not enough to justify the additional costs of noncondensable gas injection. The results of these studies indicate that the response to noncondensable gas injection depends on reservoir and operating conditions and that a more comprehensive study is needed to determine the true potential of gas-steam injection to improve steamflood performance. Our study was carried out to determine if noncondensable gas injection can indeed accelerate and/or increase oil recovery and, if so, which recovery mechanisms contribute to the improved steam flood performance. A thermal compositional simulator was used for this study. Reservoir and Fluid Models Reservoir Grid. The reservoir model was an areal 7 × 4 grid system representing one-eighth of a repeated five-spot pattern (Fig. 1). The area is 5.3 acres [21 450 m 2]; the distance between the injector and producer is 340 ft [ 104 m]. The area was divided into seven blocks in the x direction, parallel to the line between the injecter and the producer, and four blocks in the y direction. Apex cells in the three corners of the triangle were combined with blocks adjoining them, resulting in a total of 22 active blocks in each layer. The massive 100-ft [31-m] sand was divided equally into four layers, all of which were open to injection and production. production. Reservoir Properties. Two types of reservoirs were considered for this study: heavy oil and light oil. The heavy-oil reservoir is characterized by shallow depth (less than 1,000 ft [305 m]), high permeability (4,000 md), and a spongy formation (compressibility: 2 × 10 -3 psi -1 [2.9 × 10–4 kPa -1). The initial oil saturation is 60 %; the initial oil in place (OIP) in the one-eighth of a five-spot is 100,000 bbl [15 900 m3] for the 100-ft [31-m] thick sand. The light-oil reservoir, on the other hand, is typified by a greater depth (more than 2,500 ft [762 m]), lower permeability (40 md), and a less compressible formation permeability (40 md), and a less compressible formation (compressibility: 5 × 10 -5 psi -1 [7.25 × 10 -6 kPa -1 ]). The initial oil saturation is 40%; the initial OIP is 64,500 bbl [10 250 m3] for the 100-ft [31-m]-thick sand. Table 1 shows important reservoir parameters used in the simulation study for both reservoir types. In all cases, the reservoir properties were assumed uniform. The vertical permeability was assumed to be 50% of the horizontal permeability. permeability. JPT P. 2160