One proposed method to delay the onset of gravity segregation between water and gas in enhanced oil and gas projects and extend the period of effective macroscopic sweep in “SWAG” is by separating the injection wells into two parallel horizontal wells (Stone, 2004). In this “modified SWAG”, the water injection well is aligned at a distance above the gas injection well, and both water and gas are pumped simultaneously to displace the reservoir fluids. The significant density difference between the fluids and the ensuing counter-direction flow impede the segregation process. Initially Rossen et al. (2007) investigated the effectiveness of the technique based on 2D modelling and found that it increased the fluid segregation length. Van der Bol (2007) and Jamshidnezhad et al. (2010) broadened the scope of study and observed a non-uniform gas injection profile and volumetric sweep in 3D. Injection instability occurred in most of the simulation cases, despite the assumption of an ideally homogenous reservoir in the model. Mahalle (2013) identified several factors that triggers instability such as gas saturation and relative permeability behaviors in adjacent grid blocks. Since the instability was found to originate in the near-wellbore region, local grid refinement was applied along the entire length of the horizontal well, with more uniform gas-injection profile observed. Using a different reservoir simulator, Ranjan (2015) extended the previous study by checking again the effect of local grid refinement. The result, however, contradicts with the preceding finding. Grid refinement near the injection well did not improve stability in Ranjan’s study. The author also checked the effect of gas injection rate on the non-uniformity. Again, contradictory results were observed. While Ranjan reported that doubling the gas injection rate promotes non-uniform behavior, the earlier study obtained the opposite outcome. This thesis extends the previous studies by examining other parameters that may influence non-uniformity. We developed a method to quantify non-uniformity by calculating the coefficient of variation and max-min ratios for gas injection rate along the well. We looked at the effect of changes in well placement, reservoir properties, reservoir boundaries, reservoir fluids and operating constraints. We also fundamentally modified how the perturbation is applied along the gas-injection well by altering the skin factor while maintaining constant permeability. Results show that the type of perturbation significantly effects the non-uniformity of gas injection. We believe that perturbing permeability promotes the uniformity of gas-injection rate because of flow to neighboring grid blocks, and thereby more simulations are seen to be uniform compared the results with perturbation in skin factor. The results from this study suggests that non-uniformity is associated to the feedback between gas injection rate, water saturation and gas relative permeability, which is shown by the gas injection rate to vary more than proportionally to permeability or skin, even in relatively uniform cases. The effect of adjacent grid blocks also plays a crucial role, as we can see from the different results between the two types of perturbations. Finally, the mobility ratio of the fluids strongly influences the occurrence of the instability.