Water-alternating-gas (WAG) flooding has been successfully used as development alternative providing frontal mobility control for better conformance, potential reduction of residual oil saturations to water, particularly when miscibility is achieved, and relative permeability changes as result of the cyclic injection. Forecasting performance of the WAG process often involves the use of numerical simulation models that allow the inclusion of capillary, viscous, gravity forces, compositional and hysteresis onto the frontal movement, recovery and injection requirements. It is not uncommon to encounter models that are not necessarily built to address the specific challenges of the WAG displacement and thus require validation and modifications to improve consistency and predictive power. WAG processes are traditionally characterized through a series of laboratory and field observations, a process that starts with corefloods where injection volumes, sequence and water-gas proportions (among others) are optimized and its results are used to calibrate the necessary parameters to represent the WAG process in the dynamic models by simulating the laboratory experiments. These core-level calibrated parameters are often directly used without modification even though the parameters are unique to the scale at which they were calibrated. Investigation into practical translation and use of these core-level parameters on pilot and full field numerical models is the main objective of this paper.
This paper builds on the results of previous work which described a multi-stage upscaling process by introducing a series of 1D, 2D and 3D models of varying resolution ranging from core scale to expected full field resolution. The models are built to represent average well spacing as well as reproduce expected frontal advancement. The high-resolution models are used as reference since they are closest to the core scale and the results of the different numerical resolutions are used to determine upscaling requirements (concentrated on immiscible water-alternating-gas - iWAG) as well as investigate the impact and limitations on the magnitude of the upscaling. A realistic full field model is used to validate the impact of the upscaled iWAG parameters on different model resolutions.
Results of the investigation were in line with previous work (Talabi et al. 2019) identifying significant deviations in the oil, water and gas production for horizontal grid resolutions beyond 50ft where numerical dispersion and dilution accounted for an early water arrival and subsequent decrease trapping of the non-wetting phase. Oil recovery was consistently underestimated by 2-8% as the grid resolution increased. The upscaling strategy proposed involves the modification of trapping parameters to account for the numerical dilution effect: increase of non-wetting phase relative permeability reduction factor and three-phase wetting phase relative permeability was sufficient to reproduce the overall fine scale mobility (to overcome dispersion/dilution effects). Acceptable calibrations that allow coarse models to reproduce the fine scale results were obtained on the 1D/2D/3D models. As expected, the magnitude of the WAG parameter changes was process and heterogeneity dependent (Moreno et al. 2011) requiring fine tuning as model complexity increased.
While the complexity, heterogeneity and fluid dependence of the WAG process is recognized, a multi-stage upscaling approach as described in this paper, is offers an opportunity to better understand the process drivers and design a practical upscaling strategy suitable to accommodate the full field operation and avoid underprediction of WAG performance.