Field-Scale Modeling of Local Capillary Trapping during CO2 Injection into a Saline Aquifer

Local capillary trapping is the small-scale (10^-2 to 10⁺¹ m) CO₂ trapping that is caused by the capillary pressure heterogeneity. It occurs when CO₂ migrates upwards under buoyancy in the subsurface saline aquifer and encounters local capillary barriers (regions of rock with large capillary entry pressure). CO₂ would accumulate beneath these small barriers, and these accumulations are called local capillary traps. Local capillary trapping benefits storage because locally trapped CO₂ has a much larger saturation than residual gas. Such trapped gases cannot escape from the formation, even if leakage conduits (fractures or fault) in the seal develop during the long-term storage of CO₂. Thus quantifying the extent of local capillary trapping is valuable in design and risk assessment of geologic storage projects.

Modeling local capillary trapping is computationally expensive and may even be intractable using a conventional reservoir simulator. In this paper, we propose a novel method to model local capillary trapping by combining geologic criteria and connectivity analysis. The connectivity analysis originally developed for characterizing well-to-reservoir connectivity is adapted to this problem by means of a newly defined edge weight property between neighboring grid blocks, which accounts for the multiphase flow properties, injection rate, and gravity effect. Then the connectivity was estimated from shortest path algorithm to predict the CO₂ migration behavior and plume shape during injection. A geologic criteria algorithm is developed to estimate the potential local capillary traps based only on the entry capillary pressure field. The latter is correlated to a geostatistical realization of permeability field.

The extended connectivity analysis shows a good match of CO₂ plume computed by the full-physics simulation. We then incorporate it into the geologic algorithm to quantify the amount of LCT structures identified within the entry capillary pressure field that can be filled during CO₂ injection. Several simulations were conducted in the reservoirs with different level of heterogeneity (measured by the Dykstra-Parsons coefficient) under various injection scenarios. We find that there exists a threshold Dykstra-Parsons coefficient, below which low injection rate gives rise to more LCT; whereas higher injection rate increases LCT in heterogeneous reservoirs.

Both the geologic algorithm and connectivity analysis are very fast; therefore, the integrated methodology can be used as a quick tool to estimate local capillary trapping. It can also be used as a potential complement to the full-physics simulation to evaluate safe storage capacity.