Modeling CO2 Migration at Sleipner Using Models of Varying Complexity

Abstract:

The goal of geologic carbon sequestration (GCS) is to store carbon dioxide (CO₂) in the subsurface for time periods on the order of thousands of years. To ensure the safe storage of CO₂ in the subsurface, the migration of CO₂ and resident brine needs to be predicted. Mathematical modeling is an important tool to predict the migration of both CO₂ and brine. Many modeling approaches with different levels of complexity have been applied to the problem of GCS ranging from simple analytic solutions to full three-dimensional reservoir simulators. The choice of modeling approach is often a function of the spatial and temporal scales of the problem, reservoir properties, data availability, available computational resources, and the familiarity of the modeler with a specific modeling approach.
The Utsira Formation off the coast of Norway is the target formation of the Sleipner Project, where approximately 1 million tons of CO₂ are injected per year. The Utsira Sand consists of a Pliocene sandstone with high permeability and porosity, interbedded with thin mudstone layers that act as baffles for vertical flow. CO₂ is injected at the bottom of the formation and collects under the mudstone baffles as it migrates to the top of the formation. The layer of sandstone between the topmost mudstone baffle and caprock is termed the 9^th layer. Geometrical and petro-physical data of the 9^th layer have been made publicly available, and are the basis for this modeling study.
In this study we apply a series of models with different levels of model complexity to the 9^th layer of the Utsira Sand. The list of modeling approaches includes (from least complex to most complex): macroscopic invasion percolation model, numerical vertical-equilibrium model with sharp-interface, numerical vertical-equilibrium model with capillary transition zone, vertically-integrated model with dynamic vertical pressure and saturation reconstruction, and full three-dimensional model. The models are compared based on the predicted CO₂ plume footprints. The predicted CO₂ plumes are also compared to the actual CO₂ plume footprint from seismic surveys to determine the ability of the different models to predict the actual CO₂ plume footprint.