The Sleipner CO2 storage site: using a basin model to understand reservoir simulations of plume dynamics

When simulating CO2 storage, an accurate match to the observed CO2 plume distribution is a prerequisite to establishing the dominant flow physics, and forecasting the storage site behaviour beyond the observed and expected injection period. The scale of industrial CO2 storage pilots such as Sleipner, offshore Norway, is similar to that of small hydrocarbon fields, and lends itself to reservoir simulation. However, the reservoir conditions and dynamics are significantly different: oil and gas production are dominated by imbibition, which is suited to multi-phase Darcy flow simulation; whereas CO2 storage represents the injection of a non-wetting fluid that displaces the in situ brine. The latter process is often termed ‘drainage’, and with respect to simulation, is more typical of regional basin modelling and percolating oil and gas migration. The challenge of modelling CO2 storage is to accurately represent this drainage displacement at the reservoir scale on short decadal timescales. The advantage is the detailed observational dataset with which such models are constrained. Using a Darcy flow model, the first decade of reservoir simulations for Sleipner has been characterized by poor matches to the known plume distribution, and forecasted plume dynamics that persisted for decades-to-centuries beyond the injection period. To overcome this problem of poor simulated replication, and to test the veracity of long term plume dynamics, we applied a basin model to Sleipner, which simulated the gravity-dominated migration of a buoyant fluid using a capillary percolation method. The basin model achieved an accurate match to the observed CO2 plume distribution. This suggests that simple Darcy-based reservoir simulation forecasts are misleading. The basin modelling insights allowed us to revisit the reservoir simulations, and, focusing on benchmark models of the uppermost layers, approximate the gravity-dominated regime of percolating flow. A pressure-compensated black oil reservoir simulation accurately matches the distribution and dynamics of the uppermost layers. The reservoir simulations also indicate that dissolution of CO2 will contribute significantly to storage within decades. While both approaches have their limitations, a combination of basin modelling and reservoir simulation provide perspectives that illuminate the dominant flow physics processes within the storage site, implying that the plume is in a state of dynamic equilibrium and likely to stabilize within years of the injection ending. Two challenges remain for the benchmark reservoir simulations: (A) how to represent the trapping and breaching behaviour of thin shale barriers for percolating CO2 within a storage formation; and (B) how to address pressure field artifacts in larger regional Darcy flow models of CO2 storage.