Benchmark study of CO2 Storage in the Johansen Formation
Research output: Research › Master's Thesis
The remarkable increase of greenhouse gas in the atmosphere, particularly carbon-dioxide (CO2) is often assumed as one of the main causes of the climatic changes. A potential short-term mitigation is a reduction of anthropogenic CO2 emissions into the atmosphere via capture and geological storage. An important aspect of understanding the migration process is the simulation of large-scale projects. The aim of this study is to benchmark existing commercial and non-commercial simulators using a candidate site for large scale CO2 sequestration as a simulation model. The setup of the model involves injection of carbon-dioxide into the Johansen formation. For the simulation of this essentially two-phase process, three different simulators were used: CMG, Eclipse and OPM. The following properties are compared: change of water volume and injected gas volume in the field (under reservoir and surface conditions), pressure and saturation profiles of the two phases and simulation run-times. The proposed benchmark is subdivided into two cases: the first case considers a model where both fluids are incompressible, while the second case considers fluids with varying degree of compressibility. In both cases the two-phase fluid flow is assumed isotherm and immiscible, where geo-mechanical and geochemical effects are neglected. The aim is to compare the capabilities and limits of the reservoir simulators using the same conceptual model definitions for input. The same corner point grid, identical PVT tables and relative permeability curves for the fluid and rock model, equivalent initial conditions and well constraints were used for all simulators. Comparing the computational efforts needed for the two cases has shown that calculation time increases with the introduction of compressibility as well as the FVM-TPFA as a spatial discretization (i.e. in OPM). While using the commercial programs of Eclipse and CMG, the numerical simulations achieve a shorter run-time, however the definition of incompressible fluid flow is not exact. Another main source of dissimilarity across model setups is the implementation of boundary conditions.