Tenth Year Annual Report:
Carbon Storage: Simulation of CO2 Injection
Several tools aimed at improving numerical models for simulation of geological storage of CO2 have been developed in Jean H. Prévost's group. These tools allow capture the effects of coupling between fluid flow, thermal and geomechanical effects. Furthermore, with these advanced tools taking into account coupling between multiphase fluid flow, heat conduction and geomechanics, a mechanism leading to a previously undocumented risk scenario in CO2 injection operations has been identified. The tools have been implemented in dynaflow, which offers a modular, hierarchical approach for multiphysics simulations. Dynaflow allows increasing more complex models to be used to simulate physical processes with increasing accuracy.
Capturing Nonlinear Near-Well Behavior and Coupled Poromechanics
Numerical modeling of CO2 injection tends to focus on fluid flow and pressure evolution, whereas geomechanical effects are disregarded, or taken into account by means of simple approximations. Work by Prévost and postdoctoral fellow M. Preisig presents a framework that allows a correct representation of the stress state in the vicinity of a well. This has led to several developments or improvements of numerical methods:
- a stabilization procedure has been developed that avoids pressure oscillations at very small time steps.
- a consistent methodology has been presented allowing the establishment of accurate initial stresses in the near-well region after drilling of the well.
- a novel procedure for specifying injection boundary conditions in heterogeneous or anisotropic porous media has been developed.
- These methods have been added to the toolbox for modeling flow in porous media in the software dynaflow.
Thermo-Hydromechanical Coupling in a Practical CO2 Injection Scenario - In Salah
The design of dynaflow offers a modular, hierarchical approach for multiphysics simulations. Starting from a simple single-phase flow simulation, additional physical processes and additional phases can be added step-by-step, thus allowing adaptation of the numerical model to the degree of complexity required by each task. The full capabilities of dynaflow have been put to a test in a case study of the ongoing CO2 injection at In Salah, Algeria. Coupling all relevant physical processes in a simulation of CO2 injection is crucial for assessing feasibility, safety and productivity of such an operation. Simplified coupling strategies can be used in individual situations, but their accuracy can only be assessed by comparison with a fully coupled simulation. The Prévost Group's study with full thermo-poromechanical coupling illustrates a mechanism that may lead to tensile fractures in the sealing cap rock. Injection of a cold fluid into a reservoir at higher temperature causes thermal stresses, which in this case have been shown to create and/or re-open fractures in the cap rock perpendicular to the well (Figure 13).
Error Analyses and Sensitivity Studies
The work of Prévost and graduate student I. Goumiri focused on two aspects of interest in the field of reservoir modeling. First, while Galerkin finite element methods are well suited for solving geomechanical problems, the transport equation, which has to be solved when considering multiphase flow, is best solved with finite volume methods. To couple traditional finite volumes, where the unknowns are located at cell centers, with Galerkin methods, where they reside at vertices, two approaches are possible - one requires cell to node projections, while the one that is successfully used in dynaflow uses a vertex centered finite volume implementation to circumvent the problem. Cell to node projections introduce an unnecessary error in a simulation. Assessing this error systematically has been one goal of I. Goumiri's work.
The second study dealt with the influence capillary pressure has on the distribution of a phase invading a porous medium saturated with another phase. The investigation showed that in most cases relevant in practical CO2 injection scenarios, the capillary pressure only affects the CO2 saturation in a very limited region at the front of the invading plume. Capillary pressure can therefore be safely neglected in a reservoir-scale simulation.