Large-scale multiphysics CO2 storage simulation

There are several critical deficiencies in present CO₂ simulators:

1. Geochemical reactions are generally missing in most commercial reservoir simulators
2. The classical simulators are not built upon a full compositional framework
3. Coupling of multiphase flow and geomechanics is rarely addressed in large-scale simulation analysis. In the CompReact project funded under the INNO-CCUS program, we plan to develop a next-generation compositional CO₂ storage simulator with multiphase geochemical reactions. 

The current simulators should take advantage of the advances in high-performance computing, thermodynamic modeling, and multiphase reaction calculation to deliver dramatically improved efficiency and robustness.

The simulator developed in this project will be applicable to a wide range of geological reservoir types (aquifers and depleted petroleum reservoirs) and have dramatically improved robustness and efficiency compared to existing simulators.

This will be achieved by utilizing the novel RAND-based multiphase reaction algorithms recently developed at DTU.

These algorithms treat phase and chemical equilibrium simultaneously, avoiding the conventional sequential solution.

Furthermore, in collaboration with Stanford University, we will couple our algorithm with the GEOS simulator developed by Stanford University, LLNL (Lawrence Livermore National Laboratory), and TotalEnergies.

GEOS is a subsurface, non-compositional simulator with coupled multiphase flow and geomechanics. The collaboration will create synergy based on the complementary expertise, bringing the compositional and geochemical features to GEOS.

This PhD project belongs to work package 2 in CompReact. Its focus is to resolve the coupling of the geochemical reaction module developed in work package 1 with GEOS.

Since GEOS utilizes the modern architecture for high-performance computing, it needs to be taken into account during the integration. In addition, two other technical developments will be considered.

The first is the reduction of the compositional and geochemical treatment to avoid a dramatic increase in computation time when more physics is included.

This should be done without a significant compromise in the simulation fidelity, which is important to the long-term forecast after injection.

The second is upscaling and treatment of non-equilibrium effects, which is crucial for large time steps and grid sizes. The new GEOS simulator will be applied to long time-scale post-injection simulations with coupled geomechanics.