ParaGeo-Pflotran coupling for multiphase flow modelling

Coupling highlights

Over the last few months, we have been developing the framework and tools required for INCAST project; a research project focusing on the simulation of CO2 injection accounting for potential fault reactivation and assessment of any associated seismic events. A key development is the coupling of ParaGeo to Pflotran and PflotranOGS.The implemented interface allows to run both; 2 way and 1 way coupled simulations where Pflotran solves for the multiphase flow (and optionally thermal) fields and ParaGeo solves the geomechanics. All the advanced models and functionality within ParaGeo are made available to coupled simulations, including:

• Automatic Pflotran grid built from the ParaGeo mesh via export of the selected groups (formations) that need to be solved by Pflotran. This allows to quickly obtain the flow model grid as well as any key region to perform assignments in Pflotran (boundary conditions, material properties, etc). This feature in conjunction with ParaGeo capability of importing meshes from 3rd parties (Abaqus, Gmsh, Eclipse, Zmap) opens a very flexible model building workflow.

  

  Schematic of ParaGeo-Pflotran model building workflow

Mechanical (ParaGeo) and flow (Pflotran) model geometry and mesh/grid in a coupled simulation. Only the reservoirs and the fault are solved by Pflotran whereas ParaGeo solves the geomechanics in all formations.

• Mapping algorithms for cases with non-matching meshes (e.g. different resolution or element type between the Geomechanical and flow model)

  
  

• Wide range of constitutive models such as poro-elasticity with hysteresis, anisotropic, SR4 plasticity, viscoplastic creep, thermal expansion, etc

  
  

• Contact mechanics with specialised mapping of properties between discrete faults in the Geomechanical model (ParaGeo) and continuum faults in the flow model (Pflotran). This includes advanced constitutive laws such as modelling the evolution of fault aperture and permeability as function of frictional slip.

  
  

• Parallelisation to reduce the CPU time cost when running models discretized in several 10’s of Millions of elements

  
  

Tutorial examples on ParaGeo-Pflotran coupling

We have documented comprehensive tutorial examples within the manual which guide our users in using all the implemented functionality, from model building to the setup of the data for the simulation and to the post processing of the output results. See below some results extracted from one of our tutorial examples involving a coupled simulation of methane gas production followed by CO2 injection performed on the deepest reservoir.

Location of flow boundary conditions and some poro-elasto-plasticity SR4 properties for the reservoir formations in a coupled simulation from one of our tutorial examples.

The animation of gas saturation (which accounts for both, methane and CO2 shown on the left), mass fraction of methane (centre) and mass fraction of CO2 (right) shows that at the end of production stage there is still remaining methane accumulated at the crest of the anticlines. When CO2 is injected the plume flows towards the crest of the anticline in the deepest reservoir but due to the lighter density of methane, the latter acts as a barrier for the CO2. Thus with ongoing injection CO2 surrounds the methane and flows below it.

Animation of gas saturation (methane + CO2), mas fraction of methane (gas) and mass fraction of CO2 (solvent) in a simulation of methane production followed by CO2 injection from one of our tutorials. Top row shows both reservoirs whereas bottom row shows only the deepest reservoir.

At the end of gas production pore pressure has decreased leading to an increase in effective mean stress. This resulted in plastic compaction (porosity reduction) which in turn resulted in stiffening (increase in Youngs modulus) due to the modelled dependence of bulk modulus on both pre-consolidation pressure and effective mean stress. At the end of CO2 injection pore pressure has increased above the initial values while effective mean stress is reduced leading to elastic unloading. Note that the effective mean stress is still larger than the initial values in some locations due to the effect of pre-consolidation pressure on stiffness.

Pore pressure (top) and effective mean stress (bottom) distributions in the deepest reservoir at three key stages during the simulation. Note that in stress plots negative values indicate compression.

Porosity and Young's modulus distribution in the deepest reservoir at three key stages during the simulation.