287693 Numerical Simulation of Reservoir-Wellbore Fluid Flow Simulations At Geologic CO2 Sequestration Sites: Coupled or De-Coupled?

Thursday, November 1, 2012: 12:30 PM
301 (Convention Center )
Rajesh Pawar1, Haruko Wainwright2, Robert Dilmore3, Jens Birkholzer2, Jim Houseworth2 and Grant Bromhal4, (1)Computational Hydrology Group, Los Alamos National Laboratory, Los Alamos, NM, (2)LBL, Berkeley, CA, (3)National Energy Technology Laboratory, U.S. Department of Energy, Pittsburgh, PA, (4)National Energy Technology Laboratory, Morgantown, WV

Leakage of CO2 and brine through pre-existing wellbores is considered to be one of the main potential risks at geologic CO2 sequestration sites. To assess impacts of this potential weakness, it is critical to accurately estimate leakage rates through failed wellbores. Currently, there is very limited field data available on CO2 leakage rates through wells. In lieu of field data, numerical simulations that accurately capture the nature of multi-phase fluid flow through wellbores can be used to provide estimates of the leakage rates for a variety of wellbore leakage parameters. A few earlier studies have reported results of numerical simulations of leakage of CO2 and brine through wellbores from CO2 storage reservoirs. There are different types of approaches used for modeling wellbore flow:
  • Models that include only the wellbore flow and not the sequestration reservoir. These models assume constant pressure and/or saturation boundary conditions at the bottom hole.    
  • Models that include coupled wellbore and sequestration reservoir flow. These models solve the equations for fluid flow in the wellbore and reservoir implicitly (either semi-analytical or numerical approaches).
  • System models that include coupled wellbore and sequestration reservoir flow. Unlike the models in the second bullet above, the coupling in these models is not implicit.

System models are being increasingly used for quantitative risk assessment (QRA) of geologic CO2 sequestration operations. Typical QRAs require a probabilistic treatment of quantities such as wellbore leakage rates and need a large number of Monte-Carlo realizations of system performance to predict the likelihoods of different types of events. Most system models uses the following approach:  1) divide the full system (e.g., the sequestration reservoir to the surface) into component sub-systems, 2) use some type of approximation such as a reduced order models to capture the behavior of each sub-system, and 3) combine those component models into a large, but efficient, model of the full system. This approach assumes that the component models can be combined into the full systems model without any feedback. It is quite computationally efficient when the reduced order models are developed to produce very rapid results. Typically, significant effort is required to develop appropriate reduced order models that can accurately capture the underlying physics of a system component.

One concern for carbon storage systems, however, is the conditions under which the assumption that the separate systems model components produce negligible feedback between each other is valid. For wellbore leakage, the multi-phase fluid flow through or around the well is dependent on the evolution of pressure and saturation in the reservoir which is contact with wellbore. The reservoir pressures and saturations in turn will be affected by the amount of leakage through wellbore and the geologic properties of reservoir. In this study, we are testing the validity of the no feedback approximation of coupled wellbore-reservoir fluid flow by a model comparison study. We utilize both Finite Element Heat and Mass (FEHM), Los Alamos National Laboratory’s (LANL) fluid flow simulator, and TOUGH2-ECOM, Lawrence Berkeley National Laboratory’s (LBNL) fluid flow simulator. These simulators were used to perform detailed numerical studies of leakage rates through wellbores in a coupled wellbore-reservoir model. We also performed simulations of fluid migration through wellbores decoupled from the reservoir and without feedback, using some semi-analytical models for wellbore leakage, as well as LANL’s CO2-PENS system model for geologic CO2 sequestration sites. Appropriate wellbore flow model boundary conditions were investigated.  Results of these model calculations are compared to determine the validity of an approximation of no feedback in the system models.

Extended Abstract: File Not Uploaded