270 likes | 287 Views
The CO2-PENS system model supports risk-based decisions for CO2 sequestration. It includes a performance assessment framework and logic for reduced complexity in problem-solving. Collaborators are needed for process modules. The approach uses existing knowledge and risk-based decisions to generate distributions and predict outcomes. An example problem on reservoir injectivity showcases the effectiveness of the system model in reducing complexity and computational time.
E N D
CO2-PENSA CO2 SEQUESTRATION SYSTEM MODEL SUPPORTING RISK-BASED DECISIONS PHILIP H. STAUFFER HARI S. VISWANTHAN RAJESH J. PAWAR MARC L. KLASKY GEORGE D. GUTHRIE
Talk Outline • PART I • Description of CO2-PENS system model • Part II • Example problem: Reservoir Injectivity with reduction of complexity • problem description • logic for reduced complexity • results
Talk Outline • PART I • Description of CO2-PENS system model • Part II • Example problem: Reservoir Injectivity with reduction of complexity • logic for reduced complexity • codes used • results
Performance assessment framework for geologic sequestration • From the power plant • Into the Ground • Back toward the Atmosphere • Entire CO2 sequestration analysis • System analysis yields meaningful site comparisons • Provides consistent output for Quality assurance/Quality control
Linking Process-Level Modules to a System Model system model (probabilistic) CO2 release fluid flow geochemical reactions process-levelmodels
Big problem: collaborators needed for process modules. • Princeton – analytical well bore leakage • MIT – surface pipeline model • Atmospheric scientists • Economists • Modular design means flexibility • CO2 multiphase reactive transport codes: FEHM, FLOTRAN, TOUGH. etc. • Analytical solutions
Risk-Based Decisions • Predictions use probabilistic approach • Sampling of multidimensional solution spaces • Reduced complexity: abstraction, lookup tables • Generate distributions from experiment, modeling and expert opinion
Use existing knowledge: • Theory, experiment, lessons learned • Industry data (Kinder-Morgan), Weyburn, Sleipner • Performance assessment experience (Yucca Mountain, WIPP, Oil/gas, Los Alamos Environmental) • Economic experts • Risk theory experts
Reduced complexity reservoir injection module • Analytical single fluid approximation run as a dynamic link library from GoldSim • 2-D radial, multiphase finite volume calculations used to ‘tune’ the analytical solution
Analytical Approximation of Injection • single fluid • no relative permeability model • uses reservoir PT CO2 viscosity and density • infinite radius with pressure fixed at Pini • runs very quickly as a dynamic link library • can be coded in FORTRAN, C++ etc. • Reference C.S. Matthews and D.G. Russel, (1967). Pressure Buildup and Flow Tests in Wells, Society of Petroleum Engineers, Monograph Vol 1, New York.
FEHM 2-D Radial Simulation of Injection and Plume Growth • Control volume finite element method • Multiphase heat and mass transfer • Relative permeability (H20-CO2) • All constitutive relationships are in the code (e.g., density, viscosity, enthalpy)
Comparison of FEHM with published results 5 km x 30m deep radial grid Nordbotten et. al, (2005) FEHM
Example Problem Description • 30 m deep section • No flow top and bottom boundaries • Far-field at background pressure • CO2 coming from a 1 GW power plant for 50 years (300 Mt CO2)
Linear Effective Stress Relationshipminimum principle stress = 0.65 lithostatic • Gives • maximum injection pressure • Reservoir background pressure Two cases
Points were simulated in FEHM to span a range of permeability and porosity Porosity 0.13 0.15 0.17 + stdv 5e-14 m2 mean 1e-14 m2 Permeability - stdv 5e-15 m2 mean - stdv +stdv
FEHM simulations versus analytical solution These plots yield are used to “tune” the injector code to recreate FEHM behavior in GoldSim
Computational time • Goldsim calling the tuned analytical solution • 1000 realizations in 6.5 minutes. • Includes passing all variables through the framework, generating output and storing results. • FEHM simulations, 700 nodes • 10+ minutes each • (some issues with iterative solver used for CO2 EOS, we are implementing a lookup table approach)
Economic Risk Cold + Shallow 1 km Hot + Deep 3 km
Health/Environmental Risk Cold + Shallow 1 km Hot + Deep 3 km
Engineering RiskLeakage from the Reservoir Percent per Year Percent Total
Conclusions • Tuned analytical solution is much faster than running a reservoir solver • reduced complexity will be vital for performing risk analysis • Integrated approach shows interactions between different types of data and outcomes
THANK YOU Please contact me if you are interested in collaborating on process level modules or the system model