Solution of Benchmark Problems for CO2 Storage Min Jin, Gillian Pickup and Eric Mackay Heriot-Watt University Institute of Petroleum Engineering Outline • Introduction • Problem 1 – Leakage through an abandoned well • Problem 2 – Enhanced methane recovery • Problem 3 – Storage capacity in a geological formation • Conclusions Numerical Simulation • Simulation is a very important tool for CO2 storage • Can give estimates of – – – – migration of CO2 gas dissolution in brine build-up of pressure around injection well etc Reliability • Depends on – Input data • geological structure • rock permeability/porosity measurements • laboratory measurements • Also depends – Adequate computer models • flow equations • representation of physical processes Reservoir Simulation • Codes are complex • Various different versions available for – gridding model – calculating fluid properties – solving equations • May get slightly different answers Benchmark Problems • Compare solutions using different codes • If results are the same – gives confidence in simulation results • If they are different – indicates where more work is needed Stuttgart Workshop, April 2008 • Aim – Discuss current capabilities of mathematical and numerical models for CO2 storage • Compare results of 3 benchmark problems • Focus model development on open questions and challenges • 12 groups participating web site: http://www.iws.uni-stuttgart.de/co2-workshop/ Heriot-Watt Entry • Solutions to all 3 problems • Eclipse 300 – Reservoir simulation software package – Compositional simulation – Schlumberger Outline • Introduction • Problem 1 – Leakage through an abandoned well • Problem 2 – Enhanced methane recovery • Problem 3 – Storage capacity in a geological formation • Conclusions Problem 1 • CO2 plume evolution and leakage through an abandoned well leaky well k = 200 mD, f = 0.15 aquifer k = 0 mD, f = 0.0 aquitard k = 200 mD, f = 0.15 aquifer 1000 m Problem 1 • CO2 plume evolution and leakage through an abandoned well leaky well CO2 injector aquifer aquitard aquifer Problem 1 • CO2 plume evolution and leakage through an abandoned well leaky well CO2 injector ? aquifer aquitard aquifer Model Details • Lateral extent of model: 1000 m x 1000 m • Separation of wells: 100 m • Aquifer thickness: 30 m – perm: 200 mD, poro = 0.15 • Aquitard thickness: 100 m – impermeable • Abandoned well – model as thin column of 1000 mD, poro = 0.15 Details of Fluid Properties • Problem 1.1 – Reservoir is very deep, ~3000 m – Simplified fluid properties • constant with P and T • Problem 1.2 – Shallower reservoir, <800 m – CO2 can change state when rising – More complex fluid properties Other Inputs to Simulation • Constant injection rate – 8.87 kg/s • Pressure should stay constant at the edges of the model • No-flow boundaries top and bottom Challenges x • Gridding – Coarse over most of model – Fine near wells y Close-up of Grid Centre injector leaky well Challenges • Modelling of abandoned well a) Model as high perm column b) Model as closed well • output potential production high perm cells closed well Challenges • • Maintaining pressure constant at boundaries Eclipse designed for oil reservoirs – assumes sealed boundaries • leads to build up of pressure • We added aquifers to sides of the model – fluids could move into the aquifer – prevented build up of pressure Challenges • Fluid properties in Problem 1.2 a) User-defined b) Specified as functions of pressure and temperature • We used constant T = 34 oC – Tuned equations • density and pressure similar to specified values CO2 Distribution after 100 Days, Problem 1.2 Injector Leaky well Gas Sat 0.0 0.2 0.4 0.6 0.8 CO2 Distribution after 2000 Days, Problem 1.2 Inj leaky well Gas Sat 0.0 0.2 0.4 0.6 0.8 Results • Leakage rate for Problem 1.2 0.14 leakage volume/injection volume (%) 0.12 0.10 0.08 0.06 0.04 leaky well modelled as high perm cells 0.02 0.00 0 500 1000 1500 time (day) 2000 2500 Summary of Problem 1 • Successfully predicted well rate – Using high perm cells for leaky well • well model overestimated leakage – Our results similar to others • Leakage rate ~ 0.1% injected volume Outline • Introduction • Problem 1 – Leakage through an abandoned well • Problem 2 – Enhanced methane recovery • Problem 3 – Storage capacity in a geological formation • Conclusions Problem 2 • Enhanced recovery of CH4 combined with CO2 storage CO2 injector producer 45 m kh = 50 mD kv = 5mD f = 0.23 200 m 200 m Model Details • Two versions 1. homogeneous 2. layered • • • Temperature = 66.7 oC Depleted reservoir pressure = 35.5 bar Molecular diffusion = 6 x 10-7 m2/s Model for Problem 2.2 x P I z Perm (mD) 0 10 20 30 40 50 60 70 80 90 100 Other Inputs to Simulation • Constant injection rate for CO2 – 0.1 kg/s – inject into lower layer – produce from upper layer • Constant pressure at production well – P = 35.5 bar • No-flow across model boundaries Challenges • Mixing of gases • Changes in physical properties of gas mixture with composition – can be modelled in Eclipse 300 • Numerical diffusion – will artificially increase the molecular diffusion Result for Problem 2-1 Results – Homogeneous Model • Mass Flux of CH4 and CO2 3000.00 mass flux (kg/d) 2500.00 2000.00 1500.00 1000.00 500.00 0.00 0 200 400 600 800 1000 1200 time (day) CH4 CO2 1400 1600 1800 2000 Results – Layered Model • Mass Flux of CH4 and CO2 3000 mass flux (kg/d) 2500 2000 1500 1000 500 0 0 200 400 600 800 1000 1200 time (day) CH4 CO2 1400 1600 1800 2000 Results and Summary • Assume well is shut down when CO2 production reaches 20% by mass Problem Model Shut-in time (days) Recovery Efficiency (%) 2.1 homogeneous 1727 59 2.2 layered 1843 64 • Relatively easy problem Outline • Introduction • Problem 1 – Leakage through an abandoned well • Problem 2 – Enhanced methane recovery • Problem 3 – Storage capacity in a geological formation • Conclusions Problem 3 • Storage capacity in a geological model y Inj x porosity 0.17 0.19 0.21 0.23 0.25 z Model Details • Lateral dimensions – 9600 m x 8900 m • Formation thickness – between 90 and 140 m • Variable porosity and permeability • Depth ~ 3000 m • Temperature = 100 oC Challenges • Simulation of system after injection has ceased – CO2 continues to rise due to buoyancy – Brine moves into regions previously occupied by CO2 – Brine can occupy small pores, trapping CO2 in larger pores • additional trapping mechanism • hysteresis Challenges • Trapping of CO2 by hysteresis CO2 displacing brine Plume of rising CO2 brine displacing CO2 after Doughty, 2007 CO2 Distribution after 25 Years X with hysteresis Y Gas Sat 0.0 0.2 0.5 0.8 CO2 Distribution after 50 Years X with hysteresis Y Gas Sat 0.0 0.2 0.5 0.8 Results • Mass of CO2 in formation over time 1.4E+10 Mass of CO2 (kg) 1.2E+10 1.0E+10 8.0E+09 total free dissolved 6.0E+09 4.0E+09 2.0E+09 0.0E+00 0 5000 10000 Time (days) 15000 20000 Results • Leakage of CO2 across the boundaries CO2 inter-region mass flow rate for Problem 3 1.6 no hysteresis P3-1 P3-2 1.4 Mass Flow rate (kg/s) 1.2 1 0.8 with hysteresis 0.6 0.4 0.2 0 0 2000 4000 6000 8000 10000 Time (day) 12000 14000 16000 18000 20000 Summary of Problem 3 • CO2 did not move towards the fault – moved up-dip – leaked across model boundary • Hysteresis did make difference, but not much difference in this example • About 0.2 of the injected CO2 dissolved after 50 years Outline • Introduction • Problem 1 – Leakage through an abandoned well • Problem 2 – Enhanced methane recovery • Problem 3 – Storage capacity in a geological formation • Conclusions Conclusions • Benchmark solutions highlight difficulties – Adaptation of simulator for oil/gas reservoirs to CO2 storage – Difficulties are surmountable – Schlumberger created new module for CO2 storage • Participation in the workshop – Giving us confidence in simulations Acknowledgements • We thank Schlumberger for letting us use the Eclipse simulation software Solution of Benchmark Problems for CO2 Storage Min Jin, Gillian Pickup and Eric Mackay Heriot-Watt University Institute of Petroleum Engineering