Storing Carbon Dioxide in Coalbeds a Laboratory and Simulation Investigation

A.R. (Tony) KovscekStanford UniversityPetroleum Engineering DepartmentStanford, CA 94305-2220

kovscek@pangea.stanford.edu

Overview

• Stanford’s Sequestration Program• Experimental Program ECBM• Conceptual/numerical model development• Results• Permeability vs Gas Composition (prelim)• Conclusions

Geologic Storage of CO2GCEP: Global Climate and Energy Project

Project Components:

Site selection and evaluation: effective methods to assess the integrity of geologic seals that limit CO2 migration.

Fluid migration: efficient methods for predicting the flow paths and long-term fate of injected CO2.

Monitoring: appropriate tools for monitoring the state of injection projects at each stage.

Project Components:

Site selection and evaluation: effective methods to assess the integrity of geologic seals that limit CO2 migration.

Fluid migration: efficient methods for predicting the flow paths and long-term fate of injected CO2.

Monitoring: appropriate tools for monitoring the state of injection projects at each stage.

Site Selectionand Evaluation Mark Zoback and coworkers

• How will CO2 injection affect the reservoir seal?

• Did production and depletion affect the seal?

• What are the geomechanical mechanisms governing capacity?

Trap and Seal Framework

Fluid Migration : ECBM Prediction Lynn Orr, Tony Kovscek and Coworkers

• Such methods are useful to– predict where CO2 is likely to flow– interpret the volume and space contacted– optimize injection and recovery operations

• What assumptions necessary?

50 x 50 x 10 blocks

AcquisitionStrategies

Modeling &Simulation

Imaging &Inversion

GCEP 4D:• Continuous imaging, e.g.,

weekly or daily• Adaptable to dynamic

storage conditions• Exploits models for

porous media and flow prediction

ConventionalTime-lapse

High Spatial ResolutionLow Temporal Resolution

GCEP4-D

Lower Spatial ResolutionHigher Temporal Resolution

Monitoring in Real TimeJerry Harris and coworkers

Experimental Objective

• Establish a feasible CO2/CH4/N2adsorption/desorption model and displacement mechanism so that modeling of CBM production and CO2 storage on field scale is improved.

• Probe:•adsorption/desorption behavior•permeability with respect to gas composition•recovery efficiency and CO2 storage

Flow Through Apparatus

• Unconfined pack• Gas mixtures made in the lab by weight

Coal Holder

Flow-MeterBack Pressure Regulator

Gas Container

poverburdenp2

pi

To Vent

∆p

Relevant Coalpack DataPowder River Basin, WY Coal

As received After grinding

Coalpack #1 #2 #3 Size of coal particles, meshes <60 <60 <60 Length of the coalpack, cm 25.0 25.0 25.0 Diameter of the coalpack, cm 4.25 4.25 4.25 Porosity, % 44 37 34 Permeability, md 144.0 43.0 26.3 Weight of coal, g 218.0 231.0 239.4Swi, % 0 0 8-10

Coalpack

Mixed Gas Injection Data

Temperature: 72oF

Pressure: 600 psia

Initial Water Saturation: 0

Gas Injection Rate: 0.5 cc/min

Injection-gas composition:

-100% CO2, 0% N2 (c100n0)

-85% CO2, 15% N2 (c85n15)

-46% CO2, 54% N2 (c46n54)

-24% CO2, 76% N2 (c24n76)

-0% CO2, 100% N2 (c0n100)

CO2 Displacing CH4Piston like advance of CO2

coal

& C

H4

CO2

CH4 +CO2

gas analyzer

Chromatographic Separation of CO2 & N2co

al &

CH

4

46% CO2 54% N2

CH4 +CO2 + N2

gas analyzer

Adsorption/Desorption/Transport Modeling

• 1D, 2 phase:

• extended Langmuir:

• primary and secondary (grain): φ= φ1+φ2– instantaneous equilibrium– 2%< φ2 <8%

• PR-EOS with

• Finite difference solution

φ∂Ci

∂t+ (1− φ)

∂ai

∂t+ ∇⋅ (vCi) = qi

ai =αiβ iyi

1+ βkykk∑

Vads = zii∑ bi

CO2, CH4, N2 Adsorption/DesorptionPowder River Basin, WY Coal

• Correct for volume of adsorbate

• All adsorption data (pure) is well fit by Langmuir isotherm

• CO2 adsorbs preferentially

• adsorption hysteresis for all gases

• scanning loops are evident

22 °C

3 CO2 : 1 CH4

Binary Gas SystemsExperiment and Simulation, φ2=7.4%N2

CH4 + N2

CO2

CH4 + CO2

Ternary Gas SystemsExperiment and Simulation, φ2=7.4%

76%N2+24% CO2

CH4 + N2 + CO2

CH4 + N2 + CO2

54%N2+46% CO2

Ternary Gas SystemsExperiment and Simulation, φ2=7.4%

15%N2+85% CO2

CH4 + N2 + CO2

Permeability Evolution of CoalpackSteady State

Experimental and Simulation Discrepancy• mass transfer limitations

– no discrepancy in binary results– coal particles are small and times for diffusion

are less than 0.1 s• coal shrinkage/swelling

– no discrepancy in binary results– N2 did not affect permeability

• prediction of sorption– hysteresis, scanning-loops, multicomponent– extended Langmuir

• constant selectivity• selectivity independent of total pressure and gas-

species concentration

Sensitivity SimulationTernary Gas System

CH4 + N2 + CO2

54%N2+46% CO2

• Case 1: Base case• Case 2: Use only desorption curve for N2

Permeability Versus Gas Composition (CH4, N2, CO2)

Core Holder

coalpack: 1 inch diameter, 12 inches long

Flow Through Apparatus

• Net overburden pressure=400 psi, Pore pressure: 60~1100 psi• Gas mixtures made in the lab by weight

Coal Holder

Flow-MeterBack Pressure Regulator

Gas Container

poverburdenp2

pi

To Vent

∆p

Steady-State PermeabilityPconfining-Ppore = 400 psi

Permeability Reduction Correlates With Surface Coverage

Pure Gases

Coverage vs Permeability ReductionPure and Mixed Gases

Permeability ReductionPhase Behavior (CO2/N2)

experimental temperature, T

T > Tcno phase behaviorgas-solid interactions dominant

T < Tcphase behaviorvapor-vapor interactions dominant

Tc(°C)CH4 -83N2 -147CO2 31

Coverage vs Permeability ReductionPure and Mixed Gases

T > Tcno phase behaviorgas-solid interactions dominant T < Tc

phase behaviorvapor-vapor interactions dominant

Conclusions• The interplay of gas sorption and transport in coal

yields rich dynamical behavior that is sensitive to the details of adsorption

• The transient behavior of binary gas systems is well represented by the numerical model.

• The qualitative behavior of ternary gas systems is well represented in simulation results; however, quantitative agreement of breakthrough times and the elution profiles for gas species remains to be proven.

• Permeability decreases as pore space fills with immobile adsorbed gas– reduction in porosity available for flow– possible correlation to surface coverage– some N2 appears to preserve permeability

• Gas injection enhances CBM recovery significantly (at least 94% OGIP)

Acknowledgment

• This work was supported by Stanford University Global Climate and Energy Project (GCEP). Funding is acknowledged gratefully.

• Collaborators:– Lynn Orr– Kristian Jessen– Tom Tang– Wenjuan Lin– Tanmay Chaturvedi

Sequestration Time Scales

• Convection–flow between wells (detailed geological description necessary?)

• Dissolution of CO2 in brine• Gravity redistribution• Gravity overturn of dense CO2-brine• Chemical reactions in brine phase• Geologic formation to surface

transport–if applicable (detailed and/or stochastic geological description necessary?)

SL

SL

Reality check: the volumes are very large!

• At a CO2 density of 500 kg/m3 (1000 m depth at 50˚C), injection of 1 billion tonnes/yr of CO2 is equivalent to ~ 35 million barrels/day.

• At a CO2 density of 700 kg/m3, 1 Gt CO2 is equivalent to ~ 25 million barrels/day.

• Worldwide emissions of CO2, ~ 25 Gt/yr:~ 625 – 825 million barrels/day!

• World oil production is currently ~80 million barrels/day.

• These volumes are large enough that it is unlikely that CO2storage in geologic formations alone will solve the problem of CO2 emissions to the atmosphere.

expe

rimen

tsi

mul

atio

nMethane Recovery

Why is there hysteresis?Surface geometry heterogeneity–Seri-Levy and Avnir (1993)

• Monte Carlo simulation• gas-surface interactions

greater than gas-gas• rough surfaces

• various hysteresis curves obtained by modifying surface geometry

Why is there hysteresis?Surface geometry heterogeneity–Seri-Levy and Avnir (1993)

rough surface

p/p0=0.13 p/p0=0.13

p/p0=0.61 p/p0=0.61

adsorption

desorption

Experimental SetupAdsorption/Desorption CO2/N2/CH4

Electronic balance

Coreholder

Pressure gauge

Pressure gauge

Reference cell

Gas Cylinder

Vacuum pump

CH4 Scanning LoopsInitial pressure influences desorption hysteresis

22 °C

Adsorption on Moist Coal (uncorrected for volume of adsorbate)

22 °C

Effect of T on Coalpack Permeability

Data Acquisition Strategies

CO2

Seal

Fault

Injection Well

Our 4-D approach:

• Embedded detectors• Embedded low-power sources• Coded, continuous signals• Adaptive synthetic apertures

- e.g., the Stanford Cross

The Stanford Cross Array