Oklahoma State University

School of Chemical Engineering

Modeling of Gas Adsorption

on Coalbeds
K. A. M. Gasem
Z. Pan
J. E. Fitzgerald
M. Sudibandriyo
R. L. Robinson, Jr.

Oklahoma State University

Sponsored by the
U.S. Department of Energy

March 6, 2002 1
 Oklahoma State University
School of Chemical Engineering

RATIONALE
• Modeling of the adsorption behavior of coalbed gases
(methane, CO2, nitrogen) is essential in CBM
production and in CO2 sequestration.

• Further, knowledge of the competitive adsorption of

CBM gases is required to elucidate mechanisms for
enhanced recovery of CBM and CO2 sequestration
processes.

• Reliable adsorption predictions cannot be generated

using simple, empirical models. Accurate models
require sound theory, judicious approximations, and
accurate experimental information.

March 6, 2002 2
 Oklahoma State University
School of Chemical Engineering

Coalbed Adsorption Phenomenon

Cleat system where

free gas resides

Micropores where
Adsorbed gas resides

Zoom-in view
of model micropore

March 6, 2002 3
 Oklahoma State University
School of Chemical Engineering

Primary Recovery Water

• Reduce cleat pressure by producing water Methane
• Methane desorbs from matrix to cleats
• Methane and water flow to well bore
Free Gas Diffusion
To Well- Through Cleats
Bore
Diffusion to
Cleats

Coal
Matrix Desorption

March 6, 2002 4
 Oklahoma State University
School of Chemical Engineering

Absolute Adsorption on Fruitland Coal at 115°F

1.4

1.2
CO2
1.0
Adsorption, mmol/g

Methane
0.8

0.6
Nitrogen
0.4

0.2

0.0
0 500 1000 1500 2000
March 6, 2002 Pressure, psia 5
 Oklahoma State University
School of Chemical Engineering

CO2 Enhanced Recovery Water

• CO2 displaces methane after injection CO2
Injection Methane
• No initial breakthrough of CO2 CO2
• Higher cleat pressure results in faster flow

To Well
Bore
Diffusion

Coal
Matrix
Methane CO2
Desorption Adsorption

March 6, 2002 6
 Oklahoma State University
School of Chemical Engineering

Theory / Practice
• Theory: Improve our understanding of high-pressure
adsorption through rigorous methodologies.
• Practice: Provide reliable equilibrium adsorption
models for optimum CBM production and
CO2 sequestration.
• Strategy: Use rigorous methodologies to develop
reliable adsorption models for industrial
practice.
• Goal: Develop reliable coal-structure-based
generalized predictions using simple,
accessible characterizations.
March 6, 2002 7
 Oklahoma State University
School of Chemical Engineering

Current Issues
• Adsorption modeling
• Coal characterization
• Coal structure-based model generalizations
• Estimates for adsorbed-phase density
• Effect of moisture on modeling adsorption capacity
• Matrix swelling
• Binary and ternary pvT data
• Balancing computational efficiency and reliability
March 6, 2002 8
 Oklahoma State University
School of Chemical Engineering

Current Issues: Adsorbed-Phase Density

• Why do we need adsorbed-phase density?

• Current estimation methods:
! Traditional
! Experimental approximation
! Model-based:
2D equations of state, SLD theory, Ono Kondo theory
• What is the impact?

March 6, 2002 9
 Oklahoma State University
School of Chemical Engineering

Molecular Interactions:
Mean Field Approximation
1.6

1.4 Bulk Gas

1.2
Local Density, g/cc

1.0

0.8

0.6

0.4

0.2 Adsorbate
0.0
0.0 0.1 0.2 0.3 0.4 0.5
Normalized Slit Width
March 6, 2002 10
 Oklahoma State University
School of Chemical Engineering

Excess and Absolute Adsorption

• The excess adsorption is defined as follows

nGibbs = ∫ (ρads − ρbulk ) dV

V

or
(
nGibbs = V ρads − ρbulk )
• The absolute adsorption is defined as

nGibbs
nabs = Vads ρads =
ρ
1 − bulk
ρads
March 6, 2002 11
 Oklahoma State University
School of Chemical Engineering

Estimation of Phase Volume and Density

0.35
Adsorption, g CO 2/g-carbon

CO2 on activated carbon at 113oF

0.30
0.25
y = -0.360x + 0.365
0.20 R2 = 0.994
0.15

0.10
Phase Density
0.05 1.02 g/cc

0.00
0 0.2 0.4 0.6 0.8 1 1.2
Density, g/cc

0 400 800 1200 1600 2000 9200

Pressure, psia
March 6, 2002 12
 Oklahoma State University
School of Chemical Engineering

CO2 and Ethane Adsorption on

Activated Carbon (OSU)
Carbon Dioxide, Gibbs
10 Carbon Dioxide, Absolute
Ethane, Gibbs
Ethane, Absolute
8
Adsorption (mmol/g)

0
0 2 4 6 8 10 12 14
Pressure (MPa)
March 6, 2002 13
 Oklahoma State University
School of Chemical Engineering

Impact of Adsorbed-Phase Density:

CO2 on Activated Carbon at 113 oF
10.0
Adsorption (mmol/g AC)

8.0

6.0 ZGR -- 0.98 g/cc

OK -- 1.00 g/cc
Experimental -- 1.02 g/cc
4.0
OKv -- variable
SLDv -- variable
2.0 ZGRv -- variable
Traditional -- 1.18 g/cc
Gibbs Adsorption
0.0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Pressure (psia)

March 6, 2002 14
 Oklahoma State University
School of Chemical Engineering

Current Issues: Adsorption Modeling

• We seek simple, reliable adsorption equilibrium

models that are suitable for generalized predictions
and reservoir simulations.

• Such models should be capable of

! Precisely representing pure and mixture isotherms
! Facilitating a priori predictions

March 6, 2002 15
 Oklahoma State University
School of Chemical Engineering

Equilibrium Modeling: Three Methods

1. Enhanced forms of the Langmuir isotherms
-- provide simple data correlation

2. Two-dimensional equations of state (2-D EOS)

(a) Cubic EOS (b) Segment-Segment EOS

-- facilitate generalized simulations

3. Simplified-Local-Density (SLD) models

-- account for surface structure and near-critical behavior

March 6, 2002 16
 Oklahoma State University
School of Chemical Engineering

The Langmuir & Loading Ratio Correlation

(LRC)

ωi
=
(B iPy i ) i
η

1 + ∑ (B iPy i )
ηi
Li
j

P = pressure
yi = gas phase mole fraction of component “i”
ηi = LRC exponent for component “i”
ωi = amount adsorbed of component “i”
Li, Bi = Langmuir/LRC model coefficients

March 6, 2002 17
 Oklahoma State University
School of Chemical Engineering

LRC Current Capability

The loading-ratio correlations:
• Represent absolute pure-component and mixed-gas
total adsorption precisely
• Yield reasonable predictions for these systems
• Represent individual-component adsorption in
mixtures less precisely, especially the less-
adsorbed ones
• Require adsorbed-phase density estimates

March 6, 2002 18
 Oklahoma State University
School of Chemical Engineering

LRC Representations: Illinois-6 Coal

1.2
Pure CO2 Mixture is 60/40 CH4/CO2
Absolute Adsorption (mmol/g coal)

1.0 Mixture Total

Pure CH4
0.8 CO2 in Mixture
CH4 in Mixture
0.6 LRC

0.4

0.2

0.0
0 200 400 600 800 1000 1200 1400 1600 1800

Pressure (psia)
March 6, 2002 19
 Oklahoma State University
School of Chemical Engineering

2-D Equations of State

(OSU, 1992)

 


Aπ +
αω 2
2
1 + Uβω + W ( βω ) 
1 − [
( βω ) m
= ωRT ]
α = ∑ ∑ x i x j α ij β = ∑ ∑ x i x jβ ij
i j i j

α ij = (α i + α j ) / 2 β ij = β i β j

EOS m U W
VDW 1 0 0
SRK 1 1 0
PR 1 2 -1
Eyring 1/2 0 0
ZGR 1/3 0 0
March 6, 2002 20
 Oklahoma State University
School of Chemical Engineering

2D EOS Current Capability

2-D EOS models:
• Describe CBM pure-component and mixed-gas total
adsorption data with sufficient precision
• Yield reasonable predictions for these systems
• Represent individual-component adsorption less
precisely, especially the less-adsorbed ones
• Employ inadequate repulsive terms
• Do not account for variations in coal structure

March 6, 2002 21
 Oklahoma State University
School of Chemical Engineering

2D EOS Representations: Illinois-6 Coal

1.2

Pure CO2 Mixture is 60/40 CH4/CO2

Mixture Total
Absolute Adsorption (mmol/g coal)

1.0
Pure CH4
CO2 in Mixture
0.8 CH4 in Mixture
ZGR
0.6

0.4

0.2

0.0
0 200 400 600 800 1000 1200 1400 1600 1800

Pressure (psia)
March 6, 2002 22
 Oklahoma State University
School of Chemical Engineering

The EOS-SLD Adsorption Model

Coal Surface Gas Molecule

z L-z

µ fs (z ) = µ fs1 (z ) + µ fs 2 (L − z )
• The fluid-solid interaction potential equals the sum of the
potentials between the gas molecule and the two sides of
the slit.
March 6, 2002 23
 Oklahoma State University
School of Chemical Engineering

EOS-SLD Current Capability

The EOS-SLD models:

• Account for variations in coal structure

• Provide a viable framework for generalized predictions

• Have produced promising preliminary results for

mixture adsorption modeling

• Employ inadequate repulsive terms

March 6, 2002 24
 Oklahoma State University
School of Chemical Engineering

CO2 Adsorption Using Modified SLD-PR

8

Experimental Data OSU

7 Original SLD
Modified SLD

6
Gibbs Excess (mmol/g)

0
0 2 4 6 8 10 12 14
Pressure (MPa)

March 6, 2002 25
 Oklahoma State University
School of Chemical Engineering

Sample Pure-Gas Adsorption Model Results

%AAD
Nitrogen Methane CO2 Ethane
Dry Activated Carbon

LRC 0.3 0.6 6.1 5.8

ZGR 0.4 0.7 5.2 5.6
ZGR Gibbs 0.4 0.5 1.3 3.3
Original SLD 0.5 0.8 14.9 29.7
Modified SLD 0.4 0.6 2.2 6.1

Wet Fruitland Coal

LRC 1.1 0.7 3.3

ZGR 1.9 0.7 3.1
Original SLD 1.5 0.6 3.9
Modified SLD 1.1 0.6 3.6
March 6, 2002 26
 Oklahoma State University
School of Chemical Engineering

Conclusions
• 2-D EOS and the EOS-SLD models are better
equipped than Langmuir-type correlations for modeling
CBM adsorption isotherms.

• The EOS-SLD models appear both accurate and

amenable to structure-based generalization.

• Improved mixing rules and additional mixture data are

required to improve predictions for individual-
component adsorption.

• More efforts should be dedicated to structure-based

model generalizations.
March 6, 2002 27
 Oklahoma State University
School of Chemical Engineering

Modeling Work in Progress at OSU

• Extend the EOS-SLD and Ono Kondo models to
mixture predictions.

• Implement other potential models for fluid-solid

Interactions.

• Incorporate other geometries within the EOS-SLD

framework.

• Develop theoretically-based equations of state that

feature more accurate fluid-fluid repulsive terms.

March 6, 2002 28