Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
pubs.acs.org/EF
General Gas Permeability Model for Porous Media: Bridging the Gaps
Between Conventional and Unconventional Natural Gas Reservoirs
Peng Cao,*,† Jishan Liu,†,‡ and Yee-Kwong Leong†
†
School of Mechanical and Chemical Engineering, The University of Western Australia, 35 Stirling Highway, Perth, WA 6009,
Australia
‡
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of
Sciences, Wuhan 430071, PR China
ABSTRACT: Many field observations have indicated that permeabilities of both conventional and unconventional gas reservoirs
are not constant when gas pressure drops. For conventional reservoirs, permeability will decrease while for unconventional gas
rocks, the apparent permeability may increase as gas pressure decreases to a lower magnitude. Evolution trends of permeability
for different natural gas reservoirs are distinct. These differences are observed by laboratory experiments of sandstones, coals, or
shales. In this study, we present a general permeability model to bridge the gaps between conventional and unconventional gas
Downloaded by STANFORD UNIV at 12:56:23:237 on July 05, 2019
reservoirs. This model coupled three critical factors namely effective stress, adsorption, and flow regimes to reflect dynamic
from https://pubs.acs.org/doi/10.1021/acs.energyfuels.6b00683.
performances of permeability. On the basis of specific reservoirs properties, the model degenerates into four reduced types. The
first reduced model is applicable for reservoirs with lower adsorption capacity. The second reduced model is adopted by
unconventional reservoirs like coal seams when the intrinsic permeability is big and adsorption capacity is high. For the third
reduced model, effective stress is the dominating factor for permeability evolution, which means that it is applicable for
conventional reservoirs like sandstones. Unconventional gas reservoirs with low adsorption capacity like gas shales can apply the
fourth reduced model because the flow regimes dominate the evolution. These reduced models are verified against the
experimental data. Results show that effective stress is the main reason for the change of permeability for conventional gas
reservoirs. Both effective stress and flow regimes together determine the apparent permeability of unconventional gas reservoirs.
The impact of adsorption on permeability is relatively small. Permeability evolution trends can be classified into different zones
for conventional and unconventional gas reservoirs. When the gas is depleted from reservoirs, the gas permeability has two
bounds. For the upper bound, permeability is only affected by flow regimes and the apparent permeability will increase when gas
pressure drops. For the lower bound, permeability is only affected by effective stress and the apparent permeability will decrease
when the gas is depleted from the reservoirs.
Figure 1. Schematic of impacts of stress, flow regime, and adsorption on the apparent permeability.
gas reservoirs. In terms of the flow regime, it is related to one unchanged, which is impossible in the real process of gas
dimensionless parameter Knudsen number. For the traditional production. This assumption is another gap between conven-
Darcy law, it is applicable that permeability is big and the tional gas reservoirs and unconventional gas reservoirs. On the
Knudsen number is less than 0.01. But this condition is other hand, adsorption phenomenon has two effects on gas
impossible for unconventional natural gas reservoirs. For reservoirs. First, the adsorption induced swelling can change the
example, for gas shale reservoirs, the pores are concentrated effective stress. When adsorbed gas injects into porous media, it
within several nanometers to hundreds nanometers, which will induce the swelling of the matrix and both the volumes of
means the Knudsen number is over 0.01 and the flow regime pores and grains will be changed. If it is under the free swelling
belongs to the slip flow. condition, the effective stress is a constant and intrinsic
It is acceptable that permeability of gas flow is bigger than permeability is unchangeable. However, if the boundary is fixed,
that of liquid flow in the same porous media especially for tight the swelling will increase the stress and compact the pores,
porous media.12−14 The apparent permeability is bigger than which leads to increase of effective stress and decrease of
the intrinsic permeability for porous media because of slippage intrinsic permeability. Some scholars presented different
effect (flow regimes). There is a close relationship between models to reflect the impact of adsorption on permeabil-
intrinsic permeability and apparent permeability.15,16 Many ity.26−28 Second, the adsorption will occupy the space of the
scholars established apparent permeability models to reflect this pores, which can decrease the permeability. The adsorption on
effect by adding slippage factor to Darcy’s equation, which the surface of pores can reduce the porosity within the
called Klinkenberg-corrected permeability model.3,17−20 Be- connected pores network, which means that the average pore
sides, many researchers thought flow regime like Knudsen radius should deduct the adsorption layer thickness29 as shown
diffusion will play a more important role in apparent in Figure 2. When the pores of gas reservoirs are very small
permeability and adopted Knudsen number to represent the (less than 50 nm), the thickness of the adsorption layer should
flow regime. Based on this theory, a lot of Knudsen number- be taken into account.
corrected permeability models have been developed.16,21 Overall, there are three main factors namely effective stress,
Additionally, some scholars have proposed several apparent adsorption and flow regime that affecting permeability of gas
permeability models based on the idea that slip flow and
reservoirs as shown in Figure 1. Until now, all of these physical
Knudsen diffusion together lead to the apparent permeability
for shales, which means slip flow can appear independently
without considering the diffusion.22−24
These models have good agreements with experiments for
porous media with nanopores or microchannels under different
fluids conditions like helium, methane, water, and decane. They
focus on the variation of apparent permeability when using
different fluids under the same confining pressure conditions.25
Based on these studies, it is easy to understand the impacts of
flow regimes on the apparent permeability compared to the
intrinsic or absolute permeability. However, for these
permeability models, there is a common assumption or Figure 2. Difference between average pore radius and effective pore
experimental condition that the stress for shale solid is radius from adsorption effect.
where KB is the Boltzmann constant, T is the temperature (K) Substituting eqs 14 and 16 into eq 15, the rigorous
of gas, and σ̃ is collision diameter (m). From eq 10, the mathematical formulation for the general gas permeability
characteristic length re can be obtained as model for porous media is given as
k∞ eτh (r − dads)2 ϕ ⎛ 4Kn ⎞
re = 2 2 kapp = (1 + ζKn)⎜1 + ⎟
ϕ (13) 8 τ ⎝ 1 + Kn ⎠ (18)
where τh is the tortuosity of porous media. Therefore, eq 11 where subscript 0 refers to the initial value for shale rocks. Then
becomes eq 18 becomes
(r0exp( −ct[σ ̅ − σ0̅ − (p − p0 )]) − dads)2
KBT ϕ kapp =
Kn = 8
2 πσ ̃ 2p 2 2k∞ eτh (14)
ϕ0exp( −ct[σ ̅ − σ0̅ − (p − p0 )])
Flow regimes can be classified into four types based on the (1 + ζKn)
τ
Knudsen number. The four types are viscous (continuum) flow
⎛ 4Kn ⎞
(Kn < 0.01), slip flow (0.01 < Kn < 0.1), transition flow (0.1 < ⎜1 + ⎟
Kn < 10), and Knudsen’s (free molecular) flow (Kn > 10). ⎝ 1 + Kn ⎠ (19)
2.4. Apparent Permeability for Porous Media. Because This formulation contains the effects of mechanical
of complex flow mechanisms, the apparent permeability of deformations (effective stress), adsorption, and flow regime
porous media deviates significantly from the intrinsic on the gas flow within reservoirs. This general gas permeability
permeability. Therefore, we developed an intrinsic permeability model can be applicable for both conventional and unconven-
model as eq 10 for porous media. The value of the intrinsic tional natural gas reservoirs.
permeability k∞ does not depend on the type of fluid used or
the flow conditions in the experiments. It is just a property of 3. MODEL ANALYSES AND VERIFICATIONS
the porous medium. However, the permeabilities measured by
3.1. Four Types of Reduced Models. The general
using gas such as He, CH4, and CO2 through ultralow
apparent permeability model for porous media contains the
permeability rock samples are different from permeabilities
three critical factors namely effective stress, adsorption and flow
measured by liquids like water because of the effect of flow
regime. This model is applicable for different testing conditions
regime. The apparent permeability is a function of both
like changing the fluids or changing the fluid pressure. It is
intrinsic permeability and flow regime. The apparent
difficult to verify this model when considering the three factors
permeability model for gas reservoirs is given by the following
together because the interactions between the factors are
expression. We call it the BK (Beskok−Karniadakis) model,
complex and the parameters are hard to measure in the
which is based on a unified Hagen−Poiseuille-type formula-
experiments. However, in most cases, this complex model can
tion.3,18,21
be degenerated to simple models when adopting the specific
kapp = k∞ ef (Kn) (15) conditions. Then we can use these reduced models to analyze
the experimental and field data, which can verify the general
where f(Kn) is slippage incremental factor as a function of the model. For example, when we use nonadsorbed gas to measure
Knudsen number Kn, the dimensionless rarefaction coefficient permeability of porous media, the effect of the adsorption can
ζ, and the slip coefficient b. It is defined as be neglected and the model can remove the adsorption term.
Then we can compare the experimental data with the results of
⎛ 4Kn ⎞⎟
f (Kn) = (1 + ζKn)⎜1 + this reduced model. Thus, based on reservoirs properties, the
⎝ 1 − bKn ⎠ (16) model can be reduced into four types.
According to previous studies, the slip coefficient b is an (1) First reduced model: when we adopt nonadsorbed gas
empirical parameter. Beskok and Karniadakis suggested its like He to analyze the evolution of permeability or when porous
value as b = −1 and that it is independent of the type of gas. media do not contain the adsorbed materials, the model just
The value of the dimensionless rarefaction coefficient ζ varies: considers the effects of effective stress and flow regime. (2)
0 < ζ < ζ0 for 0 < Kn < ∞ where ζ0 is an asymptotic limit Second reduced model: when the average pore radius is big and
value.21 A correlation is presented by Civan et al.3 the Knudsen number is less than 0.01, the flow regime belongs
to conventional flow and the model ignores the flow regime
ζ0 term. (3) Third reduced model: when the Knudsen number is
ζ=
1+
A small and the nonadsorbed gas is used to measure permeability,
Kn B (17) the flow regime and adsorption can be neglected together and
where A = 0.17, B = 0.4348, and ζ0 = 1.358. the final model just is the function of the effective stress. (4)
As the bulk modulus K is commonly several orders of Fourth reduced model: in most cases the effective stress has
magnitude larger than the pore volume modulus Kp, we obtain great impact on the evolution of permeability but for the
1 1 1 1 ultralow permeability porous media and the flow regime plays a
− K ≈ − K , then define the compressibility ct = K , which
K p p p more significant role in the variation of permeability. The
is a real property for rocks through experimental data. change of intrinsic permeability is very small and we only
However, it is accepted that the compressibility is not constant consider the effect of flow regime.
and McKee’s model 35 is adopted in this work to 3.2. Verification for the First Reduced Model. As
1 − e−α0(Δσ̅ −Δp) discussed above, in some cases the adsorption effect is small
presentct = ct0 α0(Δσ ̅ − Δp)
, and this dynamic model has been and adsorption will be ignored in the general model. There two
verified by many researchers in recent years.27 main cases for this simplified model. One is that the
5495 DOI: 10.1021/acs.energyfuels.6b00683
Energy Fuels 2016, 30, 5492−5505
Energy & Fuels Article
nonadsorbed gases like He and N2 are used to analyze the data fitting for the model, we could obtain the initial
evolution of permeability. Another case is that porous media compressibility ct0 of sample 31A is 0.055 MPa−1 and the
has small capacity to adsorb gas like sandstones without the decline rate α0 for sample 31A is 0.02 MPa−1. The decane
clay and kerogen. It is noted that many unconventional gas permeability is the intrinsic permeability for the shale sample,
reservoirs have big capacity of the gas like coals, which means which ignores the effect of flow regimes. For the same shale
this reduced model is not applicable for these reservoirs. rock 31A, helium permeabilities were measured at the gas
The first reduced model without the effect of the adsorption pressure of 6.89 MPa and the results showed that permeability
is reorganized as were nearly 1 order of magnitude higher than permeability of
The stress−strain constitutive relationship (eq 1) becomes liquid oil under the same stress condition. The reason for the
difference between oil permeability and gas permeability is the
1 ⎛ 1 1 ⎞ α
εij = σij − ⎜ − ⎟σ δ + pδij effect of flow regime, which indicates the coupled impact of
2G ⎝ 6G 9K ⎠
kk ij
3K (21) effective stress and flow regimes on the unconventional gas
Equation 2 becomes reservoirs. The helium permeability declined from 3500 to 780
nanodarcy (nD) when the effective stress increased from 12 to
1 60 MPa as shown in Figure 3. There are strong mechanical,
εv = − (σ ̅ − αp)
K (22) chemical and physical interactions between the fluid molecules
Then eq 5 can be rewritten as and pores surface. In summary, both mechanical deformation
and flow regime have significant impacts on the apparent
⎛ p⎞ permeability and the reduced model has been well validated by
σ ̅ − p = −K ⎜εv + ⎟ Cui’s data.4 This reduced model is applicable for gas reservoirs
⎝ Ks ⎠ (23) without the adsorption phenomenon.
We verify this reduced model based on the experimental data 3.3. Verification for Second Reduced Model. For some
of Cui.4 In the experiments, permeability of shale rocks was gas reservoirs, the pores are big and the corresponding
measured with helium. The same pressure was used for the axial Knudsen number is less than 0.01 and traditional Darcy law
and radial confining in the core cell.36 The axial confining is applicable. The impact of the flow regime can be removed
pressure equaled to the radial confining pressure pr* = pz* = pc*. from the general model and this reduced model is acceptable
Therefore, the apparent permeability model for porous media for conventional gas reservoirs like sandstones.
can be simplified as The second reduced model without the effect of the flow
regime can be written as
kapp = k∞ 0exp( − 3ct[(pc* − pc0
* ) − (p − p )])(1 + ζKn)
0 2
⎛
⎜1 +
4Kn ⎞
⎟
k∞ =
(r exp(−c [σ̅ − σ̅ − (p − p )]) − D
0 t 0 0
p
CH4 p + P
L
)
⎝ 1 + Kn ⎠ (24) 8
where p*c is the mean confining pressure and ct denotes variable ϕ0exp( −ct[σ ̅ − σ0̅ − (p − p0 )])
compressibility. Based on eq 23, the effective stress can be τ (25)
considered as Δσ̅ − Δp = (pc* − pc0 * ) − (p − p0). The detailed
validation process can refer to the previous study.12,30 The data For the matrix deformation equation, it is the same as the
for shale sample no. 31A4 is selected to verify the model. general model and eqs 1, 2, and 11 are applicable in this
As shown in Figure 3, permeability calculated using the new simplified model. In the past decades, there are different models
model shows a decline trend when the effective stress changes to describe the evolution of permeability of coal without the
from 3.4 to 45 MPa. Based on experimental data, the decane consideration of flow regime.26,37,38 Cui and Bustin26 presented
permeability changed from 2 × 10−3 to 4.5 × 10−5 mD when the dynamic permeability model including the effect sorption
the effective stress increased from 3.4 to 45 MPa. Based on the induced volumetric strain as the following.
⎛ ⎛ ⎞
1 1 ⎞⎟
k∞ = k∞ 0exp⎜⎜3⎜⎜ − ⎟ [σ ̅ − σ0̅ − (p − p0 )]⎟⎟
⎝ ⎝K Kp ⎠ ⎠ (26)
Compared the reduced model to the CB model (Cui−Bustin
model),26 there are the same structure of formulations. If we
remove the thickness of the adsorption layer dads and the initial
r2 ϕ
intrinsic permeability is defined as k∞ 0 = 80 τ0 , the two
expressions will be the same. However, it is noted that the
mathematic methods to derive the models are different and the
general model comes from the poroelasticity theory. It is
pointed that the CB model did not consider the effect of the
flow regime. According to the study of Cui and Bustin, the CB
model can degenerate to the same structure as PM model
(Palmer−Mansoori model) and SD model (Shi−Durucan
model)37,38 under the conditions of uniaxial strain and constant
loading zone.
Figure 3. Comparison of the model with test data for permeability of The degenerated CB model for the coal reservoir under these
shale with decane and helium. conditions is
5496 DOI: 10.1021/acs.energyfuels.6b00683
Energy Fuels 2016, 30, 5492−5505
Energy & Fuels Article
⎧ ⎡ (1 + v) 2E
k∞ = k∞ 0exp⎨3ct⎢ (p − p0 ) −
⎩ ⎣ 3(1 − v) 9(1 − v)
⎤⎫
(εs − εs0)⎥⎬
⎦⎭ (27)
The SD model for the coal reservoir under the same
conditions is
⎧ 3 ⎡ v
⎪ E
k∞ = k∞ 0exp⎨ ⎢ (p − p0 ) −
⎩ p
K ⎪
⎣ 1 − v 3(1 − v)
⎤⎫ ⎪
(εs − εs0)⎥⎬
⎦⎭ ⎪
(28)
The PM model for the coal reservoir is
Figure 4. Comparison of field data for coal seam with the prediction of
(1 − 2v)(1 + v) 2 ⎛ 1 − 2v ⎞⎟ the reduced model.
ϕ = ϕ0 + (p − p0 ) − ⎜
E(1 − v) 3⎝ 1 − v ⎠
(εs − εs0) (29)
adsorption together. Specifically, when porous media has high
This reduced model is similar to all these models and it can permeability (over 1 mD) and the mineral compositions do not
be verified by experimental data of coal permeability. The coal include the clay and other adsorbed materials like kerogen, the
is a typical unconventional gas reservoir and permeability is model is applicable for describing the dynamic performance of
about several millidarcy, which means the effect of flow regime permeability. Conventional gas reservoirs can adopt this
can be ignored. In this study, we select the filed data of coal assumption because of high intrinsic permeability and mainly
seams38 to verify this reduced model. Also, we adopt the composed of sandstones. Under this condition, the effective
variable compressibility in the reduced model. 35 The stress is the main reason for the variation of permeability and
parameters for this reduced model based on data fitting are permeability is the function of the effectives stress.
listed as Table 1. The third reduced model without the effects of flow regime
and adsorption can be formulated as
Table 1. Main Parameters for the Second Reduced Model
k∞ = k∞ 0exp( −3ct[σ ̅ − σ0̅ − (p − p0 )]) (30)
Based on Experimental Data
parameters value The stress−strain constitutive relationship becomes
Possion’s ratio 0.25 1 ⎛ 1 1 ⎞ α
εij = σij − ⎜ − ⎟σ δ + pδij
Young’s modulus (MPa) 3585 2G ⎝ 6G 9K ⎠ kk ij
3K (31)
Langmuir pressure (MPa) 8.27
maximum adsorption induced volumetric strain 0.022 Equation 2 becomes
initial compressibility (MPa−1) 0.12 1
decline rate of the compressibility (MPa−1) 0.007 εv = − (σ ̅ − αp)
K (32)
initial reservoir pressure (MPa) 5.516
Then eq 5 can be rewritten as
⎛ p⎞
The predicted results for the coal seam have good σ ̅ − p = −K ⎜εv + ⎟
agreements with the field data as shown in Figure 4. Both ⎝ Ks ⎠ (33)
the effective stress and adsorption have impacts on the
evolution of permeability of the reservoir. When the gas It is acceptable that when the testing fluid pressure changes
produced from the well, the reservoir pressure will decrease and or the confinement pressure changes, permeability of porous
the effective stress will increase. If we do not consider the media will be changed. In the laboratory, the effective stress can
adsorption effect, permeability of the reservoir will decrease be obtained by the confining pressure and the gas pressure.27
because of the matrix deformation. However, the filed data Then eq 15 can be rewritten as
shows that permeability undergoes an increasing trend over the k∞ = k∞ 0exp( −3ct[(pc* − pc0
* ) − (p − p )])
0 (34)
whole period of production. Thus, it is seen that desorption of
the gas will induce the matrix shrinkage and permeability will where pc* is the mean confining pressure and ct denotes
become bigger when the gas pressure declines. In summary, for compressibility. In eq 15, the effective stress can be considered
this special case, the effective stress and adsorption are the key as Δσ̅ − Δp = (p*c − p*c0) − (p − p0). Also, we adopt the
factors that affecting the variation of permeability. The effect of dynamic compressibility model in this reduced model. We
the flow regimes can be ignored because of the small Knudsen select the experimental data of core sample for the coal
number. permeability39 to verify this reduced model. In the experiments,
3.4. Verification for the Third Reduced Model. In the N2 was used to measure permeability and the effect of
comparison with the first and second reduced models, the third adsorption can be neglected because of the very small
reduced model ignores the effects of flow regime and adsorption capacity for N2. The gas pressure for the Anderson
5497 DOI: 10.1021/acs.energyfuels.6b00683
Energy Fuels 2016, 30, 5492−5505
Energy & Fuels Article
is very tiny, the general model has the same trend as the
reduced model and this evolution also is verified by
experimental data in the above section. When we increase the
capacity of adsorption, the evolution curve will deviate that of
the reduced model. The first reduced model is one kind of
degenerated models for the porous medium with low
adsorption capacity. The main factors for gas permeability
evolution are effective stress and flow regimes. For this kind of
model, when the gas pressure drops, the apparent permeability
will undergo the decreasing trend. But the gas permeability will
Figure 7. Simulation model and conditions for the first and fourth be rebounded to a higher value while the gas pressure decreases
reduced models. to a lower magnitude as shown in Figures 8 and 9.
4.3. Comparison between CB Model of Coals and
m × 0.1 m. The point A is located in the center of the General Model. Coal seam gas is one kind of unconventional
geometry. The upper boundary and the right boundary are natural gas over the world. The gas methane is generated and
constant stress. The displacements at the left and bottom sides stored in the coals matrix. Permeability of coal plays a critical
are constrained in the horizontal and vertical directions, role in the production of coal seam gas from coal reservoirs. CB
respectively. A distributed overburden load of 10 MPa is model is a popular permeability model with considerations of
applied and remains unchanged during the whole process. The stress conditions and adsorption swelling for coal, which has
initial gas pore pressure is 10 MPa and gas pressure on the right been verified by many experiments of coal matrix. We adopt
side remains 0.1 MPa as shown in Figure 7. Zero fluxes are four cases of simulation to study the evolution behaviors with
specified on the other boundaries. In this section, the reduced different permeability models. The first case with the second
5499 DOI: 10.1021/acs.energyfuels.6b00683
Energy Fuels 2016, 30, 5492−5505
Energy & Fuels Article
boundary conditions are the same as second type as shown in model is a function of effective stress. If we adopt the dynamic
Figure 10. Point A is located in the center of the geometry. pressure sensitivity coefficient as the reduced model, the trends
There are four cases for the simulations. The first case is the for David model and reduced model will be the same. Through
simulation with the third reduced permeability model. The these simulation results, we can see that the impact of effective
second case adopts the general model for the porous medium stress has a close relationship with the gas pressure. When the
with high permeability and low adsorption capacity. But the intrinsic permeability is high, the effect of flow regimes is lower.
third case uses the general model with ultralow permeability For example, permeability of conventional gas reservoirs is
model and high adsorption parameters. The fourth case adopts higher than that of shale reservoir, which means that the effect
David model of sandstones. Both the David model and the of flow regime can be ignored. The adsorption capacity for
reduced model focus on the evolution of permeability from conventional gas reservoirs is also lower compared with
effective stress. Under this condition, the effective stress is the unconventional gas reservoirs like coals and shales. The main
main factor that changes the intrinsic permeability. According dominated factor for conventional gas reservoirs is the effective
to the results, we can know the reduced model is useful for the stress. Overall, this reduced model is applicable for conven-
conventional gas reservoirs like sandstones. But for the tional gas reservoirs like David model.
unconventional gas reservoirs, the evolution curve is different 4.5. Comparison between Javadpour Model of Shales
from that of the reduced model. For the first case, the initial and Fourth Reduced Model. In recent years, many scholars
intrinsic permeability is about 3.75 mD when we assume the emphasize the importance of the flow regimes for unconven-
average pore radius for the porous medium is 1000 nm. In tional gas reservoirs and many experiments show that lower gas
terms of the second case with general model, the Langmuir pressure will lead to big apparent permeability. The impact of
volume is 0.0005 m3/kg and the maximum strain is just 0.001. effective stress on the gas permeability is low when the variation
Also, the initial intrinsic permeability of the second case is of gas pressure is small. Javadpour13 proposed an apparent
approximately 3.75 mD. For the third case, the simulation permeability model for gas shales and this model has been used
adopts that Langmuir volume is 0.005 m3/kg and the maximum by many scholars. The Javadpour model can be expressed as
adsorbed strain is 0.02. The intrinsic permeability is smaller
than that of the above two cases and the value is about 0.056 2r ⎛⎜ 8RT ⎞⎟
0.5
μ ⎛ ⎛ 8πRT ⎞0.5 μ ⎛ 2 ⎞⎞ r 2
kapp = + ⎜1 + ⎜ ⎟ ⎜ − 1⎟⎟
mD. For the fourth case, high initial intrinsic permeability 3.75 3 ⎝ πM ⎠ p ⎝ ⎝ M ⎠ pr ⎝ β ⎠⎠ 8
mD is used and the other parameters are the same as the other
cases. (42)
It is clearly seen that the third reduced model experiences the where M is gas molar mass, R denotes the gas constant, T is the
same trend as the general model with high initial intrinsic absolute temperature in Kelvin, r is the average pore radius, μ
permeability and low adsorption capacity. Under these denotes the gas viscosity, and β is the tangential momentum
conditions, the impacts from flow regime and adsorption can accommodation coefficient varied from 0 to 1.
be neglected and the apparent permeability will decrease when Javadpour model takes into account the effect of Knudsen
the effective stress increases as shown in Figure 12. The David diffusion and slip flow on the gas flux. The intrinsic
permeability for Javadpour model can be assumed to be a
constant. As discussed above, the general permeability model
for unconventional gas reservoirs like shales can be reduced
into the fourth type because the variation of effective stress is
small. We adopt the gas helium to do these simulations, which
can ignore the effect of adsorption. The simulation model
geometry and boundary conditions are the same as the first
reduced model as shown in Figure 7. Point A is located in the
center of the geometry. To some extent, the decrease of gas
pressure means the increase of the Knudsen number and the
apparent permeability will undergo an increasing trend in the
gas depletion process. In this section, we propose two cases for
numerical simulation with the same geometry and boundary
conditions. The first case adopts the reduced model without
considering the variable intrinsic permeability and the
adsorption capacity. The second case adopts the Javadpour
model to study the evolution of gas permeability for shales. The
Figure 12. Evolution of permeability ratio for point A under different initial intrinsic permeability for the first case is 300 nD. The
conditions. second case is based on the Javadpour model with constant
intrinsic permeability 300 nD. The initial gas pressure is 10
MPa and the boundary gas pressure is just 0.1 MPa.
model shows a good agreement with the reduced model at the It is clearly seen that the fourth reduced model experiences
early stage of gas depletion. Permeability ratio from reduced the same trend as the Javadpour model of shales with ultralow
model reaches 0.83 when the gas pressure drops to only 4 MPa intrinsic permeability. Under these conditions, the impacts of
while the ratio from David model is almost 0.8 at the same effective stress and adsorption on the gas permeability can be
pressure. Permeability from the David model is a little lower neglected and the apparent permeability will increase with the
than that of the reduced model at the lower pressure gas depletion from the reservoirs as shown in Figure 13. When
magnitude. The reason is that the pressure sensitivity the gas pressure decreases from 10 MPa to only 0.1 MPa, the
coefficient γ is a constant but the compressibility in the general apparent permeability increases more than 60 times of the
5501 DOI: 10.1021/acs.energyfuels.6b00683
Energy Fuels 2016, 30, 5492−5505
Energy & Fuels Article
Table 3. Properties of Porous Media for Eight Cases with the General Model
adsorption capacity compressibility (MPa−1) initial intrinsic permeability (mD) possible gas reservoirs
3
case one Langmuir volume 0.005 m /kg 0.02 0.006 unconventional, coals
maximum adsorption strain 0.02
case two Langmuir volume 0.005 m3/kg 0.02 1 unconventional, coals
maximum adsorption strain 0.02
case three Langmuir volume 0.005 m3/kg 0.001 0.006 unconventional, shales
maximum adsorption strain 0.02
case four Langmuir volume 0.005 m3/kg 0.001 1 unconventional, coals
maximum adsorption strain 0.02
case five Langmuir volume 0.0005 m3/kg 0.02 0.006 unconventional, shales
maximum adsorption strain 0.001
case six Langmuir volume 0.0005 m3/kg 0.02 1 conventional, sandstones
maximum adsorption strain 0.001
case seven Langmuir volume 0.0005 m3/kg 0.001 0.006 unconventional, shales
maximum adsorption strain 0.001
case eight Langmuir volume 0.0005 m3/kg 0.001 1 conventional, sandstones
maximum adsorption strain 0.001
■ AUTHOR INFORMATION
Corresponding Author
(17) Li, J.; Liu, D.; Yao, Y.; Cai, Y.; Xu, L.; Huang, S. Control of CO2
permeability change in different rank coals during pressure depletion:
an experimental study. Energy Fuels 2014, 28 (2), 987−996.
*E-mail: peng.cao@research.uwa.edu.au. (18) Florence, F. A.; Rushing, J.; Newsham, K. E.; Blasingame, T. A.
Notes Improved permeability prediction relations for low permeability sands.
The authors declare no competing financial interest. Presented at Rocky Mountain Oil & Gas Technology Symposium,
■
Denver, CO, USA, April 16−18; Society of Petroleum Engineers,
2007.
ACKNOWLEDGMENTS (19) Klinkenberg, L. The permeability of porous media to liquids and
This work is supported by Australian International Post- gases. In Drilling and Production Practice; American Petroleum
graduate Research Scholarship (IPRS), Australian Postgraduate Institute, 1941.
Award (APA), and partially funded by National Natural Science (20) Yao, J.; Sun, H.; Fan, D.; Wang, C.; Sun, Z. Numerical
Foundation of China (51474204). This support is gratefully simulation of gas transport mechanisms in tight shale gas reservoirs.
Pet. Sci. 2013, 10 (4), 528−537.
acknowledged.
■
(21) Beskok, A.; Karniadakis, G. E. Report: a model for flows in
channels, pipes, and ducts at micro and nano scales. Microscale
REFERENCES Thermophys. Eng. 1999, 3 (1), 43−77.
(1) Loucks, R. G.; Reed, R. M.; Ruppel, S. C.; Jarvie, D. M. (22) Darabi, H.; Ettehad, A.; Javadpour, F.; Sepehrnoori, K. Gas flow
Morphology, genesis, and distribution of nanometer-scale pores in in ultra-tight shale strata. J. Fluid Mech. 2012, 710, 641−658.
siliceous mudstones of the Mississippian Barnett Shale. J. Sediment. Res. (23) Li, J.; Liu, D.; Lu, S.; Yao, Y.; Xue, H. Evaluation and modeling
2009, 79 (11−12), 848−861. of the CO2 permeability variation by coupling effective pore size
(2) Heller, R.; Vermylen, J.; Zoback, M. Experimental investigation of evolution in anthracite coal. Energy Fuels 2015, 29 (2), 717−723.
matrix permeability of gas shales. AAPG Bull. 2014, 98 (5), 975−995. (24) Naraghi, M. E.; Javadpour, F. A stochastic permeability model
(3) Civan, F.; Rai, C. S.; Sondergeld, C. H. Shale-gas permeability for the shale-gas systems. Int. J. Coal Geol. 2015, 140, 111−124.
and diffusivity inferred by improved formulation of relevant retention (25) Tanikawa, W.; Shimamoto, T. Comparison of Klinkenberg-
and transport mechanisms. Transp. Porous Media 2011, 86 (3), 925− corrected gas permeability and water permeability in sedimentary
944. rocks. Int. J. Rock Mech. Min. Sci. 2009, 46 (2), 229−238.
(4) Cui, A.; Wust, R.; Nassichuk, B.; Glover, K.; Brezovski, R.; (26) Cui, X.; Bustin, R. M. Volumetric strain associated with methane
Twemlow, C. In a nearly complete characterization of permeability to desorption and its impact on coalbed gas production from deep coal
hydrocarbon gas and liquid for unconventional reservoirs: a challenge seams. AAPG Bull. 2005, 89 (9), 1181−1202.
to conventional thinking. Presented at the Unconventional Resources (27) Pan, Z. J.; Connell, L. D. Modelling permeability for coal
Technology Conference, Denver, CO, USA, August 12−14, 2013. reservoirs: A review of analytical models and testing data. Int. J. Coal
(5) Kwon, O.; Kronenberg, A. K.; Gangi, A. F.; Johnson, B.; Herbert, Geol. 2012, 92, 1−44.
B. E. Permeability of Illite-bearing shale: 1. Anisotropy and effects of (28) Zhang, H. B.; Liu, J. S.; Elsworth, D. How sorption-induced
clay content and loading. J. Geophys. Res.: Solid Earth 2004, 109 (B10), matrix deformation affects gas flow in coal seams: A new FE model.
B10205. Int. J. Rock Mech. Min. Sci. 2008, 45 (8), 1226−1236.
(6) Soeder, D. J. Porosity and permeability of eastern Devonian gas (29) Sakhaee-Pour, A.; Bryant, S. Gas permeability of shale. SPE
shale. SPE Form. Eval. 1988, 3 (1), 116−124. Reservoir Eval. Eng. 2012, 15 (4), 401−409.
(7) Li, S.; Tang, D.; Xu, H.; Yang, Z.; Guo, L. Porosity and (30) Cao, P.; Liu, J.; Leong, Y. K. A fully coupled multiscale shale
permeability models for coals using low-field nuclear magnetic deformation-gas transport model for the evaluation of shale gas
resonance. Energy Fuels 2012, 26 (8), 5005−5014. extraction. Fuel 2016, 178, 103−117.
(8) Bhandari, A. R.; Flemings, P. B.; Polito, P. J.; Cronin, M. B.; (31) Ghanizadeh, A.; Clarkson, C.; Aquino, S.; Ardakani, O.; Sanei,
Bryant, S. L. Anisotropy and stress dependence of permeability in the H. Petrophysical and geomechanical characteristics of Canadian tight
Barnett shale. Transp. Porous Media 2015, 108 (2), 393−411. oil and liquid-rich gas reservoirs: I. Pore network and permeability
(9) David, C.; Wong, T.-F.; Zhu, W.; Zhang, J. Laboratory characterization. Fuel 2015, 153, 664−681.
measurement of compaction-induced permeability change in porous (32) Moghaddam, R. N.; Aghabozorgi, S.; Foroozesh, J. Numerical
rocks: Implications for the generation and maintenance of pore
simulation of gas production from tight, ultratight and shale gas
pressure excess in the crust. Pure Appl. Geophys. 1994, 143 (1−3),
reservoirs: flow regimes and geomechanical effects. Presented at
425−456.
EUROPEC 2015, Madrid, Spain, June 1−4, 2015.
(10) Liu, J.; Chen, Z.; Elsworth, D.; Qu, H.; Chen, D. Interactions of
(33) Detournay, E.; Cheng, A. H. D. Fundamentals of poroelasticity.
multiple processes during CBM extraction: a critical review. Int. J. Coal
Geol. 2011, 87 (3), 175−189. In Comprehensive Rock Engineering: Principles, Practice and Projects,
(11) Raghavan, R.; Chin, L. Productivity changes in reservoirs with Fairhurst, C, Ed.; Pergamon Press, 1993; Vol. 2, Chapter 5, pp 113−
stress-dependent permeability. SPE Reservoir Eval. Eng. 2004, 7 (4), 171.
308−315. (34) Yuan, W. N.; Pan, Z. J.; Li, X.; Yang, Y. X.; Zhao, C. X.; Connell,
(12) Cao, P.; Liu, J.; Leong, Y. K. Combined impact of flow regimes L. D.; Li, S. D.; He, J. M. Experimental study and modelling of
and effective stress on the evolution of shale apparent permeability. J. methane adsorption and diffusion in shale. Fuel 2014, 117, 509−519.
Unconv. Oil Gas Resour. 2016, 14, 32−43. (35) McKee, C.; Bumb, A.; Koenig, R. Stress-dependent permeability
(13) Javadpour, F. Nanopores and apparent permeability of gas flow and porosity of coal and other geologic formations. SPE Form. Eval.
in mudrocks (shales and siltstone). J. Can. Pet. Technol. 2009, 48 (8), 1988, 3 (1), 81−91.
16−21. (36) Connell, L. D.; Lu, M.; Pan, Z. J. An analytical coal permeability
(14) Ziarani, A. S.; Aguilera, R. Knudsen’s permeability correction for model for tri-axial strain and stress conditions. Int. J. Coal Geol. 2010,
tight porous media. Transp. Porous Media 2012, 91 (1), 239−260. 84 (2), 103−114.
(15) Garrouch, A. A.; Smaoui, N. Application of artificial neural (37) Palmer, I.; Mansoori, J. How permeability depends on stress and
network for estimating tight gas sand intrinsic permeability. Energy pore pressure in coalbeds: a new model. Presented at SPE Annual
Fuels 1996, 10, 1053−1059. Technical Conference and Exhibition, Denver, CO, October 6−9;
(16) Zhang, L. J.; Li, D. L.; Lu, D. T.; Zhang, T. A new formulation of Society of Petroleum Engineers, 1996.
apparent permeability for gas transport in shale. J. Nat. Gas Sci. Eng. (38) Shi, J. Q.; Durucan, S. Drawdown induced changes in
2015, 23, 221−226. permeability of coalbeds: a new interpretation of the reservoir