You are on page 1of 29

Journal of Geophysics and Engineering

ACCEPTED MANUSCRIPT

Numerical solution of fractured horizontal wells in shale gas reservoirs


considering multiple transport mechanisms
To cite this article before publication: Yu-long Zhao et al 2017 J. Geophys. Eng. in press https://doi.org/10.1088/1742-2140/aa93b2

Manuscript version: Accepted Manuscript


Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”

This Accepted Manuscript is © 2017 Sinopec Geophysical Research Institute.

During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.

After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0

Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.

View the article online for updates and enhancements.

This content was downloaded from IP address 134.117.10.200 on 28/01/2018 at 13:05


Page 1 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 Numerical solution of fractured horizontal wells in shale gas reservoirs
4

pt
5 2 considering multiple transport mechanisms
6
7
3 Yu-long Zhaoa*, Xu-chuan Tangb, Lie-hui Zhanga*,Hong-ming Tanga,Zheng-Wu Taoc
8
9 a
4 State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation, Southwest Petroleum

cri
10
11
5 University, Chengdu, Sichuan, P. R. China
12
13 b
6 Xinjiang Oilfield, PetroChina, Karamay 834000, P. R. China
14
15 7 c
Research Institute of Exploration and Development, Tarim Oilfield Company, PetroChina, Korla,
16

us
17 8 Xinjiang 841000, China
18
19 9 * Corresponding author: Y.L. Zhao and L.H., Zhang
20
21 10 E-mail address: 373104686@qq.com and swpuzhao@swpu.edu.cn
22
23
24
25
26
27
28
11

12

13
an
Abstract: The multiscale pore size and specific gas storage mechanism in organic-rich shale gas

reservoirs make gas transport in such reservoirs complicated. Therefore, a model that fully

incorporates all transport mechanisms and employs an accurate numerical method is urgently
dM
29 14 needed to simulate the gas production process. In this paper, a unified model of apparent
30
31 15 permeability was first developed, which took into account multiple influential factors including
32
33 16 slip flow, Knudsen diffusion, surface diffusion, effects of the adsorbed layer, permeability stress
34
35 17 sensitivity, and ad-/desorption phenomena. Subsequently, a comprehensive mathematical model,
36
37 18 which included the model of apparent permeability, was derived to describe gas production
38
39 19 behaviors. Thereafter, on the basis of unstructured perpendicular bisection grids and finite volume
pte

40
41 20 method, a fully implicit numerical simulator was developed using Matlab software. The validation
42
43 21 and application of the new model were confirmed using a field case reported in the literature.
44
45 22 Finally, the impacts of related influencing factors on gas production were analyzed. The results
46
23 showed that Knudsen diffusion resulted in a negligible impact on gas production in the proposed
47
48
ce

24 model. The smaller the pore size was, the more obvious the effects of the adsorbed layer on the
49
50 25 well production rate would be. Permeability stress sensitivity had a slight effect on well
51
52 26 cumulative production in shale gas reservoirs. Adsorbed gas made a major contribution to the later
53
Ac

54 27 flow period of the well; the greater the adsorbed gas content, the greater the well production rate
55
56 28 would be. This paper can improve the understanding of gas production in shale gas reservoirs for
57
58
59
1
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 2 of 28

1
2
3 1 petroleum engineers.
4

pt
5 2 Keywords: shale gas; apparent permeability; production rate; PEBI grid; fractured horizontal well
6
7 3 1. Introduction
8
9 4 With the dramatic decline in the production rate and reserves of conventional natural gas resources,

cri
10
11 5 shale gas has gradually provided an increasing contribution to the world’s natural gas supply in
12
13 6 recent years. The development of shale gas reservoirs has attracted much interest and has played
14
15 7 an important role in the energy industry of North America. However, in comparison to
16

us
17 8 conventional gas reservoirs, shale gas reservoirs are characterized by multiple gas transport
18
19 9 mechanisms during the development process. Hence, an accurate model for simulating and
20
21 10 forecasting well production performance is urgently needed.
22
11 Typical organic-rich shale gas reservoirs generally have ultralow porosity and permeability.
23
24
25
26
27
28
12

13

14
an
Specifically, the porosity of shale formations is usually less than 10% and their permeability varies

from the nanodarcy to the microdarcy range (Ghanizadeh et al., 2014). Nanoscale pores are

dominant in shale gas reservoirs and pores with a pore radius of less than 10 nm account for 42%
dM
29
30
15 of the total pore space (Akkutlu and Fathi, 2012). Organic pores usually have an even smaller
31
32 16 radius (Adesida et al., 2011). In general, shale gas is stored as a bulk gas (free gas) in the matrix
33
34 17 pore spaces and an adsorbed gas on the surface of organic material (Jenkins and Boyer, 2008; Sun
35
36 18 et al., 2013). Therefore, the specific pore sizes and forms of gas storage in shale gas reservoirs
37
38 19 mean that gas transport in such reservoirs comprises a combination of several transport
39
pte

40 20 mechanisms such as viscous flow, slip flow, Knudsen diffusion, and surface diffusion, as well as
41
42 21 the adsorption/desorption of adsorbed gas with depressurization during production.
43
44 22 In the past, a large number of papers have been published that concentrated on correcting the
45
46 23 apparent permeability for gas transport in shale gas reservoirs. Beskok and Karniadakis (1999)
47
48 24 developed a unified Hagen–Poiseuille-type equation, which has been extensively used to address
ce

49
50 25 the fundamental mechanisms of fluid transport in tight porous media, including continuum flow,
51
52 26 slip flow, transition flow, and free molecular flow. Javadpour et al. (2007) described gas flow in
53
27 micropores in terms of Darcy flow and in nanopores in terms of Knudsen diffusion, and a later
Ac

54
55
56 28 paper (Javadpour, 2009) presented an apparent model based on a linear superposition of slip flow
57
58 29 and Knudsen diffusion for gas flow in shale gas formations. Civan (2010) developed a
59
2
60
Page 3 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 permeability model for gas flow in tight porous media, which considered rarefaction effects and
4

pt
5 2 slippage effects on the basis of the Beskok–Karniadakis model. Civan et al. (2011) presented an
6
7 3 improved model to accurately describe the apparent permeability and diffusivity of shale gas.
8
9 4 Azom and Javadpour (2012) incorporated slip flow and Knudsen diffusion into a model of

cri
10
11 5 apparent permeability to simulate dual-continuum shale gas reservoirs. Xiong et al. (2012)
12
13 6 analyzed the impacts of non-Darcy flow effects and surface diffusion of an ideal gas on apparent
14
15 7 permeability. Singh et al. (2014) proposed that a combination of Darcy flow and Knudsen flow
16
8 (without considering the Klinkenberg effect) can describe gas flow for a range of Knudsen flow

us
17
18
9 conditions applicable to a shale gas system. Wu et al. (2015a,2015b, 2015c) combined viscous
19
20
10 flow and Knudsen diffusion based on weighting coefficients to study real gas transport in
21
22 11 nanopores. Wang et al. (2015a) discussed the influence of multilayer adsorption on surface
23
24
25
26
27
28
12

13

14
an
diffusion and the contribution of surface diffusion to apparent permeability. Wu et al. (2016)

developed a comprehensive model of apparent permeability by coupling multiple mechanisms,

including non-Darcy effects, real gas effects, geomechanical effects, and the effects of adsorption
dM
29
30 15 layers on gas transport. However, the impacts of geomechanical effects on effective pore radius
31
32 16 and porosity were treated separately in this paper. Song et al. (2016) developed a unified model of
33
34 17 apparent permeability for both organic matrices and inorganic matrices by coupling multiple
35
36 18 mechanisms separately.
37
38 19 Furthermore, both gas adsorption/desorption phenomena (Sigal, 2013;Yao et al., 2013a, 2013
39
20 b; Negara et al., 2016; Negara and Sun, 2016) and geomechanical effects (Yu and Sepehrnoori,
pte

40
41
42 21 2014; Cao et al., 2016; Song et al., 2016) can significantly lead to dynamic changes in the
43
44 22 effective pore radius of the shale matrix during the depressurization process in shale gas reservoirs,
45
46 23 which can further affect apparent permeability. Real gas effects can also distinctly affect gas
47
48 24 transport at relatively high gas pressures (Wang et al., 2008; Ma et al., 2014; Zhao et al., 2015).
ce

49
50 25 Therefore, the influencing factors mentioned above must be considered in forecasts of numerical
51
26 gas flow and productivity.
52
53
27 None of the studies mentioned above comprehensively considers the impacts of factors that
Ac

54
55
28 influence gas transport on the apparent permeability and production behaviors of shale gas,
56
57 29 including gas transport mechanisms (viscous flow, slip flow, Knudsen diffusion, and surface
58
59
3
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 4 of 28

1
2
3 1 diffusion), adsorption/desorption phenomena of adsorbed gas, real gas effects, and geomechanical
4

pt
5 2 effects. Furthermore, the influence that these factors exert on the productivity performance of
6
7 3 shale gas wells has rarely been quantitatively analyzed.
8
9 4 In this paper, a more comprehensive model of apparent permeability that considers multiple

cri
10
11 5 influencing factors was first developed. In this model, the impacts of both the adsorbed layer and
12
13 6 geomechanical effects on the effective pore radius, real gas effects, and different gas transport
14
15 7 mechanisms are considered simultaneously in modeling the apparent permeability. Next, the
16
8 governing mass balance equation was derived and a fully implicit numerical discretization scheme

us
17
18
9 was presented based on the finite volume method with perpendicular bisection (PEBI) grids.
19
20
10 Finally, sensitive analyses were performed to study the effects of the related influencing factors on
21
22 11 the well production behaviors of multiple fractured horizontal wells in shale gas reservoirs.
23
24
25
26
27
28
12

13

14
an
2. Gas transport mechanisms in organic-rich shale gas reservoirs

During shale gas exploitation, shale gas transport mechanisms mainly consist of viscous flow,

slip flow, and Knudsen diffusion for bulk gas and surface diffusion for adsorbed gas, as well as
dM
29
30
15 gas adsorption and desorption phenomena in the depressurization process (Fig. 1).
31
32
Viscous flow
33
34 Slip flow
35 Knudsen diffusion
36
Surface diffusion
37
38 Adsorption
39
Desorption
pte

40 16
41 17 Fig. 1 Schematic diagram of gas transport in nanoscale shale gas reservoirs
42
18 2.1 Effect of adsorption/desorption on effective pore radius
43
44
19 Adsorbed gas on the surface of an organic matrix usually obeys a Langmuir isotherm. The
45
46 20 relationship between the adsorbed gas volume and pressure for a real gas at a constant temperature
47
48
ce

21 is expressed as follows (Civan et al., 2013):


49
50 p Z
51 22 Ga = VL (1)
52 pL + p Z
53
23 The coverage can be defined as the ratio of the adsorbed gas volume to the Langmuir volume.
Ac

54
55
56 24 The coverage for a real gas can be expressed as follows:
57
58
59
4
60
Page 5 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 p Z
1 θ= (2)
4 pL + p Z

pt
5
6 2 Hence, considering the pore space occupied by adsorbed gas molecules owing to
7
8 3 adsorption/desorption phenomena, the effective pore radius of a shale matrix can be expressed as
9

cri
10 4 follows:
11
12 5 reff = r0 − θ d m (3)
13
14
6 2.2 Influence of geomechanical effects on effective pore radius
15
16
7 The influence of geomechanical effects in a shale matrix on permeability can be expressed as

us
17
18 8 follows (Wasaki and Akkutlu, 2015; Song et al., 2016;Cai, et al., 2017):
19
20 9
21 3
22   σ eff  m 
k = k0  1 −  
23
24
25
26
27
28
10

11

12
an   p1  

σ eff = pc − α p

The relationship between the permeability, porosity, pore radius, and tortuosity of a shale
(4)

(5)
dM
29
30 13 matrix can be expressed as follows:
31
32 φr2
33 14 k= (6)

34
35
15 When the effective stress varies from σeff_0 to σeff, the corresponding permeability varies from
36
37
16 k0 to k. Simultaneously, the tortuosity is regarded as constant. By eqs. (4) and (6), the ratio
38
39 17 between k and k0 can be expressed as follows:
pte

40
41 3
42   σ eff  m 
1 −  
43
φr2   p1  
44 18 =   (7)
φ0 r02   σ m 3

eff _ 0 
45
46 1 −   
  p1  
47  
48
ce

19 The porosity is directly proportional to the second power of the pore radius (Florence et al.,
49
50 20 2007; Xiong et al., 2012, Cai, et al., 2017):
51
52 2
53  reff 
21 φeff = φ0   (8)
 r0 
Ac

54
55
56 22 Eq. (7) can be simplified as follows:
57
58
59
5
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 6 of 28

1
2
3
3   σ eff  m 
4 1 −  
  p1  

pt
5 r4  
6 1 = (9)
r04   σ m 3

7 eff _ 0 
1 −   
8   p1  
 
9

cri
10 2 The effective pore radius considering geomechanical effects can be expressed as follows:
11
12 3
13  σ 
m
4
14  1 −  eff  
  p1  
15 3 reff _ stress = r0  m  (10)
16  1 −  σ eff _ 0  
  p 

us
17 
  1  
18
19
4 Hence, considering the pore space occupied by adsorbed gas molecules and geomechanical
20
21
5 effects, the effective pore radius of a shale matrix can be expressed as follows:
22
23
24
25
26
27
28
6

8
2.3 Effect of gas properties an
reff = reff _ stress − θ d m (11)

Shale gas reservoirs are generally under relatively high-pressure conditions. The influence of
dM
29
30 9 the volume of gas molecules and gas intermolecular interactions cannot be neglected. Hence, gas
31
32 10 under shale reservoir conditions cannot be regarded as an ideal gas. In this paper, a gas deviation
33
34 11 factor is used as calculated in the literature (Dranchuk et al., 1973). The gas viscosity is as
35
36 12 calculated in the literature (Lee et al., 1966).
37
13 2.4 Bulk gas transport
38
39
14 The mean free path of molecules for a real gas can be expressed as follows (Civan, 2010):
pte

40
41
π ZRT µ g
42 15 λ= (12)
43 2M p
44
45 16 The Knudsen number for a real gas in nanopores can be expressed as follows:
46
47 λ
17 Kn = (13)
48
ce

reff
49
50
18 2.4.1 Slip flow
51
52
19 When the Knudsen number Kn is less than 0.1, the pore size of the shale matrix is larger than
53
Ac

54 20 the mean free path of gas molecules. Gas transport is dominated by collision between gas
55
56 21 molecules. This regime is referred to as slip flow. Thus, the mass flux of a real gas considering slip
57
58
59
6
60
Page 7 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 flow can be expressed as follows (Civan et al., 2011):
4

pt
5 ρ g kv
6 2 Fv = − ∇p (14)
µg
7
8
3 Where
9

cri
10
 4 Kn 
11 4 kv = k0 (1 + α ( Kn ) Kn ) 1 +  (15)
12  1 + Kn 
13
14
tan −1 ( 4 Kn0.4 )
128
15 5 α ( Kn ) = (16)
15π 2
16

us
17 6 By combining eqs. (6) and (11), eq. (15) can be rearranged as follows:
18
19 φeff reff2  4 Kn 
20 7 kv =

(1 + α ( Kn ) Kn ) 1 +  (17)
21  1 + Kn 
22
23
24
25
26
27
28
8

10

11
2.4.2 Knudsen diffusion

an
When the Knudsen number Kn is larger than 1, the pore size of the shale matrix is smaller

than the mean free path of gas molecules. Gas transport is dominated by collisions between gas

molecules and matrix pore walls. This regime is referred to as Knudsen diffusion. Thus, the mass
dM
29
30
31 12 flux of a real gas considering diffusion can be expressed as follows (Javadpour, 2009):
32
33 13 Fk = − MDk ∇c (18)
34
35 14 Where
36
37 φeff 2reff 8ZRT
38 15 Dk = (19)
τ 3 πM
39
pte

40 16 In terms of the molarity of a real gas, c can be expressed as follows:


41
42 p
43 17 c= (20)
44 ZRT
45
18 By substituting eq. (20) into eq. (18), we get
46
47
48  p 
Fk = − MDk ∇ 
ce


49 19  ZRT  (21)
50 = − Dk ρ g Cg ∇p
51
52 20 Where
53
Ac

54 1 1 ∂Z
55 21 Cg = − (22)
p Z ∂p
56
57
58 22 2.4.3 Total bulk gas flux
59
7
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 8 of 28

1
2
3 1 Considering the typical conditions in shale reservoirs, the bulk gas transport mechanisms in
4

pt
5 2 shale matrix pores include continuum flow, slip flow, and transition flow (Civan et al., 2013). On
6
7 3 the basis of two transport mechanisms, namely, slip flow and Knudsen diffusion, respectively, Wu
8
9 4 (2015a) sets the ratios of the frequency of gas intermolecular collisions and the frequency of

cri
10
11 5 collisions between gas molecules and pore walls to the total frequency of collisions as the
12
13 6 weighting factors for slip flow and Knudsen diffusion, respectively. Then, the total mass flux of
14
15 7 the bulk gas for a real gas can be expressed as follows:
16
8 Fb = ε v Fv + ε k Fk (23)

us
17
18
19 9 Where
20
21 1
10 εv = (24)
22 1 + Kn
23
24
25
26
27
28
11

12 2.5 Transport of adsorbed gas


an
εk =
1
1 + 1 Kn
(25)
dM
29 13 Under the initial conditions in shale gas reservoirs, gas is stored in nanopores in the form of
30
31 14 bulk gas and free gas. During the depressurization process, adsorption/desorption phenomena
32
33 15 accompany the production of gas. At the same time, the residual adsorbed gas on the surface of
34
35 16 organic particles is transported via surface diffusion.
36
37 17 2.5.1 Surface diffusion
38
39 18 The concentration gradient of the adsorbed gas is generally regarded as the driving force of
pte

40
41 19 surface diffusion (Zhao et al., 2016a). Thus, the mass flux of surface diffusion for a real gas can be
42
43 20 expressed as follows (Wu, 2015b):
44
45 21 Fs = −MDs ∇cµ (26)
46
47 22 Where
48
ce

49 ρr
50 23 cµ = Ga (27)
Vstd
51
52
53 24 By substituting eq. (27) into eq.(26), we get
Ac

54
55 ρrVL ∂θ
25 Fs = −MDs ∇p (28)
56 Vstd ∂p
57
58
59
8
60
Page 9 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 Where
4

pt
5 ∂θ p ( pL Z )
6 2 = Cg (29)
∂p (1 + p ( pL Z ) )2
7
8
9 3 2.5.2 Total flux of adsorbed gas

cri
10
11 4 The total mass flux of adsorbed gas for a real gas can be expressed as follows:
12
13 5 Fa = Fs (30)
14
15 6 2.6 Model of apparent permeability
16

us
17 7 The total mass flux for a real gas can be expressed as the sum of the total mass flux of the
18
19 8 bulk gas and the total mass flux of the adsorbed gas:
20
21 9 F = Fb + Fa (31)
22
23
24
25
26
27
28
10

11
  ρ g kv
F = − εv 
  µg
 



an
By substituting eqs. (23) and (30) into eq. (31), we get

ρ V ∂θ
∇p  + ε k ( Dk ρ g C g ∇p ) + MDs r L
Vstd ∂p

∇p 


(32)
dM
29 12 Eq. (32) can be rearranged in the form of Darcy’s law:
30
31 ρ g kapp
32 13 F =− ∇p (33)
µg
33
34
35 14 where kapp is the apparent permeability, which can be expressed as follows:
36
37 µ g ρ rVL ∂θ
15 k app = ε v kv + ε k Dk µ g C g + MDs (34)
38 ρ g Vstd ∂p
39
pte

40 16 According to Eq.(34), with considering weighting factors, the gas flow conductance of slip
41
42 17 flow, Knudsen diffusion, and surface diffusion can be expressed as follows:
43 18 kv′ = ε v kv (35)
44
45 19 kk′ = ε k Dk µ g Cg (36)
46
47 µ g ρ rVL ∂θ
48 20 k s′ = MDs (37)
ce

ρ g Vstd ∂p
49
50 21 2.7 Development of the continuity equation
51
52 22 On the basis of the newly developed model of shale gas apparent permeability, which can
53
Ac

54 23 accurately describe gas transport in shale pores by coupling multiple mechanisms, the continuity
55
56 24 equation can be expressed as follows:
57
58
59
9
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 10 of 28

1
2
3  ρ g kapp  ∂ ∂
4 1 ∇⋅ ∇p  + ρ g Q = (φeff ρ g ) + ( Mcµ ) (38)
 µ  ∂ ∂

pt
  t t
5 g

6
7 2 3. Numerical solution
8
9 3 3.1 Mesh generation

cri
10
11 4 The traditional method of modeling fractured shale gas reservoirs using regular Cartesian
12
13 5 grids could have some disadvantages: (1) inflexibility in representing complex geologies; (2)
14
15 6 inadequacy in capturing the elliptical flow geometries expected around the fracture tips; (3)
16

us
17 7 problems caused by grid orientation effects (Olorode, 2011). Perpendicular bisection (PEBI) grids,
18
19 8 which are also known as Voronoi grids, can overcome these disadvantages. Owing to the ultralow
20
21 9 permeability of shale gas reservoirs, the multiple fractured horizontal well has been proved to be a
22
23
24
25
26
27
28
10

11

12

13
an
common way to achieve commercial value in the process of shale gas extraction (Zhao et al., 2013;

Zhao et al., 2014; Zhao et al., 2016b). In this paper, 2D PEBI grids are used to simulate the

productivity behaviors of fractured shale gas reservoirs. Fig. 2 shows PEBI grids of a fractured

horizontal well with 28 fractures. Fig. 3a represents a logarithmic radial refinement of the grid
dM
29
30
14 around the fracture tips and Fig. 3b represents a logarithmic refinement of the grid along the
31
32
15 direction of the horizontal well.
33
34
35
36
37
38
39
pte

40
41
42 16
43 17 Fig. 2 Plane view of the mesh used for application of the model
44
18
45
46
47
48
ce

49
50
51
52
53 (a) (b)
19
Ac

54
55 20 Fig. 3 Plane view of the refinements of the grids
56 21 3.2 Discretization of continuity equation
57
58
59
10
60
Page 11 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 Using the finite volume method with an implicit scheme, eq. (38) and the supplemental
4

pt
5 2 equations can be numerically discretized as follows:
6
∆Vi
(
−∑ Tij ( pi − p j ) )
n +1
+ ( ρ g qg ) − ( M in +1 − M in )
7 n +1
3 =0 (39)
8 j
i ∆t
9

cri
10 4 Mi is the total mass density of the gas in block i, which can be expressed as follows:
11
12
5 M i = (φeff ρ g + Mcµ ) (40)
13 i

14
15 6 Tij is the transmissibility of the interface between block i and block j. On the basis of an
16

us
17 7 upstream weighting scheme, Tij can be expressed as follows:
18
19  Aij  ρg 
20   kapp  pi ≥ p j

21  Di + D j  µg i
8 Tij =  (41)
22  Aij  ρg 
pi < p j
23
24
25
26
27
28
9

10
D + D
 i j

an 
 µg
kapp 

j

Because of the strong nonlinearity of some parameters and variables with pressure in eqs. (39)

–(41), the Newton–Raphson method is used to solve these equations. Eq. (39) can be simplified as
dM
29
30 11 follows:
31
32
33
12
j
(
−∑ Fijn +1 ( pin +1 , p nj +1 ) + Qin +1 ( pin +1 ) − M in +1 ( pin +1 ) − M in ( pin ) = 0 ) (42)

34
35 ∆Vi n +1 ∆V
( )
n +1
where Fijn +1 = Tij ( pi − p j ) ; Qin +1 = ( ρ g q g ) ; M in +1 =
n +1
36 13 M i ; M in = i M in .
i ∆t ∆t
37
38 r
39 14 Thus, the residual vector of the Newton–Raphson method Ri can be expressed as follows:
pte

40 r
41
42
15
j
(
Ri = −∑ Fijn +1 ( pin +1 , p nj +1 ) + Qin +1 ( pin +1 ) − M in +1 ( pin +1 ) − M in ( pin ) ) (43)

43
r
44 16 When pi ≥ pj, the elements of the sparse Jacobian matrix J can be expressed as follows:
45
46  ∂Fijn +1 ∂Qin +1 ∂M in +1
J ik =  ∑
47  − n +1
+ n +1 − , k =i
48 17 j ∂pi ∂pi ∂pin +1 (44)
ce

49 
0, other
50
51 r
18 When pi < pj, the elements of the sparse Jacobian matrix J can be expressed as follows:
52
53
Ac

54
55
56
57
58
59
11
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 12 of 28

1
2
3 ∂Qin +1 ∂M in +1
4  n +1 − , k =i
 ∂pi ∂pin +1

pt
5
 ∂F n +1
6 
J ik =  − n +1 , k= j
ij
1 (45)
7  ∂ p j
8 0, other
9 


cri
10
11
r
12 2 After calculating the sparse Jacobian matrix J , the sparse linear system eq. (46) can be
13
14 3 solved for each iteration l:
15
16 r r r
4 J ⋅ δ pln +1 = − R (46)

us
17
18 r r
19 5 When the solution is convergent on condition that R < tol , pln +1 can be calculated by eq.
20
21 6 (47).
22
23
24
25
26
27
28
7

9
4. Model verification
r

an r r
pln +1 = pln−+11 + δ pln +1 (47)

To ensure the reliability and validity of the simulation, verification was performed by history
dM
29
30 10 matching of production data from the Barnett shale gas reservoir. The parameters used in the
31
11 simulation are shown in Table 1. The reservoir parameters were obtained from the literature
32
33
12 (Al-Ahmadi and Wattenbarger, 2011; Wang et al., 2015b; Wasaki and Akkutlu, 2015). The
34
35
13 parameters of the hydraulic fractures were obtained by matching production data. The results of
36
37 14 matching historic production data are shown in Fig. 4. It is obvious that the results of the
38
39 15 simulation show good agreement with historic production data for the Barnett Shale.
pte

40
41 16 Table 1 Parameters used for simulation of Barnett Shale
42
43 Parameter Value Units
44 Model dimensions 916 × 290 × 90 m×m×m
45 Grid block number 16158 −
46 Initial reservoir pressure, pi 21 MPa
47 Bottom hole pressure, pwf 3.5 MPa
48
ce

Well radius, rw 0.1 m


49
50 Reservoir temperature, T 340 K
51 Reservoir porosity, ϕ0 0.06 −
52 Initial gas saturation, Sg 0.7 −
53 Initial permeability, k0 1.4 × 10-19 m2
Ac

54 Tortuosity, τ 1.5 −
55 Number of hydraulic fractures 28 −
56
Hydraulic fracture half-length 40 m
57
58 Hydraulic fracture spacing 30 m
59
12
60
Page 13 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 Hydraulic fracture height 90 m
4 Hydraulic fracture width 0.003 m

pt
5 Hydraulic fracture conductivity 2 × 10-16 m3
6 Langmuir volume, VL 2.72 × 10-3 m3/kg
7
Langmuir pressure, pL 4.48 MPa
8
9 Surface diffusion coefficient, Ds 1 × 10-7 m2/s

cri
10 Gas molecular mass, M 0.016 kg/mol
11 Molar volume in standard conditions, Vstd 0.0224 m3/mol
12 Shale density, ρr 2580 kg/m3
13 Gas molecular diameter, dm 3.8 × 10-10 m
14 p1 179 MPa
15
Confining pressure, pc 103 MPa
16
m 0.5

us
17 −
18 α 0.5 −
19 1
20 30
21
Field data
22 Simulation results
23 25

an
Gas production rate, 104m3/day

24
25 20
26
27 15
28
dM
29
10
30
31
32 5
33
34 0
35 0 400 800 1200 1600
2 Time, day
36
37 3 Fig. 4 Comparison between simulation data and field data for the Barnett Shale
38 4 As shown in Fig. 5, Fig. 6 and Fig. 7, it is clearly seen that when a well is open for
39
pte

40 5 production for about 100 days, gas flows from the shale matrix into hydraulic fractures linearly
41
42 6 (Fig. 5). This is referred to as the formation linear flow period. With the continuation of the
43
44 7 depressurization process, the compound linear flow period may be observed, as shown in Fig. 6.
45
46 8 Finally, a long period of production decline is required before pseudo-elliptical flow appears,
47
48 9 owing to the ultralow permeability of shale reservoirs (Fig. 7).
ce

49
50 10 From these figures, we can see that with the progress of production the pressure wave spreads
51
52 11 to more areas. The pressure at different locations varies dynamically, which can further affect the
53
12 paths available for gas transport according to eq. (11). Therefore, the porosity and permeability
Ac

54
55
56 13 change during the process of gas production.
57
58
59
13
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 14 of 28

1
2
3
4

pt
5
6
7
8
9 1

cri
10 2 Fig. 5 Gas pressure distribution in Barnett Shale at 100 days (in MPa)
11
12
13
14
15
16

us
17
18 3
19 4 Fig. 6 Gas pressure distribution in Barnett Shale at 1000 days (in MPa)
20
21
22
23
24
25
26
27
28
5
6
7
an
Fig. 7 Gas pressure distribution in Barnett Shale at 10000 days (in MPa)

5. Results and discussion


dM
29
30
31 8 On the basis of the results of matching Barnett Shale data, a base case model was established
32
33 9 to perform sensitive analyses. The parameters of the base case are identical to the data listed in
34
35 10 Table 1. In this study, to better understand the behavior of gas flow in shale gas reservoirs during
36
37 11 gas production, analyses of the impacts of gas transport mechanisms (including slip flow, Knudsen
38
39 12 diffusion, and surface diffusion), the effects of the adsorbed layer, geomechanical effects,
pte

40
41 13 ad-/desorption phenomena, surface diffusion coefficients, and real gas effect are conducted in this
42
43 14 section sequentially.
44
45 15 Fig. 8 shows the gas production rate and cumulative gas production performance considering
46
47 16 slip flow (SF), Knudsen diffusion (KD), and surface diffusion (SD). Geomechanical effects (G)
48
ce

17 are involved in all cases in Fig. 8. As shown in Fig. 8, cumulative gas production without
49
50
18 considering Knudsen diffusion increases by 1.46% in comparison to the base case. Without
51
52 19 considering surface diffusion, cumulative gas production decreases by 0.93%. From Fig. 9, we can
53
Ac

54 20 see that Knudsen diffusion and surface diffusion contribute smaller on gas transport share for the
55
56 21 base case. And slip flow dominates in the whole period of pressurization. Thus, Knudsen diffusion
57
58
59
14
60
Page 15 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 and surface diffusion may have small effects on the gas production rate and cumulative gas
4

pt
5 2 production in the present model. When the relationship between slip flow and Knudsen diffusion
6
7 3 uses linear superposition, cumulative gas production increases by 17.28%, which means that linear
8
9 4 superposition of slip flow and Knudsen diffusion may result in a forecast of larger cumulative gas

cri
10
11 5 production.
12 30 1.4
13 Gas rate SF+KD+SD+G (Base case)
Gas rate SF+SD+G
Gas rate SF+KD+G
14 25
Gas rate SF+KD+SD+G (Linear relationship)
Gas production SF+KD+SD+G (Base case) 1.2
Gas production SF+SD+G
15

Cumulative gas production, 108m3


Gas production SF+KD+G
Gas production rate, 104m3/day

Gas production SF+KD+SD+G (Linear relationship)

16 1.0

us
20
17 3.0
0.8
18 2.5
15
19

Gas production rate, 10 4m 3/day


2.0 0.6
20
10 1.5
21 0.4
22 1.0

23
24
25
26
27
28
6
7
5

0
0 2000

Fig. 8 Gas production rate and cumulative gas production performance considering multiple transport mechanisms
1.0
an
4000
Time, day
0.5

0.0
1000

6000
2000 3000
Time, day

8000
4000 5000

10000
6000
0.2

0.0
dM
29
30
0.8
31
32 k0=0.014nd, r0=0.63nm
33 0.6 k0=0.14nd, r0=2nm
kv'/kapp

k0=1.4nd, r0=6.3nm (Base case)


34 k0=14nd, r0=20nm
35 0.4
k0=140nd, r0=63nm

36
37
0.2
38
39
pte

40 0.0
0 5 10 15 20 25 30
41 8 p, MPa
42 9 (a) Slip flow
43 1.0 1.0
44 k0=0.014nd, r0=0.63nm
45 k0=0.014nd, r0=0.63nm
k0=0.14nd, r0=2nm
k0=1.4nd, r0=6.3nm (Base case)
46 0.8 k0=0.14nd, r0=2nm 0.8
k0=14nd, r0=20nm
k0=1.4nd, r0=6.3nm (Base case)
47 k0=14nd, r0=20nm
k0=140nd, r0=63nm

48
ce

k0=140nd, r0=63nm
0.6 0.6
49
ks'/kapp
kk'/kapp

50
0.4 0.4
51
52
53 0.2 0.2
Ac

54
55 0.0 0.0
56 0 5 10 15 20 25 30 0 5 10 15 20 25 30
10 p, MPa p, MPa
57
58 11 (b) Knudsen diffusion (c) Surface diffusion
59
15
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 16 of 28

1
2
3 1 Fig. 9 Contribution share with pressure for different transport mechanism
4 2 Fig. 10 and Fig. 11 illustrate the impacts of Knudsen diffusion and surface diffusion,

pt
5
6 3 respectively, on cumulative gas production for different pore radii. As shown in Fig. 10,
7
8 4 cumulative gas production undergoes a notable improvement with an increase in the pore radius of
9

cri
10 5 the shale matrix, and Knudsen diffusion has a small influence on cumulative gas production for
11
12 6 different pore radii in comparison to the base case. This is because that under a high reservoir
13
14 7 pressure when pore radius is smaller, the surface diffusion and slip flow restricts each other. When
15
16 8 pore radius is larger, the contribution shares of surface diffusion and slip flow do not dominate, as

us
17
18 9 shown in Fig. 9. From Fig. 11, we can see that the impact of surface diffusion on cumulative gas
19
20 10 production increases as the pore size decreases, and there is a remarkable impact on cumulative
21
22 11 gas production when the pore radius of the shale matrix is less than 5 nm, without considering
23
24
25
26
27
12 surface diffusion of adsorbed gas.
3.0
an
SF+KD+SD+G (Base case)
SF+SD+G
10

Difference of cumulative gas production, %


Difference of cumulative gas production
2.5
28 6
Cumulative gas production, 108m3

dM
29 2.0
4

30 2
31 1.5 0
32
-2
33
1.0
34 -4

35 -6
0.5
36 -8
37
0.0 -10
38 0 10 20 30 40 50 60 70

39 13 Nanopore radius, nm
pte

40 14 Fig. 10 Impact of Knudsen diffusion on cumulative gas production for different pore radii
41 3.0 10
SF+KD+SD+G (Base case)
42 SF+KD+G
Difference of cumulative gas production, %

Difference of cumulative gas production


43 2.5 0
Cumulative gas production, 108m3

44
45 2.0 -10
46
47 1.5 -20
48
ce

49 1.0 -30
50
51 0.5 -40
52
53
0.0 -50
Ac

54 0 10 20 30 40 50 60 70

55 15 Nanopore radius, nm

56 16 Fig. 11 Impact of surface diffusion on cumulative gas production for different pore radii
57 17 Fig. 12 shows the impacts of adsorbed gas molecules and geomechanical effects on gas
58
59
16
60
Page 17 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 production performance. As shown in Fig. 12, in comparison to the base case, cumulative gas
4

pt
5 2 production increases by 11.31% when the adsorbed layer is ignored, which means that the
6
7 3 adsorbed layer has a notable impact on gas production forecasts. In addition, it may cause a
8
9 4 significant increase in production forecasts that do not consider the influence of the adsorbed layer.

cri
10
11 5 Moreover, geomechanical effects also have a small influence on cumulative gas production, as the
12
13 6 increase in cumulative gas production is about 2.98% when this effect is not considered. This is
14
15 7 because that when reservoir pressure decreases, pore spaces increase. Thus, when geomechanical
16
8 effects is neglected, cumulative gas production increases. Furthermore, this may lead to forecasts

us
17
18
9 of higher production when geomechanical effects and impacts of the adsorbed layer are ignored
19
20
10 simultaneously, owing to an increase in cumulative gas production of about 14.47% in comparison
21
22 11 to the base case.
23
24
25
26
30

25
Gas rate SF+KD+SD+G (Base case)
Gas rate SF+KD+SD+G

Gas rate SF+KD+SD


an
(Without considering molecules occupying space)

(Considering molecules occupying space)


Gas rate SF+KD+SD
(Without considering molecules occupying space)
Gas production SF+KD+SD+G (Base case)
Gas production SF+SD+G
(Without considering molecules occupying space)
Gas production SF+KD+SD
(Considering molecules occupying space)
Gas production SF+KD+SD
(Without considering molecules occupying space)
1.6

1.4

3
27

Cumulative gas production, 10 m


Gas production rate, 10 m /day

1.2

8
28
3

20
4

1.0
dM
29
30 15
3.0

0.8
31
2.5
Gas production rate, 10 m /day
3

2.0

32 0.6
4

10 1.5

33 1.0 0.4
34 5 0.5

35 0.0
0.2
1000 2000 3000 4000 5000 6000

36 0
Time, day

0.0
37 0 2000 4000 6000 8000 10000
12 Time, day
38
39 13 Fig. 12 Gas production rate and cumulative gas production performance considering the impacts of adsorbed gas
pte

40 14 molecules and geomechanical effects


41 15 Fig. 13 and Fig. 14 show the impacts of the adsorbed layer and geomechanical effects,
42
43 16 respectively, on cumulative gas production for different pore radii. As shown in Fig. 13, the impact
44
45 17 of the adsorbed layer on cumulative gas production increases as the pore size decreases.
46
47 18 Furthermore, when the pore radius of the shale matrix is less than 15 nm, adsorbed gas molecules
48
ce

49 19 may exert a notable negative effect on cumulative gas production. With a decrease in pore size,
50
51 20 this phenomenon will be more prominent because there is a greater similarity in size between the
52
53 21 pore radius and the diameter of gas molecules. According to Fig. 14, it can be observed that
Ac

54
55 22 geomechanical effects have a slight impact on cumulative gas production for different pore radii.
56
57
58
59
17
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 18 of 28

1
2
3.0 200
3 SF+KD+SD+G (Base case)
SF+KD+SD+G (Without considering molecules occupying space)
4 175

Difference of cumulative gas production, %


Difference of cumulative gas production

pt
2.5
5 150

3
Cumulative gas production, 10 m
8
6 125
2.0
7 100
8
1.5 75
9

cri
10 50
1.0
11 25

12 0
0.5
13 -25
14
0.0 -50
15 0 10 20 30 40 50 60 70

16 1 Nanopore radius, nm

us
17 2 Fig. 13 Impact of the adsorbed layer on gas production for different pore radii
18 3.0 10
SF+KD+SD+G (Base case)
19 SF+KD+SD (Considering molecules occupying space)
8

Difference of cumulative gas production, %


Difference of cumulative gas production
20 2.5
6
Cumulative gas production, 108m3

21
4
22 2.0

23
24
25
26
27
28
1.5

1.0

0.5
an 2

-2

-4

-6
dM
29 -8

30 0.0 -10
0 10 20 30 40 50 60 70
31 3 Nanopore radius, nm
32 4 Fig. 14 Impact of geomechanical effects on gas production for different pore radii
33
34 5 In the base case, the Langmuir volume VL and Langmuir pressure pL are 2.72 × 10-3 kg/m3
35
36 6 and 4.48 MPa, respectively. The Langmuir pressure is regarded as constant, whereas the Langmuir
37
38 7 volume is changeable. Three scenarios, namely, 0 kg/m3 (no adsorption), 2.72 × 10-3 kg/m3 (base
39
8 case), and 5 × 10-3 kg/m3 are analyzed. As shown in Fig. 15, it can easily be found that cumulative
pte

40
41
42 9 gas production increases with an increase in the Langmuir volume. The reason for this is that an
43
10 increase in the Langmuir volume means an increase in adsorbed gas reserves. When adsorbed gas
44
45
11 is ignored, gas production is underestimated by 18.62% in comparison to the base case after 10000
46
47
12 days. Thus, the Langmuir volume is an important parameter in forecasts of gas production.
48
ce

49 13 Then, the Langmuir volume is treated as a constant, whereas the Langmuir pressure is
50
51 14 assigned different values, namely, 2.48 MPa, 4.48 MPa (base case), and 6.48 MPa. From Fig. 16,
52
53 15 we can see that cumulative gas production increases by 5.78% and 8.94% for increases in the
Ac

54
55 16 Langmuir pressure from 2.48 MPa to 4.48 MPa and 6.48 MPa, respectively, after 10000 days.
56
57 17 Thus, from Fig. 15 and Fig. 16, we can see that gas desorption has a significant effect on the later
58
59
18
60
Page 19 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 flow period of gas production owing to the relatively low reservoir pressure.
4

pt
30 1.4
5 Gas rate VL=0 m3/kg
Gas rate VL=2.72×10-3 m3/kg (Base case)
6 Gas rate VL=5×10-3 m3/kg
1.2
25 Gas production VL=0 m3/kg
7

3
Gas production VL=2.72×10-3 m3/kg (Base case)

Cumulative gas production, 10 m


Gas production rate, 10 m /day
8

8
Gas production VL=5×10-3 m3/kg
1.0

3
9 20

cri
10 0.8

11 15 3.0

12 0.6
2.5

Gas production rate, 10 4 m 3/day


2.0

13 10
1.5 0.4
14 1.0

15 5
0.5 0.2
16 0.0

us
2000 4000 6000

17
Time, day

0 0.0
0 2000 4000 6000 8000 10000
18 2 Time, day
19 3 Fig. 15 Impact of Langmuir volume on gas production rate and cumulative gas production performance
20 30 1.4
21 Gas rate pL=2.48×106 Pa
Gas rate pL=4.48×106 Pa (Base case)
22 Gas rate pL=6.48×106 Pa 1.2
23
25

an
Gas production pL=2.48×106 Pa

3
Cumulative gas production, 10 m
Gas production pL=4.48×106 Pa (Base case)
Gas production rate, 104m3/day

8
24 Gas production pL=6.48×106 Pa 1.0
20
25
26 3.0
0.8
15
27 2.5
G as production rate, 10 4 m 3 /day

0.6
28 2.0
dM
29 10 1.5

0.4
30 1.0

31 5 0.5
0.2
32 0.0
1000 2000 3000 4000 5000 6000
Time, day

33 0
0 2000 4000 6000 8000 10000
0.0

34 4 Time, day

35 5 Fig. 16 Impact of Langmuir pressure on gas production rate and cumulative gas production performance
36
37 6 Fig. 17 shows the influence of the surface diffusion coefficient on well production
38
7 performance. Four scenarios are presented to study its impact after 10000 days. As shown in Fig.
39
pte

40
8 17, cumulative gas production increases as Ds increases. When Ds is assigned a value of 1 × 10-5
41
42
9 m2/s, cumulative gas production increases by about 48.9% in comparison to the base case. The
43
44 10 effect of Ds on cumulative gas production can be negligible when the surface diffusion coefficient
45
46 11 is less than 1 × 10-7 m2/s. Thus, a reasonable surface diffusion coefficient for real gas reservoirs is
47
48
ce

12 very important for predicting gas production performance.


49
50
51
52
53
Ac

54
55
56
57
58
59
19
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 20 of 28

1
2
30 1.8
3 Gas rate Ds=1×10-5m2/s
Gas rate Ds=1×10-6m2/s
Gas production Ds=1×10-5m2/s
Gas production Ds=1×10-6m2/s

4 -7 2
Gas rate Ds=1×10 m /s (Base case) Gas production Ds=1×10 m /s (Base case)
-7 2
1.6

pt
Gas rate Ds=1×10-8m2/s Gas production Ds=1×10-8m2/s
25
5

Cumulative gas production, 108m3


1.4

Gas production rate, 104m3/day


6
7 20 1.2

8 1.0
9 15 3.0

cri
0.8
10 2.5

Gas production rate, 104 m3 /day


11 10 2.0
0.6

12
1.5

1.0
0.4
13 5
0.5

0.2
14 0.0
1000 2000 3000 4000 5000 6000
Time, day

15 0 0.0
0 2000 4000 6000 8000 10000
16 1 Time, day

us
17 2 Fig. 17 Impact of surface diffusion coefficient on gas production rate and cumulative gas production performance
18
19 3 Fig. 18 illustrates the impact of real gas effects on the gas production rate and cumulative gas
20
21 4 production performance. As shown in Fig. 18, when the gas deviation factor Z and gas viscosity µg
22
5 are assigned values of 1 (for an ideal gas) and 2 × 10-5 Pa s, respectively, cumulative gas
23
24
25
26
27
28
6

8
production is underestimated by 11% in comparison to the base case after 10000 days. If we

define Z as unity, cumulative gas production will decrease by 6.98%. In summary, the gas

deviation factor and gas viscosity play crucial roles in predictions of gas production. Ignorance of
an
dM
29
30 9 real gas effects may result in dramatic underestimates of well production.
31
32 30
Gas rate (Base case) Gas production (Base case)
1.2

33 Gas rate Z=1


Gas rate Z=1, µg=2×10-5Pa.s
Gas production Z=1
Gas production Z=1, µg=2×10-5Pa.s

34 25 1.0
Cumulative gas production, 108m3

35
Gas production rate, 104m3/day

36 20 0.8

37 3.0

38 15
2.5
0.6
Gas production rate, 10 4m3/day

39 2.0
pte

40 10 1.5
0.4

41 1.0

42 5 0.5 0.2

43 0.0
1000 2000 3000
Time, day
4000 5000 6000

44 0
0 2000 4000 6000 8000 10000
0.0

45 10 Time, day

46 11 Fig. 18 Impact of real gas effects on gas production rate and evolution of cumulative gas production
47
48
12 7. Conclusion
ce

49
50 13 In this paper, a unified model of apparent permeability was established. A comprehensive
51
14 mathematical model, which included the model of apparent permeability, was derived to simulate
52
53
15 the shale gas production process. Validation of the mathematical model and numerical
Ac

54
55
16 discretization was conducted by matching production data from the Barnett Shale. Finally, a
56
57 17 sensitive analysis was performed to quantitatively estimate the impacts of different influencing
58
59
20
60
Page 21 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 factors on gas production. According to this study, the following conclusions can be deduced:
4

pt
5 2 (1) The gas transport mechanisms play a key role in the gas transport process. A linear
6
7 3 superposition relationship of slip flow and Knudsen diffusion may lead to a notable increase in gas
8
9 4 production forecasts. When the pore radius is greater than 5 nm, surface diffusion can be ignored.

cri
10
11 5 Knudsen diffusion may result in a negligible impact on gas production in the proposed model.
12
13 6 (2) The impact of the adsorbed layer increases as the pore size decreases. The impact of the
14
15 7 adsorbed layer can be neglected when the pore radius is greater than 15 nm. Geomechanical
16
8 effects in shale gas reservoirs have slight impacts on cumulative gas production.

us
17
18
9 (3) The Langmuir parameters significantly affect the evolution of cumulative gas production.
19
20
10 Cumulative gas production increases as the Langmuir volume VL and Langmuir pressure pL
21
22 11 increase. The desorption process contributes more to gas production in the late period of gas
23
24
25
26
27
28
12

13

14
production.
an
(4) An artificial increase in the surface diffusion coefficient Ds, which may be inconsistent

with real gas reservoir conditions, can result in significant overestimates in production forecasts,
dM
29
30 15 and the impact of surface diffusion can be neglected when the surface diffusion coefficient is less
31
32 16 than the order of 10-7 m2/s.
33
34 17 Acknowledgments
35
36 18 This work was supported by the National Natural Science Foundation of China (Key Program)
37
38 19 (Grant No. 51534006), National Natural Science Foundation of China (Grant No. 51704247 and
39
20 51674211) and the Scientific Research Starting Project of SWPU (No. 2015QHZ003). The authors
pte

40
41
21 would also like to appreciate the reviewers and editors whose critical comments were very helpful
42
43
22 in preparing this article.
44
45
46 23 Nomenclature
47 24 Aij area of the interface between block i and block j, m2
48
ce

49 25 C molarity of real gas, mol/m3


50
51 26 cµ molarity of adsorbed gas, mol/m3
52
53 27 Cg gas compressible coefficient, 1/Pa
Ac

54
55 28 dm molecular diameter of gas, m
56
57 29 Di distances from the nodes of block i to the interface shared by blocks i and j, m
58
59
21
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 22 of 28

1
2
3 1 Dj distances from the nodes of block j to the interface shared by blocks i and j, m
4

pt
5 2 Dk Knudsen diffusion coefficient, m2/s
6
7 3 Ds surface diffusion coefficient, m2/s
8
9 4 F total mass flux of gas for real gas, kg/(m2 s)

cri
10
11 5 Fa total mass flux of adsorbed gas for real gas, kg/(m2 s)
12
13 6 Fb total mass flux of bulk gas for real gas, kg/(m2 s)
14
15 7 Fk mass flux by Knudsen diffusion for real gas, kg/(m2 s)
16
8 Fs mass flux of gas by surface diffusion for real gas, kg/(m2 s)

us
17
18
9 Fv mass flux by slip flow for real gas, kg/(m2 s)
19
20
10 Ga amount of adsorbed gas, m3/kg
21
22 11 i block i
23
24
25
26
27
28
12

13

14
j

Jik
r
J
block adjacent to block i
an r
element of row i and column k in sparse Jacobian matrix J

sparse Jacobian matrix


dM
29
30
31 15 k permeability of shale matrix, m2
32
16 k0 permeability at zero effective stress, m2
33
34
17 kapp apparent permeability of shale gas reservoirs, m2
35
36
18 kv permeability of slip flow for real gas, m2
37
38
19 k k′ gas flow conductance of Knudsen diffusion, m2
39
pte

40
41 20 k s′ gas flow conductance of surface diffusion, m2
42
43 21 kv′ gas flow conductance of slip flow, m2
44
45 22 Kn Knudsen number for real gas in nanopores
46
47 23 m associated with geometry of the conductive pore space
48
ce

49 24 M gas molecular mass, kg/mol


50
51 25 Mi total mass density of the gas in block i, kg/m3
52
53 26 n current time level
Ac

54
55 27 n+1 next time level
56
57 28 p pressure of shale gas reservoir, Pa
58
59
22
60
Page 23 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 p1 effective stress when pores are closed completely, Pa
4

pt
5 2 pc confining pressure, Pa
6
7 3 pi pressure of block i, Pa
8
9 4 pj pressure of block j, Pa

cri
10
11 5 pL Langmuir pressure, Pa
12
13 6 pwf bottom hole pressure, Pa
14
15 7 qg gas production or injection rate, m3/s
16
8 Q gas production/injection rate, kg/s

us
17
18
9 r pore radius of shale matrix, m
19
20
10 r0 initial pore radius of shale matrix, m
21
22 11 rw well radius, m
23
24
25
26
27
28
12

13

14
reff

reff_stress

R
an
effective pore radius of shale matrix, m

effective pore radius of shale matrix considering geomechanicals effects, m

general gas constant, J/(K∙mol)


dM
29
30 15 Sg initial gas saturation
31
32 16 t elapsed time, s
33
34 17 Tij transmissibility of the interface between block i and block j, kg/(Pa∙s)
35
36 18 tol calculating tolerance, Pa
37
38 19 T reservoir temperature, K
39
Langmuir volume in reservoir conditions, m3/kg
pte

40 20 VL
41
42 21 Vstd molar volume in standard conditions (273.15 K and 1.01325 MPa), m3/mol
43
44 22 Z gas deviation factor
45
46 23 ∆t time step, s
47
48 24 ∆Vi volume of block i, m3
ce

49
50 25 α effective stress coefficient
51
52 26 φ porosity of shale matrix
53
27 φ0 initial porosity of shale matrix
Ac

54
55
28 φeff effective porosity of shale matrix
56
57
29 ρr shale density, m3/kg
58
59
23
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 24 of 28

1
2
3 1 σeff effective stress in shale matrix, Pa
4

pt
5 2 σeff_0 initial effective stress in shale matrix, Pa
6
7 3 εv weighting factor of slip flow for real gas
8
9 4 εk weighting factor of Knudsen diffusion for real gas

cri
10
11 5 θ gas coverage of real gas
12
13 6 ߬ tortuosity of shale matrix
14
15 7 λ mean free path of molecules for real gas, m
16
8 µg gas viscosity, Pa∙s

us
17
18
19 9 ρg gas density, kg/m3
20
21 10 References
22
23
24
25
26
27
28
11

12

13

14
an
Adesida A., Akkutlu I., Resasco D., Rai C., 2011. Kerogen pore size distribution of Barnett shale

using DFT analysis and Monte Carlo simulations. Paper SPE 147397, SPE Annual Technical

Conference and Exhibition, Denver, Colorado, 30 October–2 November.

Akkutlu I.Y., Fathi E., 2012. Multiscale gas transport in shales with local kerogen heterogeneities.
dM
29
30
15 SPE Journal, 17:1002–1011.
31
32
16 Al-Ahmadi H.A., Wattenbarger R.A., 2011. Triple-porosity models: one further step towards
33
34
17 capturing fractured reservoirs heterogeneity. In: SPE/DGS Saudi Arabia Section Technical
35
36 18 Symposium and Exhibition, Society of Petroleum Engineers.
37
38 19 Azom P.N., Javadpour F., 2012. Dual-continuum modeling of shale and tight gas reservoirs. In:
39
pte

40 20 SPE Annual Technical Conference and Exhibition, Society of Petroleum Engineers.


41
42 21 Beskok A., Karniadakis G.E., 1999. Report: a model for flows in channels, pipes, and ducts at
43
44 22 micro and nano scales. Microscale Thermophysical Engineering, 3:43–77.
45
46 23 Cai J, Hu X, Xiao B, Zhou Y, Wei W. Recent developments on fractal-based approaches to
47
48
ce

24 nanofluids and nanoparticle aggregation. INT J HEAT MASS TRAN. 2017;105:623-37.


49
50 25 Cao P., Liu J., Leong Y.-K., 2016. A fully coupled multiscale shale deformation-gas transport
51
52 26 model for the evaluation of shale gas extraction. Fuel, 178:103–117.
53
27 Civan F., 2010. Effective correlation of apparent gas permeability in tight porous media. Transport
Ac

54
55
56 28 in Porous Media, 82:375–384.
57
58 29 Civan F., Devegowda D., Sigal R.F., 2013. Critical evaluation and improvement of methods for
59
24
60
Page 25 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 determination of matrix permeability of shale. In: SPE Annual Technical Conference and
4

pt
5 2 Exhibition, Society of Petroleum Engineers.
6
7 3 Civan F., Rai C.S., Sondergeld C.H., 2011. Shale-gas permeability and diffusivity inferred by
8
9 4 improved formulation of relevant retention and transport mechanisms. Transport in Porous

cri
10
11 5 Media, 86:925–944.
12
13 6 Dranchuk P.M., Purvis R., Robinson D., 1973. Computer calculation of natural gas compressibility
14
15 7 factors using the Standing and Katz correlation. In: Annual Technical Meeting, Petroleum
16
8 Society of Canada.

us
17
18
9 Florence F.A., Rushing J., Newsham K.E., Blasingame T.A., 2007. Improved permeability
19
20
10 prediction relations for low permeability sands. In: Rocky Mountain Oil & Gas Technology
21
22 11 Symposium, Society of Petroleum Engineers.
23
24
25
26
27
28
12

13

14
an
Ghanizadeh A., Aquino S., Clarkson C.R., Haeri-Ardakani O., Sanei H., 2014. Petrophysical and

geomechanical characteristics of Canadian tight oil and liquid-rich gas reservoirs. In:

SPE/CSUR Unconventional Resources Conference–Canada, Society of Petroleum Engineers.


dM
29
30 15 Javadpour F., 2009. Nanopores and apparent permeability of gas flow in mudrocks (shales and
31
32 16 siltstone). Journal of Canadian Petroleum Technology, 48:16–21.
33
34 17 Javadpour F., Fisher D., Unsworth M., 2007. Nanoscale gas flow in shale gas sediments. Journal
35
36 18 of Canadian Petroleum Technology, 46:55–61.
37
38 19 Jenkins C.D., Boyer C.M., 2008. Coalbed and shale-gas reservoirs. Journal of Petroleum
39
pte

40 20 Technology, 60:92–99.
41
42 21 Lee A.L., Gonzalez M.H., Eakin B.E., 1966. The viscosity of natural gases. Journal of Petroleum
43
44 22 Technology, 18:997–1000.
45
46 23 Ma J., Sanchez J.P., Wu K., Couples G.D., Jiang Z., 2014. A pore network model for simulating
47
48 24 non-ideal gas flow in micro- and nano-porous materials. Fuel, 116:498–508.
ce

49
50 25 Negara A., Elgassier M., Saad B., 2016. Numerical simulation of natural gas flow in shale
51
26 reservoirs with thermodynamic equation of state: A comparative study. In: SPE Europec, 78th
52
53
27 EAGE Conference and Exhibition, Society of Petroleum Engineers.
Ac

54
55
28 Negara A., Sun S., 2016. Effects of multiple transport mechanisms on shale gas production
56
57 29 behavior. In: SPE Kingdom of Saudi Arabia Annual Technical Symposium and Exhibition,
58
59
25
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 26 of 28

1
2
3 1 Society of Petroleum Engineers.
4

pt
5 2 Olorode O.M., 2011. Numerical modeling of fractured shale-gas and tight-gas reservoirs using
6
7 3 unstructured grids. In, Texas A&M University.
8
9 4 Sigal R.F., 2013. The effects of gas adsorption on storage and transport of methane in organic

cri
10
11 5 shales. In: SPWLA 54th Annual Logging Symposium, Society of Petrophysicists and
12
13 6 Well-Log Analysts.
14
15 7 Singh H., Javadpour F., Ettehadtavakkol A., Darabi H., 2014. Nonempirical apparent permeability
16
8 of shale. SPE Reservoir Evaluation & Engineering, 17:414–424.

us
17
18
9 Song W., Yao J., Li Y., Sun H., Zhang L., Yang Y., Zhao J., Sui H., 2016. Apparent gas
19
20
10 permeability in an organic-rich shale reservoir. Fuel, 181:973–984.
21
22 11 Sun H., Yao J., Gao S.-H., Fan D.-Y., Wang C.-C., Sun Z.-X., 2013. Numerical study of CO2
23
24
25
26
27
28
12

13

14
of Greenhouse Gas Control, 19:406–419.an
enhanced natural gas recovery and sequestration in shale gas reservoirs. International Journal

Wang J., Liu H., Wang L., Zhang H., Luo H., Gao Y., 2015a. Apparent permeability for gas
dM
29
30 15 transport in nanopores of organic shale reservoirs including multiple effects. International
31
32 16 Journal of Coal Geology, 152:50–62.
33
34 17 Wang J., Luo H., Liu H., Ji Y., Cao F., Li Z., Sepehrnoori K., 2015b. Variations of gas flow
35
36 18 regimes and petro-physical properties during gas production considering volume consumed
37
38 19 by adsorbed gas and stress dependence effect in shale gas reservoirs. In: SPE Annual
39
pte

40 20 Technical Conference and Exhibition, Society of Petroleum Engineers.


41
42 21 Wang M., Lan X., Li Z., 2008. Analyses of gas flows in micro- and nanochannels. International
43
44 22 Journal of Heat and Mass Transfer, 51:3630–3641.
45
46 23 Wasaki A., Akkutlu I.Y., 2015. Permeability of organic-rich shale. SPE Journal, 20:1384–1396.
47
48 24 Wu, K., X. Li, C. Wang, Z. Chen and W. Yu, 2015a. A model for gas transport in microfractures of
ce

49
50 25 shale and tight gas reservoirs. AIChE Journal. 61: 2079-2088.
51
26 Wu, K., Z. Chen and X. Li, 2015b. Real gas transport through nanopores of varying cross-section
52
53
27 type and shape in shale gas reservoirs. Chemical Engineering Journal. 281: 813-825.
Ac

54
55
28 Wu, K., X. Li, C. Wang, W. Yu and Z. Chen, 2015c. Model for Surface Diffusion of Adsorbed Gas
56
57 29 in Nanopores of Shale Gas Reservoirs. Industrial & Engineering Chemistry Research. 54:
58
59
26
60
Page 27 of 28 AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1

1
2
3 1 3225-3236.
4

pt
5 2 Wu, K., Z. Chen, X. Li, C. Guo and M. Wei, 2016. A model for multiple transport mechanisms
6
7 3 through nanopores of shale gas reservoirs with real gas effect–adsorption-mechanic coupling.
8
9 4 International Journal of Heat and Mass Transfer. 93: 408-426.

cri
10
11 5 Xiong X., Devegowda D., Villazon M., German G., Sigal R.F., Civan F., 2012. A fully-coupled
12
13 6 free and adsorptive phase transport model for shale gas reservoirs including non-Darcy flow
14
15 7 effects. In: SPE Annual Technical Conference and Exhibition, Society of Petroleum
16
8 Engineers.

us
17
18
9 Yao, J., Sun, H., Huang, Z., Zhang, L., Zeng, Q., Sui, H., Fan, D., 2013a. Key mechanical
19
20
10 problems in the developmentof shale gas reservoirs. Sci Sin-Phys Mech Astron, 43:1527–
21
22 11 1547. doi: 10.1360/132013-97 (in Chinese)
23
24
25
26
27
28
12

13

14
an
Yao J., Sun H, Fan D., Wang C., Sun Z., 2013b, Numerical simulation of gas transport

mechanisms in tight shale gas reservoirs. Petroleum Science, 10(4): 528-537.

Yu W., Sepehrnoori K., 2014. Simulation of gas desorption and geomechanics effects for
dM
29
30 15 unconventional gas reservoirs. Fuel, 116:455–464.
31
32 16 Zhao J., Li Z., Hu Y., Ren L., Tao Z., 2016a. The impacts of microcosmic flow in nanoscale shale
33
34 17 matrix pores on the gas production of a hydraulically fractured shale-gas well. Journal of
35
36 18 Natural Gas Science and Engineering, 29:431–439.
37
38 19 Zhao Y.L., Zhang L.H., Luo J.X., Zhang B.N., 2014. Performance of fractured horizontal well
39
pte

40 20 with stimulated reservoir volume in unconventional gas reservoir. Journal of Hydrology,


41
42 21 512:447–456.
43
44 22 Zhao Y.L., Zhang L.H., Zhao J.Z., Luo J.X., Zhang B.N., 2013. “Triple porosity” modeling of
45
46 23 transient well test and rate decline analysis for multi-fractured horizontal well in shale gas
47
48 24 reservoirs. Journal of Petroleum Science and Engineering, 110:253–262.
ce

49
50 25 Zhao, Y.L., Zhang, L.H., Liu, Y.H., Hu, S.Y., Liu, Q.G., 2015. Transient pressure analysis of
51
26 fractured well in bi-zonal gas reservoirs. Journal of Hydrology, 524, 89-99.
52
53
27 Zhao Y., Zhang L., Xiong Y., Zhou Y., Liu Q., Chen D., 2016b. Pressure response and production
Ac

54
55
28 performance for multi-fractured horizontal wells with complex seepage mechanism in
56
57 29 box-shaped shale gas reservoir. Journal of Natural Gas Science and Engineering, 32:66–80.
58
59
27
60
AUTHOR SUBMITTED MANUSCRIPT - JGE-101461.R1 Page 28 of 28

1
2
3 1
4

pt
5 2
6
7
8
9

cri
10
11
12
13
14
15
16

us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
39
pte

40
41
42
43
44
45
46
47
48
ce

49
50
51
52
53
Ac

54
55
56
57
58
59
28
60

You might also like