Chemical Engineering Science 172 (2017) 230–238

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: www.elsevier.com/locate/ces

A complex model for the permeability and porosity of porous media

Xiao-Hua Tan a,⇑, Li Jiang a, Xiao-Ping Li a,⇑, Yue-Yang Li b, Kai Zhang b
a
State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation, Southwest Petroleum University, Chengdu 610500, China
b
CNPC Southwest Oil and Gas Field Exploration and Development Research Institute, 610041, China

h i g h l i g h t s

A complex model for the permeability and porosity of porous media is developed.

The characters of pore microstructure and fluid in porous media are considered.

This model reveals the all kinds of fluids flow in variety of porous media.

The predictions of the model show good agreement with available experimental data.

The proposed model can be used to characterize the flow in porous media.

a r t i c l e i n f o a b s t r a c t

Article history: Transport phenomenon in porous media is essential in various scientific and engineering fields. The per-
Received 17 February 2017 meability and porosity of porous media serve important functions in transport phenomena. In this study,
Received in revised form 19 June 2017 a complex model that considers the characters of pore microstructure and fluid in porous media is devel-
Accepted 21 June 2017
oped for the permeability and porosity of porous media. The flow rate expressions of this model are com-
Available online 22 June 2017
pared with those of previous models. As the proposed models relate the properties of fluids to the
structural parameters of porous media, the various expressions of the developed model can reveal the
Keywords:
phenomena of all types of fluid flow in a variety of porous media. The predicted permeabilities and
Complex model
Permeability
porosities show good agreement with the available experimental data and illustrate that the proposed
Porosity model can be used to characterize flow in porous media.
Porous media Ó 2017 Elsevier Ltd. All rights reserved.
Probability density function
Fractal

1. Introduction transport procession and obtain the transport parameters, such

Transport phenomena in porous media have steadily received
considerable attention in many fields, such as hydraulics
(Escobar et al., 2011; Pia and Sanna, 2014a; Pia et al., 2014), chem-
ical (Cai et al., 2012), petroleum engineering (Dong et al., 2005; Xu
et al., 2017). The subject of transport phenomena includes three
closely related topics: fluid dynamics (Xiong et al., 2017), heat
transfer (Miao et al., 2016), and mass transfer (Yang et al., 2017).
Fluid dynamics involves the transport of momentum, heat transfer
deals with the transport of energy, and mass transfer is concerned
with the transport of mass of various chemical species (Bird, 2002).
Transport phenomena in porous media are still not well-
understood. Because of the disordered and extremely complicated
microstructures of porous media, it is difficult to describe the
⇑ Corresponding authors.
E-mail addresses: xiaohua-tan@163.com (X.-H. Tan), ncixphm@126.com
(X.-P. Li).
as permeability (Guarracino et al., 2014; Liu et al., 2016; Xu
et al., 2016), thermal conductivity (Huai et al., 2007; Pia and
Sanna, 2014b), mechanical behavior (Gou and Schwartz, 2013). In
this case, fractal theory is introduced to describe the microstruc-
tures of porous media and obtain the permeability and porosity
of porous media. Because fractures in nature are random and dis-
order in spaces and lengths, and real porous media have been
shown to have the statistically self-similar and fractal characteris-
tic. Tremendous attention has also been directed toward its meth-
ods, theories, and progresses in these areas (Xu et al., 2017). The
permeability and porosity of porous media serve important func-
tions in transport phenomena, which are usually studied via exper-
iments, numerical simulations, and analyses. Capillary bundle
models have been used to predict the permeability of porous
media with parameters obtained from experimental data, avoiding
the tedious calculations of numerical simulations (Cao et al., 2016;
Yang et al., 2014a).

http://dx.doi.org/10.1016/j.ces.2017.06.041
0009-2509/Ó 2017 Elsevier Ltd. All rights reserved.
 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238 231

Purcell (1949) assumed that porous media are composed of pressure Sb1 , displacement pressure pd, and pore shape factor F. He
numerous parallel cylindrical capillaries with equal length but dif- represented the equation for the permeability of porous media as
ferent radii, and that total flow rate through porous media is equiv-  2:14
alent to the sum of the contributions made by each capillary. All Sb1
K¼A ; ð6Þ
these assumptions laid the foundation of the capillary bundle pd
model. Purcell took advantage of Poiseuille’s equation
where A could not be determined because of the limitations of tech-
pr4 Dp nology at that time.
q¼ ; ð1Þ
8lL Twenty years later, Thomeer (1983) found that parameter A is
related to pore shape factor and took advantage of experimental
where q is the flow rate of a single capillary, r is the radius of a cap- data to modify the power-law index from 2.14 to 2. The expression
illary, Dp is the pressure gradient along a tortuous capillary, l is the is given as follows:
fluid viscosity, and L is the length of a capillary. The capillary pres-
 2
sure equation is Sb1
K ¼ 3:8068F 1:3334 : ð7Þ
pd
2r cos h
pc ¼ ; ð2Þ
r Swanson (1981) also made an enormous contribution to the
permeability expression of porous media. On the basis of the
where pc is the capillary pressure of a single capillary, r is the inter-
experimental data of 24 clean sandstone samples from 21 forma-
facial tension, and h is the angle of contact.
tions, Swanson first proposed the correlation as
Then, according to the basic assumption of the capillary bundle
model, Purcell proposed the flow rate expression of porous media  2:109
Sb
shown as follows: K ¼ 431 : ð8Þ
pc
ðr cos hÞ2 Dp XN
Vi Afterwards, according to 32 carbonate samples representing 13 for-
Q¼ ; ð3Þ
2lL 2
p2
i¼1 ci mations, Swanson gave the second expression as

where Q is the flow rate through porous media, Vi is the pore vol-
 1:901
Sb
ume of each capillary, and N is the number of capillaries in porous K ¼ 290 : ð9Þ
pc
media. The permeability expression of porous media is
In the end, the overall relationship based on all the experimen-
F/ðr cos hÞ2 XN
Si tal data was presented as
K¼ 4 2
; ð4Þ
2  10 p
i¼1 ci  2:005
Sb
K ¼ 355 : ð10Þ
where K is the permeability of porous media, F is the pore shape fac- pc
tor, / is the porosity of porous media, and Si is the saturation of each
Wells and Amaefule (1985) proposed a new relationship
capillary.
between the parameters and the permeability for tight gas sands.
Therefore, Purcell established the basic permeability form of the
This relationship led to a more accurate predicted permeability
capillary bundle model, which contains the pore geometrical fac-
below 10 microdarcies:
tor, the porosity of porous media, the saturation, and the capillary
pressure. His study was continued by many researchers, as shown  1:56
Sb
in Table 1. The parallel expression was given as K ¼ 30:5 : ð11Þ
pc
 B
Sb Buiting and Clerke (2013) considered the tortuosity and fractal
K¼A ; ð5Þ
pc character of capillary bundles to develop their capillary bundle
model. The model can be expressed similar to the formulation of
where Sb and pc are the respective saturation and capillary pressure
the Swanson permeability expression with new parameters.
of porous media. Some researchers (Buiting and Clerke, 2013;
Thomeer, 1983, 1960) used saturation at infinite capillary pressure pffiffi
 1:87
Sb
Sb1 and displacement pressure pd instead of Sb and pc, but the K ¼ 4316 e5:67 F
; ð12Þ
pc
expression Sb1 =pd presented the similar relationship between
injected fluid saturation and capillary pressure. The parameters A where F presents the variability of pore throats in porous media. A
and B, which matched experimental data, presented the pore struc- large F value means high pore throat variability within the pore sys-
ture of porous media. tem, resulting in a low permeability of porous media.
Thomeer (1960) showed the capillary pressure curves based on Kozeny (1927) derived the expression of permeability of porous
experiments and proposed the relationship between capillary media by assuming that the porous media comprise a bundle of
pressure pc and saturation Sb, using saturation at infinite capillary capillaries with the same diameter. Carman (1937) modified

Table 1
The permeability expressions of Purcell-based models.

Model name Value of a Value of b Dataset description Reference

Thomeer 1 – 2.14 – Thomeer (1960)
1.3334
Thomeer 2 3.8068F 2 165 siliciclastic and 114 carbonate samples in 54 fields Thomeer (1983)
Swanson 1 431 2.109 24 sandstone samples in 21 formations Swanson (1981)
Swanson 2 290 1.901 32 carbonate samples in 13 formations Swanson (1981)
Swanson 3 355 2.005 24 sandstone and 32 carbonate samples in 21 formations Swanson (1981)
Wells and Amaefule 30.5 1.56 35 tight gas sandstone samples in 2 fields Wells and Amaefule (1985)
pffiffi
Buiting and Clerke 4316 e5:67 F 1.87 More than 500 samples Buiting and Clerke (2013)
232 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238

Kozeny’s model by multiplying it by the tortuosity s and then pro-
posed the Kozeny–Carman equation:
c0 r 2 /
K¼ ; ð13Þ
s2
where c0 is a coefficient called Kozeny’s constant, and s is the tor-
tuosity of porous media.
Some studies with similar results have been made based on the
Kozeny–Carman equation (Dastidar et al., 2007; Kolodzie, 1980;
Pittman, 1992), as shown as Table 2.
K ¼ R/c
Eq. (14) reveals the relationship between the permeability and
porosity of porous media. R is the function of the pore radius of por-
ous media, and c is the power-law index.
Winland’s (Gunter et al., 2014) correlation was refined using
approximately 300 samples from multiple formations spanning
different geological ages, lithologies, and diagenetic fabrics. Win-
land described the relationship among permeability, porosity,
and r 35 ; which is the effective pore throat radius at 35% saturation.
K ¼ 0:2880r 1:7007
35 /1:4694 :
Kolodzie (1980) developed the model to estimate the perme-
ability and porosity from well logs and calculate pore throat size.
The relationship, as refined by Kolodzie, was suitable for the por-
ous media with a minimum pore throat size of 0.5 mm.
K ¼ 0:1954r 1:8028
35 /1:6284 :
Pittman (1992) used the injection capillary pressure curves for
sandstone to investigate the relationship of the permeability and
porosity of various parameters. Instead of r 35 , Pittman used r apex ,
the radius associated with the critical length, to describe the rela-
tionship as
0:2084
K ¼ 4:6042r 2:1053
apex / : ð17Þ
On the basis of the multiple linear regression analysis method,
Dastidar et al. (2007) established an empirical correlation among
the permeability, porosity, and rWGM ; the weighted geometric mean
of the pore throat radius, for tight sand reservoirs:
K ¼ 0:08127r WGM
1:64
/3:06 ;
as well as developed a correlation among permeability, porosity,
and r 35 :
1:35 3:06
K ¼ 0:04832r 35 / :
Dastidar et al. (2007) indicated that the predicted permeability
using the r WGM correlation, Eq. (18), is closer to the actual perme-
ability than the permeability determined using r35 correlation,
Eq. (19).
The capillary bundle model can describe both simple phase flow
and various flow situations in porous media. Basing on the capil-
lary bundle model that considers adsorptive surface forces and liq-
uid films at low potentials, Tuller et al. (1999) researched the
phenomenon of adsorption and capillary condensation in porous
media. New expressions they established for capillary behavior
further assisted the development of conventional expressions.
Tuller and Or (2002) developed this model to studied the unsatu-
rated flow in porous media. A parameter estimation scheme was
developed to calculate liquid saturation and predict sample-scale
unsaturated hydraulic conductivity. Effects of the non-
equilibrium state between matrix and structural pore domains
on the hydraulic conductivity were discussed. Dong et al. (2005)
analyzed the immiscible displacement processes in porous media
by using a developed interacting capillary bundle model. The evo-
lutions of the prediction of saturation profiles were consistent with
both numerical simulation and experimental results. Peters and
ð14Þ
Durner (2008) studied the unsaturated flow by presenting a new
capillary bundle model combined with a simple film flow function.
The new model achieved the best performance using a modified
version of Akaike’s information criterion. To predict the hydraulic
conductivity of unsaturated porous media, Lebeau and Konrad
(2010) derived a new capillary bundle model for analyzing the
hydraulic conductivity of unsaturated porous media. The model
considered both capillary and thin film flow processes in a capillary
and well validated the hydraulic conductivity data they collected.
Yu and Cheng (2002) introduced the fractal distribution of capillar-
ð15Þ ies into the capillary bundle model, and their following research
described various types of flow phenomena in porous media. The
presented analyses were helpful in understanding the fundamental
mechanisms of flow in fractal porous media (Miao et al., 2015; Tan
et al., 2015b; Xu et al., 2013; Yang et al., 2014b).
Consequently, the previous studies greatly improved the model
ð16Þ and enabled it to describe the flow character under certain condi-
tions. However, these studies could not represent various flow phe-
nomena based on the same model. Having just one model to
describe the flow character under different conditions would be
more convenient and practical. Therefore, a complex model should
be established to describe all types of flow phenomena in porous
media. This study attempts to establish a complex model for the per-
meability and porosity of porous media. Hence, the basic physical
porous media model based on fractal theory will be introduced in
Section 2.1 and the mathematical models for the permeability and
porosity of porous media will be derived in Section 2.2. All the func-
tions used in the proposed mathematical models will be presented
in Section 2.3. Then, the expression of proposed model will be com-
pared with those of previous models in Section 3.1. A series of mea-
ð18Þ
sured data will be used to verify the reliability of proposed models in
Section 3.2. Finally, conclusions are given in Section 4.
2. Complex model
ð19Þ
In this section, a complex model is established. The physical
model is proposed first (Section 2.1). Afterwards, the mathematical
model composed of three functions (Section 2.2) is presented.
Finally, three functions are introduced and explained (Section 2.3).
2.1. Physical model
Porous media are assumed to be composed of a bundle of tortu-
ous capillaries with different diameters. The space among the cap-
illaries is a solid frame, as shown in Fig. 1.

Table 2
The permeability expressions of Kozeny-Carman-based models.

Model name Value of R Value of c Dataset description Reference

Winland 0:2880r 1:7007
35
1.4694 56 sandstone and 26 carbonate samples in 14 formations Kolodzie (1980)
Kolodzie 0:1954r 1:8028
35
1.6284 82 samples in 12 fields Kolodzie (1980)
Pittman 4:6042r 2:1053
apex
0.2084 202 samples in 16 formations Pittman (1992)
Dastidar et al. 1 0:08127r 1:64
WGM
3.06 150 samples in 12 formations Dastidar et al. (2007)
Dastidar et al. 2 0:04832r 1:35
35
3.06 150 samples in 12 formations Dastidar et al. (2007)
 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238 233

Z
Capillaries are assumed to be smooth, and so the wall thickness lLN kmax
K¼ qðkÞf ðkÞdk: ð24Þ
of the capillary is ignored. The porous media are saturated with a A Dp kmin
single-phase fluid. The flow in porous media is isothermal and fol-
lows Darcy’s law, which ignores the impact of gravity and capillary The pore volume of porous media can be expressed by fractal
forces. The permeability and porosity of porous media do not theory (Tan et al., 2014):
change with pressure. Z kmax Z kmax
Fp 2 FpN
Vp ¼ k Lt ðkÞdN ¼ Lt ðkÞf ðkÞk2 dk; ð25Þ
kmin 4 4 kmin
2.2. Mathematical model
where V p is the pore volume of porous media, Lt ðkÞ is the length
Given the definition of the model, the flow rate of porous media function of a capillary, and F p is the shape factor of a capillary.
is given by the sum of the flow rate of all capillaries and can be For a capillary with a circular cross section, F p equals p:
expressed as the integral of all flow rates from the minimum cap- The porosity of porous media can be presented by
illary diameter to the maximum one.
Vp
Z /¼ ; ð26Þ
kmax V
Q¼ qðkÞdN; ð20Þ
kmin where / is the porosity of porous media, V is the volume of porous
media and the characteristic volume, V ¼ L3 :
where Q is the flow rate of porous media, qðkÞ is the function of the When Eq. (25) is inserted into Eq. (26), the porosity expression
flow rate through a capillary, N is the total number of capillaries in of the complex model is given by
porous media, and kmax and kmin are the diameters of the maximum Z kmax
and minimum capillaries respectively. FpN
/¼ Lt ðkÞf ðkÞk2 dk: ð27Þ
Based on the statistical principles, dN can be given as 4V kmin

dN ¼ f ðkÞNdk; ð21Þ
2.3. Functions in the model
where f ðkÞ is the probability density function of capillaries in the
cross-sectional area of porous media. 2.3.1. Flow rate function of a capillary
When Eq. (21) is inserted into Eq. (20), the flow rate expression qðkÞ is the flow rate function of a single tortuous capillary with
of the complex model is expressed as diameter k; it describes the flow characters of different fluids in a
Z kmax capillary. qðkÞ includes the following major functions.
Q ¼N qðkÞf ðkÞdk: ð22Þ
kmin
2.3.1.1. Newtonian fluid function. Newtonian fluid is defined to
According to Darcy’s law, the permeability of porous media can exhibit a direct proportionality between stress s and strain rate c_
be expressed as in laminar flow. Its constitutive equation is
Q lL s_ ¼ lc_ ; ð28Þ
K¼ ; ð23Þ
ADp
where s_ is the shear stress, l is the fluid viscosity, and c_ is the shear
where K is the permeability of porous media, l is the viscosity of rate.
fluid in porous media, Dp is the pressure differential, L is the flow The flow rate for Newtonian fluid flowing in a capillary is given
distance, and A is the cross-sectional area of porous media. L and by (Bear, 1972)
A can also be considered as the characteristic length and character-
F K k 4 Dp
istic area respectively, and the relationship between the above two qðkÞ ¼ ; ð29Þ
128lLt
factors is A ¼ L2 .
When Eq. (22) is inserted into Eq. (23), the permeability expres- where F K is the permeability factor of a capillary. F K = p represents a
sion of the complex model is obtained as capillary with a circular cross section.

Fig. 1. The schematic of the complex model for the porous media.
234 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238

2.3.1.2. Power-law fluid function. Power-law fluid is one of the non- 2.3.3.2. Gamma distribution function. Gamma distribution is one of
Newtonian fluids. Its constitutive equation is the most common distributions whose probability density function
can be expressed as
s ¼ lc_ n ; ð30Þ
kkg 1 ek=hg
where n is the power-law index. f ðk; kg ; hg Þ ¼ ð39Þ
The flow rate for power-law fluid through a capillary is hkgg Cðkg Þ
expressed as (Aziz and Govier, 1972)
with the shape parameter kg , scale parameter hg , and Gamma func-
F K k3þ1=n Dp1=n tion Cðkg Þ.
qðkÞ ¼ : ð31Þ
ð3 þ 1=nÞ23þ2=n l1=n L1=n
t
2.3.3.3. Fractal distribution function. The probability density func-
tion of fractal distribution was proposed by Yu and Cheng (2002)
2.3.1.3. Bingham fluid function. The constitutive equation of Bing- as follows:
ham fluid is
kðDf þ1Þ ;
D
f ðk; Df Þ ¼ Df kmin
f
ð40Þ
s_ ¼ s_ 0 þ lc_ ; ð32Þ
where Df is the fractal dimension.
where s_ 0 is the yield stress.
For Bingham fluid, the flow rate in a capillary can be described as
  3. Results and discussion
F K k4 Dp 4 s_ 0 =k
qðkÞ ¼ 1 : ð33Þ
128lLt 3 Dp=Lt In this section, the validity of the proposed model is discussed
basing on the advantages of the model comparison (Section 3.1)
2.3.1.4. Herschel–Bulkley fluid function. The constitutive equation of and the case analysis (Section 3.2).
Herschel–Bulkley fluid is given by
3.1. Model comparisons
s_ ¼ s_ 0 þ lc_ n ; ð34Þ
and the flow rate in a capillary is expressed as The flow rate expressions are compared among the complex
  model and the previous models.
F K k3þ1=n Dp1=n n þ 4 þ 1=n s_ 0 =k Newtonian fluid function [Eq. (29)], tortuosity function [Eq.
qðkÞ ¼ 1  : ð35Þ
ð3 þ 1=nÞ23þ2=n l1=n L1=n
t
1 þ 3n þ 2n2 Dp=Lt (36)], and fractal distribution function [Eq. (40)] are chosen for
the flow rate function of a capillary qðkÞ; the length function of a
capillary Lt ðkÞ; and the probability density function of capillaries
2.3.2. Length function of a capillary
f ðkÞ; respectively. The total number of capillaries in porous media
The length function of a capillary, Lt ðkÞ; is used to describe the
is given if the fractal distribution function is chosen.
relationship between the actual length and the straight length of
 D
a capillary, including some functions as follows. kmax f
N¼ : ð41Þ
kmin
2.3.2.1. Tortuosity function. Tortuosity is defined as the ratio of
actual length and straight length. The relationship among the With Eqs. (29), (36), (40), and (41) inserted into the flow rate
above three factors is expression of the complex model, Eq. (22), and all capillaries in
porous media considered as circular, the flow rate expression is
Lt ¼ sL; ð36Þ
pDf k4max Dp
where s is the tortuosity of a capillary. s equals 1 if the capillary is Q¼ : ð42Þ
128lsLð4  Df Þ
straight.
Eq. (42) is the same as the flow rate expression of porous media pro-
2.3.2.2. Tortuosity fractal dimension function. Based on the tortuos- posed by Tan et al. (2015a).
ity fractal theory proposed by Yu and Cheng (2002), the actual Newtonian fluid function [Eq. (29)], tortuosity fractal dimension
length of a capillary is a function of the tortuosity fractal dimen- function [Eq. (37)], and fractal distribution function [Eq. (40)] are
sion, the diameter, and the straight length of a capillary. chosen for qðkÞ; Lt ðkÞ; and f ðkÞ; respectively. With Eqs. (29), (37),
Lt ¼ k1Dt LDt ; ð37Þ (40), and (41) inserted into Eq. (22) and all capillaries in porous
media considered as circular, the flow rate expression can be
where Dt is the tortuosity fractal dimension of a capillary. expressed as

2.3.3. Probability density function of capillaries pDf k3þD

max Dp
f
Q¼ : ð43Þ
The probability density function of capillaries f ðkÞ presents the 128lð3 þ Dt  Df ÞLDt
distribution of capillaries in the cross-sectional area, including the
following major functions. Eq. (43) is identical to the flow rate expression of porous media
presented by Yu and Cheng (2002).
2.3.3.1. Normal distribution function. In probability theory, the nor- Power-law fluid function [Eq. (31)], tortuosity fractal dimension
mal (or Gaussian) distribution is a common continuous probability function [Eq. (37)], and fractal distribution function [Eq. (40)] are
distribution. The probability density function of normal distribu- selected for qðkÞ; Lt ðkÞ; and f ðkÞ; respectively. With Eqs. (31), (37),
tion is (40), and (41) inserted into Eq. (22) and all capillaries in porous
ðkln Þ
2 media considered as circular, the flow rate expression can be given
1 
f ðk; ln ; rn Þ ¼ pffiffiffiffiffiffi e 2rn ; ð38Þ
2
by
r n 2p
pDf k3þD t =n
Dp1=n
where ln is the expectation of distribution, and rn is the standard Q¼ max
: ð44Þ
deviation. ð3 þ 1=nÞð3  Df þ Dt =nÞ23þ2=n l1=n LDt =n
 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238 235

Eq. (44) is the same as the flow rate expression of porous media where S is the saturation of the injected fluid and is expressed as
derived by Yun (2014).
VL
Bingham fluid function [Eq. (33)], tortuosity fractal dimension S¼ ; ð54Þ
Vp
function [Eq. (37)], and fractal distribution function [Eq. (40)] are
selected for qðkÞ; Lt ðkÞ; and f ðkÞ; respectively. With Eqs. (33), and r is the radius of a capillary and given by
(37), (40), and (41) inserted into Eq. (22) and all capillaries in por-
k
ous media considered as circular, the flow rate expression can be r¼ : ð55Þ
2
given by
With Eqs. (54) and (55) inserted into Eq. (53), Eq. (53) can be rewrit-
pDf k3þD
max Dp
f
32ps_ 0 Df kDmax
f
ten as
Q¼  : ð45Þ
128lð3 þ Dt  Df ÞL Dt 3 lð3  D fÞ
dS 2:3026k dðV L Þ
¼ : ð56Þ
Eq. (45) is consistent with the flow rate expression of porous d logðrÞ Vp dðkÞ
media proposed by Cai (2014).
Herschel–Bulkley fluid function [Eq. (35)], tortuosity fractal Based on Eq. (25),
dimension function [Eq. (37)], and fractal distribution function dðV L Þ 1
[Eq. (40)] are selected for qðkÞ; Lt ðkÞ; and f ðkÞ; respectively. With
¼ F p Nk2 Lt ðkÞf ðkÞ; ð57Þ
dðkÞ 4
Eqs. (35), (37), (40), and (41) inserted into Eq. (22) and all capillar-
ies in porous media considered as circular, the flow rate expression Eq. (56) can be rewritten again, with Eq. (57) inserted into Eq. (56):
can be given by dS 0:57565sLNF p k3 
ðkln Þ
2

¼ pffiffiffiffiffiffi e 2rn : ð58Þ

2

Q¼
pDf k3þD
max
t =n Dp1=n
d logðrÞ Vp rn 2p
l
ð3þ1=nÞð3Df þDt =nÞ23þ2=n 1=n LDt =n
t þDt =n Dp1=n1
ð46Þ The distribution factor of porous media X K is defined as
ps_ 0 Df ðnþ4þ1=nÞLDt ð11=nÞ k3D
Q¼  ð3þ1=nÞð1þ3nþ2n max
2 Þð3D D þD =nÞ23þ2=n l1=n
f t t
0:57565sLNF p 0:57565sNF p
Xf ¼ ¼ : ð59Þ
Eq. (46) is identical with the flow rate expression of porous Vp /A
media presented by Yun et al. (2010).
Eq. (53) can be expressed as
3.2. Case analysis ðkln Þ 2
dS k3 
¼ X f pffiffiffiffiffiffi e 2rn : ð60Þ
2

d logðrÞ rn 2p
The permeability and porosity of porous media calculated by
the presented models are compared with those from the available Eq. (60) is applied to calculate the permeability factor of porous
experimental data (Ghadami et al., 2015), with the advantages of media, X K ; and the porosity factor of porous media, X p : First, Eq.
the pore distributions of six samples in the same formation taken (60) is used to predict the pore distributions, setting the average
into consideration. Newtonian fluid function [Eq. (29)], tortuosity value of the porosity / ¼ 0:2 and tortuosity s ¼ 1:5: Then, the dis-
function [Eq. (36)], and normal distribution function [Eq. (38)] tribution factor of porous media, X f ; the expectation of the distri-
are selected for qðkÞ; Lt ðkÞ; and f ðkÞ; respectively. With Eqs. (29), bution, ln ; and the standard deviation, rn ; in Eq. (60) are
(36), and (38) inserted into Eq. (24), the flow rate expression is adjusted repeatedly to ensure that the predictions match the
Z ðkln Þ 2
experimental data. The comparisons between the predictions and
kmax
NF K k4 
the experimental data of the pore distributions are revealed in
K¼ pffiffiffiffiffiffi e 2rn dk: ð47Þ
2

128sA kmin rn 2p Figs. 2–7. Next, the total number of capillaries in the characteristic
area N=A is calculated by X f from Eq. (59). Afterward, X K and X p are
The permeability factor of porous media, X K ; is defined as
obtained by N=A from Eqs. (48) and (51) respectively. Finally, the
NF K results of six samples in the same formation are shown in Table 3.
XK ¼ : ð48Þ
128sA The differences between the predictions and the experimental
Thus, Eq. (47) can be rewritten as data of the pore distributions of samples 1–3 (shown as Figs. 2–
Z kmax ðkln Þ 2 4) are smaller than those of samples 4 and 5 (shown as Figs. 5
k4 
and 6), and the difference of sample 6 (shown as Fig. 7) is the lar-
K ¼ XK pffiffiffiffiffiffi e 2rn dk: ð49Þ
2

kmin r n 2p gest. This phenomenon is attributed to fact that the pore distribu-
With Eqs. (36) and (38) inserted into Eq. (27) and all capillaries tions of samples 1–3 are unimodal functions, those of samples 4
in porous media considered as circular, the flow rate expression is and 5 are bimodal functions, and that of sample 6 is a multi-
Z modal function. However, the selected normal distribution func-
sLNF p ðkln Þ
kmax 2
k2 
tion is a unimodal function. Indeed, more distribution functions
/¼ pffiffiffiffiffiffi e 2rn dk: ð50Þ
2

4V kmin r n 2p must be applied to the description of the various pore

distributions.
The porosity factor of porous media X p is defined as
The predictions of permeability and porosity of the selected for-
sLNF p sNF p mation are derived from the calculated results in Table 3. The
Xp ¼ ¼ : ð51Þ
4V 4A ranges of the expectation of distribution and the standard devia-
Thus, Eq. (51) can be rewritten as tion for the six samples are ln = 0.3–2.4  106 m and rn = 0.2–
Z ðkf 1 Þ2
0.7  106 m respectively, and the average values of X K ¼ 0:001
kmax
k2 
2f 2 and X p ¼ 0:077 are used. To begin with, the extremes of ln and
/ ¼ Xp pffiffiffiffiffiffi e 2 dk: ð52Þ
kmin r n 2p rn are selected to predict the permeability and porosity respec-
The relationship between the saturation and the pore radius can tively. Points A (ln = 0.3, rn = 0.2), B (ln = 2.4, rn = 0.2), C (ln =
be expressed as 2.4, rn = 0.7), and D (ln = 0.3, rn = 0.7) are drawn in Fig. 8. After-
ward, set the value of rn as its extremes and change ln in the range
dS dðSÞ so that Lines AB (ln = 0.3–2.4, rn = 0.2) and CD (ln = 0.3–2.4, rn =
¼ 2:3026 r ; ð53Þ
d logðrÞ dðrÞ 0.7) can be drawn. To draw lines BC (ln = 2.4, rn = 0.2–0.7) and AD
236 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238
Fig. 2. The matching result of the pore size distribution of the porous media
(Sample 1).
Fig. 3. The matching result of the pore size distribution of the porous media
(Sample 2).
Fig. 4. The matching result of the pore size distribution of the porous media
(Sample 3).
(ln = 0.3, rn = 0.2–0.7), the value of ln is set as the extreme and rn
is changed in the range. Finally, the predictions of permeability and
porosity are presented through the surface consisting of four
points and four lines. The experimental data from the same forma-
tion are also marked in Fig. 8 as the solid points. Good agreement
between the permeabilities and the porosities calculated by the
presented model and those of the experimental data is illustrated
Fig. 5. The matching result of the pore size distribution of the porous media
(Sample 4).
Fig. 6. The matching result of the pore size distribution of the porous media
(Sample 5).
Fig. 7. The matching result of the pore size distribution of the porous media
(Sample 6).
in Fig. 8. Most experimental points are on the prediction surface.
The relationship between the permeabilities and the porosities of
the predictions also has the same tendency as that of the experi-
mental data. The variation and value of permeability both increase
with the increase in porosity.
Subsequently, the effect of the distribution parameters of
capillaries (the expectation of distribution ln and the standard
 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238 237

Table 3
The calculations of six samples.

ln ð106 mÞ rn ð106 mÞ X f ð106 mÞ N=A ð1012 mÞ X p ð1012 mÞ X K ð1012 mÞ

Sample 1 2.40 0.50 0.30 0.00686 0.00651 0.00014

Sample 2 0.30 0.20 13.00 0.29719 0.28229 0.00603
Sample 3 1.25 0.55 1.00 0.02286 0.02171 0.00046
Sample 4 1.00 0.50 1.00 0.02286 0.02171 0.00046
Sample 5 0.90 0.70 1.00 0.02286 0.02171 0.00046
Sample 6 0.30 0.35 5.00 0.11431 0.10857 0.00232
Total value 6.15 2.80 21.30 0.48694 0.46252 0.00987
Average value 1.03 0.47 3.55 0.08116 0.07709 0.00145

Fig. 8. Comparison of the permeabilities and porosities of the porous media Fig. 10. The permeability- porosity relationship versus the standard deviation rn .
between the predictions and experimental data.

porosity both increase with the increase in average capillary diam-

deviation rn ) on the permeability–porosity relationship of porous
media is also analyzed.
Fig. 9 illustrates how the expectation of distribution ln affects
the permeability–porosity relationship. Permeability K increases
with the increase in porosity /. The bigger the porosity, the bigger
the growth rate of permeability. Fig. 9 also shows that the perme-
ability–porosity curve moves toward the upper right when ln
increases. ln is changed from 0.3  106 m to 2.4  106 m, caus-
ing the permeability to vary from 0.034–0.86  1015 m2 to 34.56–
50.83  1015 m2, and that of porosity to vary from 0.0099–0.036
to 0.45–0.48. This phenomenon can be interpreted by the theory
of probability density function. The expectation of distribution ln
represents the average capillary diameter of porous media in the
normal distribution function. The average capillary diameter
becomes bigger when the ln increases. The permeability and
eter, but the increase rate of permeability is bigger than that of
porosity.
The influence of the standard deviation rn on the permeability–
porosity relationship is expounded in Fig. 10. A parallel trend of the
permeability–porosity relationship with the curves in Fig. 9 is
noted. The curve in Fig. 10 moves slowly toward the top direction
with the increase in rn : rn changes from 0.2 to 0.7, which makes
permeability K increase from 0.034–34.56  1015 m2 to 0.86–
50.83  1015 m2, and porosity / increase from 0.0099–0.45 to
0.036–0.48. This outcome is attributed to the distribution range
of the capillary diameter being represented by rn : The bigger the
rn ; the wider the distribution range of the capillary diameter. In
addition, the effect of rn on the permeability–porosity relationship
is much less than that of ln (see Fig. 10).

4. Conclusions

First, previous studies on the permeability correlations based on

Fig. 9. The permeability- porosity relationship versus the expectation of the
distribution ln .
the model are reviewed. Second, the physical model and mathe-
matical model of a complex model are proposed with the assump-
tion that porous media consist of a bundle of capillaries. All the
functions comprising the proposed model are explained in detail.
The flow rate expressions of this model are likewise compared
with those of previous models. As the proposed models relate
the properties of fluids to the structural parameters of porous
media, various expressions of this model can reveal the phenom-
ena that all types of fluids flow in various porous media. Addition-
ally, in the case analysis, the permeability–porosity relationships
calculated by the proposed model are compared with available
experimental data. Good agreement between the predicted perme-
ability–porosity relationships and the experimental data is
obtained. This result verifies the validity of the complex model.
In the end, the proposed model for permeability and porosity can
238 X.-H. Tan et al. / Chemical Engineering Science 172 (2017) 230–238

reveal additional mechanisms that affect fluid transport, with the Miao, T., Cheng, S., Chen, A., Yu, B., 2016. Analysis of axial thermal conductivity of
dual-porosity fractal porous media with random fractures. Int. J. Heat Mass
characters of microstructure and the fluid flow in porous media
Transf. 102, 884–890.
considered. Miao, T., Yu, B., Duan, Y., Fang, Q., 2015. A fractal analysis of permeability for
fractured rocks. Int. J. Heat Mass Transf. 81, 75–80.
Acknowledgements Peters, A., Durner, W., 2008. A simple model for describing hydraulic conductivity in
unsaturated porous media accounting for film and capillary flow. Water Resour.
Res. 44.
This work is jointly supported by PetroChina Innovation Foun- Pia, G., Sanna, U., 2014a. An intermingled fractal units model and method to predict
dation (2016D-5007-0209) and the National Natural Science Foun- permeability in porous rock. Int. J. Eng. Sci. 75, 31–39.
Pia, G., Sanna, U., 2014b. An intermingled fractal units model to evaluate pore size
dation of China (51474181, 51504202). distribution influence on thermal conductivity values in porous materials. Appl.
Therm. Eng. 65, 330–336.
References Pia, G., Sassoni, E., Franzoni, E., Sanna, U., 2014. Predicting capillary absorption of
porous stones by a procedure based on an intermingled fractal units model. Int.
J. Eng. Sci. 82, 196–204.
Aziz, K., Govier, G.W., 1972. The Flow of Complex Mixtures in Pipes. Van Nostrand
Pittman, E.D., 1992. Relationship of porosity and permeability to various
Reinhold Company, New York.
parameters derived from mercury injection-capillary pressure curves for
Bear, J., 1972. Dynamics of Fluids in Porous Media. Elsevier, New York.
sandstone. AAPG Bull. 76, 191–198.
Bird, R.B., 2002. Transport Phenomena. JohnWiley & Sons, New York.
Purcell, W.R., 1949. Capillary pressures – their measurement using mercury and the
Buiting, J.J.M., Clerke, E.A., 2013. Permeability from porosimetry measurements:
calculation of permeability therefrom. J. Petrol. Technol. 1, 39–48.
derivation for a tortuous and fractal tubular bundle. J. Petrol. Sci. Eng. 108, 267–
Swanson, B.F., 1981. A simple correlation between permeabilities and mercury
278.
capillary pressures. J. Petrol. Technol. 33, 2498–2504.
Cai, J., 2014. A fractal approach to low velocity non-Darcy flow in a low permeability
Tan, X.-H., Li, X.-P., Liu, J.-Y., Zhang, G.-D., Zhang, L.-H., 2014. Analysis of
porous medium. Chin. Phys. B 23, 044701.
permeability for transient two-phase flow in fractal porous media. J. Appl.
Cai, J., Hu, X., Standnes, D.C., You, L., 2012. An analytical model for spontaneous
Phys. 115, 113502.
imbibition in fractal porous media including gravity. Colloids Surf. A 414, 228–
Tan, X.-H., Li, X.-P., Liu, J.-Y., Zhang, L.-H., Fan, Z., 2015a. Study of the effects of stress
233.
sensitivity on the permeability and porosity of fractal porous media. Phys. Lett.
Cao, L., Li, X., Luo, C., Yuan, L., Zhang, J., Tan, X., 2016. Horizontal well transient rate
A.
decline analysis in low permeability gas reservoirs employing an orthogonal
Tan, X.-H., Liu, J.-Y., Li, X.-P., Zhang, L.-H., Cai, J., 2015b. A simulation method for
transformation method. J. Nat. Gas Sci. Eng.
permeability of porous media based on multiple fractal model. Int. J. Eng. Sci.
Carman, P.C., 1937. Fluid flow through granular beds. Chem. Eng. Res. Des. 75, S32–
95, 76–84.
S48.
Thomeer, J.H., 1983. Air permeability as a function of three pore-network
Dastidar, R., Sondergeld, C.H., Rai, C.S., 2007. An improved empirical permeability
parameters. J. Petrol. Technol. 35, 809–814.
estimator from mercury injection for tight clastic rocks. Petrophysics 48.
Thomeer, J.H.M., 1960. Introduction of a pore geometrical factor defined by the
Dong, M., Dullien, F.A., Dai, L., Li, D., 2005. Immiscible displacement in the
capillary pressure curve. J. Petrol. Technol. 12, 73–77.
interacting capillary bundle model Part I. Development of interacting capillary
Tuller, M., Or, D., 2002. Unsaturated hydraulic conductivity of structured porous
bundle model. Transp. Porous Media 59, 1–18.
media. Vadose Zone J. 1, 14–37.
Escobar, F.-H., Zambrano, A.-P., Giraldo, D.-V., Cantillo-Silva, J.-H., 2011. Pressure
Tuller, M., Or, D., Dudley, L.M., 1999. Adsorption and capillary condensation in
and pressure derivative analysis for non-newtonian pseudoplastic fluids in
porous media: liquid retention and interfacial configurations in angular pores.
double-porosity formations. CT&F-Ciencia, Tecnología y Futuro 4, 47–57.
Water Resour. Res. 35, 1949–1964.
Ghadami, N., Rasaei, M.R., Hejri, S., Sajedian, A., Afsari, K., 2015. Consistent porosity–
Wells, J.D., Amaefule, J.O., 1985. Capillary Pressure and Permeability Relationships
permeability modeling, reservoir rock typing and hydraulic flow unitization in a
in Tight Gas Sands. Society of Petroleum Engineers.
giant carbonate reservoir. J. Petrol. Sci. Eng. 131, 58–69.
Xiong, J., Liu, X., Liang, L., Zeng, Q., 2017. Adsorption of methane in organic-rich
Gou, X., Schwartz, J., 2013. Fractal analysis of the role of the rough interface
shale nanopores: an experimental and molecular simulation study. Fuel 200,
between Bi2Sr2CaCu2Ox filaments and the Ag matrix in the mechanical behavior
299–315.
of composite round wires. Supercond. Sci. Technol. 26, 202–208.
Xu, C., Kang, Y., You, L., You, Z., 2017. Lost-circulation control for formation-damage
Guarracino, L., Rötting, T., Carrera, J., 2014. A fractal model to describe the evolution
prevention in naturally fractured reservoir: mathematical model and
of multiphase flow properties during mineral dissolution. Adv. Water Resour.
experimental study. SPE J.
67, 78–86.
Xu, P., Sasmito, A.P., Yu, B., Mujumdar, A.S., 2016. Transport phenomena and
Gunter, G., Spain, D., Viro, E., Thomas, J., Potter, G., Williams, J., 2014. Winland Pore
properties in treelike networks. Appl. Mech. Rev. 68, 040802.
Throat Prediction Method-A Proper Retrospect: New Examples From
Xu, P., Yu, B., Qiao, X., Qiu, S., Jiang, Z., 2013. Radial permeability of fractured porous
Carbonates and Complex Systems, SPWLA 55th Annual Logging Symposium.
media by Monte Carlo simulations. Int. J. Heat Mass Transf. 57, 369–374.
Society of Petrophysicists and Well-Log Analysts.
Yang, S., Fu, H., Yu, B., 2017. Fractal analysis of flow resistance in tree-like branching
Huai, X., Wang, W., Li, Z., 2007. Analysis of the effective thermal conductivity of
networks with roughened microchannels. Fractals 25, 1750008.
fractal porous media. Appl. Therm. Eng. 27, 2815–2821.
Yang, S., Liang, M., Yu, B., Zou, M., 2014a. Permeability model for fractal porous
Kolodzie Jr., S., 1980. Analysis of Pore Throat Size and Use of the Waxman-Smits
media with rough surfaces. Microfluid. Nanofluid., 1–9
Equation to Determine Ooip in Spindle Field. Society of Petroleum Engineers,
Yang, S., Yu, B., Zou, M., Liang, M., 2014b. A fractal analysis of laminar flow
Colorado.
resistance in roughened microchannels. Int. J. Heat Mass Transf. 77, 208–217.
Kozeny, J., 1927. Uber die kapillare Leitung des Wassers im Boden. Akad. Wiss.
Yu, B., Cheng, P., 2002. A fractal permeability model for bi-dispersed porous media.
Wien.
Int. J. Heat Mass Transf. 45, 2983–2993.
Lebeau, M., Konrad, J.M., 2010. A new capillary and thin film flow model for
Yun, M., 2014. Seepage characteristics study on power-law fluid in fractal porous
predicting the hydraulic conductivity of unsaturated porous media. Water
media. Math. Probl. Eng. 2014.
Resour. Res. 46.
Yun, M., Yu, B., Lu, J., Zheng, W., 2010. Fractal analysis of Herschel-Bulkley fluid flow
Liu, R., Li, B., Jiang, Y., 2016. A fractal model based on a new governing equation of
in porous media. Int. J. Heat Mass Transf. 53, 3570–3574.
fluid flow in fractures for characterizing hydraulic properties of rock fracture
networks. Comput. Geotech. 75, 57–68.