ARTICLE IN PRESS

Computers & Geosciences 35 (2009) 1159–1166

Contents lists available at ScienceDirect

Computers & Geosciences

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

Fractal characterization of seepage-pores of coals from China:

An investigation on permeability of coals
Yanbin Yao , Dameng Liu, Dazhen Tang, Shuheng Tang, Wenhui Huang, Zhihua Liu, Yao Che
School of Energy Resources, China University of Geosciences, No. 29 Xueyuan Road, Haidian District, Beijing 100083, PR China

a r t i c l e in fo abstract

Article history: To better understand the characteristics of seepage-pores (pore radius larger than 100
Received 9 January 2008 nanometers) and their influence on the permeability of coals, we have conducted fractal
Received in revised form analyses for 34 fresh coal samples (mean maximum vitrinite reflectance Ro,max from
11 September 2008
0.43% to 4.21%) from North, Northwest and Northeast China. Mercury porosimetry data
Accepted 16 September 2008
indicate that the coals are fractal, with pore radius ranging from 0.1 to 50 mm. Calculated
fractal dimensions of these coals range from 2.61 to 2.98, higher than those from other
Keywords: kinds of rocks such as sandstone, shale, and carbonate. The data suggest that the coals
Fractal characterization have more complicated and inhomogeneous pore structures than other rocks.
Permeability
The fractal dimension has a negative correlation with the petrologic permeability of
Coal rank
coals, particularly for higher rank coals (with 1.47–4.21% Ro,max), from which a strong
Coalbed methane (CBM)
Mercury porosimetry negative linear correlation (R2 ¼ 0.85) between fractal dimension and permeability is
observed. A ‘U-shaped’ trend between fractal dimensions and coal ranks is observed,
with the minimum fractal dimensions occurring at 1.1–1.3% Ro,max. The sub-bituminous,
high volatile bituminous, semi-anthracite, and anthracite have higher fractal dimen-
sions. The effects of coal rank upon fractal dimensions are mainly due to the variety of
micropore contents and aromaticity of coals with coalification.
& 2008 Elsevier Ltd. All rights reserved.

1. Introduction (macropores) are important permeable pathways for gas and

water migration during production.
The geometry of the pore space (pore size distribution or Various methods such as mercury porosimetry, scan-
pore structure) of coals has significant influences on the ning electron microscopy (SEM), transmission electron
behavior of coals as a reservoir rock for gas. It affects not only microscopy (TEM), nuclear magnetic resonance (NMR),
the gas transportation in coal seams, but also the mechan- and small angle neutron scattering (SANS) have been used
isms of adsorption and storage of gas (Clarkson and Bustin, to study the seepage-pore structure (Radlinski et al.,
1999; Gilman and Beckie, 2000; Karacan and Okandan, 2001; 2004; Sing, 2004; Yao and Liu, 2006) of coals, among
Radovic et al., 1997; Yao and Liu, 2007). Based on their which the mercury injection is the most commonly used
contribution to coalbed methane (CBM) storage capacity method (Fu et al., 2005; S- enel et al., 2001; Xu et al., 2005;
and recoverability, pores in coals are divided into adsorption- Yao et al., 2006). To quantify the heterogeneity of seepage-
pores including micropores (o 10 nm) and mesopores pores, many fractal methods have been applied by
(10–100 nm), and seepage-pores (4100 nm) (Hodot, 1966; researchers (Bernal and Bello, 2001; Wang et al., 2002;
Shi and Durucan, 2005; Yao et al., 2006). Seepage-pores Zhao et al., 1998). The fractal data collected from mercury
porosimetry analysis provides additional textural infor-
mation that may be added to the results of traditional
 Corresponding author. Tel.: +86 10 82320892; fax:+86 10 82326850. determinations (Friesen and Ogunsola, 1995; Zhang
E-mail address: yaoyanbin@126.com (Y. Yao). et al., 2006). As demonstrated in our preliminary study

0098-3004/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cageo.2008.09.005
 ARTICLE IN PRESS
1160 Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166

(Yao et al., 2008a), fractal characterization of adsorption- Mercury porosimetry analyses and petrologic perme-
pores can add very useful information about the CH4 ability analyses were conducted at the Gas Research
adsorption characteristics of coals. However, fractal analy- Center, Langfang Branch of Research Institute of Petro-
sis of seepage-pores and their influence on the perme- leum Exploration and Development, China, following the
ability of coals have not been adequately studied. This SY/T 5346-1994 (by Chinese Oil and Gas Industry) and
study investigates the relationship between pore fractal SY/T 5336-1996 standard procedures.
dimension and permeability of coals and the influence of The experiments for mercury porosimetry were per-
coal rank and components on fractal dimensions. formed using an Autopore III 9420 Instrument (Micro-
metics, US), which automatically registers pressure, pore
radius, injection volume, and surface area. Volume
2. Coal samples and experiments injection curves were obtained for every sample, at the
pressure intervals of 0.0074–7.35 MPa, which correspond
A total of 34 coal samples from several potential to a pore radius range of 100 and 0.1 mm. Before
coalbed methane production areas in North, Northeast porosimetry analysis, all samples were dried at 75 1C
and Northwest China were investigated in this study for 48 h.
(Fig. 1). Most of the selected coal samples are large blocks The petrologic permeability analyses were selectively
of about 2 kg. All the coal samples were directly collected carried out for 20 large coal blocks due to the difficulties
from 23 active underground mines using the chip channel in sample preparation. Before permeability analysis, each
method, carefully packed, and then instantly carried to the coal block was cut parallel to the bedding surface for a
laboratory for experiments. horizontal cylindrical core (with 3.75 cm in diameter).
Vitrinite reflectance measurements (500 points), prox- Coal permeability to air was measured under simulated
imate and ultimate analyses were performed following in-situ stress conditions, using a tri-axial cell (isotropic
the GB/T 6948-1998, GB/T 212-2001, and GB/T 476-2001 ambient pressure of 2.5 MPa). Determining absolute
standard procedures, respectively. Results show that the permeability by flowing air through the core sample until
selected coals represent a broad range in lithotype the variation of permeability becomes relatively stable.
composition and coal rank (Table 1). These coals are
sub-bituminous to anthracite, with mean maximum 3. Results and discussion
vitrinite reflectance in oil (Ro,max, %) ranging from 0.43%
to 4.21%. Carbon contents of coals are from 50.68% to 3.1. Mercury porosimetry analysis and fractal dimension
89.06% and hydrogen contents range from 1.84% to 5.04%.
Ash contents of the coals vary from 0.22% to 9.34% and
Mercury injection porosimetry (mercury porosimetry)
inherent moisture contents are from 0.40% to 3.92%. Coal
is an important and widely used technique for analyzing
macerals are mainly characterized by intermediate to high
the pore structure of a variety of porous materials
vitrinite (50–85% volume) that corresponds to intermedi-
(Mahamud, 2006; Mahamud et al., 2003, 2004; Yao and
ate to low inertinite (10–50%), plus minor mineral matters
Liu, 2006; Yao et al., 2006). It is based on the gradual
(o12%) and various liptinite (0–25%) (Table 1).
injection of liquid mercury into an evacuated pore system
with external pressures. Increasing pressure makes smal-
ler pores accessible to mercury (Ritter and Drake, 1945).
Sampling spots Based on the assumption of cylindrical pores, the pore size
distribution can be calculated by the Washburn Equation
r c ¼ ð2s cos yÞ=Pc , (1)
Beijing 1 where Pc is the absolute injection pressure, MPa; rc is the
Qiadam Basin 7
66
6 pore radius (mm) when mercury enters at the pressure Pc
8 2 (MPa); y is the contact angle between mercury and the
3 4 pore surface (assumed to be 1401 in the experiment); and
Ordos Basin 9
5 s is the interfacial tension of mercury (set to 0.48 J/m2).
N After substitution, we have
North China Basin P c ¼ 0:735=r c (2)
Eq. (2) enables cumulative injection volume to be
inferred from the measured mercury injection curve.
Fractals and fractal geometry are used to describe
Coalfield: highly disordered systems characterized by their invaria-
1, Hongyang 2, Anhe bility with scales. Fractal systems are characterized by the
3, Xinggong 4, Yongxia fractal dimension (D): a fractal surface could have any
5, Huannan 6, Helanshan fractal dimension value between two and three—the more
7, Fugu-Baode 8, Hedong it approached three, the more complex is the surface
9, Weibei (Reich et al., 1992). It has been proven that fractals of coals
can be obtained from the data of mercury porosimetry
Fig. 1. Sample location in investigated CBM districts. analysis (Friesen and Ogunsola, 1995; Fu et al., 2005;
 ARTICLE IN PRESS
Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166 1161

Table 1
Details of coal samples, also with their coal rank, coal macerals and results of proximate and ultimate analyses.

Coal samples Ro (%) Proximate and ultimate (wt%) Coal maceral composition (%)

Sample no. Coalfield Coal mine Coal age Cad Had Mad Aad V I L Mm

HK81 Ordos Huating J 0.43 70.66 4.43 8.50 3.102 34.30 54.10 9.70 1.90
DMG12 Qaidam Dameigou J 0.55 64.68 3.42 9.34 8.74 31.40 60.40 5.20 3.00
NT56 Ordos Shenbei J 0.57 71.14 3.06 6.63 4.20 74.10 19.50 4.60 1.80
TG59 Ordos Yulin J 0.64 52.91 3.21 5.57 28.36 71.40 19.90 1.70 7.00
BDWJY-3 Fugu-Baode Baode C 0.68 76.92 4.43 2.20 4.30 72.30 16.90 10.50 0.30
FGDYG-2 Fugu-Baode Fugu P 0.70 74.85 4.50 3.88 4.06 68.30 25.20 6.00 0.50
BDXX-3 Fugu-Baode Baode C 0.75 70.56 4.20 2.15 11.79 72.20 15.90 11.80 0.10
HY12-1 Hongyang Hongyang C 0.77 50.08 3.19 5.40 30.20 82.50 9.50 6.50 1.50
LS12-2 Hongyang Linsheng C 0.83 71.34 4.34 0.56 17.28 72.00 15.50 1.80 10.70
ZJ11-2 Huainan Zhangji P 0.84 64.32 3.61 1.60 20.59 54.80 18.70 20.80 5.70
FGDYG-5 Fugu-Baode Fugu C 0.85 69.80 4.33 3.08 9.64 90.60 0.70 4.90 3.80
BDWJY-1 Fugu-Baode Baode C 0.86 72.91 4.18 2.36 9.66 66.70 24.40 8.90 x
PPT2 Hedong Panpanta P 0.86 77.37 5.04 0.98 8.38 63.60 27.50 7.40 1.50
PAN11-2 Huainan Panji P 0.87 73.22 4.22 1.44 12.76 54.20 18.70 24.60 2.50
ML25 Qaidam Muli J 1.01 85.42 4.86 0.58 2.04 50.10 48.90 0.70 0.30
LL3 Hedong Liulin P 1.20 81.98 4.79 0.38 7.65 66.90 29.80 x 3.30
LS5-1 Hongyang Linsheng P 1.25 79.78 4.60 0.95 7.54 64.00 31.00 2.20 2.80
HJ6 Ordos Hejin P 1.47 81.60 4.15 0.70 8.30 66.20 32.60 x 1.20
HC3 Weibei Hancheng C 1.70 87.30 4.48 0.22 3.74 71.00 28.50 x 0.50
JY5-4 Weibei Jinyu C 1.72 55.08 2.21 0.24 31.25 58.50 30.20 x 11.30
XC10-2 Weibei Xinchang C 1.86 81.22 3.78 0.46 7.52 72.70 26.80 x 0.50
HBSK4 Anhe Hebi-3 P 1.87 79.93 3.64 0.38 10.86 68.70 30.70 x 0.60
HY7-1 Hongyang Hongyang P 1.90 74.62 3.94 1.73 12.96 86.80 10.70 x 2.50
XSH3-2 Weibei Xiangshan P 1.98 74.58 3.50 0.36 16.19 77.10 22.00 x 0.90
XSH3-4 Weibei Xiangshan P 2.09 69.46 3.34 0.48 21.32 61.50 35.50 x 3.00
XSH3-3 Weibei Xiangshan P 2.10 58.88 2.86 0.62 32.62 16.30 80.20 x 3.50
RQ69 Helanshan Ruqigou J 2.48 86.60 3.08 1.30 5.54 60.10 38.10 x 1.80
XM12-3 Hongyang Xima C 2.60 81.85 3.26 1.22 8.81 72.80 25.60 x 1.60
XM13-1 Hongyang Xima C 2.68 87.82 3.40 1.04 3.58 73.60 26.20 x 0.20
GD4 Yongxia Gedian P 2.88 80.36 3.04 0.83 10.89 80.10 10.40 x 9.50
CSL3 Yongxia Chensilou P 2.89 82.61 3.02 1.37 8.44 80.30 15.40 x 4.30
CSL4 Yongxia Chensilou P 3.02 80.36 2.98 1.13 11.28 78.10 20.60 x 1.30
XM13-2 Hongyang Xima C 3.04 89.06 3.52 1.11 2.17 87.90 11.80 x 0.30
XY1 Xinggong Xingyang C 4.21 87.76 1.84 3.22 4.24 80.20 19.10 x 0.70

Coal age: P ¼ Permian, C ¼ Carboniferous (Pennsylvanian), J ¼ Jurassic.

X ¼ present in content o0.1%.
Coal maceral composition: V ¼ Vitrinite, I ¼ Inertinite, L ¼ Liptinite, Mm ¼ Mineral matter.

Mahamud et al., 2003; Zhang et al., 2006). Mathematical where Vr is the cumulative injection volume at a given
models for fractal dimension analysis were provided by pore radius r and Ds is the surface fractal dimension.
Friesen and Mikula (1987), Neimark (1990), Pfeifer and Following the work of Pfeifer and Avnir (1983), Friesen and
Avnir (1983), and Zhang and Li (1995). The characteristics Mikula (1987) combined Eqs. (2) and (3), and obtained a
and differences among these mathematical models have relation between the cumulative injection volume (Vp)
been discussed in detail elsewhere (e.g., Bernal and Bello, derivative with respect to pressure (p) and the surface
2001; Le0 on y Le0 on, 1998; Mahamud, 2006; Zhang and fractal dimension (Ds)
Li, 1995). In this study, the relatively popular and conve-  
dV p
nient method by Friesen and Mikula (1987) was chosen to log  að4  Ds Þ logðpÞ, (4)
dp
calculate the fractal dimension of seepage-pores of coals.
The following paragraphs give a brief summarization of where Vp is the cumulative injection volume at a given
the methods for obtaining the fractal dimension. pressure p. Using those relations, the surface fractal
A coal contains pores of varying sizes, so it can be simu- dimension can be calculated without pore surface area
lated by the Menger sponge fractal model (Mandelbrot, calculations. The logarithm of derivative diff injection
1983), which has a fractal surface and an infinite range of volume (log(dV/dp)) is plotted vs. pressure logarithm
pore radius. For further study, Pfeifer and Avnir (1983) (log(p)) and shown in Fig. 2. If the pore surface is fractal
stated that the volume injection curve of a pore structure within certain ranges of pore radius, a linear trend should
with fractal surface must obey the relation be observed from the slope (A) of the logarithm line
  (Fig. 2), and D is calculated by D ¼ 4+A.
dV r It is also important to note that Eq. (4) is valid only for
log  að2  Ds Þ logðrÞ, (3)
dr a subset of pores with radius between rmax and rmin,
 ARTICLE IN PRESS
1162 Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166

o,max o,max
3

3
2 2

o,max o,max

3
3

2 2

Fig. 2. Plots showing representative examples of log(dV/dp) vs. log(p).

although the pore radius scale applied for mercury The calculated fractal dimensions of coals is in the
porosimetry is broad (Friesen and Ogunsola, 1995; range of 2.61–2.98 (Table 2), which is consistent with the
Fu et al., 2005; Suuberg et al., 1995; Wang et al., 2002; previous reports from coals collected from North and
Zhao et al., 1998; Zhang et al., 2006). Friesen and Ogunsola Northeast China (Wang et al., 2002; Zhao et al., 1998). A
(1995) indicated that two different fractal dimensions coal with a higher fractal dimension means a rougher
exist at a low-pressure range (o10 MPa) and a high- surface and more complicated pore structures.
pressure range (410 MPa), and the fractal dimension at
higher pressure represents a combination of pore-filling
and coal compressibility. Mahamud et al. (2003) also 3.2. Permeability variation with fractal dimension
indicated that at the higher pressure, the compressibility
of coal samples is important for correcting mercury The permeability data from the petrologic analysis
intrusion volumes. Due to the fact that the maximum show that the Chinese coals have medium to very low
pressures, considered in this study, are under 8 MPa, the permeability, with highly variable values ranging from
compressibility effect could be negligible. For our mercury 0.016 to 18.95 mD (Table 2). One sample (sample PPT2)
porosimetry results, only within scales of seepage-pores has an anomalously high value of 72.55 mD, which is
with pore radius ranges from 0.1 to 50 mm, the experi- likely caused by artificially induced fracturing. Although
mental data appear to give a better fit to a straight line all results obtained, in this study, were measured under an
(four representative coal samples with various ranks are ambient pressure that is unnecessary, the same as those in
shown in Fig. 2). It means that coals are fractal at this pore the subsurface, the data provide relative valid perme-
range interval. For this reason, the values of calculated ability values among the coals analyzed.
fractal dimensions for all samples are only valid within The relationship between fractal dimensions and
the pore radius of 0.1–50 mm. permeability is shown in Fig. 3. In general permeability
 ARTICLE IN PRESS
Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166 1163

Table 2
Analyzed petrological permeability results and calculated fractal dimension for selected coal samples.

Sample no. kp A D Goodness of fit (R2) Sample no. kp A D Goodness of fit (R2)

HK81 1.105 1.1210 2.8790 0.95 HJ6 1.402 1.1966 2.8034 0.89
n n
DMG12 1.0895 2.9105 0.97 HC3 1.1761 2.8239 0.81
n
NT56 3.987 1.1380 2.8620 0.99 JY5-4 1.1091 2.8909 0.98
TG59 18.950 1.2287 2.7713 0.93 XC10-2 0.848 1.0976 2.9024 0.99
BDWJY-3 4.777 1.1183 2.8817 0.99 HBSK4 0.026 1.0341 2.9659 0.99
n
FGDYG-2 4.402 1.1144 2.8856 0.99 HY7-1 1.0927 2.9073 0.98
n n
BDXX-3 1.1784 2.8216 0.96 XSH3-2 1.0525 2.9475 0.96
n
HY12-1 1.2904 2.7096 0.98 XSH3-4 0.016 1.0675 2.9325 0.99
n
LS12-2 1.648 1.1396 2.8604 0.99 XSH3-3 1.0386 2.9614 0.98
ZJ11-2 0.121 1.1792 2.8208 0.97 RQ69 0.254 1.0774 2.9226 0.90
n
FGDYG-5 1.1276 2.8724 0.99 XM12-3 0.020 1.0367 2.9633 0.93
n
BDWJY-1 0.241 1.1316 2.8684 0.89 XM13-1 1.0560 2.9440 0.93
n n
PPT2 1.3438 2.6562 0.91 GD4 1.0239 2.9761 1.00
n
PAN11-2 1.3002 2.6998 0.93 CSL3 0.041 1.0568 2.9432 0.99
ML25 0.111 1.2803 2.7197 0.95 CSL4 0.265 1.0715 2.9285 0.99
n n
LL3 1.3918 2.6082 0.88 XM13-2 1.0442 2.9558 0.96
LS5-1 0.409 1.3004 2.6996 0.99 XY1 0.548 1.1369 2.8631 0.97

kp ¼ petrological permeability (mD), n ¼ not analyzed, f ¼ the sample with artificially induced fractures, A ¼ slope of the log(dV/dp) vs. log(p),
D ¼ calculated fractal dimension.

100 1.5

R2=0.16 R2=0.85

10
1.0

1.0
0.5

0.1
0.0

0.01
2.6 2.7 2.8 2.9 3.0 2.75 2.8 2.85 2.9 2.95 3.0

Fig. 3. Relationships between permeability and fractal dimension for coals with (a) 0.43–4.21% Ro,max and (b) 1.47–4.21% Ro,max.

decreases with fractal dimensions, but the two para- fractal dimension should reflect the pore structure and the
meters show a poor to slightly positive correlation for permeability of coals. Fig. 3 shows that coals with higher
coals with 0.43–4.21% Ro,max (Fig. 3a). For the coals tested, fractal dimensions are commonly less permeable for gas,
permeability decreases with coal ranks. High volatile and such a relationship is apparent for coals of higher
bituminous have permeability values higher than 1 mD ranks.
(mostly higher than 4 mD). In contrast, low volatile The origin of the scatter in Fig. 3a is uncertain, but may
bituminous and anthracite have permeability values be attributable to several factors. First, the scatter may
generally lower than 1 mD (Table 2). In order to eliminate represent the natural variation of permeability with
the effects of coal rank upon measured permeability, fractal dimension. Second, it may be caused by the
samples are divided into two groups: low to medium rank combined effect of coal rank and fracture on the perme-
coals including sub-bituminous and high volatile bitumi- ability. Third, artificially induced fractures or pitting near
nous with 0.45–1.25% Ro,max and high rank coals with or at the measuring point may have induced variations in
Ro,max4 1.25%. For the low and medium rank coals, there the measured permeability and then reduced the correla-
is no correlation or only a slightly negative correlation tions. Fourth, the sample volume cut for petrological
between the permeability and fractal dimensions. For high analysis may not be completely representative of the
rank coals with 1.47–4.21% Ro,max, a strong negative linear volume measured for mercury porosimetry. Finally, some
correlation (R2 ¼ 0.85) is observed. sample points that have high fractal dimension but high
The fractal dimension of a surface is supposed to be permeability may be due to the macro- or microscopic
independent of its size or macroscopic shape. Instead, the fracturing that may result in higher permeability.
 ARTICLE IN PRESS
1164 Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166

Although cares were taken to avoid artificial fracturing dimension are possibly due to the structure changes
and to choose points where the coal appeared to be resulting from physical and chemical changes (e.g.,
homogeneous in composition, this task is proven to be compression of pore structure and increase in aromaticity
difficult. with coalification) during coalification.
At the sub-bituminous to early high volatile bitumi-
nous stage (stage I in Fig. 4, with 0.43–0.64% Ro,max),
3.3. Effects of coal rank upon fractal dimension
textures of coals are mainly composed of the methoxyl
and carboxylic functional group (Han, 1996). At this stage,
The fractal dimension of coals represents an intrinsic the pores of coals have high moisture contents (46%,
nature of the pore surface and pore structure of coals, Table 1 and Fig. 5a) and high fractal dimension (generally
which is also related to the rank and composition of coals. 42.85, Fig. 5). A positive relationship between the
Scatter plots and regression analyses reveal a polynomial moisture content and fractal dimension (Fig. 5b) suggests
distribution between fractal dimension and coal rank that the fractal properties of coal may condition the
(Fig. 4). The fractal dimension values exhibit an expo- capacity for retaining water molecules in its porous
nential decay with increasing coal rank from 2.88 network.
(0.43% Ro,max) to 2.61 (1.2% Ro,max), a rapid exponential At the high volatile bituminous A to early medium
increase with increasing coal rank from 2.61 (1.2% Ro,max) volatile bituminous stage (stage II in Fig. 4, with 0.6–1.2%
to 2.96 (2.1% Ro,max), and maintaining higher values at the Ro,max), the fractal dimension decreases with increasing
semi-anthracite and anthracite stage (Ro,max 42.5%). coal rank. Interestingly, a rapid decrease of moisture
Variable fractal dimension values for coals at the high contents (from 5.57% to 0.38%), with increasing coal rank
volatile bituminous stage (0.43–1.2% Ro,max) are in good is also seen (stage II in Fig. 5a). Since the moisture
agreement with the SANS- and N2-adsorption results by contents and fractal dimensions show a slightly positive
Prinz et al. (2004). The effects of coal rank upon fractal correlation as shown in Fig. 5b, it seems that the decrease
of fractal dimension in stage II is indicated by the loss of
moisture within coal components (Fig. 5a). However, the
3.0 loss of moisture coincides with the greatest changes in
pore surface and pore diameter distribution, it is con-
2.9 ceivable that the decrease of micropore volume with
Francal dimension, D

increasing coal rank most likely caused the decrease of

IV
I fractal dimension at this stage. Bustin and Clarkson (1998)
2.8
have shown that micropore capacity reaches a minimum
III at high volatile bituminous coal rank. It is also noted by
2.7 Gürdal and Yalc- ın (2001) that the micropore capacity
decreases with increasing vitrinite reflectance up to a
2.6 value of 1.0–1.1% Ro,max and starts to increase with
II increasing rank. Fractal dimension and coal rank data
show the same ‘U-shaped’ trend as fractal dimensions and
2.5
0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 4.4 micropore volumes have (Gürdal and Yalc- ın, 2001; Prinz
Ro,max,% and Littke, 2005), indicating a close relationship between
fractal change and micropore volume. The decrease of
Fig. 4. Relationship between fractal dimension and coal rank. micropores makes the pore surface more flat and the pore

12 3.0

2.9
9

I 2.8 R2=0.33
6

II 2.7

3 V
IV 2.6

0 2.5
0 1 2 3 4 5 6 0 2 4 6 8 10
o,max,

Fig. 5. Relationships between (a) moisture content and coal rank and (b) moisture content and fractal dimension.
 ARTICLE IN PRESS
Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166 1165

diameter distribution more regular, and, therefore, leads medium coal ranks. Such effects may be due to the
to the decrease of fractal dimension. ‘two-phase’ physico-chemical model of coals. In this
The rapid increase of fractal dimension at the medium model, coal structure consists of two components, a
to low volatile bituminous stage (stage III in Fig. 4, with macromolecular network and a molecular component
1.2–2.1% Ro,max) may also be related to the change of the (Clarkson and Bustin, 1996; Rouzaud and Oberlin, 1990).
pore structure of coals. The increase of fractal dimension The macromolecular network is built up of cross-linked
exactly occurs at the transitional stage from the second to aromatic carbon structures (so-called crystallites), which
the third coalification jump (Han, 1996), when micropores have a turbostratic or random layer lattice order (Prinz
and inner surface areas begin to increase rapidly (Bustin et al., 2004). These crystallites contain a significant
and Clarkson, 1998; Gürdal and Yalc- ın, 2001). During this amount of aliphatic side chains by which they are linked
transition, increasing surface area led to coarser pore and form macromolecules. In general, the higher is the
surface and increasing micropore content, resulting in carbon content, the more complicated and anisotropic of
more anisotropic pore diameter distribution. At this stage, the crystallites structure would be. Thus, coals with
it seems that the moisture content as an indicator for relatively high carbon content generally have higher
fractal dimension change is weak. This is because the fractal dimensions at similar coal ranks.
moisture contents of coals are generally low in stage III, The negative correlation between permeability and
with the values between 0.2 and 1.0% (Fig. 5a). fractal dimension means that coals with higher fractal
The semi-anthracite and anthracite stage (stage IV in dimensions are not favorable for CBM production. Fractal
Fig. 4, with 2.1–3% Ro,max) represents the transition from dimension also changes with coal ranks, displaying a
the third to the fourth coalification jump (Han, 1996). ‘U-shaped’ trend. It means that a higher fractal dimension
During this stage, the meso- and micropore contents have is always observed in the sub-bituminous, high volatile
reached a maximum, and thus the fractal dimensions bituminous, semi-anthracite and anthracite in China.
maintain at higher values. Furthermore, when it comes to These relationships may account for a reason for the
the meta-anthracite stage, a slight decrease in fractal ‘‘low permeability (o1 millidarcy), low production (aver-
dimension is expected to be associated with the decrease agely about 2000 m3, per day and per well), and high
of porosity and pore surface area. However, due to the heterogeneity (Su et al., 2005; Yao et al., 2007; Yao et al.,
limited data, this relationship is not clearly demonstrated 2008b) of the high rank coal reservoirs in China.
in this study.
Relationships between coal ranks and fractal dimen-
sions indicate that coal rank has important influences on 4. Conclusions
the fractal dimension of coals. However, coals with similar
coal ranks may have significantly different fractal dimen- Mercury porosimetry is a commonly used method to
sions. Taking Samples BDXX-3 and HY12-1 as an example, study the seepage-pores (pore radius larger than 100 nm)
the difference in Ro,max for these samples is only 0.02, but of coals, adding additional textual information to the
the difference in D is higher than 0.1. Such variations in D results of traditional determinations. Using mercury
are most likely caused by the different carbon contents in injection data of 34 Chinese coals, the fractal dimensions
coals, as shown in Fig. 6. For coals with low-medium rank of pores with 0.1–50 mm in radius have been calculated.
(0.6–1% Ro,max), the fractal dimension increases/decreases The fractal dimension values are very high, indicating that
with increasing/decreasing carbon contents at the similar coals are highly heterogenetic in their pore structures and
coal rank. It is clear that the carbon content of coals has surfaces. The heterogeneity from the fractal analyses can
effects on the fractal dimension at identical low to be used for the estimation of the permeability of coals.
The relationship between fractal dimension and per-
meability shows that for coals with 0.43–4.21% Ro,max,
Carbon content Fractal dimension only a slight negative relationship is observed; in contrast,
3.0 for high rank coals with 1.47–4.21% Ro,max, a strongly
85 negative linear correlation (R2 ¼ 0.85) is observed. So
2.9 coals with higher fractal dimensions are unfavorable for
Frascal dimension, D

gas production.
Carbon content, %

75
2.8 Further investigation shows that the fractal dimension
of coals is mainly controlled by coal rank, coal compo-
65 nents (especially the change of aromaticity with coalifica-
2.7 tion), and pore structure of coals. The fractal dimensions
show a ‘U-shaped’ trend with coal ranks, with minimum
55 2.6 values occurring at 1.1–1.3% Ro,max. The sub-bituminous,
high volatile bituminous, semi-anthracite and anthracite
have higher fractal dimensions than other coals. Effects of
45 2.5
0.43 0.68 0.83 0.86 1.25 1.86 2.09 2.68 3.04 coal rank upon fractal dimensions are mainly due to
changes in micropore contents during coalification.
Ro,max,%
It is suggested that for Chinese coals, the fractal
Fig. 6. Variations of fractal dimension and carbon content with dimension is an another piece of evidence that supports
increasing coal rank. the fact that very high rank coals possess low permeability
 ARTICLE IN PRESS
1166 Y. Yao et al. / Computers & Geosciences 35 (2009) 1159–1166

(o1 mD), thus have low gas production rates. However, it Prinz, D., Littke, R., 2005. Development of the micro- and ultramicropor-
is still unclear whether this is the common case for coals ous structure of coals with rank as deduced from the accessibility to
water. Fuel 84, 1645–1652.
of the other coal basins in the world. Prinz, D., Pyckhout-Hintzen, W., Littke, R., 2004. Development of the
meso- and macroporous structure of coals with rank as analysed
with small angle neutron scattering and adsorption experiments.
Acknowledgments Fuel 83, 547–556.
Radlinski, A.P., Mastalerz, M., Hinde, A.L., Hainbuchner, M., Rauch, H.,
Baron, M., Lin, J.S., Fan, L., Thiyagarajan, P., 2004. Application of
This research was funded by the National Key Basic SAXS and SANS in evaluation of porosity, pore size distribution and
Research Program of China (Grant nos. 2006CB202202, surface area of coal. International Journal of Coal Geology 59,
2002CB211702, 2009CB219604), the National Natural 245–271.
Radovic, L.R., Menon, V.C., Le0 on y Le0 on, C.A., Kyotani, T., Danner, R.P.,
Science Foundation of China (No. 40572091), the Anderson, S., Hatcher, P.G., 1997. On the porous structure of coals:
China Geological Survey (Grant nos. 20021010004, evidence for an interconnected but constricted micropore system
1212010534702), and PetroChina Innovation Foundation. and implications for coalbed methane recovery. Adsorption 3,
221–232.
The paper was oral presented at the IAMG2007. Prof. Reich, M.H., Snook, I.K., Wagenfeld, H.K., 1992. A fractal interpretation of
Ganqing Jiang from University of Nevada at Las Vegas is the effect of drying on the pore structure of Victorian brown coal.
greatly appreciated for his assistance in polishing the Fuel 71 (6), 669–672.
manuscript. The anonymous reviewers are acknowledged Ritter, H.L., Drake, L.C., 1945. Pore size distribution in porous materials. I.
Pressure porosimeter and determination of complete macropore size
for their valuable comments that improved the quality of distributions. Industrial Engineering Chemical Analysis Edition 17,
this paper. 782–786.
Rouzaud, J.-N., Oberlin, A., 1990. Advanced Methodologies in Coal
Characterization. Elsevier, Amsterdam, p. 311.
References + uz,
S- enel, G., Gur + A.G., Yucel,
+ H., 2001. Characterization of pore structure
of Turkish coals. Energy & Fuel 15, 331–338.
Shi, J.Q., Durucan, S., 2005. Gas storage and flow in coalbed reservoirs:
Bernal, J.L.P., Bello, M.A., 2001. Fractal geometry and mercury porosi- implementation of a bidisperse pore model for gas diffusion in a coal
metry comparison and application of proposed models on building matrix. SPE (Society of Petroleum Engineers) Reservoir Evaluation &
stone. Applied Surface Science 185, 99–107. Engineering 8 (2), 169–175.
Bustin, R.M., Clarkson, C.R., 1998. Geological controls on coalbed Sing, K.S.W., 2004. Characterization of porous materials: past, present
methane reservoir capacity and gas content. International Journal and future. Colloids and Surfaces A: Physicochemical and Engineer-
of Coal Geology 38, 3–26. ing Aspects 241, 3–7.
Clarkson, C.R., Bustin, R.M., 1996. Variation in micropore capacity and Su, X., Liu, X., Liu, S., Zhao, M., Song, Y., 2005. Geology of coalbed methane
size distribution with composition in bituminous coal of the Western reservoirs in the southeast Qinshui Basin of China. International
Canadian Sedimentary Basin: implications for coalbed methane
Journal of Coal Geology 62, 197–210.
potential. Fuel 75 (13), 1483–1498.
Suuberg, E.M., Deevi, S.C., Yun, Y., 1995. Elastic behaviour of coals studied
Clarkson, C.R., Bustin, R.M., 1999. The effect of pore structure and gas
by mercury porosimetry. Fuel 74 (10), 1522–1530.
pressure upon the transport properties of coal: a laboratory and
Wang, W., Xu, L., Fu, X., 2002. Study on pore texture of coal with fractal
modeling study. 1. Isotherms and pore volume distributions. Fuel 78,
theory. Coal Geology of China 14 (2), 26–33 (in Chinese with English
1333–1344.
abstract).
Friesen, W.I., Mikula, R.J., 1987. Fractal dimensions of coal particles.
Xu, H., Zhang, S., Leng, X., Tang, D., Wang, M., 2005. Analysis of pore
Journal of Colloid Interface Science 120 (l), 263–271.
system model and physical property of coal reservoir in the Qinshui
Friesen, W.I., Ogunsola, O.I., 1995. Mercury porosimetry of upgraded
Basin. Chinese Science Bulletin 50, 52–58.
western Canadian coals. Fuel 74 (4), 604–609.
Yao, Y., Liu, D., 2006. Developing features of fissure system in Henan coal
Fu, X., Qin, Y., Zhang, W., Wei, C., Zhou, R., 2005. Fractal classification and
reserves seams and research on mining of coalbed methane. Coal
natural classification of coal pore structure based on migration of
Science and Technology 34 (3), 64–68 (in Chinese with English
coalbed methane. Chinese Science Bulletin 50, 66–71.
abstract).
Gilman, A., Beckie, R., 2000. Flow of coal-bed methane to a gallery.
Transport in Porous Media 41, 1–16. Yao, Y., Liu, D., 2007. Adsorption characteristic of coal reservoirs in North
Gürdal, G., Yalc- ın, M.N., 2001. Pore volume and surface area of the China and its influencing factors. Journal of China University of
Carboniferous coals from the Zonguldak basin (NW Turkey) and their Mining & Technology 36 (3), 308–314 (in Chinese with English
variations with rank and maceral composition. International Journal abstract).
of Coal Geology 48, 133–144. Yao, Y., Liu, D., Huang, W., Tang, D., Tang, S., 2006. Research on the pore-
Han, D., 1996. Coal Petrology in China. China University of Mining and fractures system properties of coalbed methane reservoirs and
Technology Press, Xuzhou, p. 599 (In Chinese). recovery in Huainan and Huaibei coal-fields. Journal of China Coal
Hodot, B.B., 1966. Outburst of Coal and Coalbed Gas (Chinese Transla- Society 31 (2), 163–168 (in Chinese with English abstract).
tion). China Industry Press, Beijing, p. 318. Yao, Y., Liu, D., Tang, D., Tang, S., Huang, W., 2007. Reserving and
Karacan, C.O., Okandan, E., 2001. Adsorption and gas transport in coal recovering characteristics of coalbed methane in coal reservoirs in
microstructure: investigation and evaluation by quantitative X-ray North China. Petroleum Exploration and Development 34 (6),
CT imaging. Fuel 80, 509–520. 664–668 (in Chinese with English abstract).
Le on y Le0 on, C.A., 1998. New perspectives in mercury porosimetry.
0 Yao, Y., Liu, D., Tang, D., Tang, S., Huang, W., 2008a. Fractal characteriza-
Advances in Colloid and Interface Science 76–77, 341–372. tion of adsorption-pores of coals from North China: an investigation
Mahamud, M.M., 2006. Textural characterization of active carbons using on the CH4 adsorption capacity of coal. International Journal of Coal
fractal analysis. Fuel Processing Technology 87 (10), 907–917. Geology 73, 27–42.
Mahamud, M., López, Ó., Pis, J.J., Pajares, J.A., 2003. Textural characterization Yao, Y., Liu, D., Tang, D., Huang, W., Tang, S., Che, Y., 2008b. A
of coals using fractal analysis. Fuel Processing Technology 81, 127–142. comprehensive model for evaluating coalbed methane reservoirs in
Mahamud, M.M., López, Ó., Pis, J.J., Pajares, J.A., 2004. Textural China. Acta Geologica Sinica 82 (6), 1253–1270.
characterization of chars using fractal analysis. Fuel Processing Zhang, B., Li, S., 1995. Determination of the surface fractal dimension for
Technology 86, 135–149. porous media by mercury porosimetry. Industrial & Engineering
Mandelbrot, B.B., 1983. The Fractal Geometry of Nature (Updated and Chemistry Research 34 (4), 1383–1386.
Augmented Edition). W.H. Freeman, San Francisco, CA, p. 468. Zhang, B., Liu, W., Liu, X., 2006. Scale-dependent nature of the surface
Neimark, A.V., 1990. Calculating surface fractal dimensions of adsor- fractal dimension for bi- and multi-disperse porous solids by
bents. Adsorption Science & Technology 7 (44), 210–219. mercury porosimetry. Applied Surface Science 253, 1349–1355.
Pfeifer, P., Avnir, D., 1983. Chemistry in non integer dimensions between Zhao, A., Liao, Y., Tang, X., 1998. Quantitative analysis of pore structure by
2 and 3, I: fractal theory of heterogenous surface. Journal of Chemical fractal analysis. Journal of China Coal Society 23 (4), 439–442 (in
Physics 79 (7), 3558–3565. Chinese with English abstract).