Fuel 158 (2015) 239–247

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

A new laboratory method for accurate measurement of the methane

diffusion coefficient and its influencing factors in the coal matrix
Hao Xu ⇑, Dazhen Tang, Junlong Zhao, Song Li, Shu Tao
School of Energy Resources, China University of Geosciences (Beijing), Beijing 100083, China
Coal Reservoir Laboratory of National Engineering Research Center of Coalbed Methane Development & Utilization, Beijing 100083, China

h i g h l i g h t s

 A new laboratory method was proposed to measure the CBM diffusion coefficient.
 The tested diffusion coefficient was the order of magnitude of 10
11
–109 m2/s.
 The influences on the methane diffusion coefficient were analyzed.

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

Article history: The diffusion coefficient is a key parameter for evaluating the methane diffusion properties of a coal
Received 3 March 2015 matrix. However, all of the current measurement technologies for methane diffusion coefficients, includ-
Received in revised form 4 May 2015 ing the particle method, the steady state method and the inverse diffusion method, have certain limita-
Accepted 20 May 2015
tions. For example, the pore structure of the sample is easily destroyed, and the test time is long.
Available online 29 May 2015
Moreover, the results from different methods lack comparability under the various test principles and
conditions. In this work, a new laboratory method for measuring the methane diffusion coefficient in coal
Keywords:
matrixes was proposed, where coal matrix flakes instead of coal particles were used as the samples to be
Coal matrix
Diffusion coefficient
measured, and the factors influencing methane diffusion coefficients such as gas pressure, coal rank and
Measurement method moisture content were also analyzed. The results indicate that this new method can keep the intrinsic
Gas pressure space structure of the coal matrix and yields methane diffusion coefficients on the order of 1011–
Coal rank 109 m2/s in coal matrix, which are close to realistic reservoir values. Under the same temperature,
Coalbed methane methane diffusion coefficients increase with increasing gas pressure regardless of the water saturation
and coal rank. Additionally, the adsorption capacity increases when the metamorphic degree of coal
increases, while the methane diffusion coefficient in coal matrixes exhibit a trend of first dropping and
then rising (‘‘U’’ shape) with an increase in coal rank. Moreover, moisture has an important effect on
the methane diffusion coefficient. Increasing moisture can reduce the matrix adsorption capacity for
methane, which makes it difficult to form a larger concentration gradient because the diffusion coeffi-
cients of dry samples are higher than those of the saturated water samples.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction
It is of great significance for the methane gas diffusion coeffi-
cient estimation since it controls the late production rate of the
coal and shale reservoirs [1–5]. Diffusion refers to the phenomenon
where molecules migrate from high concentration areas to low
concentration areas until they are evenly distributed [6–8]. For
coal reservoirs, methane cannot simultaneously be in contacted
⇑ Corresponding author at: School of Energy Resources, China University of
Geosciences (Beijing), Beijing 100083, China. Tel.: +86 10 82320106; fax: +86 10
82326850.
E-mail address: Xuhao600@163.com (H. Xu).
with and adsorbed onto all of the pores and fissure surfaces
because of the nonpolar characteristics of the methane molecules,
which results in concentration and pressure gradients in the coal
body. Therefore, methane adsorption in the coal reservoir is an
integrated process of seepage, diffusion and adsorption [2,9]. The
production rate of coalbed methane (CBM) is mainly controlled
by diffusion in the matrix and the permeability within the cleat
system [10]. Thus, the understanding of methane diffusion mecha-
nism in coal matrix and the accurate determination the methane
diffusion coefficients are of great significance for CBM production
planning and economical reserve estimation [11].
The diffusion coefficient is the one of the key parameters to con-
trol the gas transport dynamics in of coal matrix [12]. The methane

http://dx.doi.org/10.1016/j.fuel.2015.05.046
0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
240 H. Xu et al. / Fuel 158 (2015) 239–247
diffusion coefficient determination methods include the particle
method, the steady state method and the inverse diffusion method
[10,11,13–15]. Many scholars have extensively reviewed the influ-
encing parameters on the diffusion coefficient. Zhao [16], through
measuring diffusion coefficients in coal micropores, discovered
that reservoir pressure, methane concentration and water satura-
tion substantially affect methane diffusion coefficients. Shi and
Durucan [1] thought that the diffusion coefficients depend on the
gas concentration and the adsorption state of the inner coal sur-
face. Through studying the diffusion rates of coal particle samples
of different ranks, Clarkson and Bustin [17] speculated that the
absorption capacity differences of coal samples are mainly due to
the different proportions of macropores, mesopores and microp-
ores, while the pore size distribution is related to the coal rank
[18]. Furthermore, Busch et al. [19] demonstrated that diffusion
coefficient decreases with increasing of particle size. Wang et al.
[20] analyzed the diffusion dynamics in the process of CBM desorp-
tion under the different coal ranks and moisture contents. Xu et al.
[21] revealed the influence of CBM bearing boundary conditions on
methane diffusion among coal particles. Additionally, the diffusion
coefficients of different gas composition are different under the
same conditions, and the sensitivities to various influencing factors
are also different. Saghafi et al. [22] determined the carbon dioxide
and methane diffusion coefficients of the coals from the Sydney
Basin by concentration difference experiments, and demonstrated
that the carbon dioxide diffusion coefficient is about twice as large
as that of methane. Pan et al. [10] studied the relationship between
the diffusion of carbon dioxide and diffusion of methane under dif-
ferent moisture contents, which further indicated that the
methane diffusion coefficient is more easily affected by moisture
than that of carbon dioxide and that the influences of pore size
on the diffusion coefficient are not identical. With the coals from
Lorraine Basin in France, Charriere et al. [23] compared the carbon
dioxide and methane diffusion coefficients, which indicated that
temperature and pressure are the two key factors controlling the
diffusion coefficient.
However, these results lack comparability with each other
under the different test principles and conditions [24], and these
methods have the following disadvantages: (1) Although the parti-
cle method is the most widely adopted test technology in the CBM
industry, the coal briquettes are smashed into tiny particles which
results in the particles being easily accumulated and leading to the
formation of intergranular pores. Meanwhile, the specific space
structures of the coal matrix are destroyed. (2) With the steady
state method, a cylindrical coal sample is chosen to measure the
diffusion coefficient under a very small differential pressure.
However, the requirements of the sample are demanding (i.e.,
the sample must be without fractures) and the test time is too long.
Moreover, the results are sensitive to the environment tempera-
ture, which limits the wide application of the method.
Additionally, it is difficult to ensure that only diffusion without
seepage exists during the test process. (3) The inverse diffusion
method can ensure that the experimental process is not influenced
by pressure differences. Nevertheless, the diffusion coefficient of
the high-pressure environment cannot be measured. Hence, it is
necessary to explore a suitable measurement and analysis technol-
ogy for methane diffusion estimation.
In this work, we took into account that the space structure of
the coal matrix cannot be damaged, the instrument can measure
the methane diffusion instead of the adsorption and desorption,
also can simulate the actual underground temperature and pres-
sure condition, and the experiment can be conducted under safe
conditions and no other gases interfere. A new laboratory method,
which can overcome the deficiencies of current methods was pro-
posed to measure the methane diffusion coefficient in coal
matrixes. Furtherly, because gas pressure, coal rank and moisture
content are essential factors for the adsorption, desorption and dif-
fusion properties, their influences on methane diffusion coefficient
in the coal matrix were analyzed, which is great importance for the
CBM production.
2. Experimental work
2.1. Sample preparation and experimental setup
The bulk coal samples with different metamorphic degree were
obtained from three active mine areas (Yimin, Hancheng and
Jincheng) located in the Hailar, Ordos and Qinshui Basins of
China, respectively. A total of six active coal mines (Liaoyuan,
Shengjie, Nangou, Xiangshan, Wumuchang and Tang’an) were
investigated in this research (Fig. 1).
Prior to the start of the matrix diffusion coefficient test, the
flake coal samples were polished and prepared from the bulk coals.
The size (length  width  thickness) is approximately 20 mm 
15 mm  3 mm (Fig. 2) and the largest fracture width must be less
than 3 lm. Then, the flake coal samples were cleaned with an
ultrasonic cleaner and the volumes were measured by the water
drainage method. Moreover, the thicknesses of the samples were
also measured with a vernier caliper. Then, the samples were dried
completely in a constant temperature chamber at 110 °C. The dried
samples were used to measure the coal matrix diffusion coefficient.
To compare the diffusion coefficients of the dry samples and the
water saturated samples, all the dry samples were put into distilled
water for saturation before they were tested. In this work, the
moisture treatment for each sample was processed for 48 h. This
situation is much closer to actual underground condition. Then,
the diffusion coefficients under the wetted condition were
obtained. The test gas used in this work was methane.
Additionally, broken coal samples from the three mine areas
were collected, and some were ground to 60–80 mesh for adsorp-
tion isotherm experiments and the others were measured for vit-
rinite reflectance, proximate analysis and coal composition.
Fig. 3 is a schematic illustration of the experimental setup for
measuring methane diffusion in the coal matrix, including the coal
matrix, the scaffold, the seal diffusion chamber, the thermostat, the
constant pressure device, the gas source (methane), the counting
device and the evacuating device. The functions of these devices
are described as follows:
(1) The constant pressure device is composed of a metering
pump, which is connected to the seal diffusion chamber
and the gas source through the sealing catheter. This device
is used for detecting the gas pressure in the diffusion cham-
ber at each time and judging whether the gas pressure is the
tested pressure (0.1–10 MPa). If the gas pressure is lower
than the tested pressure, the extra gas will be injected into
the seal diffusion chamber with the sealing catheter until
the gas pressure is the same as the tested pressure.
(2) The thermostatic device is made up of a water pool and is
connected to the seal diffusion chamber and is used to keep
the seal diffusion chamber at the rated temperature (20–
60 °C).
(3) The scaffold is used for supporting the coal matrix to ensure
that the coal sample is hung in the seal diffusion chamber.
(4) The counting device is composed of a data acquisition card
and a computer. The data acquisition card, which is used
to record the injection volume of gas into the seal diffusion
chamber at each time interval and send it to the computer,
is connected to the constant pressure device. According to
the tested pressure, rated temperature and the quantity of
the injected gas which is recorded by the data acquisition
 H. Xu et al. / Fuel 158 (2015) 239–247 241

Fig. 1. Location of coal samples studied in this paper.

Fig. 2. Schematic illustration and photos of prepared samples ((A) is for the front view; (B) is for the side view; and (C) is for the prepared sample).

card, the diffusion coefficient of gas in the coal matrix can be
calculated with the computer.
(5) The evacuating device is connected to the constant pressure
device by the sealing catheter, which is used for evacuating
the sealing diffusion chamber and the sealing catheter
before starting the detection of the gas pressure.
Additionally, a filter is installed between the seal diffusion
chamber and the constant pressure device, which is used to filter
out the dust in the gas flowing from the diffusion chamber to the
constant pressure device. Another filter, which is installed between
the constant pressure device and the gas source, is used to filter out
the dust in the gas from the gas source.
2.2. Experimental design
Using the experimental setup for measuring the diffusion coef-
ficient in the coal matrix mentioned in Section 2.1, the steps are as
follows:
(1) Hanging a prepared coal sample wrapped by the barbed wire
in the seal diffusion chamber (as shown in Fig. 3) to ensure
that all of the surfaces of the coal matrix contact well with
the methane in the seal diffusion chamber, and then evacu-
ating the seal diffusion chamber for approximately 40 min,
which allows the coal sample to degas sufficiently.
(2) Putting the diffuser into the thermostat (water bath), where
the water is heated up to the specified temperature (40 °C in
this work) and keeping the temperature at constant value
(40 °C).
(3) Injecting the methane into the seal diffusion chamber at the
constant gas pressure (1 MPa, 2 MPa, 3 MPa, 4 MPa, 5 MPa,
and 6 MPa in this work) by the constant pressure device (a
metering pump) and then shutting down the valve. With
the pressure sensor, which carries the pressure sensitive ele-
ment and converts the pressure to a voltage, the change of
the gas pressure in the seal diffusion chamber is detected
and recorded continuously over time (where the time step
is 1 s) until the coal is gas saturated and the gas pressure
242 H. Xu et al. / Fuel 158 (2015) 239–247

Fig. 3. Schematic illustration of the experimental setup.

does not decrease. Before measuring the pressure drop, if the
gas pressure is lower than the tested pressure in the seal dif-
fusion chamber, then the extra gas is injected into the cham-
ber through the sealing catheter until the gas pressure is the
same as the tested pressure. However, because the gas pres-
sure change is very small in the seal diffusion chamber, the
pressure sensor converts the gas pressure change into a volt-
age change, which further improves the detection accuracy
compared to directly measuring the adsorption quantity
change [5,10].
(4) Computing the diffusion coefficient of methane in the coal
matrix by the calculating device (computer and data acqui-
sition card) with the plate diffusion model and the rated
temperature, tested pressure and the methane quantity
injected into the seal diffusion chamber at every time step.
This new method of measuring the diffusion coefficient is able
to keep the space structure of the coal matrix because it uses the
same block matrix samples as the underground coal.
Furthermore, the operation of the vacuum can ensure that the
experiments are conducted under safe conditions and that no other
gases interfere. Thus, the measured diffusion coefficient is much
closer to the real value underground in the coal matrix.
The samples for isothermal adsorption were crushed to
0.18–0.25 mm (60–80 mesh), and then 100–125 g samples were
weighed for the moisture-equilibrium treatment. The
moisture-equilibrium treatment for each sample was processed
for at least four days. After these pretreatments, the coal samples
were inserted into the sample cell of the IS-100 for the adsorption
isotherm experiments. The experimental temperature and equilib-
rium pressure were 30 °C and up to 10 MPa, respectively.
Additionally, the mean vitrinite reflectance measurements and
maceral analyses (500 points) were performed on the same pol-
ished section of the coal samples using a Leitz MPV-3 photometer
microscope, according to ISO 7404.3-1994 [25] and ISO
7404.5-1994 [26], respectively. Samples were analyzed for proxi-
mate analysis, including ash yield, moisture content and volatile
matter, following Chinese national standard GB/T 212-2008 [27].
Table 1 lists the properties of the experimental samples and
Table 2 reports the measured values of vitrinite reflectance, prox-
imate analysis, coal composition and isothermal adsorption.
2.3. Calculation models
The CBM in a coal matrix system exists mainly in the adsorbed
and free gas states. The adsorbed gas obeys the Langmuir equation,

Table 1
Properties of the experimental samples from three coal mine districts.

Mine area Sample number Coal seam Coal mine Coal lithotype Size
Volume (cm3) Thickness (mm)
Yimin WMC 2 Wumuchang Dull 0.9462 3.54
Hancheng SJ 3 Shengjie Semi-bright 1.0921 3.82
XS 5 Xiangshan Semi-bright 0.6470 2.40
NG 11 Nangou Bright 1.3522 3.53
YC 11 Yuchang Bright 1.1057 3.34
Jincheng TA 3 Tang’an Bright 1.3612 3.58
 H. Xu et al. / Fuel 158 (2015) 239–247 243

Table 2
Results of vitrinite reflectance, proximate analysis, coal composition and isothermal adsorption.

Sample number Ro,ran (%) Proximate analysis (%) Coal composition (%) Isothermal adsorption
Mad Aad Vad FCad Vitrinite Inertinite Exinite Mineral VL (m3/t) PL (MPa)
WMC 0.42 12.35 19.02 51.27 29.97 76.71 4.74 1.75 16.80 3.51 3.87
SJ 1.85 0.61 11.18 11.63 76.58 77.44 11.26 0.00 11.30 22.94 2.20
XS 1.79 0.87 9.81 11.34 77.98 78.57 14.63 0.00 6.80 21.30 2.08
YC 1.90 0.43 12.54 10.38 76.65 83.68 11.52 0.00 4.80 24.05 2.24
NG 1.68 0.52 17.63 9.99 71.86 80.88 13.31 0.00 5.80 25.95 2.20
TA 3.20 0.13 12.43 9.68 77.76 70.61 23.5 0.00 5.89 43.57 3.75

Mad = moisture content (wt.%, air dry basis), Aad = ash yield (wt.%, air dry basis), and Vad = volatile matter (wt.%, air dry basis).

while the free gas obeys the ideal gas state equation. The diffusion where qt is the gas adsorption quantity at time t; q1 is the gas sat-
behavior can be represented by Fick’s Second Law, and the overall urated adsorption quantity; and qn represents the consecutive
process of desorption–diffusion–adsorption is non-equilibrium non-zero positive roots of equation tanðqÞ þ kq ¼ 0.
and pseudo steady. To quantify gas diffusion in a coal matrix, For the gas,
Crank [28] developed the unipore diffusion model based on the qt p  p1
Fick’s Second Law and assuming spherical symmetric flow. 1 ¼ ð5Þ
q1 p2  p1
Ruckenstein et al. [29] developed the bidisperse model to account
for both macropore and micropore diffusion in the matrix using a where p is the gas pressure at time t; p2 is the initial gas pressure;
simplified approach in which the adsorbent is spherical and con- and p1 is the gas saturated pressure.
tains microporous spherical particles separated by inter-particle Substituting Eq. (5) into Eq. (4),
macropores. However, the unipore model may be adequate only X1  
p  p1 2að1 þ aÞ 4q2 Dt
for some bright coals [17] and inadequate for materials with ¼ 2 2
exp  n2
heterogeneous pore structures, and thus would tend to overpredict p2  p1 n¼1 1 þ a þ a qn L
 
the diffusivities. Note that to derive the above bidisperse model, a 2að1 þ aÞ 2
4q1 Dt
linear adsorption isotherm is assumed, which does not account for  exp  ð6Þ
1 þ a þ a2 q21 L2
isothermal adsorption, although it is used to match the data well
[5,10]. Therefore, to avoid the disadvantages of the above models, Taking natural logarithms on both sides of Eq. (6),
coal matrix flakes instead of coal particles were used as the sam-   
2að1 þ aÞ 4q2 Dt
ples to be measured, and the experimental data were analyzed lnðp  p1 Þ ¼ ln 2 2
ðp2  p1 Þ  12 ð7Þ
1 þ a þ a q1 L
with the plate diffusion model in this work.
The isothermal diffusion in the thin plate coal sample (the By linear fitting of lnðp  p1 Þ and t, the diffusion coefficient D of
thickness is L) can be described by Fick’s Second Law when the methane in the coal matrix can be computed with the slope k
sample is placed in the seal diffusion chamber with a certain vol- 2
ume filled with methane [30]: kL
D¼ h i ð8Þ
2að1þaÞ
2 4q21 ln ðp2  p1 Þ
@c @ c 1þaþa2 q21
¼D 2 ð1Þ
@t @x The above calculation process was programmed with
where c is the gas concentration; x is the space coordinate; t is the Mathematica 5.0, and when the thickness of the coal sample L,
time coordinate; and D is the diffusion coefficient. the volume Vsa and the total injected gas volume Vg are imported,
At the initial moment, the diffusion chamber is under vacuum the diffusion coefficient can be obtained directly after the experi-
and the coal surface has not absorbed the gas at x = 0. The initial ment is over by the computer. However, before the coal becomes
condition of Eq. (1) is expressed as saturated with the gas, the diffusion coefficient changes with the
accumulative adsorption quantity because the gas is injected into
cðx; 0Þ ¼ 0; 0 < x < L ð2Þ
a container of certain volume under the set temperature. Thus, it
When the diffusion on the surface of the coal sample is in bal- is necessary to match all of the data in the adsorption process for
ance with the inner diffusion chamber (i.e., the rate of gas release
from the coal surface into the diffusion chamber is equal to that of
the gas from the diffusion chamber into the coal sample), the
boundary conditions of Eq. (1) are expressed as
8 @c
< @t ¼ 0;
> x ¼ 12 L; t > 0
kL @c @c
2 @t
¼ D @x ; x ¼ 0; t > 0 ð3Þ
>
: V
a ¼ aALg
where Vg is the total gas volume in the chamber; A is the largest face
area of the flake coal sample, the sample volume Vsa is equal to the
product of A and L; a is the partition coefficient, and a = 1.
Combining Eqs. (1)–(3), the solution of Eq. (1) under the initial
condition (2) and boundary conditions (3) is expressed as
X
1  
qt 2að1 þ aÞ 4q2 Dt
1 ¼ 2 2
exp  n2 ð4Þ Fig. 4. A representative differential pressure (ln) – time curve of the YC dry sample
q1 n¼1 1 þ a þ a qn L at an injection pressure of 3 MPa and a diffusion coefficient of 3.26  1010 m2/s.
244 H. Xu et al. / Fuel 158 (2015) 239–247

Table 3
Results of methane diffusion coefficients at 40 °C and different gas injection pressures.

Injection pressure (MPa) Diffusion coefficient

WMC SJ XS NG YC TA
Dry Wet Dry Wet Dry Wet Dry Wet Dry Wet Dry Wet
1 2.04 – 2.07 1.24 1.12 0.95 1.02 0.82 1.45 0.55 2.79 1.86
2 3.35 – 2.45 1.85 1.83 1.15 2.29 1.09 1.86 1.07 5.16 3.12
3 5.11 – 3.09 2.11 2.58 1.99 3.85 1.85 2.09 1.57 6.99 3.96
4 5.81 – 3.96 2.85 3.86 2.37 4.8 2.58 2.8 2.42 9.07 6.17
5 6.92 – 4.8 3.46 6.47 2.98 6.48 3.26 3.56 3.57 11.9 7.29
6 9.32 – 5.27 4.32 7.93 3.97 7.63 3.63 4.86 3.83 15.5 9.17

an average diffusion coefficient. Fig. 4 is a representation of the dif-
ferential pressure (ln) – time curves of YC dry samples at the injec-
tion pressure of 3 MPa. Table 3 reports the measurements of the
methane diffusion coefficient for 11 samples from six coal mines
under 40 °C and different gas injection pressures.
3. Results and discussion
3.1. Effect of gas pressure on the diffusion coefficient
Gas pressure is an important factor that affects the adsorption,
desorption and diffusion properties of the coal matrix. Generally
speaking, methane uptake nonlinearly increases with pressure
increase at a certain temperature. However, when the gas pressure
reaches a certain value, the adsorption quantity will not increase
with the pressure change, which means that the coal is gas satu-
rated [31]. Fig. 5 shows the diffusion coefficient curves for four typ-
ical samples under different injection pressures (the other samples
are similar). Here, different samples have different diffusion coeffi-
cients. The magnitudes of the diffusion coefficients range from
1011 to 109 m2/s. For the same sample, as the gas pressure
increases from 1 MPa to 6 MPa, a strong positive correlation exists
between the gas pressure and the diffusion coefficient. This law is
applicable to the dry samples (WMC-D, SJ-D, TA-D), saturated water
samples (SJ-S), low-rank coal sample (WMC-D), medium-rank coal
samples (SJ-D, SJ-S) and high-rank coal sample (TA-D).
Fig. 5. Diffusion coefficients in the matrix at different gas pressures (-D represents
the dry sample and -S represents the water saturated sample).
The diffusion rate of methane in the coal matrix under different
pressures depends on the diffusion mechanism. On the basis of the
methane molecular mean free path and the micropore diameter of
the coal matrix, the diffusion modes can be divided into three
types: when the pore diameter is 10 times larger than the molec-
ular mean free path, Fickian diffusion is the main diffusion mode;
when the mean free path is 10 times greater than the pore diame-
ter, the primary diffusion mode is Knudsen diffusion; when the
mean free path is somewhere in between, it is known as transi-
tional diffusion [32,33]. In addition to the above three types of
gas phase diffusion, surface diffusion and crystal diffusion are also
two diffusion modes, but because the amount of diffusion due to
the latter two modes is small, they are not usually considered in
CBM studies and will not be considered here [34]. The most promi-
nent diffusion mechanism of a gas in a porous media can be quan-
titatively determined through the relationship between the mean
free path and the micropore diameter [35] if the mean free path
obeys the Maxwell distribution, which is expressed as
KT
k ¼ pffiffiffi 2 ð9Þ
2pd p1
where K is the Boltzmann constant, 1.38  1023 J/K; T is the exper-
imental temperature in K; and d is the effective diameter of gas
molecules in nm.
With Eq. (9), the methane mean free paths and the correspond-
ing pore diameter ranges of different diffusion modes in the coal
matrix at the different equilibrium pressures at the temperature
of 40 °C (313 K, 0 °C  273 K) can be obtained (Table 4). Taking
the equilibrium pressure 1 MPa as an example (the other cases
are similar), when the pore diameter is greater than 67.3 nm, the
primary diffusion mode is Fickian diffusion, while the primary dif-
fusion mode is Kundsen diffusion if the pore diameter is smaller
than 0.67 nm. However, the effective molecular diameter of
methane is approximately 0.33 nm, the effective distance inter-
acted with the coal surface is close to 0.55 nm (the corresponding
pore diameter should be greater than 1.1 nm), and the distance of
the adsorption potential well is approximately 0.36 nm [36]. This
means that, due to the effects of the coal surface, the methane
molecules cannot move freely and exist as the adsorption phase,
resulting in surface diffusion when the range of pore diameters is
between 0.33 nm and 1.1 nm. Thus, under the experimental condi-
tions, the diffusion modes are mainly Fickian diffusion and transi-
tional diffusion because the corresponding pore diameter of
Kundsen diffusion is below 0.7 nm, which indicates that Kundsen
diffusion will not occur.

Table 4
Mean free paths and the corresponding pore size ranges of different diffusion types in the coal matrix under different equilibrium pressures at 40 °C.

Equilibrium pressure (MPa) 1 2 3 4 5 6 Range of pore diameter

Mean free path (nm) 6.73 3.37 2.24 1.68 1.35 1.12
Diffusion modes Fickian diffusion >67.3 >33.7 >22.4 >16.8 >13.5 >11.2 d > 10k
Transitional diffusion 0.67–67.3 0.34–33.7 0.22–22.4 0.17–16.8 0.14–13.5 0.11–11.2 10k > d > 0.1k
Kundsen diffusion <0.67 <0.34 <0.22 <0.17 <0.13 <0.11 d < 0.1k
 H. Xu et al. / Fuel 158 (2015) 239–247
Fig. 6. Sorption isotherms of different rank coals.
3.2. Effect of coal rank on the diffusion coefficient
The effects of the coal rank on the methane diffusion coeffi-
cient in the coal reservoirs are mainly related to the differences
of coal adsorption capacity caused by the different coal pore struc-
tures [37]. Table 2 and Fig. 6 show the isothermal adsorption
Fig. 7. Diffusion coefficients in the matrix of different rank coals.
245
results and curves of the high-rank coal sample from the
Jincheng mine area, medium-rank coal samples from the
Hancheng mine area and low-rank coal sample from the Yimin
mine area. It can be observed that the coal adsorption capacities
with different ranks are different. With an increase in coal rank,
the adsorption capacity increases. Here, the Jincheng sample with
high coal rank has the greatest Langmuir volume, while
the Langmuir volume of the Yimin sample with low coal rank is
the smallest. The different rank coals have different adsorption
capability, the higher rank coal can absorb more gas within a
shorter time than the lower rank coal, meanwhile, the higher gas
pressure is, the more gas is adsorbed.
Fig. 7 shows the relationship between methane diffusion coeffi-
cient and coal rank. On the whole, the methane diffusion coeffi-
cient exhibits a ‘‘U’’ shape with the coal rank, namely the
diffusion coefficient first decreases and then increases with the
increase in the coal rank. The methane diffusion coefficient of the
Jincheng high-rank coal sample is the largest, followed by the
Yimin coal sample with low rank. The diffusion coefficients of
the Yuchang, Shengjie, Nangou and Xiangshan samples are close
to each other and are the smallest. Gürdal and Namik [38] charac-
terized the pore structures based on coal samples with 40 different
ranks from the Zonguldak Basin in Turkey, and confirmed that the
microporous surface and the coal rank also exhibited a ‘‘U’’ shape.
246 H. Xu et al. / Fuel 158 (2015) 239–247
The surface area is at a minimum when the coal rank is located in
the range of 1.0–1.1%. The surface area high means either the pore
is concentrated in very small pore range with high frequency or the
fractal number of the pore tends high. Therefore, combining the re-
search results of Gürdal and Namik [38], it can be inferred that dif-
fusion is ‘‘structural’’ controlled transport process, and the
diffusion coefficient of the high-rank coal sample is the largest
under identical temperature and gas pressures because it has the
highest micropore content, largest specific surface, most complex
pore structures and strongest adsorption capacity. This is consis-
tent with Clarkson and Bustin [17], where the micropore content
of the high-rank coal is up to 80% while the macropore content
of the low-rank coal is higher. Nonetheless, the reason why the dif-
fusion coefficient of the low-rank coal sample is higher than that of
the middle-rank coal samples is due to the developed pore net-
work structure [17,39], which is caused by the lower compaction
degree during its evolution and could results easier and more obvi-
ous gas molecular collision with the other gas molecular or the
pore wall [40].
3.3. Effect of moisture on the diffusion coefficient
When the water molecules get into the coal rock pores, not only
can they occupy certain pores, reducing pore surface area which -
will result in reduced methane adsorption, but also the desorp-
tion performance of methane in the coal micropores will be
significantly reduced because of the high capillary pressure con-
straint [41]. Pan et al. [10] demonstrated that the gas adsorp-
tion amount of wet coal samples were clearly smaller than the
dry coal samples. Yuan et al. [5] further indicated that the exis-
tence of water would make the methane adsorption capac-
ity decrease by more than 50% in the shale samples. Fig. 8 shows
the methane diffusion experimental results for the Hancheng
dry and water saturated samples in the coal matrix (the results of
the other samples are similar). Clearly, the diffusion coefficients
of water saturated coal samples are much lower than those of
the dry samples under the same pressure due to the impacts of
Fig. 8. Diffusion coefficients in the matrix of dry and water saturated coals.
external moisture on adsorption and diffusion in the coal matrix.
This is in accord with the result of Day et al. [42], who compared
dry coal samples with wet ones. The reasons for these observations
are that the polar functional groups on the coal surface have sur-
face activities, and the wettability to water is higher. When the
water in the coal matrix increases, the coal rock expands [43] and
then the adsorption effects of methane molecules on the coal
matrix are reduced substantially, which makes it hard for the
methane molecule to be absorbed and forms a large concentration
gradient. Therefore, the methane diffusion coefficients of wet sam-
ples are small and the diffusion rate is slow, while the diffusion
coefficients of dry samples are high and the diffusion rate is fast.
It is so difficult to prepare the water saturated coal samples for
the low-rank coal in the Yimin mine district that the diffusion coef-
ficient of the dry sample cannot be compared with that of the wet
sample. The main reason for this is that the coal samples of
low-rank coal have high clay mineral content, and they are easily
destroyed when they expand because of water. It can be seen from
Table 2 that the proportion of mineral content of the Yimin sam-
ple is up to 16.8%, and most of the clay mineral content is
water-expandable, such as illite and montmorillonite which can
change the original pore volume and pore structure when they
expand, resulting in the destruction of samples [44]. However,
from the comparison of medium-rank and high-rank coal samples,
it can be inferred that the diffusion coefficients of low-rank water
saturated coal samples may also be lower than the diffusion coeffi-
cients of the corresponding dry samples.
4. Conclusions
The methane storage and diffusion behaviors in the coal matrix
are of great significance for CBM production because gas transport
in the matrix is one of the rate limiting factors for gas production.
With the new laboratory method proposed to measure the
methane diffusion coefficient in the coal matrix, the effects on
the methane diffusion coefficient of gas pressure, coal rank and
 H. Xu et al. / Fuel 158 (2015) 239–247
moisture content were analyzed and the following conclusions
were drawn:
(1) Three improvements can overcome the disadvantages of the
current measuring methods and improve the measurement
accuracy of the methane diffusion coefficient, namely (1)
keeping the intrinsic space structure of the coal matrix by
using coal matrix flakes instead of coal particles; (2) measur-
ing the voltage instead of the adsorption quantity; (3) con-
ducting the experiments under the safety environment
with no other gases interfering. The methane diffusion coef-
ficients were on the order of 1011–109 m2/s in coal
matrixes.
(2) Under the same temperature, the methane diffusion coeffi-
cients increase with increasing gas pressure regardless of
the water saturation and coal rank.
(3) The adsorption capacity increases when the coal rank
increases, with the Ro ranging from 0.42% to 3.20%, the VL
increases from 3.51 m3/t to 43.57 m3/t. While the methane
diffusion coefficients in coal matrixes exhibit a trend of first
rise and then drop (‘‘U’’ shape) with an increase in coal rank.
(4) Increased moisture can reduce the adsorption capacity of a
matrix for methane, making it difficult to form a larger con-
centration gradient, and the diffusion coefficients of satu-
rated water samples are lower than dry samples.
(5) This method can be applied widely for accurate measure-
ment of the diffusion coefficients in the coal matrix, except
that the flake coal sample cannot be polished and prepared
under the influence of composition and structure.
Acknowledgments
This work was financially supported by the Key Project of the
National Science & Technology (2011ZX05038-001), the National
Natural Science Foundation Project (41272175), the Beijing
Higher Education Young Elite Teacher Project and the
Fundamental Research Funds for the Central Universities. The
authors are grateful to the anonymous reviewers for their careful
reviews and detailed comments.
References
[1] Shi JQ, Durucan S. A bidisperse pore diffusion model for methane displacement
desorption in coal by CO2 injection. Fuel 2003;82:1219–29.
[2] Pillalamarry M, Harpalani S, Liu SM. Gas diffusion behavior of coal and its
impact on production from coalbed methane reservoirs. Int J Coal Geol
2011;86:342–8.
[3] Shi JT, Zhang L, Li YS, Yu W, He XN, Liu N, et al. Diffusion and flow mechanisms
of shale gas through matrix pores and gas production forecasting. In: Prepared
for presentation at the SPE unconventional resources conference-Canada held
in Calgary, Alberta, Canada, 5–7 November, 2013; SPE 167226.
[4] Etminan SR, Javadpour F, Mainia BB, Chen ZX. Measurement of gas storage
processes in shale and of the molecular diffusion coefficient in kerogen. Int J
Coal Geol 2014;123:10–9.
[5] Yuan WN, Pan ZJ, Li X, Yang YX, Zhao CX, Connell LD, et al. Experimental study
and modelling of methane adsorption and diffusion in shale. Fuel
2014;117:509–19.
[6] King GR. Numerical simulation of the simultaneous flow of methane and water
through dual porosity coal seams during the degasification process. PhD
dissertation, Pennsylvania State University; 1985.
[7] Staib G, Sakurovs R, Gray EMA. Dispersive diffusion of gases in coals. Part I:
Model development. Fuel 2014;143:612–9.
[8] Staib G, Sakurovs R, Gray EMA. Dispersive diffusion of gases in coals. Part II: An
assessment of previously proposed physical mechanisms of diffusion in coal.
Fuel 2014;143:620–9.
[9] Staib G, Sakurovs R, Gray EMA. A pressure and concentration dependence of
CO2 diffusion in two Australian bituminous coals. Int J Min Sci Tech 2013;116–
117:106–16.
247
[10] Pan ZJ, Connell LD, Camilleri M, Connelly L. Effects of matrix moisture on gas
diffusion and flow in coal. Fuel 2010;89:3207–17.
[11] Li YB, Xue S, Wang JF, Wang YC, Xie J. Gas diffusion in a cylindrical coal sample
– a general solution, approximation and error analyses. Int J Min Sci Tech
2014;24:69–73.
[12] Busch A, Gensterblum Y. CBM and CO2-ECBM related sorption processes in
coal: a review. Int J Coal Geol 2011;87:49–71.
[13] Smith DM, Williams FL. Diffusion models for gas production from coals:
application to methane content determination. Fuel 1984;63:251–5.
[14] Kumar AJ. Methane diffusion characteristics of Illinois coals. MS thesis.
Carbondale: Southern Illinois University; 2007.
[15] Chhajed P. Diffusion characterization of coal for enhanced coalbed methane
production. MS thesis. Carbondale: Southern Illinois University; 2008.
[16] Zhao X. An experimental study of methane diffusion in coal using transient
approach. PhD dissertation. The University of Arizona; 1991.
[17] Clarkson CR, Bustin RM. The effect of pore structure and gas pressure upon the
transport properties of coal: a laboratory and modeling study: 2. Adsorption
rate modeling. Fuel 1999;78:1345–62.
[18] Laxminarayana C, Crosdale PJ. Role of coal type and rank on methane sorption
characteristics of Bowen Basin, Australia coals. Int J Coal Geol
1999;40:309–25.
[19] Busch A, Gensterblum Y, Krooss BM, Littke R. Methane and carbon dioxide
adsorption–diffusion experiments on coal: upscaling and modeling. Int J Coal
Geol 2004;60:151–68.
[20] Wang K, Zang J, Feng YF, Wu YQ. Effects of moisture on diffusion kinetics in
Chinese coals during methane desorption. J Nat Gas Sci Eng 2014;21:1005–14.
[21] Xu LH, Jiang CL, Tian SX. Experimental study of the gas concentration
boundary condition for diffusion through the coal particle. J Nat Gas Sci Eng
2014;21:451–5.
[22] Saghafi A, Faiz M, Roberts D. CO2 storage and gas diffusivity properties of coals
from Sydney Basin, Australia. Int J Coal Geol 2006;70:240–54.
[23] Charriere D, Pokryszka Z, Behra P. Effect of pressure and temperature on
diffusion of CO2 and CH4 into coal from the Lorraine basin (France). Int J Coal
Geol 2010;81:373–80.
[24] Karaiskakis G, Gavril D. Determination of diffusion coefficients by gas
chromatography. J Chromatogr A 2004;1037:147–89.
[25] ISO 7404.3-1994. Methods for the petrographic analysis of bituminous coal
and anthracite – Part 3: Method of determining maceral group composition;
1994.
[26] ISO 7404.5-1994. Method for the petrographic analysis of bituminous coal and
anthracite – Part 5: Method of determining microscopically the reflectance of
vitrinite. MOD; 1994.
[27] GB/T 212-2008. Chinese national standard. Coal, proximate analysis of coal;
2008 [in Chinese].
[28] Crank J. The mathematics of diffusion. 2nd ed. London: Oxford University
Press; 1975.
[29] Ruckenstein E, Vaidyanathan AS, Youngquist GR. Sorption by solids with
bidisperse pore structures. Chem Eng Sci 1971;26:1305–18.
[30] Rong ZM, Vadgama P. Simple expressions for diffusion coefficient
determination of adsorption within spherical and cylindrical absorbents
using direct simulation method. J Col Int Sci 2006;303:75–9.
[31] Smith DM, Williams FL. Diffusional effects in the recovery of methane from
coalbeds. SPE J 1984;24:529–35.
[32] Yang RT. Gas separation by adsorption processes. London: Imperial College
Press; 1997 [chapter 4].
[33] Reeves S, Pekot L. Advanced reservoir modeling in desorption-controlled
reservoirs. In: SPE rocky mountain petroleum technology conference.
Keystone, Colorado, 21–23 May 2001; SPE 71090.
[34] Nie BS, He XQ, Wang EY. Mechanism and modes of gas diffusion in coal seams.
China Saf Sci J 2000;10:24–8 [in Chinese with English abstract].
[35] Do DD. Adsorption analysis: equilibria and kinetics. 1st ed. London: Imperial
College Press; 1998.
[36] Jiang WP, Cui YJ, Zhang Q, Li YH. The quantum chemical study on the coal
surface interacting with CH4 and CO2. J China Coal Soc 2006;31:237–40 [in
Chinese with English abstract].
[37] Gan H, Nandi SP, Walker Jr PL. Nature of the porosity in American coals. Fuel
1972;51:272–7.
[38] Gürdal G, Namik YM. Pore volume and surface area of the Carboniferous coals
from the Zonguldak basin (NW Turkey) and their variations with rank and
maceral composition. Int J Coal Geol 2001;48:133–44.
[39] Pan Z, Connell LD. Modelling permeability for coal reservoirs: a review of
analytical models and testing data. Int J Coal Geol 2012;92:1–44.
[40] Zhang SY, Sang SX. Adsorption–diffusion coefficient analysis of coal-bed
methane in different rank coals. Coal Geol China 2009;21:24–7.
[41] Van Bergen F, Spiers C, Floor G, Bots P. Strain development in unconfined coals
exposed to CO2, CH4 and Ar: effect of moisture. Int J Coal Geol 2009;77:43–53.
[42] Day S, Sakurovs R, Weir S. Supercritical gas sorption on moist coals. Int J Coal
Geol 2008;74:203–14.
[43] Fry R, Day S, Sakurovs R. Moisture-induced swelling of coal. Int J Coal Pre Util
2009;29:298–316.
[44] Zhang TW, Ellis GS, Ruppel SC, Milliken K, Yang RS. Effect of organic-matter
type and thermal maturity on methane adsorption in shale-gas systems. Org
Geochem 2012;47:120–31.