See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323248256

Experimental study and isotherm models of water vapor adsorption in shale

rocks

Article  in  Journal of Natural Gas Science and Engineering · February 2018

DOI: 10.1016/j.jngse.2018.02.002

CITATIONS READS

22 241

6 authors, including:

Weijun Shen Shangwen Zhou

Chinese Academy of Sciences Researh institute of petroleum exploration and development,PetroChina
35 PUBLICATIONS   274 CITATIONS    40 PUBLICATIONS   402 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fluid Flow and Transport Dynamics in Unconventional Reservoirs--Special Issue in Geofluids View project

National Natural Science Foundation of China (51206016 and 51676027) View project

All content following this page was uploaded by Weijun Shen on 07 June 2018.

The user has requested enhancement of the downloaded file.

 Journal of Natural Gas Science and Engineering 52 (2018) 484–491

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

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

Experimental study and isotherm models of water vapor adsorption in shale T

rocks
Weijun Shena,b,∗, Xizhe Lic, Xiaobing Lua,b, Wei Guoc, Shangwen Zhouc, Yujin Wanc
a
Institute of Mechanics, Chinese Academy of Sciences, Beijing 100190, China
b
School of Engineering Science, University of Chinese Academy of Sciences, Beijing 100049, China
c
PetroChina Research Institute of Petroleum Exploration and Development, Langfang 065007, China

A R T I C L E I N F O A B S T R A C T

Keywords: The understanding of water vapor adsorption and equilibrium in the low permeability matrix of gas shale is
Water adsorption crucial for predicting and optimizing gas productivity in shale gas reservoirs. In this study, water vapor ad-
Shale rocks sorption isotherms for gas shale samples from the Lower Silurian Longmaxi Formation in Southern China were
Isotherm models measured gravimetrically at two temperatures (30 °C and 50 °C) under the relative humidity ranging from 11.1%
Monolayer-multilayer adsorption
to 97.0%, and four different isotherm models were used to fit the experimental data and to analyze water vapor
Capillary condensation
adsorption on shale rocks. The experimental results showed that water vapor adsorption for shale rocks followed
Water retention
a typeⅡsigmoid shape over the humidity range. At the lower humidity range, the monolayer-multilayer ad-
sorption was the dominant process while capillary condensation and temperature effects became significant with
relative humidity increasing. As the amount of total organic carbon increases, water adsorption weakens while
calcite has an inhibitory effect. Through quantifying the average relative error (ARE), coefficient of determi-
nation (R2) and chi-square (χ2) of the isotherm models relative to data, the GAB isotherm model was identified to
be the best-fitting isotherm to describe the water adsorption process in shale rocks. Moreover, the FHH plot was
used to analyze and distinguish the states of water retention by adsorption and capillary condensation.

1. Introduction water will be adsorbed on the sites of clay minerals and free surfaces,
which results in a water film predominately formed through surface
Despite the success of deep horizontal drilling and hydraulic frac- hydration. Water adsorption by shale clays can cause swelling and
turing to yield large production from shale gas reservoirs, uncertainties damage shale matrix permeability (Chenevert, 1970a,b; Scott et al.,
associated with basic transport processes require understanding, which 2007). With the growing of water film, capillary and osmotic hydration
improves efficiency and minimizes environmental impacts. The hy- will occur (Roshan et al., 2015, 2016). Such water adsorption is par-
draulic fracturing process introduces large volumes of water into shale ticularly important in wellbore stability analysis (Chenevert, 1970a,b,
gas reservoirs. Less than half of the fracturing water is recovered as the Chenevert and Pernot, 1998; Al-Awad and Smart, 1996). And water
flow back fluid with the production of gas wells, and most of the in- retained in the shale matrix will reduce the relative permeability to gas,
jected fluid remains in the matrix or fractures of the reservoirs to in- which impedes gas production. The water vapor adsorption of gas
terfere with gas production (Engelder et al., 2014; Makhanov et al., shales is essential for understanding the alteration of shale properties
2014; Shen et al., 2016, 2017). Organic rich shale rocks are char- and predicting water blocks in gas recovery because it provides some
acterized by small pore size and extremely ultra-low permeability, important information on how water saturation of shale pores depends
which can consist of a high amount of clay minerals (Boyer et al., 2006; on the vapor pressure of water. Therefore, the determination of water
Alexabder et al., 2011; Shen et al., 2015, 2018). Because of these spe- vapor adsorption and equilibrium in shale rocks is significant for pre-
cial features, shale rocks are subject to these changes such as hydration, dicting gas productivity and for optimizing extraction conditions in
swelling and instability (Chenevert, 1970a,b; Lomba et al., 2000; Chen shale gas reservoirs.
et al., 2003; Rojas et al., 2006). Due to water adsorption and subsequent The adsorption phenomenon of water taking place in porous media
swelling, the intrinsic properties of gas shale will be affected, which is a complex process, which is associated with the characterization of
may lead to shale failure. When a solution is injected into shale rocks, porous materials (Everett, 1988; Sing, 1991; Willems et al., 1988).

∗
Corresponding author. Institute of Mechanics, Chinese Academy of Sciences, Beijing 100190, China.
E-mail address: [email protected] (W. Shen).

https://doi.org/10.1016/j.jngse.2018.02.002
Received 12 September 2017; Received in revised form 26 January 2018; Accepted 2 February 2018
Available online 17 February 2018
1875-5100/ © 2018 Elsevier B.V. All rights reserved.
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

Adsorption equilibrium is usually described by an isotherm model, Table 1

which provides the valuable information on pore structure and capacity Some measured characteristics of shale samples used in this study.
in porous materials (Kwiatkowski, 2007). There are various isotherm
Sample Depth (m) TOC (%) Surface Area (m2/g)
models developed over the years to analyze and describe the experi-
mental adsorption equilibrium data, such as the Langmuir, Freundlich, S1 1911.45 2.70 18.25
Brunauer-Emmett-Teller (BET), Dubinin–Radushkevich, Guggenheim- S2 1948.44 4.00 26.68
S3 1955.67 4.80 23.43
Anderson-de Boer (GAB), and Frenkel-Halsey-Hill (FHH) isotherms
S4 1984.21 5.40 23.46
(Allen et al., 2003; Foo and Hameed, 2010). These models are used for S5 1986.13 6.30 29.38
describing and predicting adsorption equilibrium, and the isotherm S6 1987.85 1.60 16.16
parameters. The underlying thermodynamic assumptions of these
model often provides some insight into both the surface properties and
sorption mechanism of the sorbent (Ho et al., 2002). In the last few
years, many studies have focused on methane adsorption isotherms in
shale (Rexer et al., 2013; Heller and Zoback, 2014). However, there is
little work on water vapor adsorption on shale rocks and the feasibility
of isotherm models. Although Al-Awad and Smart (1996) and Dosunmu
and Okoro (2012) successively proposed that the GAB isotherm model
could be successfully applied to fit water adsorption on shale rocks, a
detailed comparative of the isotherm models was somewhat lacking. In
addition, water vapor adsorption of gas shale is not fully understood,
and many uncertainties still exist in the process. Thus, there is a ne-
cessity to understand the water vapor adsorption process and determine
Fig. 1. Schematic diagram for apparatus used to measure water vapor adsorption iso-
adsorption isotherms for water on shale rocks so as to enhance the re-
therm (① — incubator; ② — desiccated container; SI (I = 1, 2, 3, 4, 5, 6) — jars with
covery of shale gas reservoirs. different shale rocks; ③ — plate to support the jars; ④ — saturated solution plus the salt to
In this study, the measurements of water vapor adsorption isotherm control the humidity).
for gas shale samples from the Lower Silurian Longmaxi Formation in
Southern China were performed using gravimetric measurements under
Table 2
controlled relative humidity at 30 °C and 50 °C, respectively. The water Saturated salt solutions used to control relative humidity. Relative humidity values are
vapor adsorption process and the effects of temperature on shale rocks from Greenspan (1977).
were analyzed, and the type of isotherms for the shale samples was
determined according to the International Union of Pure and Applied Salt Relative Humidity (%)

Chemistry (IUPAC) classification. Four different isotherm models in- 30 °C 50 °C

cluding Langmuir, Freundlich, GAB and FHH isotherms were applied to
fit the experimental isotherm data, and the isothermal parameters ob- LiCl 11.28 11.10
tained from these models were used to understand the water adsorption MgCl2 32.44 30.54
NaBr 56.03 50.93
equilibrium on shale rocks. In order to evaluate different isotherm
KI 67.89 64.49
models, three metrics, including the average relative error (ARE), NaCl 75.09 74.43
coefficient of determination (R2) and chi-square test (χ2), were applied. KCl 83.62 81.10
Furthermore, the FHH plot was used to determine the states of water K2SO4 97.00 95.82
absorbed by adsorption and capillary condensation.

in Fig. 1. Saturated salt solutions shown in Table 2 were used to provide

2. Materials and methods
the controlled-humidity environments (Tokunaga et al., 2003;
Greenspan, 1977) within the desiccated container. These salt solutions
2.1. Shale samples
controlled relative humidity (RH) at values ranging from 11.1% to
97.0%. The RH is the ratio of the water vapor pressure to the saturation
Shale samples used in this study were obtained from four different
vapor pressure (Tokunaga et al., 2003). The shale samples were first put
sites within the Lower Silurian Longmaxi Formation in Southern China
in weighing jars and dried overnight at 110 °C, then were weighted
at depths ranging from 1911.45 m to 1987.85 m. These samples were
after cooling to room temperature in a desiccator containing silica gel.
crushed to pass a 800 μm sieve, and the particle size range of
The jars containing shale samples were periodically removed from
500–800 μm was used for water vapor adsorption measurements. Total
constant humidity desiccated container and weighed to monitor pro-
organic carbon (TOC) analysis was obtained on the fraction < 250 μm
gression toward adsorption equilibrium. Weights of shale samples were
using a Shimadzu TOC-V. The BET surface areas of the 500–800 μm
measured daily for a specific humidity until adsorption equilibrium
fractions were measured with nitrogen adsorption at 77 K. The X-ray
when the weights became constant, after which shale samples were
diffraction analyses show that quartz, plagioclase and pyrite were
transferred to a higher humidity container. And vapor adsorption
common in all the samples, and that calcite was also detected except in
equilibrium was reached in 10–20 d, after which shale rocks were again
the sample S5. The characteristics of the shale rocks are summarized in
oven dried. The measurement at 50 °C was carried out after oven drying
Table 1.
samples from the 30 °C measurements, following the same procedure.
2.2. Adsorption isotherm measurements
3. Adsorption isotherm models
Water vapor adsorption isotherms were measured to characterize
water retention in shale rocks using the gravimetric techniques at 30 °C In order to describe the isotherm phenomenon, a number of ad-
and 50 °C. The shale samples were put in controlled-humidity en- sorption isotherm models have been developed over the years
vironments maintained at a constant temperature in an incubator (Adamson, 1990; Talu and Meunier, 1996). The isotherm models de-
(Tokunaga et al., 2003). A schematic diagram of the method used for scribe the equilibrium relationship between the adsorbed amount and
obtaining water vapor adsorption equilibrium in shale rocks was shown the pressure of the adsorbate at a constant temperature (Limousin et al.,

485
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

2007; Shang et al., 1995; Talu and Meunier, 1996). The isotherm predicts too little adsorption at low pressures and too much adsorption
parameters and their thermodynamic relations are often used to inter- at high pressures (Adamson, 1990).
pret adsorption mechanism and the properties of the adsorbent. How- The GAB isotherm can be represented as a polynomial expression as
ever, the macroscopic adsorption measurements do not yield under-
aw k 1 c − 2⎞ 1
standing on molecular level mechanisms underlying adsorption = G⎛ + 1⎞·a w2 + ⎜⎛ G
⎜ ⎟ ⎟·a w +

processes (Sposito, 1984). Such insight, accessible through spectro- q qm ⎝ cG ⎠ ⎝ cG qm ⎠ cG k G qm (6)

scopy, is beyond the scope of the present study. In this study some
common isotherm models were used to fit the experimental data and to
3.4. FHH isotherm
describe and understand the adsorption process at a macroscopic level.
Based on an assumed variation of adsorption potential with distance
3.1. Langmuir isotherm from the surface, the FHH isotherm, developed by Frenkel (1946),
Halsey (1948) and Hill (1949), is used to describe the multilayer ad-
The Langmuir adsorption isotherm developed by Langmuir (1916), sorption on the surface of the flat solid as well as capillary condensa-
was used to describe gas adsorption onto solids when the continuous tion. The FHH isotherm may be expressed as
monolayer coverage can occur in activated carbon. The Langmuir
n
model assumes that the monolayer adsorption from the gas phase oc- ⎛⎜ q ⎞⎟ = A
curs on specific homogeneous sites of the solid adsorbent, which has q
⎝ ⎠
m − ln(p / p0 ) (7)
been successfully used to describe adsorption in microporous solids
ε0
(Sing et al., 1985). It may be expressed as A=
RTx mn (8)
qm kL p
q= where n and A are the FHH constants related to adsorbent and gas at a
1 + kL p (1)
particular temperature; ε0 is the potential of the solid surface of the
where q is the adsorbed amount; qm is the maximum amount adsorbed adsorption; x mn is the film thickness at the monolayer point; T is the
at monolayer coverage; kL is the Langmuir constant (kL = 1/ pL ), pL is the temperature; R is the gas constant.
Langmuir pressure; p is the pressure of the adsorbate. For fitting the data, the FHH isotherm can be simplified into the
To determine the constant kL andqm , the Langmuir equation (1) can following form,
be expressed in its linear form as
B
qn =
1 1 1 1 −ln(p / p0 ) (9)
= +
q qm kL qm p (2)
B= Aqmn (10)
Hence the values of qm and kL were obtained from the intercept ( 1 )
qm
1 Then the FHH isotherm can be described as follows
and slope ( 1
) by plotting versus 1 .
kL qm q p
ln B 1
ln q = + ln(ln(p / p0 ))
3.2. Freundlich isotherm n n (11)

The Freundlich adsorption isotherm (Freundlich, 1906) is an ex- 4. Results and discussion
ponential equation, which describes the reversible, multilayer adsorp-
tion by the following form 4.1. Water vapor adsorption isotherm
1
q= kF p n (3)
Water vapor adsorption on the six shale samples was measured at
where kF is the Freundlich isotherm constant; n is adsorption intensity the relative humidity (p/p0) from 11.1% to 97.0%. The water adsorp-
which can be used to describe the heterogeneous adsorbent (Moon and tion isotherms at the temperatures of 30 and 50 °C were shown in Fig. 2,
Lee, 1983). and they exhibited fairly similar adsorption isotherms. As can be seen in
Freundlich adsorption parameters can be obtained by transforming Fig. 2, the isotherm can be divided into two sections. Within the range
the Freundlich equation (3) into the linear form of p/po < 0.65, the adsorbed water nearly linearly increased with p/po
1 at a slower rate. In the range of p/po > 0.65, the adsorbed water
ln q = ln kF + ln p
n (4)

3.3. GAB isotherm

Guggenheim (1966), Anderson (1946), and De-Boer (1953) pro-

posed the GAB isotherm based on the theory of multilayer gas ad-
sorption to surfaces. The GAB isotherm for water can be written as
qm cG kG a w
q=
(1 − kG a w )(1 − kG a w +cG kG a w ) (5)
where a w is water activity, a w = p / p0 ; cG and kG are the GAB constants,
which are related to the properties of the monolayer and multilayer
(Adamson, 1990). The GAB isotherm is an extension of the BET iso-
therm of Brunauer, Emmett, and Teller (Brunauer et al., 1938, 1943),
and when the GAB parameterkG = 1, the GAB and BET models are
identical. The BET model is widely used to determine the surface area of
adsorbents, typically with gases such as N2, Ar, or Kr, over lower ranges
Fig. 2. Measured water vapor adsorption isotherm of six shale samples at 30 °C and 50 °C.
of vapor pressure (0.05 < p/po < 0.3). And the BET model normally

486
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

inflection point of the isotherm indicates that capillary condensation

goes on after the completion of the monolayer-multilayer adsorption.
The analysis of surface area on shale rocks shows that shale rocks in-
clude micro-porous, meso-porous and macro-porous. It is reasonable
that water adsorption is complex process from monolayer-multilayer
adsorption to capillary condensation. Previous studies have shown that
water adsorption on shale exhibits the type Ⅱ isotherm with monolayer-
multilayer adsorption, and the type Ⅴ isotherm with capillary con-
densation (Chen et al., 2003; Dosunmu and Okoro, 2012; Hill, 1949).
The X-ray diffraction analyses indicate that some minerals (quartz,
plagioclase, pyrite and calcite) exist in shale rocks, and the interaction
of water molecules with clay surfaces will further take place with
temperatures increasing.

4.2. Modelling of water vapor adsorption isotherm

The Langmuir, Freundlich, GAB and FHH isotherm models were

used to fit the experimental data of water vapor adsorption on shale
Fig. 3. Effect of total organic carbon on water vapor adsorption isotherm. rocks at 30 °C and 50 °C. Fig. 4 shows the comparisons between the
experimental data and different models on the samples S1, S2, S3, S4,
increased with p/po at a much increasing rate. This change suggested S5 and S6. From the results of Fig. 4, it is clear that the Langmuir model
that capillary condensation had taken place in the later period. The does not describe water adsorption well because it was developed for
results in Fig. 2 also show that at the lower relative humidity (p/ systems where adsorption is dominated by surface site specific inter-
po < 0.65), the temperature effect on water adsorption is insignificant. actions and where multilayer adsorption is not accounted. Although the
As the relative humidity increases (p/po > 0.65), the temperature ef- Freundlich was widely used in the adsorption of a heterogeneous ad-
fect becomes considerable. Higher temperature associated with in- sorbent, it does not fit the data well either, particularly at the high
creased water retention, suggesting that the physicochemical process relative humidity. Although there is a little deviation at p/p0 > 0.93,
had occurred that requiring both higher temperature and humidity. the GAB model fits the data reasonably well. The GAB model is used to
Unlike conventional rocks, shale is a very fine-grained and clastic se- describe the monolayer and multilayer adsorption that fits a number of
dimentary rock, which is characterized by extremely low porosity, different materials (Adamson, 1990). Several studies on water adsorp-
ultra-low permeability and high clay content (Shen et al., 2015, 2016). tion of shale rocks indicated that the GAB model fit well up to 0.96 (Al-
In the study, the TOC values of shale samples ranged from 1.6% to Awad and Smart, 1996; Dosunmu and Okoro, 2012), and indicated a
6.3%. Plagioclase and pyrite are the minerals common to all samples, deviation at p/p0 > 0.93 with varying extents depending on the type of
and calcite was also detected except in the sample S5. Thus, the phy- shale rocks. The FHH model is commonly used to describe adsorption
sicochemical process will occur in shale at the higher temperature and combined with capillary condensation at the higher relative humidity
humidity, and consequently water retention increases. (Frenkel, 1946; Halsey, 1948; Hill, 1949), and the water content ad-
Compared with the properties of the shale samples, it seems that sorbed increases rapidly at p/p0 > 0.65, which indicates a capillary
water adsorption had positive correlation with organic carbon content condensation phenomenon. In this study, the plots using the FHH model
and temperatures. Fig. 3 shows the comparisons of water adsorption give a good fit through the entire range of relative pressure range for
with different total organic carbon and temperatures. With the in- some shale samples. From the result of Fig. 4, the GAB model fits the
creasing of total organic carbon, water content in shale rocks increases experimental data well in the whole stage of relative humidity. Thus the
at the higher relative humidity. The elevated temperature is beneficial GAB model is optimal to fit the water vapor adsorption of shale sam-
to water adsorption in shale rocks, which is not the same as methane ples, and can be used to describe and predict water vapor adsorption
adsorption by shale (Hu, 2014). It is notable that the sample S5 exhibits equilibrium in shale rocks.
a higher adsorption capacity than other shale samples, for it has the The isotherm parameters for different fitted models were de-
highest surface area (29.38 m2/g) shown in Table 1. We suspect that the termined at two different temperatures using the linear regression and
reason is that there is no calcite in sample S5. Calcite might have the isotherm parameters of different isotherm models are summarized
suppressed water adsorption in shale rocks. Some previous studies by in Table 3. The monolayer moisture content is the most important
Chenevert (1970a,b), Kahr and Madsen (1995) have reported the effect parameter in these models, which is a measure of adsorption capacity of
of clay minerals on water adsorption capacity on reservoir rocks. Kerisit the shale samples. As seen in Table 3, the monolayer moisture content
and Parker (2004), Rahaman et al. (2008) and Schultz et al. (2013) varies from the different models. The values with the Langmuir model
proposed that calcite is characterized by high surface area and weak are 12.31, 11.03, 8.87, 8.38, 17.95 and 15.92 mg/g for the shale sample
affinity for water. Consequently, shale rocks with calcite will have a S1, S2, S3, S4, S5 and S6, respectively. With the temperature increasing,
relatively low water capacity in the shale samples. However, the shale the calculated maximum Langmuir monolayer coverage capacity in-
sample S5 has the highest water vapor adsorption due to the highest creases. The monolayer moisture content calculated from the Freun-
organic carbon (OC), and the higher OC strengthens shale water vapor dlich model is very small, and it would decrease with temperatures
adsorption. Therefore, it can be seen that the water vapor adsorption on increasing. From the results of the GAB model, the maximum mono-
shale rocks is closely related to the clay minerals and organic carbon. layer moisture content at 30 °C is 10.89 mg/g (sample S4), and the
According to the IUPAC classification (Sing et al., 1985), water minimum is 3.13 mg/g (sample S3). With the change of temperatures,
vapor adsorption isotherms on shale rocks exhibit a standard typeⅡ they exhibit various trends, which are related to the shale features.
sigmoid shape. The type Ⅱ isotherm is very common in a non-porous or Although the monolayer moisture content using the FHH model cannot
a macro-porous adsorbent, where multilayer adsorption occurs at the be determined, it is shown that their values have a tendency to decrease
high relative humidity (Schrodt and Ocampo, 1984; Rouquerol et al., with increasing temperatures from the values of B and n. The decrease
2014). The isotherm is concave at the low relative humidity, then of the monolayer moisture content (qm) indicates a reduction in the
nearly linearly increasing, and finally convex to the p/p0 axis. The number of active sites as a result of physical and chemical changes by
temperatures (Dosunmu and Okoro, 2012). The prediction of qm is very

487
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

Fig. 4. Comparisons of four isotherm models fitted experimental data of different shale samples.

important since it gives information on solving the problems of shale 4.3. Estimation of the best-fitting isotherm model
failure and wellbore instability (Al-Awad and Smart, 1996; Dosunmu
and Okoro, 2012). Within recent decades, the linear regression is frequently considered
as one of the most important methods determining the best-fitting
isotherm and estimating the isotherm parameters (Ho et al., 2005).

488
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

Table 3
Isotherm parameters for fitting four different isotherm models with the experimental data.

Sample NO. Temperature (°C) Langmuir Freundlich GAB FHH

qm (mg/g) kL (10−4Pa−1) kF (10−2 mg/g) n qm (mg/g) cG kG B=A(qm)n (mg/g)n n

S1 30 12.31 8.19 8.64 1.71 4.25 47.86 0.72 223.54 3.16

50 23.25 0.62 0.23 1.09 3.56 9.55 0.84 11.07 1.76
S2 30 11.03 7.93 6.52 1.65 3.70 38.16 0.75 100.74 2.94
50 15.82 0.91 0.25 1.12 3.11 9.65 0.85 8.70 1.74
S3 30 8.87 6.72 3.13 1.37 2.57 16.84 0.85 11.91 2.06
50 72.99 0.13 0.05 0.93 4.32 3.42 0.80 7.00 1.54
S4 30 8.38 10.89 10.74 1.93 3.24 84.12 0.69 173.24 3.55
50 14.73 0.79 0.32 1.19 2.49 16.58 0.82 9.51 2.04
S5 30 17.95 6.22 8.60 1.62 5.36 61.82 0.74 519.70 3.17
50 44.25 0.38 0.21 1.05 4.46 9.65 0.85 15.42 1.71
S6 30 15.92 7.12 9.88 1.69 5.71 34.18 0.67 741.47 3.42
50 23.09 0.84 0.51 1.19 4.53 11.53 0.80 25.21 2.01

Fig. 5. Comparisons of average relative errors with different isotherm models at 30 °C (a) and 50 °C (b).

Fig. 6. Comparisons of coefficient of determination (R2) with different isotherm models at 30 °C (a) and 50 °C (b).

Fig. 7. Comparisons of chi-square test (χ2) with different isotherm models at 30 °C (a) and 50 °C (b).

However, the linearization of isotherm models may alter the error determination (R2) (Subramanyam and Das, 2014) and chi-square
structure and violate the error variance (Allen et al., 2003). Thus, in analysis (χ2) (Ho, 2004) between the calculated and experimental data,
order to determine and evaluate the best suitable isotherm model fitting were examined respectively.
the experimental data, three common error functions, including the The average relative error (ARE) is often used to compare the ex-
average relative error (ARE) (Kapoor and Yang, 1989), coefficient of perimental and calculated data, which provides a tendency to estimate

489
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

humidity, which is very common in soil science (Prost et al., 1998). The
FHH curves can be obtained by plotting log(water content (mg/g)) as a
function of log(log(1/RH)) , as shown in Fig. 8. From the result of Fig. 8,
it can be seen that the FHH plot curve provides a useful plot for dif-
ferentiating three domains of water retention in shale rocks, including
monolayer adsorption, multilayer adsorption and capillary condensa-
tion. At the lowest values of RH, the domain 1 corresponds to the
monolayer adsorption process. Water will be adsorbed on the hydro-
philic sites of shale. With relative humidity increasing, the multilayer
adsorption in the domain 2 will occur on free surfaces, which are the
walls of shale pores or surfaces that allow to expand freely (Ho, 2004;
Prost et al., 1998). The domain 3 is observed for the higher RH and
corresponds to capillary condensation. The process occurs in pores that
correspond to the shale fabric, which results in adsorbed water in-
crease. Jurinak (1963) and Pierce (1960) successively applied the FHH
model to break out the domains of water retention in kaolinite, quart
and Al2O3 samples. In the study the hypothesis is that adsorption is the
predominant process, and it mainly exists in the domain 1 and 2. At
Fig. 8. The FHH plot demonstrates the water ways held in pores.
higher relative humidities more than 0.65, capillary condensation will
become dominant in shale pores. Thus the FHH model allow us to
the experimental data. It minimizes the error distribution across the analyze water retention by adsorption and capillary condensation in
entire range. According to Eq. (12), Fig. 5 shows the distribution of the shale rocks.
average relative error (ARE) with different isotherm models. From the
result of Fig. 4, the GAB isotherm model is better to describe water 5. Conclusions
adsorption on shale rocks compared with other models.
In this work, the measurements of water vapor adsorption isotherms
1 n mcal − mexp
average relative error =
n
∑i =1 mexp
× 100% on shale rocks from the Lower Silurian Longmaxi Formation in
i (12) Southern China were carried out at 30 and 50 °C for over a relative
humidity range of 11.1%–97.0% using the gravimetric method. The
where mcal is the calculated value with the isotherm model; mexp is the
water adsorption of all shale samples exhibited a typeⅡ sigmoid shape,
experimental value; n is the number of the experimental data.
indicating the complex processes including monolayer-multilayer ad-
The correlation coefficient (R) and the coefficient of determination
sorption, and capillary condensation. At the relative humidity less than
(R2) are the proportion of the variance in the dependent variable (Ho,
0.65, adsorption is the dominant process and the temperature effect is
2004). The coefficient R2 has been used to determine the relationship
small. When the relative humidity is greater than 0.65, capillary con-
between the experimental data and the calculated data. The coefficient
densation becomes significant and the adsorbed water contents become
R2 may vary from 0 to 1. The results of different isotherm models are
considerable with temperature increasing. According to the experi-
summarized in Fig. 6 using Eq. (13). It is seen that the all isotherm
ments, it is shown that the amount of organic carbon and clay minerals
models (Langmuir, Freundlich, GAB and FHH) provide the high coef-
has a great effect on water adsorption. With the amount of total organic
ficient of determination (R2). However, the Langmuir and FHH iso-
carbon increasing, water adsorption will strengthen while calcite will
therms are not expected to describe water adsorption well according to
inhibit the water adsorption capacity in shale rocks. The GAB isotherm
the result of Fig. 4, especially high relative humidity. It implies that
model provided the best fit to the experimental data, which could de-
using the coefficient R2 to estimate the fitting degree sometimes causes
scribe and predict water vapor adsorption in shale rocks. And the FHH
errors.
plot allows us to determine the three different statues of water adsorbed
SSres by monolayer and multilayer adsorption, and capillary condensation.
R2 = 1 −
SStot (13)
Acknowledgements
where SSres is the sum of squares of residuals; SStot is the total sum of
squares (proportional to the variance of the data).
This work was supported by the National Science and Technology
The chi-square analysis (χ2) is an important statistical tool to esti-
Major Project of the Ministry of Science and Technology of China
mate the best fit of an isotherm model. The value χ2 indicates the si-
Project (NO. 50150503-12 and NO. 2016ZX05037006), and by the
milarities between the calculated and the experimental data. The
Project of PetroChina Research Institute of Petroleum Exploration &
smaller the value is, the better the agreement among them is. From
Development (NO. RIPED-LFFY-2017-JS-118). We also thank the sup-
Fig. 7, the GAB isotherm model is better than other models since the
port from the Youth Foundation of Key Laboratory for Mechanics in
value χ2 is minimum. Based on the above analysis, it can be known that
Fluid Solid Coupling Systems, Chinese Academy of Sciences.
the GAB model provides the best fits to the measured water vapor ad-
sorption isotherms on shale rocks.
References
(mexp − mcal )2
χ2 = ∑ Adamson, A.W., 1990. Physical Chemistry of Surfaces. John Wiley, New York.
mcal (14)
Al-Awad, M.N.J., Smart, B.G.D., 1996. Characterization of shale-drilling fluid interaction
mechanisms related to wellbore instability. J. King Saud Univ. 8 (2), 187–215.
Alexander, T., Baihly, J., Boyer, C., Clark, B., Waters, G., et al., 2011. Shale gas revolu-
4.4. States of water retention tion. Oilfield Rev. 23 (3), 40–55.
Allen, S.J., Gan, Q., Matthews, R., Johnson, P.A., 2003. Comparison of optimised iso-
therm models for basic dye adsorption by kudzu. Bioresour. Technol. 88 (2),
The FHH model is commonly used to describe the multilayer ad- 143–152.
sorption on the surface of the flat solid as well as capillary condensa- Anderson, R.B., 1946. Modifications of the brunauer, Emmett, and teller equation. J. Am.
tion. In general, the FHH plot allows us to analyze and determine the Chem. Soc. 68 (4), 686–691.
Boyer, C., Kieschnick, J., Suarez-Rivera, R., Lewis, R.E., Waters, G., 2006. Producing gas
states and location of water adsorbed in shale rocks verses relative

490
W. Shen et al. Journal of Natural Gas Science and Engineering 52 (2018) 484–491

from its source. Oilfield Rev. 18 (3), 36–49. workflow to estimate water loss during shut-in periods after fracturing operations. J.
Brunauer, S., 1943. The Adsorption of Gases and Vapors. Princeton University Press, Unconv. Oil Gas. Resour 7, 22–32.
Princeton. Moon, H., Lee, W.K., 1983. Intraparticle diffusion in liquid phase adsorption of phenols
Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases in multimolexular layers. with activated carbon in a finite batch adsorber. J. Colloid Interface Sci. 96 (1),
J. Am. Chem. Soc. 60 (2), 309–319. 162–170.
Chen, G.Z., Chenevert, M.E., Sharma, M.M., Yu, M.J., 2003. A study of wellbore stability Pierce, C., 1960. The Frenkel-Halsey-Hill adsorption isotherm and capillary condensation.
in shales including poroelastic, chemical, and thermal effects. J. Petrol. Sci. Eng. 38 J. Phys. Chem. 64 (9), 1184–1187.
(3), 167–176. Prost, R., Koutt, T., Benchara, A., Huard, E., 1998. State and location of water adsorbed
Chenevert, M.E., 1970a. Shale control with balanced-activity oil-continuous muds. J. on clay minerals: consequences of the hydration and swelling-shrinkage phenomena.
Petrol. Technol. 22 (10), 1309–1316. Clay Clay Miner. 46 (2), 117–131.
Chenevert, M.E., 1970b. Shale alteration by water adsorption. J. Petrol. Technol. 22 (9), Rahaman, A., Grassian, V.H., Margulis, C.J., 2008. Dynamics of water adsorption onto a
1141–1148. calcite surface as a function of relative humidity. J. Phys. Chem. 112 (6), 2109–2115.
Chenevert, M.E., Pernot, V., 1998. Control of shale swelling pressures using inhibitive Rexer, T.F.T., Benham, M.J., Aplin, A.C., Thomas, K.M., 2013. Methane adsorption on
water based muds. In: SPE-49263, SPE Annual Technical Conference and Exhibition, shale under simulated geological temperature and pressure conditions. Energy Fuels
New Orieans, Louisiana. 27 (6), 3099–3109.
De-Boer, J.H., 1953. The Dynamic Character of Adsorption. Clarendon Press, Oxford. Rojas, J.C., Clark, D.E., Greene, B., Zhang, J.G., 2006. Optimized Salinity Delivers
Dosunmu, A., Okoro, E.E., 2012. Determination of moisture adsorption isotherm of shale Improved Drilling Performance, vol. 11. pp. 1–12 AADE-06-DFHO.
from Agbada Formation using GAB model. Res. J. Eng. Sci. 1 (4), 27–33. Roshan, H., Ehsani, S., Marjo, C.E., et al., 2015. Mechanisms of water adsorption into
Engelder, T., Cathles, L., Bryndzia, L.T., 2014. The fate of residual treatment water in gas partially saturated fractured shales: an experimental study. Fuel 159, 628–637.
shale. J. Unconv. Oil Gas. Resour. 1, 33–48. Roshan, H., Andersen, M.S., Rutlidge, H., et al., 2016. Investigation of the kinetics of
Everett, D.H., 1988. Pore systems and their characteristics. In: Unger, K.K. (Ed.), water uptake into partially saturated shales. Water Resour. Res. 52 (4), 2420–2438.
Characterisation of Porous Solids. Elsevier Science, Amsterdam. Rouquerol, F., Rouquerol, F., Sing, K.S.W., 2014. Adsorption by Powders and Porous
Foo, K.Y., Hameed, B.H., 2010. Insights into the modeling of adsorption isotherm systems. Solids: Principles, Methodology and Applications. Academic Press, Oxford.
Chem. Eng. J. 156 (1), 2–10. Schrodt, J.T., Ocampo, A., 1984. Variations in the pore structure of oil shales during
Frenkel, J., 1946. Kinetic Theory of Liquids. Oxford University Press, Oxford. retorting and combustion. Fuel 63 (11), 1523–1527.
Freundlich, H.M.F., 1906. Over the adsorption in solution. J. Phys. Chem. 57, 385–471. Schultz, L.N., Andersson, M.P., Dalby, K.N., et al., 2013. High surface area calcite. J.
Greenspan, L., 1977. Humidity fixed-points of binary saturated aqueous solutions. J. Res. Cryst. Growth 371 (10), 34–38.
Natl Bur. Stand., Phys. Chem. 81A (1), 89–96. Scott, H., Patey, I.T.M., Byrne, M.T., 2007. Return permeability measurements - proceed
Guggenheim, E.A., 1966. Application of Statistical Mechanics. Clarendon Press, Oxford. with caution. In: SPE-107812, European Formation Damage Conference,
Halsey, G.D.J., 1948. Physical adsorption on non-uniform surfaces. J. Chem. Phys. 16, Scheveningen, Netherlands.
931–937. Shang, S.B., Horne, R.N., Ramery, J.H.J., 1995. Water vapor adsorption on geothermal
Heller, R., Zoback, M., 2014. Adsorption of methane and carbon dioxide on gas shale and reservoir rocks. Geothermics 24 (4), 523–540.
pure mineral samples. J. Unconv. Oil Gas. Resour. 8, 14–24. Shen, W.J., Wan, J.M., Kim, Y., et al., 2015. Porosity calculation, pore size distribution
Hill, T.L., 1949. Advances in catalysis. J. Chem. Phys. 17, 580–668. and mineral analysis within shale rocks: application of scanning electron microscopy.
Ho, Y.S., 2004. Selection of optimum sorption isotherm. Carbon 42 (10), 2115–2116. Electron. J. Geotech. Eng. 20, 11477–11490.
Ho, Y.S., Porter, J.F., Mckay, G., 2002. Equilibrium isotherm studies for the sorption of Shen, W.J., Xu, Y.M., Li, X.Z., Huang, W.G., Gu, J.R., 2016. Numerical simulation of gas
divalent metal ions onto peat: copper, nickel and lead single component systems. and water flow mechanism in hydraulically fractured shale gas reservoirs. J. Nat. Gas
Water, Air, Soil Pollut. 141 (1–4), 1–33. Sci. Eng. 35, 726–735.
Ho, Y.S., Chiu, W.T., Wang, C.C., 2005. Regression analysis for the sorption isotherms of Shen, W.J., Li, X., Xu, Y., et al., 2017. Gas flow behavior of nanoscale pores in shale gas
basic dyes on sugarcane dust. Bioresour. Technol. 96 (11), 1285–1291. reservoirs. Energies 10 (6), 11–12.
Hu, H.Y., 2014. Methane adsorption comparison of different thermal maturity kerogens in Shen, W.J., Zheng, L.G., Oldenberg, C.M., et al., 2018. Methane adsorption and diffusion
shale gas system. Chin. J. Geochem. 33, 425–430. in shale rocks-A numerical study using the Dusty Gas Model in Tough2/EOS7C-
Jurinak, J.J., 1963. Multilayer adsorption of water by kaolinite. Soil Sci. Soc. Am. J. 27 ECBM. Transport Porous Media 1–11.
(3), 269–272. Sing, K.S.W., 1991. Characterisation of porous solids: an introductory survey. In:
Kahr, G., Madsen, F.T., 1995. Determination of the cation exchange capacity and the Rodriguez-Reinoso, F. (Ed.), Characterisation of Porous Solids II. Elsevier Science,
surface area of bentonite, illite and kaolinite by methylene blue adsorption. Appl. Amsterdam.
Clay Sci. 9, 327–336. Sing, K.S.W., Everett, D.H., Haul, R.A.W., 1985. Reporting physisorption data for gas/
Kapoor, A., Yang, R.T., 1989. Correlation of equilibrium adsorption data of condensable solid systems with special reference to the determination of surface area and porosity.
vapors on porous adsorbents. Gas Separ. Purif. 3 (4), 187–192. Pure Appl. Chem. 57 (4), 603–619.
Kerisit, S., Parker, S.C., 2004. Free energy of adsorption of water and mental ions on the Sposito, G., 1984. The Surface Chemistry of Soils. Oxford University Press, New York.
{10 ̄ī 4} calcite surface. J. Am. Chem. Soc. 126 (32), 10152–10161. Subramanyam, B., Das, A., 2014. Linearised and non-linearised isotherm models opti-
Kwiatkowski, M., 2007. Numerical analysis of nitrogen adsorption isotherms on active mization analysis by error functions and statistical means. J. Environ. Health Sci. 12
carbons by an employment of the new LBET class models. J. Colloid Interface Sci. 313 (92), 1–6.
(2), 428–439. Talu, O., Meunier, F., 1996. Adsorption of associating molecules in micropores and ap-
Langmuir, I., 1916. The constitution and fundamental properties of solids and liquids. J. plication to water on carbon. AIChE J. 42 (3), 809–819.
Am. Chem. Soc. 38 (11), 2221–2295. Tokunaga, T.K., Olson, K.R., Wan, J.M., 2003. Moisture characteristics of Hanford
Limousin, G., Gaudet, J.P., Charlet, L., Szenknect, S., Barthes, V., et al., 2007. Sorption gravels: bulk, grain-surface, and intragranular components. Vadose Zone J. 2 (3),
isotherms: a review on physical bases, modeling and measurement. Appl. Geochem. 322–329.
22 (2), 249–275. Willems, H.H., Massen, C.H., Hardon, J.J., Poulis, J.A., Robens, E., 1988. Investigation of
Lomba, R.F.T., Chenevert, M.E., Sharma, M.M., 2000. The ion-selective membrane be- the micro-structure of hardened cement pastes. In: Unger, K.K. (Ed.), Characterisation
havior of native shales. J. Petrol. Sci. Eng. 25 (1–2), 9–23. of Porous Solids. Elsevier Science, Amsterdam.
Makhanov, K., Habibi, A., Dehghanpour, H., Kuru, E., 2014. Liquid uptake of gas shales: a

491

View publication stats