You are on page 1of 13

Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering


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

Gas permeability tests on core plugs from unconventional reservoir rocks T


under controlled stress: A comparison of different transient methods
Garri Gaus∗, Alexandra Amann-Hildenbrand, Bernhard M. Krooss, Reinhard Fink
Institute of Geology and Geochemistry of Petroleum and Coal, Energy and Mineral Resources Group (EMR), Lochnerstr. 4-20, RWTH Aachen University, D-52062, Aachen,
Germany

ARTICLE INFO ABSTRACT

Keywords: Accurate and routinely applicable methods to determine porosities and permeability coefficients are needed in
Gas permeability order to ensure effective hydrocarbon recovery in shale and tight sandstone plays. In this study 129 gas uptake
Pressure transient analysis measurements (“GRI method”, “inflow” experiments) were performed on core plugs from three unconventional
Klinkenberg effect reservoir lithotypes (oil shales, gas shales and tight gas sandstones) under elevated effective stress conditions.
Porosity
The results were compared to those from “flow-through” tests (standard pulse decay) under similar experimental
Unconventional reservoir
conditions, e.g. the same gas type and pore pressure range. The samples covered a porosity range from 1.3% to
12%. Equilibration times ranged from 102 s to 104 s and permeability coefficients from 10−18 to 10−21 m2. In
order to successfully determine apparent gas permeability coefficients and porosities and to reliably interpret
fluid dynamic effects from gas uptake data it is necessary to ensure a sufficiently high excess pressure drop
during the uptake tests. This can be controlled by adjustment of the reservoir to pore volume ratio and initial
differential pressure. Permeability coefficients derived from uptake tests on all six samples do not show any
systematic deviations from those obtained from flow-through measurements. Best results were achieved for a
core plug from the Lower Palaeozoic Alum Shale (Djupvik, Öland, Sweden), where Klinkenberg regressions of
inflow and flow-through differ only by 4% (slope) and 10% (y-axis intercept). Here, the gas storage capacity
ratio was ≈2.5 and excess pressure drops ranged from 0.1 to 1.2 MPa. Generally, measurement errors were
lowest when excess pressure drops during uptake were at least 0.05 MPa for all samples. Excess pressure drops of
less than 0.05 MPa resulted in coefficients of variance of single apparent gas permeabilities of ≈10%–100%,
whereas excess pressure drops of > 0.05 MPa resulted in coefficients of variance of ≈2%–15%. We show that it
is possible to adjust the initial conditions of inflow measurements such, that Klinkenberg-corrected permeability
coefficients and gas slippage factors can be readily determined. This will contribute significantly to a better
understanding of the effects of anisotropy and/or the pore structure on the transport properties of unconven-
tional lithotypes under elevated stress conditions. However, standard deviations of single apparent permeability
coefficients derived from inflow experiments are higher by up to one order of magnitude than for those obtained
from flow-through tests under all tested conditions.

1. Introduction Therefore, accurate and routinely applicable methods to determine the


pore volume and permeability are needed.
The study of transport processes in low-permeable porous rocks has Different laboratory methods, typically using cylindrical core or
received increasing attention in research over the past four decades. plugs, are used to determine permeability coefficients and porosity of
This is due to their relevance in geotechnical applications such as porous rocks. These can be subdivided into steady state and non-steady
carbon dioxide storage, nuclear & water waste disposal, and mining state tests. The latter can be designed either as flow-through or as in-
(Gensterblum et al., 2015) but also in the exploitation of conventional flow experiments. A classical “flow-through” type measurement is the
and unconventional hydrocarbon resources (gas shales and tight sand- pressure pulse decay (PPD) technique (Brace et al., 1968; Dicker and
stones). The economical exploitation of unconventional reservoirs often Smits, 1988; Jones, 1997). This method is less time consuming than a
requires horizontal drilling and hydraulic fracturing. Due to the high typical steady state flow test and, within a well-calibrated system, of
costs of these technologies an optimal reservoir assessment is essential. high accuracy and reproducibility. The experimental PPD protocol


Corresponding author.
E-mail address: garri.gaus@emr.rwth-aachen.de (G. Gaus).

https://doi.org/10.1016/j.jngse.2019.03.003
Received 24 November 2018; Received in revised form 1 March 2019; Accepted 2 March 2019
Available online 07 March 2019
1875-5100/ © 2019 Elsevier B.V. All rights reserved.
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

consists of the creation of a pressure difference across the faces of a the logarithm of the pressure difference between the reservoirs
cylindrical sample plug separating two closed reservoirs (upstream and (log( P) ) versus time (t). The correction factor f1 takes into account the
downstream reservoir) of known volume, and continuous monitoring of time delay resulting from the filling of the pore system of the sample
the successive pressure transient. Permeability coefficients are obtained (inflow effect). It depends on the gas storage capacity ratios of the
by solving the diffusion equation, which is derived by combining the sample pore volume and the upstream and downstream compartment:
differential form of Darcy's law and the continuity equation (Jones, 2
1997; Ghanizadeh et al., 2014). An “inflow” type of experiment ana- f1 = 1
a+ b (2)
lyses the gas uptake rate into a sample, which is placed between two or
one calibrated reservoir(s) (plane-parallel sheet, Crank, 1975; Cui and Here, 1 is the first solution of tan( ) = 2
(a+b)
, with a and b being
ab
Bustin, 2010; Yang et al., 2016; Peng and Loucks, 2016). This approach gas storage ratios of the sample and upstream or downstream com-
is comparable to the “Gas Research Institute” (GRI) method usually Vpore Vpore
partments (a= V b= V ). If the reservoir volumes are much larger
applied to crushed rocks (cuttings; assuming spherical particle geo- top bot
than the sample pore volume, then the contribution of f1 becomes
metry) when cores or plugs are not available. Here the pressure drop
negligible.
observed after initial pressurisation is recorded and the permeability
Alternatively, k gas can be determined by opening the downstream
coefficient is obtained from the combined endmember solution for the
reservoir or keeping it at constant pressure by appropriate measures. As
diffusion in a plane sheet and effective gas diffusivity equation as given
a result, the compressive storage of the downstream reservoir is effec-
in Crank (1975). As a by-product, the porosity can be determined from
tively infinite and equation (1) becomes:
these experiments. A detailed review of methods to measure low per-
meability rocks is given in Sander et al. (2017). s1 µL
k gas =
Publications on inflow tests with rock cuttings are relatively abun-
dant (e.g. Luffel and Guidry, 1992; Cui and Bustin, 2009; Heller et al.,
f1APmean ( )1
Vtop (3)
2014; Ghanizadeh et al., 2015; Peng and Loucks, 2016; Blount et al., 2
2017; Fisher et al., 2017; Fink et al., 2017b). Usage of crushed rock where f1 a
1
as b 0
samples may offer advantages over intact core measurements with re-
spect to the time needed to perform a measurement and the costs. 2.2. Inflow method – modified GRI method
However, so far it was not clearly demonstrated that crushed rock
permeability is reproducible using different apparatuses and evalua- Cui et al. (2009) describe an approach to determine the gas uptake
tions (Fisher et al., 2017). Additionally, fluid dynamic effects (slip flow) rate into rock samples placed in a defined volume by monitoring the
usually are not accounted for and measurements are performed at re- decline of a gas pressure pulse. The permeability coefficients are com-
latively small pore pressures and ambient conditions (Peng and Loucks, puted under the assumption of a defined sample size and geometry (e.g.
2016). Comparisons with conventional core analysis methods often spherical particles with identical diameters). In the present study, the
reveal large gaps, which is not surprising due to severe differences in gas phase was allowed to enter cylindrical samples from the two faces
boundary conditions. Only a few authors have addressed the inflow while the sides of the cylinder were sealed. The gas uptake was inter-
tests on intact cores (Cui and Bustin, 2010; Yang et al., 2016; Peng and preted in terms of (pressure) diffusion into a plane sheet from a stirred
Loucks, 2016) and a detailed methodological study was either not solution of limited volume (Crank, 1975; Yang et al., 2016).
possible due to the limited amount of experimental data or because of
non-comparable measurement conditions (e.g. gas type, pore pressure
k gas =
()
L 2
2
µcg s2
range). Accordingly, the primary aim of this study was to compare the 2
1 (4)
results obtained by these two different types of experiments (flow-
through vs. inflow). For flow-through and inflow measurements, the Here L is the sample length [m], the porosity, cg the gas com-
same apparatus and gas type was used, and experiments were per- pressibility [Pa−1] and 1 the first positive root of tan( n) = K c n . K c
formed successively on the same sample and within the same pressure is the ratio of the system void volume (Vref , Vtop & Vbot ) and the pore
V
(effective stress) regime to yield best comparability. Key-parameters for volume (K c = Vvoid ). The parameter s2 is the slope of the plot of the
pore
comparison are the apparent and Klinkenberg-corrected permeability logarithm of the fractional gas uptake ( Fr = 1
(K c + 1)*( max t)
) versus
coefficients. max min
time ( log(Fr) ). Here, max is the average initial gas density in the void
t
2. Theory and evaluation procedures volume (Vref , Vtop & Vbot ) after expansionfrom Vref to Vtop and Vbot, min
denotes the initial gas density in Vtop , Vbot and Vpore prior to expansion
2.1. Transient flow-through test and t is the density of the gas at experimental time (t). A detailed
explanation of the balance equations is given in Cui et al. (2009) and
Brace et al. (1968) describe a transient procedure to determine gas Yang et al. (2016).
permeability coefficients on low-permeable rocks. This method was
later modified by Dicker and Smits (1988) and Jones (1997). A pressure 2.3. Klinkenberg effect
pulse is imposed across a rock sample placed between two reservoirs of
known volume. The evaluation is based on the mass balance equation Deviations of the apparent permeability coefficient from the in-
and Darcy's law (Dicker and Smits, 1988; Jones, 1997; Ghanizadeh trinsic permeability coefficient in fine-grained porous media are mainly
et al., 2014). The pressure transient is evaluated to obtain the apparent due to slip flow (“Klinkenberg effect”). Slip flow occurs when molecu-
gas permeability coefficient: le–wall collisions become more frequent as compared to molecule-
molecule collisions. This is the case when the mean free path length of
s1 µL
k gas = the gas molecules (distance travelled between two intermolecular col-
f1APmean ( 1
Vtop
+
1
Vbot ) (1) lisions) is in the order of the pore diameters as will typically occur at
low gas pressures and/or gas flow in very narrow (sub-μm) transport
Here, μ is the dynamic gas viscosity [Pa s], L and A are the sample pores (Kundt and Warburg, 1875; Klinkenberg, 1941). Therefore, the
length and cross-sectional area [m], Pmean is the mean pressure of the phenomenon becomes increasingly relevant for low-permeable rocks
Ptop + P bot
gas [Pa] (Pmean = 2
) and Vtop and Vbot are the upstream and with gas transport in micro- and meso-pores (Ghanizadeh et al., 2014).
downstream compartments [m³]. The value of is the slope of the plot of Under these conditions the gas can no longer be considered as a

225
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Table 1
Porosity, specific pore volume, total organic carbon content (TOC), vitrinite reflectance (VRr) and mineral composition (XRD meas.) for the samples used in this
study.
Sample Porositya Specific pore volumea TOC VRr Q+F Clays Carbonates

[%] [cm³g−1] [wt.%] [%] [wt.%]

Alum Shale1 11.5 0.052 8.2 0.5 57 33 0.2


Bossier Shale2 7.7 0.031 2.2 2.2 30 56 10
Eagle Ford Shale3 11.2 0.050 4.5 0.9 26 8 62
Carbonaceous Shale, Garau Formation4 2.0 0.007 0.9 1.1 12 3 82
Kimmeridge Shale5 7.0 0.045 45 0.5 8 40 10
Westphalian D sandstone 2.3 0.009 – – 70 27 3

Geochemical and mineralogical data from 1Ghanizadeh et al. (2014), 2Fink et al. (2017b), 3Gasparik et al. (2013), 4Shabani et al. (2018), 5Li et al. (2017). [Q + F]:
quartz + feldspar.
a
Measured by He-pycnometry on unconfined sample plugs.

Fig. 1. Thermostated (30 °C ± 0.3 °C) experimental set-up for specific pore volume (Vsp ) and apparent gas permeability measurements under controlled effective
stress. Permeability experiments can be performed with the inflow or flow-through method. Enclosed volumes Vtop , Vbot and Vref , (volume between valves V1,V2 and
V3) are calibrated by helium expansion.

uniform fluid phase in terms of continuum fluid mechanics (with de- the pressure and temperature readings using the equation of state
fined viscosity or density). The “apparent” permeability (k gas ) can be ( = RT ) within the GERG-2004 software (Kunz et al., 2007).
PM

related to the “intrinsic” or “liquid phase” gas permeability (k ) ac- In addition to the pore volume measurements under controlled
cording to the Klinkenberg equation: stress, He-pycnometry measurements were performed on the un-
confined plugs. The apparatus and procedure have been described in
4Cl̄ b detail previously (Ghanizadeh et al., 2014). Briefly, the skeletal
k gas = k 1+ =k 1+
r Pmean (5) (“solid”) volume of the rock samples is determined by helium expansion
from a calibrated reference volume into a sample cell of known volume
Here, b is the gas slippage factor [MPa], Pmean the mean pore pressure
under isothermal conditions. The mass balance equation yields the
[Pa], C the dimensionless Adzumi constant (0.9) and r the mean effective
following expression for the skeletal volume Vsk :
transport pore radius [m]. The additional velocity component (slip factor b)
is thus related to the mean free path of the gas molecules (l̄ ) and the ref eq
transport pore radius. The intrinsic permeability coefficient k is determined Vsk = Vsc Vref
from the y-axis intercept of the regression line of k gas versus Pmean −1.
eq sc (7)
Here, Vref and Vsc are the volumes of the reference cell and the
2.4. Stressed and unstressed pore volume measurements sample cell, respectively, ref and sc are the gas densities in the cor-
responding cells prior to expansion, and eq is the gas density after
The pore volumes (Vpore ) of the dry cylindrical plugs under defined pressure equilibration between the two volumes.
axial and confining stress conditions were determined using the fol- In both procedures, the porosity and the specific pore volume Vsp
lowing mass balance equation: are obtained by:
Vsk
ref eq =1
Vpore = Vref Vdead Vb (8)
eq x (6)
Vb Vsk Vp
Here, ref and x are the gas densities [kg/m³] in the reference volume Vsp = = =
m m (1 ) (9)
Vref and the dead volume of the set-up (Vdead = Vtop + Vbot ) before pres-
sk

sure equilibration and eq is the gas density in the entire system (Vref + With sk denoting the skeletal density. In both cases a separate de-
Vdead ) after pressure equilibration. Helium densities were calculated from termination of the bulk sample volume Vb is required.

226
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 2. Log-linear plot of normalized gas uptake (Fr) from inflow measurements for different mean pore pressures and constant confining pressures: (a) Bossier Shale
at 40 MPa confining pressure, (b) Eagle Ford Shale at 10 MPa confining pressure, (c) Alum Shale at 30 MPa confining pressure, (d) Garau Shale at 10 MPa confining
pressure, (e) Kimmeridge Shale at 40 MPa confining pressure and (f) Westphalian D sandstone at 30 MPa confining pressure.

3. Samples performed in a triaxial flow cell, which enabled loading of the cylind-
rical sample plugs with axial and confining pressure of up to 40 MPa
In this study, three unconventional lithotypes were investigated: oil (Fig. 1). In the experimental set-up the samples are placed between
shale samples from the Eagle Ford and Kimmeridge formations, potential stainless steel pistons. Porous steel discs between the pistons and the
gas shale samples from the Lower Palaeozoic carbonaceous Alum Shale sample are used to distribute the fluid over the sample surface. A
(Djupvik, Öland, Sweden; Schovsbo and Buchardt, 1994), Lower Cre- double-layered sleeve system (inner sleeve: 0.15 mm lead foil; 0.1 mm
taceous Garau (Iran), the Upper Jurassic Bossier formations (USA) and an outer sleeve aluminium tube) separates the sample from the confining
Upper Carboniferous tight gas sandstone from Germany (Pennsylvanian, pressure chamber and prevents by-passing.
Westphalian D). Porosity data, total organic carbon (TOC) contents, vi- As shown in Fig. 1, Vref (reference volume) is the volume between
trinite reflectance and mineral compositions of the samples are listed in valves V1, V2 and V3, whereas Vtop and Vbot are the volumes between V3
Table 1. Cylindrical cores with a diameter of 38 mm were drilled per- and the sample surface and V2 and the sample surface, respectively.
pendicular to bedding from each sample set, except for the Bossier shale Prior to the experiments, Vref , Vtop and Vbot were calibrated by helium
(parallel to bedding). All sample plugs were dried at 105 °C until weight expansion at experimental pressure conditions. For this purpose, a
constancy was reached, but for at least 24 h. stainless-steel cylinder was installed in place of the rock sample.
In the present study permeabilities and pore volumes were mea-
sured at 30 °C while unloading the samples from 40 to 10 MPa confining
4. Experimental set-up and testing procedure pressure. After loading of the sample to the respective confining pres-
sure, a leak test was performed with helium at the maximum pore
Single-phase gas permeability and pore volume measurements were

227
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 3. Specific pore volume and apparent permeability from inflow measurements for different pore pressure ranges and constant confining pressure: (a) Bossier
Shale at 40 MPa confining pressure, (b) Eagle Ford Shale at 10 MPa confining pressure, (c) Alum Shale at 30 MPa confining pressure, (d) Garau Shale at 10 MPa
confining pressure, (e) Kimmeridge Shale at 40 MPa confining pressure and (f) Westphalian D sandstone at 30 MPa confining pressure.

pressure. At a given confining pressure, alternating cycles of flow- confining pressures. Confining pressures were then decreased stepwise
through and inflow experiments were run with successively increasing from 40 to 10 MPa. Experimental results reported here are therefore
pore pressures to reduce stress effects. Alternatively, flow-through tests representative of the first unloading cycle. The first loading path is
with constant downstream pressure and inflow measurements were known to be dominated by initial compaction of structures, which re-
performed successively. sult from drilling, unloading to surface conditions, plug preparation and
storage (McKernan et al., 2014).
Selected experimental results are shown in Fig. 2 in terms of the
5. Results and discussion (K c + 1)*( max t)
normalized pressure decline curves (Fr; 1 ) as a func-
max min
tion of time. All data show a reasonably linear relationship between
5.1. Inflow measurements
log10(Fr) and time in the late stage, which is a prerequisite to calculate
permeability coefficients. The slope of log10(Fr) vs. time increases with
Inflow permeability measurements utilize data that are recorded
pore pressure. Scattering of the data increases as the pressures approach
during porosity determination at experimentally induced overburden
the equilibrium values, particularly for samples with small pore vo-
conditions. Therefore, they provide a simple means for determining the
lumes (Fig. 2 d & f, Table 1, porosity of ∼2%), because the limit of the
stress-dependence of gas permeability in low-permeable rocks. In this
resolution of the pressure transducer is reached. For the Westphalian D
study, 129 apparent permeability coefficients were obtained from in-
sample (Fig. 2f) more than 50% of the total gas uptake occurred during
flow measurements at increasing mean fluid pressures (up to 18 MPa)
the first few seconds. This phenomenon was not observed for other
and defined initial pressure pulses (up to 5 MPa). Equilibration times
samples.
ranged from 102 to 104 s. The flow tests were started at the maximum

228
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 4. Uncertainties of apparent permeability coefficients by pore volume determination for (a) Alum, (b) Bossier, (c) Kimmeridge and (d) Garau shale. Red error
bars describe deviation from measured apparent permeability when pore volumes are changed to an equivalent of ± 0.5% porosity (abs.).

The apparent gas permeability coefficients and specific pore vo- experimental data. For the Garau sample (Fig. 4d), however, the de-
lumes obtained from these measurements are shown in Fig. 3. Apparent termination of the pore volume yielded somewhat ambigious results:
permeability coefficients range from 1600 nD to 1 nD and decrease here the unstressed pore volume values were sometimes smaller than
significantly with increasing pore pressure. Specific pore volumes range those measured under controlled stress. This is certainly due to the
from 0.005 to 0.045 cm3g−1 (equivalent to 0.013 and 0.1 fractional small pore volume of the sample (i.e. a large K c ) and the fact that the
porosity) and generally increase slightly with increasing pore pressure accuracy limit of the experimental apparatus was reached.
and decreasing confining pressure. Similar observations are reported in Plotting single apparent permeability coefficients from each ex-
numerous publications and are attributed to a combination of fluid- pansion over the linear segment of log10(Fr), gives a measure of the
dynamic and poro-mechanic effects (e.g. Fink et al., 2017a). For two quality of each measurement in terms of the absolute standard devia-
samples the experimental data deviate from the general trend at higher tion (SD). This is exemplary shown for selected inflow experiments at
fluid pressures. Here both, apparent permeability and the specific pore given Δp, Pmean, and Pconf conditions in Fig. 5. Ideally, there should not
volume, increase for the Garau shale at pore pressures > 5 MPa (Pconf be a variation of apparent permeability coefficients for the chosen in-
= 10 MPa) and for the Kimmeridge clay at pore pressures > 10 MPa tervals, but standard deviations are one to two orders of magnitude
(Pconf = 40 MPa, Fig. 3d and e). This is a clear indication of poro-elastic lower than the mean apparent permeability coefficients. Standard de-
pore widening, superimposing the fluid dynamic effect of gas slippage viations were analyzed for all expansion series and are shown as error
(Fink et al., 2017a). Therefore, these values had to be excluded when bars in Figs. 7 and 10.
determining the Klinkenberg-corrected permeability coefficients. Normalizing the standard deviation to the corresponding mean
A major source of error when determining apparent permeability
SD
permeability, yields the coefficient of variance ( mean permeability *100 ). Low
coefficients from inflow measurements lies in the determination of coefficients of variance indicate high precision of permeability coeffi-
absolute porosity and its stress-dependence. Fig. 4 shows apparent cients. In this study the coefficient increases in the order
permeability coefficients as a function of the mean fluid pressure using Alum ≤ Kimmeridge ≤ Bossier ≪ Eagle Ford ≈ Garau ≈ Westphalian
values of the porosity measured under stress (Fig. 3) and at ambient D. This can be either related to the gas storage capacity ratio and/or the
conditions (unstressed porosity from He pycnometry). The error bars differential pressure, both of which control the excess pressure drop
indicate deviations assuming an uncertainty of ± 0.5% (abs.) for por- (theoretical system pressure after expansion from reference cell minus
osity values, which is a reasonable assumption. Fig. 4 clearly shows that equilibrium pressure). In order to investigate this in more detail, Fig. 6
an overestimation of the pore volume (unstressed conditions), will lead shows the specific variance coefficient as function of the excess pressure
to an underestimation of the gas storage capacity ratio and, in con- drop for each uptake series measured in this study. It is shown, that
sequence, of the apparent permeability coefficient. Largest deviations below a critical uptake pressure of 0.05 MPa the specific variance
were 52%, 70%, 176% and 63% for the Alum, Bossier, Kimmeridge and coefficient increases dramatically to values of up to 100%, indicating a
Garau samples. A dependence of these deviations on the gas storage large permeability uncertainty. Above total pressure uptakes of >
capacity ratio, K c , is not apparent for this data set. It should be noted 0.05 MPa the specific variance coefficients are only 2%–15%. From
that these are purely theoretical error/uncertainty considerations that Fig. 6 it is clear that measurements performed on the Westphalian D
do not necessarily imply the same level of uncertainty for the sandstone, Garau and Eagle Ford shale all fall into the region of high

229
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 5. Single apparent permeability coefficients over fractional intervals of the total linear interval for (a) Bossier, (b) Eagle Ford, (c) Alum (d) Garau, (e)
Kimmeridge and (f) Westphalian D samples (standard deviation: SD).

uncertainty, where apparent permeabilities may be grossly over or trends that provide information on the transport pore diameters in
underestimated. A comparision to other measuring techniques should (axial) flow direction.
therefore be done with caution for these samples. Fig. 7 shows Klinkenberg plots obtained for samples derived from
It is often implied in the literature that large K c ( Vvoid ) ratios should data of Fig. 2. A summary of all inflow measurements performed is
Vpore
be preferred for inflow measurements because the dependence of K c on given in Table 2, whereas all Klinkenberg plots are visualized in
Fr decreases (e.g. Cui et al., 2009; Yang et al., 2016). However, it was Appendix A. Linear relationships were obtained for all samples, in-
found in this study that a large K c is counterproductive as larger un- dicating gas transport in the slip-flow regime. Klinkenberg-corrected
certainties of pressure recordings and larger influences of temperature permeabilities ranged from 345 nD to 1.8 nD. At pressures > 5 MPa
fluctuations will strongly affect the data accuracy and reproducibility and > 10 MPa for the Garau and Kimmeridge samples (Fig. 7d and e),
(Fig. 6). respectively, pore-widening is more dominant than gas slippage,
The Klinkenberg formulation (Klinkenberg, 1941) is commonly leading to a distinct apparent permeability minimum.
applied to investigate the dependence of flow regimes on apparent
permeability coefficients in tight rocks. A linear relationship of ap- 5.2. Flow-through measurements
parent permeability coefficients and the reciprocal mean pore pressure
is consistent with the slip flow theory. In contrast to uptake experiments Permeability coefficients measured by flow-through (pulse-decay
with crushed or powdered samples the inflow measurements on the with two closed reservoirs or constant downstream volume) were per-
confined cylindrical rock samples yield realistic apparent permeability formed under the same experimental conditions (gas type, pore

230
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

for flow-through measurements may be negligible or not, as shown in


Fig. 9 for the Eagle Ford and Garau shales. Here, apparent gas perme-
abilities were calculated using the zero pore volume assumption (f1 = 1,
equivalent to the solution given by Brace et al., 1968) as well as cor-
rections based on pore volumes of the unstressed samples (equivalent to
the solution given by Jones, 1997) and the pore volumes under stress.
The unstressed pore volume correction uses pore volumes obtained
from He-pycnometry, whereas stressed pore volume correction is per-
formed by using a fitting function and extrapolating the pore volume
measurements from inflow data to the effective stress at which the
particular measurement was performed (Fig. 2). Error bars (in red)
show the permeability fluctuations upon stressed porosity variations
of ± 0.5% (abs.).
The unstressed pore volume of the Eagle Ford sample (∼3.4 cm³)
exceeded the volume of the two reservoirs (∼3 cm³ on average), which
corresponds to a volumetric error of 18%. The calculation is shown
Fig. 6. Coefficient of variance versus excess pressure drop for all inflow mea- exemplary in equations (10)–(13):
surements performed in this study. [N]: number of inflow measurements per-
formed for the respective sample. Vpore 3.4 cm³
a= = = 1.40
Vtop 2.42 cm³ (10)
Table 2
Vpore 3.4 cm³
Summary of inflow derived Klinkenberg-corrected permeabilities (k ), gas b= = = 0.93
slippage factors (b), average specific pore volumes (Avg. Vsp ) and coefficients of Vbot 3.65 cm³ (11)
variance (Avg. CV) for different confining pressures.
(1.40 + 0.93) 1
Sample Confining b Avg. Vsp Avg. CV
tan( 1) = ;( 1 = 1.38)
k 2
1 1.40*0.93 (12)
pressure

1.38²
[MPa] [nD] [MPa] [cm³g−1] [%] f1 = = 0.82
1.40 + 0.93 (13)
Bossier Shale 40 74.9 4.5 0.025 11
30 96.5 5.5 0.025 10 This resulted in an apparent permeability increase of 418 nD after
Eagle Ford Shale 40 116.4 5.6 0.044 26 correction at lowest mean gas pressures of 0.5 MPa (Fig. 9a). In con-
30 204.0 2.9 0.044 44 trast, the Garau shale has a much smaller pore volume of 0.3 cm³. The
20 234.4 3.2 0.044 33 corresponding volume error is 3% and led to apparent permeability
10 373.9 1.9 0.045 25
Alum Shale 40 20.0 6.6 0.043 4
underestimations of 3 nD at lowest mean gas pressures of 1.2 MPa
30 17.1 6.8 0.044 6 (Fig. 9b). Using the pore volumes of the stressed samples for volume
20 17.9 6.6 0.044 5 error calculations, the calculated volume errors decreased to 16% and
Carbonaceous shale, 40 23.8 3.6 0.007 33 1% for the Eagle Ford and Garau samples, respectively. A summary of
Garau Formation 30 43.3 1.7 0.007 37
average volume errors for each confining pressure step is provided in
20 56.9 1.7 0.007 35
10 97.8 1.5 0.007 27 Table 3 together with Klinkenberg-corrected permeabilities and gas
Kimmeridge Clay 40 1.8 2.0 0.036 8 slippage factors. Additionally, all Klinkenberg plots are displayed in
Westphalian D sandstone 40 80.7 2.3 0.005 50 Appendix A. Varying the value for the stressed pore volume by an
30 109.1 1.8 0.005 64 equivalent of ± 0.5% (abs.) porosity for each sample resulted in a vo-
20 135.9 1.8 0.006 45
10 218.8 1.3 0.006 77
lume error change < 1.5% for both samples, thus a negligible apparent
permeability change. A similar uncertainty analysis was performed for
flow-through measurements with one closed reservoir. However, as in
pressure and confining pressure range) as the inflow experiments. Flow- this case the downstream reservoir is infinitely large, only one gas
through (mostly with two reservoirs) is a common and established storage ratio needs to be accounted for.
method to measure permeability coefficients of low-permeable rocks
(e.g. Brace et al., 1968; Dicker & Smits, 1988; Jones, 1997; Ghanizadeh 5.3. Comparison of permeability coefficients from inflow and flow-through
et al., 2014).
Similar to Fig. 5, apparent permeability coefficients from each Fig. 10 shows the comparison of flow-through and inflow-derived
measurement step plotted over dimensonless time give a measure of permeabilities for all data sets using pore volumes determined under
quality and can be used to determine the precision in terms of the ab- stress as input parameters. Standard deviations as determined in Fig. 5
solute standard deviation (SD). This is shown here for the Eagle Ford for inflow measurements were included for each data point. Generally,
(Fig. 8a) and Garau shale (Fig. 8b). Generally, the standard deviations inflow permeabilities do not show any systematic deviations from flow-
are two to three orders of magnitude lower than average apparent through derived permeabilities. However, sample-specific trends were
permeability coefficients. Coefficients of variance are 0.93% and 0.36% observed: For the Bossier Shale flow-through permeabilities are con-
for the Garau and Eagle Ford shale, respectively. The Garau shale re- sistently two orders of magnitude higher than inflow permeability
vealed the largest variance coefficient of all flow-through measure- coefficients (Fig. 10a). For the Eagle Ford, Alum, Garau and Kimmer-
ments with two closed reservoirs, which is still much better than for all idge samples results are equal or similar (Fig. 10b–e). For the West-
inflow measurements. In fact, standard deviations are very small and phalian D sample, the permeability coefficients determined from inflow
negligible in flow-through experiments and will therefore not be further experiments are larger by 52% (slope) and 74% (y-axis intercept) than
reported explicitly in the following sections. those determined from flow-through tests (Fig. 10f).
Another source of error already discussed in the previous chapter is Since the specific variance coefficients (uncertainty) for the Eagle
the volume error resulting from the gas-storage capacity ratio. Ford, Garau and Westphalian D samples are very large, these data sets
Depending on the gas-storage capacity ratios, the permeability change cannot be readily compared with the results from flow-through

231
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 7. Selected Klinkenberg plots from inflow measurements: (a) Bossier at 40 MPa confining pressure, (b) Eagle Ford at 10 MPa confining pressure, (c) Alum at
30 MPa confining pressure, (d) Garau at 10 MPa confining pressure, (e) Kimmerdige at 40 MPa confining pressure and (f) Westphalian D at 30 MPa confining
pressure. Red error bars indicate standard deviations calculated as shown in Fig. 5.

Fig. 8. Single apparent permeability coefficients over dimensionless time for flow-through measurements for (a) Eagle Ford shale and (b) Garau shale.

232
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 9. Errors in permeability determination based on gas storage capacity ratios for (a) Eagle Ford shale and (b) Garau shale.

Table 3 transport can be divided into two phases: i) flow through the sample
Summary of flow-through derived Klinkenberg-corrected permeabilities (k ), from upstream to downstream reservoir until pressure equilibration is
gas slippage factors (b), and average volume errors of gas storage capacities achieved in the reservoirs and ii) a slow subsequent pressure drop until
( Avg f1 ) for different confining pressures. finally reaching a minimum equilbration pressure. Comparison of the
Sample Confining k b Avg. f1 normalized pressure decay curves (Fig. 11b) showed that pressure
pressure equilibration during inflow experiments is consistently longer. It is very
likely, that this sample has a strong pore network anisotropy, e.g.
[MPa] [nD] [MPa] [-]
parallel and perpendicular bedding flow, or a dual porosity system, e.g.
Bossier Shale 40 10985.3 0.2 0.88 fracture and matrix porosity. The Bossier sample was drilled parallel to
30 11647.6 0.2 bedding, but bedding-parallel fractures were observed, too. This implies
Eagle Ford Shale 40 216.3 2.4 0.85 that the two methods might have different effective transport pore ac-
30 215.3 2.6 0.84
cessibilities. It is reasonable to assume that flow-through favours more
20 220.8 2.6 0.84
10 424.7 2.1 0.84 easily accessible and therefore larger pores (e.g. micro-fractures). As for
Alum Shale 40 15.1 6.4 0.86 inflow, infilling of larger pores occurs first for the same reasons, but as
30 15.4 7.7 0.85 soon as infilling of all accessible larger pores has taken place, the gas
20 16.1 7.2 0.85 will enter smaller pores at a lower rate yielding lower apparent per-
10 16.4 8.0 –
meability coefficients. This is well reflected in the gas slippage factors
Carbonaceous shale, Garau 40 29.1 2.0 0.98
Formation 30 34.9 2.0 0.98 obtained from inflow (b = 4.0 MPa, Table 2) and flow-through
20 52.7 1.8 0.98 (b = 0.2 MPa, Table 3) for this sample set at 40 MPa confining pressure.
10 111.2 1.3 0.98 As thoroughly discussed by Klinkenberg (1941), Randolph et al. (1984)
Kimmeridge Clay 40 1.5 1.7 0.93
or Letham et al. (2015) the gas slippage factor is inversely proportional
Westphalian D sandstone 40 21.5 4.1 0.96
30 27.4 3.4 0.96 to the average transport pore diameter (Klinkenberg, 1941) or average
20 33.1 3.17 0.95 transport slit width (Randolph et al., 1984). Thus, the combination of
10 58.8 2.5 0.95 both methods shows, whether a dual porosity system or anisotropy are
present or not and to what extent they may affect the permeability
coefficient. Cui and Bustin (2010) reported that permeabilities differed
measurements. Thus, the only meaningful interpretation at this point is by two orders of magnitude when flow-through and inflow (ISPP: In-
that the results of the (more reliable) flow-through measurements lie Situ Permeability and Porosity) was measured on intact cores or-
within the range of variance of the inflow measurements for all three ientated parallel to the bedding. They argued that flow-through prefers
samples. A better agreement is likely to be achievable if measuring high permeability zones, since it is based on differential pressures be-
procedures for inflow tests can be improved. Averaging of apparent tween the up- and downstream reservoirs, and solely gives information
permeability coefficients for the Garau shale, as reported in Fig. 5, gave on largest pathways (e.g. fractures) in longitudinal direction (c-axis) of
approximately the same permeability coefficients as obtained with the samples. Since shales likely develop hierarchical and anisotropic
flow-through. Interestingly, the previously interpreted poro-elastic structures and hence permeability pathways, inflow most likely gives
permeability increase (Fig. 7d) was apparent in the flow-through data, information on permeability coefficients of smaller scaled pathways.
too, thus confirming that not only fluid-dynamic but also poro-me- This agrees with the findings made for the Bossier sample in this study
chanic effects can be analyzed and quantified from inflow data. As for and explains the difference of permeability coefficients from inflow and
the Kimmeridge and Alum samples, based on the linear regressions flow-through although the variance coefficient for inflow was relatively
(Fig. 10c and e), the slopes and y-axis intercepts of the Klinkenberg small (Fig. 6). It should be noted that when performing flow-through
trends differed by 29% and 14% as well as 4% and 10%, respectively, measurements with two closed reservoirs, flow perpendicular and
indicating that by improving the measuring procedure for inflow tests, parallel to bedding are competing throughout the measurement, prob-
permeability coefficients obtained from both methods will show better ably leading to an underestimation of the maximum flow-through
agreement. controlled permeability coefficients for the Bossier sample.
The large deviation observed for the Bossier sample (Fig. 10a) by Yang et al. (2016) compared three flow-through and three inflow
approximately two orders of magnitude could be explained after de- (OCPPD: One-Chamber Pressure Pulse Decay) measurements at equal
tailed data analysis. Klinkenberg-corrected permeabilities were 91 nD differential and mean pore pressures as well as confining pressures.
from inflow measurements and 11000 nD from flow-through mea- Main results were that inflow-derived permeability coefficients are
surements at 40 MPa confining pressure. Flow-through raw data (with smaller than flow-through derived permeability coefficients by ∼10%
two closed reservoirs) are shown in Fig. 11a and it is evident that gas for three apparent permeabilities. Data from our study do not support

233
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

Fig. 10. Comparison of Klinkenberg plots from inflow and flow-through measurements: (a) Bossier Shale at 40 MPa confining pressure, (b) Eagle Ford Shale at
10 MPa confining pressure, (c) Alum Shale at 30 MPa confining pressure, (d) Garau Shale at 10 MPa confining pressure, (e) Kimmerdige Shale at 40 MPa confining
pressure and (f) Westphalian D sandstone at 30 MPa confining pressure.

Fig. 11. (a) Raw data for flow-through measurement with two closed reservoirs and (b) comparison of normalized pressure decay for inflow and flow-through for the
Bossier sample.

234
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

their conclusions as apparent inflow permeabilities were not con- Acknowledgements


sistently smaller than flow-through permeabilities. It should be noted
that Yang et al. (2016) did not account for stress-dependent porosity We thank our anonymous reviewers for their constructive comments
changes, but rather used unstressed (constant) porosity as input, and to improve this paper.
thus, permeability evaluations for both methods are questionable if
stress-dependent change is relevant for their sample set. Zheng et al. Appendix A-C. Supplementary data
(2015) and Fink et al. (2017a) have shown that the largest decrease in
specific pore volume occurs upon loading the sample at low effective Supplementary data to this article can be found online at https://
stress (up to ∼ 5 MPa). Using unstressed pore volumes instead of doi.org/10.1016/j.jngse.2019.03.003.
stressed pore volumes will always result in smaller apparent perme-
ability coefficients because gas storage capacity ratios (K c ) will be un- References
derestimated as shown in Fig. 4 for the sample set of this study.
Peng and Loucks (2016) compared Klinkenberg-corrected helium Brace, W.F., Walsh, J.B., Frangos, W.T., 1968. Permeability of granite under high pres-
permeabilities from inflow (MGE: Modified Gas Expansion) with ap- sure. J. Geophys. Res. 73, 2225–2236 John Wiley & Sons Ltd.
Blount, A., Croft, T., Driskill, B., Tepper, B., 2017. Lessons learned in Permian core
parent argon permeabilities from flow-through (at 14 MPa for an Eagle analysis: comparison between retort, GRI, and routine methodologies. Petrophysics
Ford and 7 MPa for a Barnett sample) at varying confining pressures on 58 (5), 517–527.
intact cores. They found inflow-derived permeabilities 2–4 times larger Cui, X., Bustin, A.M.M., Bustin, R.M., 2009. Measurements of Gas Permeability and
Diffusivity of Tight Reservoir Rocks. Different Approaches and Their Applications.
than flow-through derived permeability coefficients. It is possible that John Wiley & Sons Ltd, Geofluids, pp. 208–223.
this difference is due to the use of different gases (inflow – He, flow- Cui, X., Bustin, R.M., 2010. A new method to simultaneously measure in-situ permeability
through – Ar), even though fluid dynamic effects were accounted for. and porosity under reservoir conditions: implications for characterization of un-
conventional gas reservoirs. In: Petroleum Conference, Calgary, Canada, 19-21
Fink et al. (2017b) showed that Klinkenberg-corrected permeability October.
coefficients of fine-grained rocks decrease in the order of Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press, pp. 56–59.
He ≫ Ar ≥ N2 ≥ CH4 ≥ CO2. Therefore, methodological comparisons Dicker, A.I., Smits, R.M., 1988. A practical approach for determining permeability from
laboratory pressure-pulse decay measurements. In: SPE Paper 17578, Presented at the
of permeability data measured with different gases are problematic.
1988 International Meeting on Petroleum Engineering, Tianjin, China, pp. 285–292.
Fink, R., Krooss, B.M., Gensterblum, Y., Amann, A., 2017a. Apparent permeability of gas
6. Conclusions shales – superposition of fluid-dynamic and poro-elastic effects. Fuel 199, 532–550.
Fink, R., Krooss, B.M., Amann, A., 2017b. Stress-dependence of Porosity and Permeability
of the Upper Jurassic Bossier Shale: an Experimental Study, vol. 454 Geological
The primary objective of this study was to analyse whether inflow- Society London Special publications 1.
derived permeability coefficients give reasonable results by comparing Fisher, Q., Lorinczi, P., Grattoni, C., Rybalcenko, K., Crook, A.J., Allshorn, S., Burns, A.D.,
them to other commonly used and well-established methods (flow- Shafagh, I., 2017. Laboratory characterization of the porosity and permeability of gas
shales using the crushed shale method: Insights from experiments and numerical
through, standard pulse decay). This was achieved by performing a modelling. Mar. Petrol. Geol. 86, 95–110 Elsevier.
series of measurements on six different unconventional lithotypes from Gasparik, M., Bertier, P., Gensterblum, Y., Ghanizadeh, A., Krooss, B.M., Littke, R., 2013.
the Eagle Ford, Kimmeridge, Alum, Westphalian D, Garau and Bossier Geological controls on the methane storage capacity in organic-rich shales. Int. J.
Coal Geol. 123, 34–51.
Formations. Ghanizadeh, A., Gasparik, M., Amann-Hildenbrand, A., Gensterblum, Y., Krooss, B.M.,
Results indicate that in order to successfully use inflow data it is 2014. Experimental study of fluid transport processes in the matrix system of the
necessary to ensure a sufficient excess pressure drop during gas uptake European organic-rich shales: I. Scandinavian Alum Shale. Mar. Petrol. Geol. 51,
79–99 Elsevier.
by the sample. This can be controlled by adjustment of the gas storage Ghanizadeh, A., Bhowmik, S., Haeri-Ardakani, O., Sanei, H., Clarkson, C.R., 2015. A
capacity ratio (sample size/quantity and measuring cell volumes), the comparison of shale permeability coefficients derived using multiple non-steady-state
initial differential pressure and sensitivity of the pressure transducers. measurement techniques: examples from the Duvernay Formation, Alberta (Canada).
Fuel 140, 371–387.
For inflow measurements, the excess pressure drop should be at least
Gensterblum, Y., Ghanizadeh, A., Cuss, R.J., Amann-Hildenbrand, A., Krooss, B.M.,
0.05 MPa. Comparison of coefficients of variance of the two methods Clarkson, C.R., Harrington, J.F., Zoback, M.D., 2015. Gas transport and storage ca-
used in this study showed that flow-through accuracy is generally better pacity in shale gas reservoirs – a review. Part A: transport processes. Journal of un-
by one order of magnitude at optimal conditions for inflow. Thus, in- conventional oil and gas resources 12, 87–122 Elsevier.
Heller, R., Vermylen, J., Zoback, M., 2014. Experimental investigation of matrix per-
flow measurements should not be used to replace flow-through mea- meability of gas shales. AAPG (Am. Assoc. Pet. Geol.) Bull. 98 (5), 975–995.
surements. However, comparing both methods can give information on Jones, S.C., 1997. A technique for faster pulse-decay permeability measurements in tight
whether flow is controlled by anisotropy, either induced by bedding rocks. SPE Form. Eval. 12, 19–25.
Klinkenberg, L.J., 1941. The Permeability of Porous Media to Liquids and Gases. Drilling
orientation or matrix/fracture porosity. and Production Practics. American Petroleum Institute, pp. 200–213.
Based on a large data set, it was possible for the first time to show Kundt, A., Warburg, E., 1875. Ueber Reibung und Wärmeleitung verdünnter Gase.
that inflow measurements on intact cores can be used to interpret fluid Annalen der Physik und Chemie, vol. 7. VEB Johann Ambrosius Barth, pp. 337–365.
Kunz, O., Klimeck, R., Wagner, W., Jaeschke, M., 2007. The GERG-2004 Wide-Range
dynamic effects and, therefore can be reliably used to obtain Equation of State for Natural Gases and Other Mixtures. Publishing House of the
Klinkenberg-corrected permeabilities. The question arises whether Association of German Engineers.
formulations used for crushed rock permeability are useful since basic Letham, E.A., Bustin, R.M., 2015. Klinkenberg Gas Slippage Measurements as a Means for
Shale Pore Structure Chracterization. Geofluids. John Wiley & Sons Ltd., pp.
fluid dynamic effects and stress-dependence, are absent or not detect-
264–278.
able. Reasons for being absent or not detectable might be wrong initial Li, X., Krooss, B.M., 2017. Influence of grain size and moisture content on the high-
boundary conditions chosen for the experiments and evaluations or pressure methane sorption capacity of Kimmeridge Clay. Energy Fuels 31,
11548–11557.
experimental conditions that do not cover these effects. The results
Luffel, D.L., Guidry, F.K., 1992. New core analysis methods for measuring rock properties
presented in this study may now be used to investigate whether inflow of devonian shale. J. Pet. Technol. 44 (11), 1184–1190 Paper SPE-20571-PA, Society
on intact cores at ambient conditions can be utilized - as a spin-off of of Petroleum Engineers.
porosity measurements - to give fast and cheap estimations of perme- McKernan, R.E., et al., 2014. SPE 14UNCV-167762-MS influence of effective pressure on
mudstone matrix permeability: implications for shale gas production. In: SPE/EAGE
ability on rock volumes that can be considered representative in terms European Unconventional Conference and Exhibition Held Vienna, pp. 13. https://
of matrix pore structure and rock fabric. doi.org/10.2118/167762-MS.
It is important to note that the mathematical formulation used in Peng, S., Loucks, B., 2016. Permeability measurements in mudrocks using gas-expansion
methods on plug and crushed-rock samples. Mar. Petrol. Geol. 73, 299–310 Elsevier.
this study requires radial sealing of the intact cores, which typically is Randolph, P.L., Soeder, D.J., Chowdiah, P., 1984. Porosity and permeability of tight
the case in triaxial flow cells with overburden but is not in pycnometry sands. In: SPE Paper 12836, Presented at the SPE/DOE/GRI Unconventional Gas
vessels at ambient conditions. Therefore, materials capable of sealing Recovery Symposium, 13-15 May, Pittburgh, USA, pp. 57–62.
Sander, R., Pan, Z., Connell, L.D., 2017. Laboratory measurment of low permeability
the rocks while not invading the pore space need to be utilized.

235
G. Gaus, et al. Journal of Natural Gas Science and Engineering 65 (2019) 224–236

unconventional gas reservoir rocks: a review of experimental methods. J. Nat. Gas organic-rich carboaceous rocks, Lurestan provincs, southwest Iran. Int. J. Coal Geol.
Sci. Eng. 37, 248–279. 186, 51–64.
Schovsbo, N., Buchardt, B., 1994. Djupvik - 1 and Djupvik – 2 Completion Report, BMFT- Yang, Z., Sang, Q., Dong, M., Zhang, S., Li, Y., Gong, H., 2016. A modified pressure-pulse
Project 032 66686 B: Pre-westphalian Source Rocks in Northern Europe. . https:// decay method for determining permeabilities of tight reservoir cores. J. Nat. Gas Sci.
www.researchgate.net/profile/Niels_Schovsbo/publication/299597690_Djupvik-1_ Eng. 1–11 Elsevier.
and_Djupvik-2_completion_report/links/57020c5a08ae1408e15eada0/Djupvik-1- Zheng, J., Zheng, L., Liu, H.H., Ju, Y., 2015. Relationships between permeability, porosity
and-Djupvik-2-completion-report.pdf. and effective stress for low-permeability sedimentary rock. Int. J. Rock Mech. Min.
Shabani, M., Moallemi, S.A., Krooss, B.M., Amann-Hildenbrand, A., Zamani-Pozveh, Z., Sci. 78, 304–318 Elsevier.
Ghalavand, H., Littke, R., 2018. Methane sorption and storage characteristics of

236

You might also like