Professional Documents
Culture Documents
Spe 131772 Pa
Spe 131772 Pa
Non-Clay Non-Clay
Grain Volume Grain Volume
Volume Volume
Volume
Connected Pore Connected Pore
Volume
Total Void
Total Void
Free Gas
Volume
Volume
Free Oil, Gas, and Water Free Oil, Gas, and Water
Vv0 must be corrected at the beginning and at the end of the pressure If the changes in void volume are not taken into account, the
step, as described in Eqs. 9 and 10 (Menon 1968): isotherm will be in error and will not be usable in engineering
calculations. An example of isotherm data with and without void-
ˆ
n1 M volume corrections is shown in Fig. 2. The aforementioned void-
Vv1 = Vv 0 − , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9) volume considerations that are necessary to accurately determine
s adsorbed-gas volumes have significant implications on the “live,”
in-situ shale pore volume available for free-gas storage. There-
ˆ
n2 M fore, the effective-porosity/gas-saturation product (Sge) derived
Vv 2 = Vv 0 − . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
s from a total pore volume that is determined under static labora-
tory conditions does not reflect live-reservoir conditions. The
reservoir total pore volume is not only consumed by water and
Accordingly, over the course of the isotherm analysis, the void
oil but also by the adsorbed gas. It is for this reason we propose
volume is further reduced for each subsequent pressure step. In
that the calculated free-gas pore volume requires a correction for
practice, it is often more convenient to determine a so-called
the adsorbed-gas fraction present under reservoir-temperature and
Gibbs isotherm in terms of number of moles of adsorbed gas, as
-pressure conditions.
in Eq. 11:
Additional evidence to support the need for the correction lies
⎛ pr 1 pr 2 ⎞ ⎛ ps1 ps 2 ⎞ in visual investigation of the complex rock fabric typical of fine-
n2′ = n1′ + Vr ⎜ − + Vv 0 ⎜ − . grained, organic-rich shales. It has recently been reported that
⎝ zr 1 RTr 1 zr 2 RTr 2 ⎠⎟ ⎝ z s1 RTs1 z s 2 RTs 2 ⎠⎟ much of the gas-storage capacity within shale is predominantly
. . . . . . . . . . . . . . . . . . . . . . . (11) associated with the organic fraction of the rock matrix (Loucks
et al. 2009; Wang and Reed 2009; Sondergeld et al. 2010).
The Gibbs isotherm can then be converted to volumes using an Therefore, the free-gas volumes measured by pycnometry using
equation of state and can be adjusted for the void volume using a nonadsorbent gas such as helium must be corrected for the
the Gibbs correction factor f /s: presence of adsorbed gas within the organics. Fig. 3 shows a 2D
FIB/SEM gas-shale image supporting the preceding observations.
Ga' In the image, the matrix is represented by the gray color, with the
Ga = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (12)
dark- and light-gray regions being the organic (kerogen) and the
1− f inorganic constituents (e.g., clay, silica, and feldspar), respectively,
s
whereas the pores are shown in black. Clearly, most of the pores
are significantly small, typically less than 100 nm, and are almost
exclusively found within the kerogen.
Using advanced imaging technologies, we have created 3D
shale segmentations showing kerogen and pore networks that
contain up to 600 such FIB/SEM images. A similar investigation
has previously been conducted by Tomutsa et al. (2007) for the sub-
micron petrophysical analysis of chalk. A typical kerogen network
of our study is shown in Fig. 4a. The network consists of large
interconnected pockets of kerogen, which make up an estimated
7.7% (volume) of the segment. Fig. 4b shows the superimposed
kerogen (bounded with yellow) and pore (on red) networks of the
same shale segment. As can be seen again from these images,
nearly all of the porosity within the system is associated with the
kerogen network. The image-computed porosity for this segmented
volume is 2.5%. A 3D network overlay showing the comparison
of the kerogen/pore networks and the gray-scale image is shown
in Fig. 3c. Orange denotes kerogen, whereas dark orange denotes
pores. These volumes are produced through gray-scale threshold-
Fig. 3—2D FIB/SEM image showing porosity and kerogen within ing. The setting of thresholds is subjective, and care should be
shale. Black depicts pores; dark gray is kerogen; light gray is taken, although the medium-sized pores are best resolved (Curtis
matrix (clay and silica). et al. 2010).
Fig. 4—3D FIB/SEM segmentations of 300 separate 2D SEM scans. Sample size is 4 m high, 5 m wide, and 2.5 m deep. The
Method for Shale Gas-in-Place Calculations The volume occupied by the sorbed gas must be accounted for after
In this paper, a new petrophysical model is proposed that alters the correction for water saturation. This is because of the porosity
the previous concept of effective porosity discussed earlier. Fig. 1 basis of the water-saturation measurement. Appendix A shows the
shows a comparison between the new model and the old model. derivation of the fundamental equation (Eq. 13). The porosity-
The new model emphasizes two distinct conceptual changes with reduction effects can be shown to vary. In deep reservoirs where
respect to the previous model. First, there exists a dependency of there is not much sorbed-gas content, the correction to the free gas
the connected pore space on the organics (Wang and Reed 2009; in place can be less than 5%. In other reservoirs with low porosity
Loucks et al. 2009; Sondergeld et al. 2010). Second, there is a and a high sorbed-gas content, the reduction could be up to 40%
dependency on the free pore space by the inclusion of a sorbed of the free gas in place or greater. Appendix B shows an example
phase. Fig. 5 shows a simple diagram of the current methodology calculation and the impact on the calculated gas in place.
used to determine gas in place vs. the proposed methodology. For
simplicity, the water and oil volumes are not considered in the Sorbed-Phase Density. In order to calculate the volume occupied
diagram. The simple illustration taken in context with the new by the sorbed phase, the density of adsorbed gas in the organic
information from the FIB/SEM images and segmentations visu- pores (s) must be known a priori. Measurement of the sorbed-
ally shows the errors that result by assuming that the sorbed gas phase density is not a trivial matter, however; the local density for
takes up no volume. methane is expected to vary across the pore and to be different
In order to properly account for the total and free gas in place, from its average bulk density because of the added interactions
the volume occupied by the adsorbed gas must be determined and between methane and the organic pore walls. Furthermore, in gas
subtracted from the free-gas calculation. We therefore propose that shales where the reservoir temperature is significantly greater than
the standard calculations used to calculate free-gas storage capacity the critical temperature of the natural gas, it is difficult to study
(i.e., Eq. 4) be modified as follows: phase transitions and determine if the adsorbate is in the form of
a liquid or vapor.
There have been several suggestions, which have been docu-
32.0368 ⎡ (1 − Sw ) 1.318 × 10 −6 M
ˆ ⎛ p ⎞⎤
Gf = ⎢ − ⎜ GsL ⎥ mented in physical-chemistry literature, to determine the density
Bg ⎣ b s ⎝ p + pL ⎟⎠ ⎦ of the adsorbed phase on solid surfaces. Dubinin (1960) suggested
that the adsorbate density is related to the van der Waals co-volume
. . . . . . . . . . . . . . . . . . . . . . . (13) constant b. Independently, Haydel and Kobayashi (1967) used an
experimental method and found the density values for methane
Old Methodology New Methodology and propane to be nearly equal to the van der Waals co-volume
constant. Later, it was argued that the sorbed-phase density is
Void Space Void Space equivalent to the liquid density (Menon 1968) and to the critical
Measured by Measured by density (Tsai et al. 1985) of the sorbed fluid. Ozawa et al. (1976)
Porosity Porosity considered the adsorbed phase as a superheated liquid with a den-
Measurement Measurement sity dependent upon the thermal expansion of the liquid. Recently
+ Ming et al. (2003) compared all of the previously stated methods
+ Sorbed Mass to a Langmuir-Freundlich adsorption model and found that there
Sorbed Mass Measured by exists a temperature dependence in the sorbed-phase density, but
Measured by Adsorption the value approaches those proposed by Dubinin (1960). These
Adsorption Experiment studies, although fundamentally important to our understanding of
Experiment -
gas adsorption in shale, do not show a clear and accurate path to
Free Gas prediction of the adsorbed-phase density of shale gas and, hence,
Volume Taken to estimation of shale gas in place. All the previous studies were
=Total GIP up by Sorbed
conducted at low pressure and temperature values (at most, 104oF
Gas
=Total GIP and 1,000 psi) or using an inorganic wall (silica gel, as in the
case of Haydel and Kobayashi). In this research, density values
Fig. 5—Comparison of the old and new methodologies in pre- in the first layer within the range 0.28–0.3 g/cc were found using
dicting shale gas-in-place; for simplicity, oil and water volumes molecular modeling and simulation. It is found that the density
are not shown. is sensitive to changes in temperature, pressure, and pore size.
Typically, at 3,043 psi and 176oF, the adsorbed-phase density of MD Simulation of Methane Adsorption in Organic Slit-Pores.
methane is 0.35 g/cc for a pore width of 2.3 nm when the Langmuir For the molecular-level investigation, methane is considered at
single-layer theoretical model is used. some supercritical conditions under thermodynamic equilibrium in
In this study, we used a numerical molecular-modeling approach 3D periodic orthorhombic pore geometry consisting of upper and
to determine the adsorbed-phase density from the first principles lower pore walls made of graphene (carbon) layers (see Fig. 6).
of Newtonian mechanics. Molecular modeling and simulation is a For comparison, we take into account three slit-like pores with a
form of computer simulation that enables us to study thermody- pore width equal to H = 3.9, 2.3, and 1.1 nm. Pore width is an
namic and transport properties of many-particle systems, in which important length scale of the study, and it is defined as the dis-
particles (atoms and molecules that make up the natural gas and tance between the innermost graphene planes. The pores maintain
pore walls) with initially known and instantaneously predicted posi- fixed dimension in the y-coordinate: Ly = 3.93 nm; however, the
tions and momenta are allowed to interact for a period of time, giv- dimension is changed in the x-coordinate such that both pores host
ing a view of the motion of the particles as trajectories in space and approximately the same number of methane molecules (400–450)
time. Numerical integration of Newton’s equations of motion make during the simulations. Hence, the dimensions in this direction are
up the core of the simulation; therefore, special algorithms (e.g., equal to Lx = 4.26 nm and to Lx = 7.67 nm for the large and small
Verlet, Leap Frog, or Beeman algorithms) have been developed and pores, respectively. While the effect of pore pressure is investi-
are commonly used during a molecular simulation study. Readers gated, the number of particles is increased from 400 to 600 within a
who are interested in details of molecular simulation are encour- 2.3-nm pore. Typically, the approximate parallel computation time
aged to consult textbooks by Frenkel and Smit (2002) and Allen was around 1,500–3,000 seconds, and 100 to 144 cores have been
and Tildesley (2002) for deterministic and stochastic treatment used during the simulations.
of the numerical-integration process. Molecular simulations have A united-atom carbon-centered Lennard-Jones potential (based
made enormous strides in recent years and are gradually becoming on the optimized potentials for liquid simulations, OPLS-UA
a commonly used tool in science and engineering (de Pablo and force field) is used as the model of a methane molecule; Table 2
Escobedo 2002). Today, molecular simulations are being widely shows the energy and distance parameters used for fluid/fluid
used to construct virtual experiments in cases where controlled and solid/solid interactions (Nath et al. 1998). The methane/solid
laboratory measurements are difficult, if not impossible, to perform. and methane/methane interactions are of the Lennard-Jones type
There exists an exhaustive literature studying the equilibrium ther- (Frenkel and Smit 2002). A Lorentz-Berthelot mixing rule [ij =
modynamics of fluids using molecular simulation involving phase (ii + jj)/2 and εij = (εiiεjj)1/2] is set up in order to describe solid/
change of bulk fluids (Harris and Yung 1995), characterization of fluid (i.e., pore-wall methane) interactions. Here, ij and εij are
porous materials using gas adsorption (Aukett et al. 1992; Sweat- the Lennard-Jones parameters accounting for interactions between
man and Quirke 2001), and multicomponent gas separation (Ghoufi a molecular site of methane species and a carbon atom of the
et al. 2009). In our case, two sets of molecular-dynamics (MD) organic wall. Lennard-Jones interactions were cut off at 4.1ii for
simulations are used to estimate and analyze the adsorbed-phase the 3.9-nm pore width, and at 3ii for the 1.1- and 2.3-nm pore
density under typical reservoir pressure and temperature conditions: width, respectively. The van der Waals interactions were cut off at
(a) runs involving bulk-phase (i.e., in the absence of pore walls) 3ii for all three systems.
methane-density measurements using a fixed number of methane The MD simulation package, DL-POLY (version 2.20), was
molecules at fixed pressure and temperature and (b) runs involving used to perform the simulations (Todorov and Smith 2008). Meth-
measurements of density of methane at the same temperature but ane bulk-density computations are made considering isobaric and
confined to a pore with organic (graphite) walls. We will consider isothermic conditions with a constant number of atoms, constant
deviations in the methane-density profile along the pore width of pressure, and constant temperature (NPT) at three constant tem-
the second set of runs relative to the uniform density value of the peratures (176, 212, and 266°F) using the Nosé- Hoover thermostat
bulk as an impact of the pore-wall effects. and barostat ensemble (Frenkel and Smit 2002). The relaxation
time used for the Nosé-Hoover thermostat is optimized to 15
pico-seconds (ps). In the case of simulations involving methane in
TABLE 2—LENNARD-JONES POTENTIAL PARAMETERS carbon slit-pores, initially the fluid system was equilibrated at a
FOR METHANE AND CARBON constant temperature, with a canonical ensemble [constant number
of atoms, constant volume, and constant temperature (i.e., NVT)].
Atom (nm) /kB (K) The equilibrium is assumed to be achieved if no drift in time was
Carbon 0.34 28 observed in the time-independent quantity, such as the total energy
of the system. The relaxation time used for the Nosé-Hoover ther-
Methane 0.373 147.9
mostat is optimized to 13, 15, and 20 ps for the pore widths of 1.1,
32 0.0
0.20
ρCH4 (g/cc)
0.0 3.8 7.6 11.4 15.2 19.
0
0.163
24
0.15 0.133 0.126 0.124
16
0.10
7.24E+00
8 5.18E+00 0.05
0 0.00
0.0 3.8 7.6 11.4 15.2 19.0 0.0 3.8 7.6 11.4 15.2 19.0
2.3, and 3.9 nm, respectively. The Nosé-Hoover algorithm needs to gas density of methane 0.124 g/cc at the center of the pore. The
be used to control the system temperature while the real canonical pore pressure is 3,043 psia, which is a quantity predicted using the
ensemble computations are carried out for the density computa- free-gas molecules and the volume at the center of the pore. The
tions (Frenkel and Smit 2002). The average quantities within observed density profile shows that the assumption of Langmuir
1 nanosecond (ns) intervals after the equilibrium is reached are adsorption theory with monolayer is reasonable to describe the
taken for the analysis. Thermodynamic equilibrium is considered equilibrium adsorption dynamics of methane in the organic pores.
to be reached when the fluctuations in the density profile along the The inset graph on the upper right shows the equivalent Langmuir
time are uniform. Typically, the total run time for a simulation with monolayer plot where the pore half-length has been divided by the
the pore was 1.5 ns and the timestep used was 0.003 ps. adsorbed and free gas. The adsorbed-layer density is estimated as
At the end of each simulation run, number density Number (num- the total of the discrete densities associated with adsorption:
ber of molecules per volume) for methane across the pore space
was computed at every z = 0.2 Å interval (for the continuous s = 0.281 + (0.163–0.124) + (0.133–0.124)
density profile) and at every Lz = 3.8 Å (for the discrete density + (0.126–0.124) = 0.33g/cm3.
profile) volume segment in the z-direction. The number density
for each volume segment is estimated by summing the number of Pore-Size Effects on Methane Adsorption. In essence, super-
molecules counted in each interval within the volume segment and critical methane in small organic pores is structured as a result
dividing the total by the number of intervals. Then, the number of pore-wall effects, showing layers of graded density across the
density is converted to the local mass density of methane using its pore. Depending on the pore size, a bulk-fluid region may exist at
molecular weight MCH4 and Avogadro’s number as follows: the central portion of the pore, where the influence of molecular
interactions with the pore walls is very small. In pores with sizes up
Number M CH4 to 50 nm, a combination of molecule/molecule and molecule/wall
CH4 = .
interactions dictates thermodynamic states of the gas and its mass
6.02252 × 10 23
transport in the pore (Krishna 2009). On the other hand, within a
As explained earlier, it is expected that the local mass density small pore (see Fig. 8), methane molecules are always under the
for methane across the pore will be different from its mean bulk influence of the force field exerted by the walls; consequently, no
density, owing to changing levels of interactions between the bulk-fluid region can be observed in the pore, and no pore pressure
methane/methane and methane/carbon bodies. The purpose of our can be measured. Therefore, behavior of the adsorbed molecules
numerical investigation was to predict a precise density profile should be considered rather than the motion of free-gas molecules.
across the pore as an indication of the presence of these interac- Furthermore, the adsorbed-phase density at the first layer is signifi-
tions and to determine an average adsorbed-gas-density value, a cantly large. The smaller the pore, the larger the adsorbed-phase
macroscopic quantity necessary for the gas-in-place calculations density is. In the limit, Langmuir theory is not suitable for the
using Eq. 13. description of physical interactions between the gas and solid.
Fig. 7 shows the density profile for methane confined to the Effect of Temperature on Methane Adsorption. The effect
3.92-nm pore at 176°F. It is clear that the predicted methane density of temperature on methane density in the large pore is shown in
is not uniform across the pores; its value is significantly greater Fig. 9. The estimated average adsorbed-methane density in the
near the wall, where adsorption takes place, and it decreases with first layer is around 0.3 g/cm3, although it decreases with tem-
damped oscillations as the distance from the pore wall increases. perature (for example, to 0.259 g/cm3 at 266°F). These values
The oscillations are caused by the presence of adsorption in show variations caused by changing levels of kinetic energy at
the pores and involve structured distribution of molecules (i.e., the microscopic scale.
molecular layers). The number of molecules is largest in the first Table 3 shows that the predicted free-gas density values at the
layer near the wall, indicating physical adsorption. The molecules center of the pore correspond to the bulk methane density values
in the second layer are still under the influence of pore walls, predicted independently using the National Institute of Standards
although intermolecular interactions among the methane molecules and Technology’s (NIST) thermophysical properties of hydro-
begin to dominate, not allowing locally high methane densities. In carbon mixtures database, SUPERTRAPP. The comparisons also
this layer, the density of methane is slightly larger than the bulk show that our numerical simulations are accurate.
50 0.35 0.2
0.124
0.1
42.80 0.30 0.292
43.01 0.0
40
0.25 0.0 3.8 7.6
ρCH4 (g/cc)
ρ* CH4 (1000Å–3 )
30 0.20 0.171
0.15 0.135
20
0.10
10 7.65 7.72
0.05
4-5 (bulk)
50 4.57E+01 0.35
0.307
4.69E+01 0.30
40
ρ* CH4 (1000Å–3 )
20 0.15
1.57E+01
0.10
10
0.05
4-5(bulk)
0 0.00
0.0 3.8 7.6 11.4 0.0 3.8
Pore Length (Å) Pore Half Length (Å)
Fig. 8—Number-density (left) and discrete-density (right) profiles for methane at 176°F (80°C) in pore widths of 2.31 nm (upper)
and 1.41 nm (lower). Density values are estimated at each 0.2-Å interval for the continuous-density profile. Discrete density cor-
responds to molecular-layer density for methane across the pore. Estimated pore pressure at the center of the pores is 3,043
psi. The inset graph in the upper right corner is the equivalent density using a Langmuir single-layer adsorption model.
Conclusions adjustments. This change in gas in place persists even when the
In this paper, we addressed the following issues in regard to the sorbed density values reported by other researchers, such as the
volume available for free gas in organic shales: liquid-methane density of 0.4223 g/cc, corresponding to 1 atm
• The sorbed phase follows Langmuir theory for most of the pores and –161°C (Mavor et al. 2004), and of 0.374 g/cc (Haydel and
and takes up a one-molecule-thick portion of a pore, although Kobayashi et al. 1967). These are clearly shown in Table 4.
there is a damped oscillation density profile with an increased In conclusion, a robust method that matches the local phys-
density in the second layer. For a 100-nm pore, the volume is ics is presented to determine an estimate of the gas in place in
fairly insignificant; however, for pores on the order of 1 nm, it organic-rich gas shale. Future work includes re-evaluation of the
is quite large. concepts and approaches that are presented and discussed in the
• The current industry standard disregards the volume consumed presence of multicomponent gases with varying pore sizes. It is
by the sorbed phase, thus inadvertently overestimating the pore expected that there will be added uncertainties in the gas-in-place
volume available for free-gas storage. We showed that with Eq. estimation because of (1) multicomponent nature of gas and its
13, a more correct volume is calculated. Through MD simulation adsorbed phase and (2) volume fraction of organic nanopores that
and Langmuir theory, we showed that the density for adsorbed are currently inaccessible for visual observation and measurement.
methane typically equals 0.34 g/cm3. This value is based on The former requires computer-simulation studies on the adsorbed-
numerical work using simple flat organic pore-wall surfaces. In phase thickness and density using binary, ternary, and quaternary
small pores, the adsorbed-layer thickness will be affected by the mixtures of gases with components such as ethane, propane, and
roughness and curvature of the solid surface, and the density will carbon dioxide. The latter uncertainty, on the other hand, can be
increase because of increased surface area of the wall. Further reduced by measuring the absolute area of the organic surfaces
analysis using more-complex pore-wall structures is necessary. using isotherm data. In addition, a new study on transport proper-
• In our calculation of the gas in place, we predict a dramatic ties of natural gases in nanoporous organic materials is necessary
change in the estimated gas-in-place values because of volume in light of the new pore-scale findings.
266°F
0.20
ρCH4 (g/cc)
24
0.15
16
0.10
8
0.05
0 0.00
Fig. 9—Number-density (left) and discrete-density (right) profiles across half-length of a 3.9-nm-width slit-pore as a function of
temperature.
TABLE 4—EFFECT OF SORBED PHASE DENSITY ON FREE AND TOTAL GIP CALCULATIONS
Adsorbed-Phase Density Free Storage Capacity Total Storage Capacity
3
(g/cm ) (SCF/ton) (SCF/ton)
Given that it is in scf, we can convert scf into a mass with the Using the new calculation method for gas in place, the free, sorbed,
ideal-gas law at standard temperature and pressure: and total storage capacities for Shale A are, respectively,
φ 0.06 0.06
Sw 0.35 0.35
So 0.0 0.0
Bg 0.0046 0.0046
M̂ 16 lb/lb-mol 16 lb/lb-mol
GsL 50 scf/ton 120 scf/ton
p 4000 psia 4000 psia
T
o o
180 F 180 F
pL 1150 psia 1800 psia
3 3
b 2.5 g/cm 2.5 g/cm
3 3
s
0.34 g/cm 0.34 g/cm