You are on page 1of 16

SPE 144317

Production Analysis of Tight Gas and Shale Gas Reservoirs Using the
Dynamic-Slippage Concept
C.R. Clarkson, University of Calgary; M. Nobakht, Fekete Associates Inc. and University of Calgary; D. Kaviani,
University of Calgary; and T. Ertekin, Pennsylvania State University

Copyright 2011, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE North American Unconventional Gas Conference and Exhibition held in The Woodlands, Texas, USA, 14–16 June 2011.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Shales and some tight gas reservoirs have complex, multi-model pore size distributions, including pore sizes in the nanopore
range, causing gas to be transported via multiple flow mechanisms through the pore structure. In 1986, Ertekin et al.
developed a method to account for dual mechanism (pressure- and concentration-driven) flow for tight formations that
incorporated an apparent Klinkenberg gas-slippage factor that is not a constant, which is commonly assumed for tight gas
reservoirs. In this work, we extend the dynamic- slippage concept to shale gas reservoirs, for which it is postulated that
multi-mechanism flow can occur. Inspired by recent studies that have demonstrated the complex pore structure of shale gas
reservoirs, which may include nanoporosity in kerogen, we first develop a numerical model that accounts for multi-
mechanism flow in the inorganic and organic matter framework using the dynamic-slippage concept. In this formulation,
unsteady-state desorption of gas from the kerogen is accounted for. We then generate a series of production forecasts using
the numerical model to demonstrate the consequences of not rigorously accounting for multi-mechanistic flow in tight
formations. Finally, we modify modern rate-transient methods by altering pseudovariables to include dynamic-slippage and
desorption effects and demonstrate the utility of this approach with simulated and field cases. The primary contribution of
this work is therefore the demonstration of the use of modern rate transient methods for reservoirs exhibiting multi-
mechanistic (non-Darcy) flow. The approach is considered to be useful for analysis of production data from shale gas and
tight gas formations as it captures the physics of flow in such formations realistically.

Introduction
The unique storage and transport properties of unconventional gas reservoirs require that conventional petroleum engineering
methods be modified to account for these unique characteristics. With the widespread successful development of low-
permeability (“tight”) and shale gas reservoirs in North America, there is a pressing need to develop new techniques for
deriving quantitative estimates of hydraulic-fracture and reservoir properties to assist with field development. Due to the
ultra-low permeability nature of these reservoir types, conventional well-test (pressure transient) analysis is not practical
because of the need to shut-in the wells for extensive periods of time to achieve quantitative results. Although alternative
methods of well-test design and analysis are being developed (ex. Barree et al. 2007), petroleum engineers are increasingly
relying on rate-transient analysis methods as a substitute for well-test analysis.
A fundamental problem with the application of conventional rate-transient analysis to ultra-low permeability reservoirs is
that current methods were derived with the assumption of viscous laminar flow, i.e. flow that can be described with Darcy’s
law. Shale reservoirs have recently been observed to contain a wide distribution of pore sizes, including in some cases
nanopores associated with organic matter (Loucks et al. 2009). Wang et al. (2009) identified four types of porous media that
may be present in gas shales: organic matter, nonorganic matrix, natural fractures and pore space created by hydraulic
fractures. Gas flow through the shale matrix may therefore be expected at several scales and by several mechanisms.
Javadpour (2009) suggested that several mechanisms for flow can occur in gas shales including advective, slip-flow and
diffusion; he demonstrated the pressure and temperature-, gas composition- and pore size-dependence of apparent gas
permeability. A further complication is that gas storage in the finest nanoporosity can occur through sorption.
Our goal is to first develop simple yet rigorous approaches to model transport at various scales for shale gas reservoirs,
as illustrated in Fig. 1, then modify rate-transient analysis techniques accordingly. For the purposes of this study we will
focus on matrix (including organic matter) transport mechanisms only.
2 SPE 144317

Macroscale (reservoir)

Mesoscale
(microfracture
network) Macrofabric

Microscale (nanopore
network)

Nanoscale (gas
desorption from
nanopore walls)

Molecular (mass
transfer from
kerogen/clay bulk
to pore surface)
Modified from Bustin, Bustin and Cui, 2008

Fig. 1 — Illustration of the impact of scale on transport mechanisms in shale gas reservoirs. Flow to the wellbore is first initiated at
the macroscale, followed by flow at progressively finer scales, including molecular transport through nanoporosity in kerogen.
Modified from Javadpour et al. (2007).

The approach used in this study for modeling matrix transport in tight gas/shales is to use the dynamic-slippage concept,
as suggested by Ertekin et al. (1986), for modeling multi-mechanistic flow in larger meso (2 – 50 nm pore diameter) and
macropores (> 50 nm pore diameter), and non steady-state diffusion to model transport at the scale of micropores (< 2 nm
pore diameter). Dynamic-slippage manifests itself as an apparent gas permeability increase with decrease in pressure. We
first develop a mathematical model to incorporate these processes (+ sorption in the microporosity), and demonstrate how
commercial simulators with options for modeling coalbed methane (CBM) reservoirs may be adapted to account for
dynamic-slippage effects. We also demonstrate how modern rate-transient analysis techniques can be modified to account
for apparent gas permeability changes due to dynamic-slippage effects and desorption through alteration of pseudovariables
(pressure and time). Finally, we test our new rate-transient analysis approach against simulated and field cases.

Theory
In some tight/shale gas reservoirs, the mean-free path of gas molecules may be comparable to or larger than the average
effective rock pore throat radius causing the gas molecules to “slip” along pore surfaces. This slip-flow creates an additional
flux which may be additive to viscous (Darcy) flow. Slip-flow can cause apparent gas permeability (ka) to be higher than
permeability derived from single-phase flow of a liquid through the same porous medium. Historically, the Klinkenberg
method (Klinkenberg, 1941) has been used to “correct” effective gas permeability to a liquid-equivalent permeability using a
“gas-slippage” factor:

⎛ b⎞
k a = k ∞ ⎜⎜1 + ⎟⎟ ……………………………………….………….…………………………………………………………..(1)
⎝ p⎠

The gas-slippage factor is usually derived experimentally from core data by plotting apparent gas permeability versus the
reciprocal mean pressure (1/ p ) – at high pressures, the slippage effect becomes negligible because the mean-free path of the
gas molecules is small. The Klinkenberg procedure generally assumes that b is constant, although b has been observed to
increase with increasing pressure. Klinkenberg (1941) derived the following expression for the slippage factor, which
illustrates the effect of mean-free path and pore radius:

b 4 K 1λ ……………………………..…..……….................................………...…………………………………………..(2)
=
p rc
SPE 144317 3

To account for experimentally-observed pressure-dependence of the gas-slippage factor, Ertekin et al. (1986) assumed
that gas is transported in tight formations under the influence of both a concentration and pressure field (dual-mechanism
flow). Furthermore, it was assumed that these two flow fields were acting in parallel. Darcy’s law was used to model
pressure-driven flow, and Fick’s law was used to model concentration-driven flow associated with gas-slippage along the
pore wall boundaries. An apparent Klinkenberg gas-slippage factor for single- and multi-phase flow was introduced:

pc g μ g D
ba = ……………...………...……………………………………………………………………………………….(3)
α c2 k∞

We note that in Ertekin et al.’s formulation (Eq. 3), the gas-slippage factor (b) is not constant, which is commonly
assumed for tight gas reservoirs, but is pressure-composition-dependent. Ertekin et al. also noted that the slippage factor is
saturation-dependent in multi-phase flow cases. The apparent gas permeability, including the dynamic-slippage factor, is
given below:

⎛ b ⎞ ……….......…………………………….………….…………………………………………………………..(4)
k a = k ∞ ⎜⎜1 + a ⎟⎟
⎝ p ⎠

Because pressure in Eq. 3 cancels upon substitution in Eq. 4, the pressure-temperature-composition dependence of
apparent permeability is associated with gas compressibility and gas viscosity – the diffusion coefficient (D) is assumed to be
independent of pressure. In their work, Ertekin et al. first considered the dual mechanism flow in tight gas sands, therefore,
did not include a desorption term. In later work, a desorption term was included for modeling dual mechanism flow in
coalbed methane reservoirs (ex. see Thararoop, 2010).
Javadpour (2009) and Civan (2010) introduced methods for calculating apparent permeability change as a function of
Knudsen number (Kn); Knudsen number, defined below, can be used (assuming pipe flow geometry) to distinguish between
continuum flow (Kn < .001), slip flow (0.001 < Kn < 0.1), transitional flow (0.1 < Kn < 10) and free molecular flow (Kn >
10):

λ ………………………………..……..……………………………...…......…………………………………………..(5)
Kn =
lc

The approaches of Javadpour (2009) and Civan (2010) are therefore rigorous as they predict the dominant flow regime as a
function of pore size, pressure, temperature and gas composition. We note that in all cases, a dynamic-slippage factor can be
derived by solving Eq. 4 for ba, as was done by Civan (2010). The reader is referred to those works for details of their
derivations.
In this work, we use Eq. 3 and Eq. 4 to predict apparent permeability changes as a function of pressure. Limited
comparisons between the Ertekin et al. (1986) and Javadpour (2009) approaches for calculating apparent permeability
changes will be shown in the Discussion section.

Modeling Approach. In the present work, we assume that the shale gas matrix (inorganic + organic matter framework) has
a multi-modal pore structure, as has been recently observed. Flow through the meso/macroporosity in the inorganic (or
organic fraction, see below) is modeled using a similar approach to Ertekin et al. (1986), where a dynamic gas-slippage factor
is incorporated into the transport equation. Gas is assumed to be adsorbed in the microporosity of the organic matter
(kerogen), and transport through the microporosity is assumed to be governed by Fick’s law. This new model for shale matrix
flow is similar to dual porosity (fracture + matrix) models developed for CBM reservoirs, except flow through the larger pore
sizes (inorganic fraction) of the shale is assumed to occur by multiple mechanisms and takes the place of fracture flow in
CBM simulation. The mathematical model used to describe flow through the matrix is given below:

∂ ⎛⎜ Ax k ∞x ∂p ⎞⎟ ∂ ⎛⎜ A y k ∞y ∂p ⎞ ⎛ ⎞
βc Δx + βc ⎟Δy + q sc + q m = Vbφ ∂ ⎜ 1 ⎟ ……….……………..…...……………….(6)
∂x ⎜⎝ μ g B g K g ∂x ⎟⎠ ∂y ⎜⎝ μ g B g K g ∂y ⎟⎠ α c1 ∂t ⎜ B g ⎟
⎝ ⎠

Eq. 6 describes flow of gas through a 2-D shale matrix (Cartesian coordinates) with an incompressible pore volume. We note
that the larger pores (meso/macro) may be associated with either the inorganic framework or organic framework; indeed, in
some cases, the matrix porosity may be primarily associated with organic matter (Wang et al., 2009). The source term qm
represents flow from the organic matter microporosity (via diffusion) to the meso/macropores associated with the
inorganic/organic matter framework, and can be calculated using the simple pseudo steady-state approach, as was done in
Clarkson et al. (2007) for CBM reservoirs, or using more sophisticated unipore/bidisperse non steady-state models as was
done in Clarkson et al. (1999). Dynamic-slippage in the larger pores is incorporated into the Kg term as follows:
4 SPE 144317

−1
⎛ b ⎞
K g = ⎜⎜1 + a ⎟⎟ ………...……………………………………………………………….…………………………………….(7)
⎝ p ⎠

Although we have used the Ertekin et al. (1986) for calculating dynamic-slippage for this work, the methods of Civan (2010)
and Javadpour (2009) can also be used for this purpose.
A computer model was developed to solve Eq. 6 numerically. Pore volume of the matrix was assumed to be static so
that the effect of gas-slippage alone on apparent permeability change could be investigated. As will be discussed later, pore
volume for shale may not be a static property, and stress dependence of porosity and permeability can be quite strong in some
cases (Thompson et al., 2010).
Because our numerical model is relatively limited in terms of gridding choices, we investigated the use of commercial
simulators to perform sensitivity runs. A commercial simulator capable of modeling dual porosity CBM behavior (including
gas sorption and matrix diffusion) was chosen; the fracture porosity was set equal to the estimated meso/macroporosity of the
shale and dynamic-slippage effects were accounted for by using a table of transmissibility multipliers (inverse of Eq. 7) as a
function of pressure. As will be shown in a later section, this approach yielded similar results to our numerical model
described by Eq. 6.

Modification of Rate-Transient Techniques. A primary objective of our work is to establish if rate-transient analysis
techniques can be adapted to account for both dynamic-slippage and desorption effects. Previous work (ex. Clarkson et al.
2007) has shown that desorption effects can be accounted for in relatively high permeability coal reservoirs through the
inclusion of a desorption compressibility term in the total compressibility calculation – use of pseudotime requires
modification for CBM material balance. Non-static permeability (relative and absolute) changes have also been accounted
for (Clarkson et al. 2009). In a recent paper (Thompson et al. 2010), non static (stress-dependent) permeability was
incorporated into pseudopressure and pseudotime calculations for shale gas type-curve analysis. In this paper, we use a
similar approach, except pressure-dependent permeability is assumed to be due to non-Darcy (slip) flow. We note that it is
possible to have multiple causes of pressure-dependent permeability in shale reservoirs. The modified pseudopressure and
pseudotime used in this work are provided below:

pi
1

* *
m ( p i ) − m ( p wf ) = 2 pdp ………...……………..……………….…….…………………………………….(8)
K g ( p)μ g z
pwf

t
dt
t a* = ( μ g ct* ) i ∫K
0
*
g μ g ct
………...…………….……………...………………………….…………………………………….(9)

Where * indicates altered variables and Kg is given in Eq. 7. Note that c t* includes desorption compressibility, which in turn
assumes that desorption is instantaneous from the micropores to the meso/macropore system. The pressure-dependent
variables in the integrand for pseudotime are evaluated at p , obtained from flowing material balance analysis. We note that
this approximation is expected to cause some error – a more rigorous approach for calculating pseudotime using the radius-
of-investigation concept is discussed in Nobakht and Clarkson (2011a and 2011b).
Eq. 8-9 can be incorporated into both type-curve analysis and flow-regime (straight-line) analysis. As discussed in detail
by Clarkson and Beierle (2010), straight-line analysis involves first the identification of flow-regimes using derivative
techniques, followed by analysis of data associated with those flow-regimes using specialty plots. These specialty plots are
[ ]
plots of m* ( pi ) − m* ( p wf ) / q g versus superposition time functions. For radial and linear flow analysis, the superposition
time functions are:

n
( q j − q j −1 )
Radial superposition time: ∑
j =1
qn
log(tn* − t *j −1 ) ………….………….………………….….……………………….(10)

n ( q j − q j −1 )
Linear superposition time: ∑
j =1
qn
(t n* − t *j −1 ) ………….………….………………….….……………………….(11)

Where time in Eq. 10 and 11 is the modified pseudotime of Eq. 9.


SPE 144317 5

For flowing material balance analysis, we use an altered version of the dry coal flowing material balance discussed in
Clarkson (2009), which accounts for free-gas + sorbed gas storage. The altered flowing material balance (accounting for gas-
[ ]
slippage) involves plotting q g / m * ( p i ) − m * ( p wf ) versus modified cumulative production, given by:

[m ( p ) − m ( p )]
*
i
*

[m ( p ) − m (p )]
Modified cumulative production: Gi …….…………….……………….….……………………….(12)
* *
i wf

The flowing material balance analysis procedure is iterative, and material balance calculations that include desorption are
required. In this work we used the Clarkson and McGovern (2005) equation which accounts for adsorbed and free gas
storage.

Simulation Results and Rate-Transient Analysis


In this section we illustrate the impact of dynamic-slippage and desorption on simulated production forecasts. We also
demonstrate the sensitivity of rate-transient analysis to the pseudovariable corrections provided in the previous sections using
simulated examples, and the impact of not correcting for gas-slippage and desorption on rate-transient analysis results.
Prior to performing the simulation sensitivities using a numerical simulator, we first confirmed the accuracy of using
transmissibility multipliers (inverse of Eq. 7) in a commercial simulator, which include the effects of dynamic-slippage, by
comparison to the mathematical model developed in this work. We note that this (transmissibility multiplier) approach was
previously used by Clarkson and McGovern (2005) to model absolute permeability changes in coalbed methane reservoirs.
A commercial simulator capable of modeling dual porosity coalbed methane behavior was selected for the comparison. The
input for one of the comparison runs (several additional runs were performed) is provided in Table 1, and the results of the
comparison run are shown in Fig. 2. Dynamic-slippage was calculated using Eq. 3 and the procedure of Ertekin et al. (1986),
and the transmissibility multipliers were calculated using the inverse of Eq. 7. Instantaneous desorption (sorption time = 0)
was assumed.

TABLE 1 — INPUT PARAMETERS FOR SIMULATION 500


COMPARISON RUN. VERTICAL WELL WITH SKIN,
COMPLETED IN A SHALE RESERVOIR 400
Gas Rate, Mscf/D

Input Parameter Parameter Value


THICKNESS (ft) 250 300

3
BULK DENSITY (g/cm ) 2.47
200
POROSITY (%) 7.0
GAS GRAVITY 0.69 100

ABSOLUTE PERMEABILITY (md) 0.005


INITIAL RESERVOIR PRESSURE 3500 0
(psia) 0 2000 4000 6000 8000 10000 12000

0.0 Time, days


INITIAL WATER SATURATION (%)
Commercial Simulator New Model
RESERVOIR TEMPERATURE (°F) 200
Fig. 2 — Comparison of numerical model forecasts using Eq. 6 and
LANGMUIR VOLUME (scf/ton, in- 89
a commercial simulator. Dynamic-slippage was incorporated into
situ)
the commercial simulator by using transmissibility multipliers.
LANGMUIR PRESSURE (psia) 535.6
DRAINAGE AREA (acres) 40
RESERVOIR DIMENSIONS (ft) 1570.2 X 1109.7
WELLBORE DIAMETER (in.) 6.25
SKIN FACTOR -1
FLOWING BOTTOMHOLE
250
PRESSURE (psia)

Agreement between the commercial simulator and the new model (Eq. 6) appears reasonable. We, therefore, use the
commercial simulator in all future runs due to the ease of case generation.
For simulation sensitivities, a vertical well completed in high- (0.005 md) and low- (0.0005 md) permeability shale gas
reservoir with an infinite conductivity hydraulic fracture was modeled. Input for the model is provided in Table 2.
Logarithmic gridding (Fig. 3) was used to properly capture pressure transients for simulation.
6 SPE 144317

TABLE 2 — INPUT PARAMETERS FOR SIMULATION


SENSITIVITIES. VERTICAL WELL WITH HYDRAULIC
FRACTURE, COMPLETED IN A SHALE RESERVOIR
Input Parameter Parameter Value
THICKNESS (ft) 250
3
BULK DENSITY (g/cm ) 2.47 Hydraulic Fracture

POROSITY (%) 7.0


GAS GRAVITY 0.69
ABSOLUTE PERMEABILITY (md) 0.005/0.0005
INITIAL RESERVOIR PRESSURE 3500
(psia)
INITIAL WATER SATURATION (%) 0.0
RESERVOIR TEMPERATURE (°F) 200
LANGMUIR VOLUME (scf/ton, in- 89
situ) Fig. 3 — Snapshot of grid block pressures obtained for one of the
simulation sensitivity cases.
LANGMUIR PRESSURE (psia) 535.6
DRAINAGE AREA (acres) 160
RESERVOIR DIMENSIONS (ft) 2639.8 X 2639.8
WELLBORE DIAMETER (in.) 7.872
HYDRAULIC FRACTURE HALF-
200
LENGTH (ft)
FLOWING BOTTOMHOLE
250/1000*
PRESSURE (psia)

*assume pwf = 250 psia unless otherwise stated

Several cases were simulated:

• Case 1: assumes no apparent gas permeability change


• Case 2: assumes apparent gas permeability change using Ertekin et al. (1986) dynamic-slippage approach
• Case 3: assumes apparent gas permeability change using the Jones-Owens (1979) static-slippage approach
• Case 4: assumes no apparent gas permeability change and no gas desorption

In all cases, except for Case 4, instantaneous desorption was assumed. Two approaches were used for calculating
permeability changes: the dynamic-slippage approach (Eq. 3) of Ertekin et al. (1986) and the static-slippage calculation of
Jones and Owens (1979):

b = 12.639(k ∞ )
−.33
………...………...…………………………………………………………………………………….(13)

We now summarize the results of the simulation runs and rate-transient analysis, starting with the high permeability (0.005
md) scenario.

High Permeability Simulation Runs and Analysis. Cases 1-4 were generated, assuming an initial matrix permeability of
0.005 md and forecasts summarized in Fig. 4. The slippage-factor calculations using Eq. 3 (Case 2) and Eq. 13 (Case 3) are
given in Fig 5a, along with the corresponding permeability multiplier used in the simulator (Fig. 5b). The dynamic-slippage
factor is non-linear, decreasing strongly with pressure below ~ 2000 psi. Apparent gas permeability remains relatively
constant until low pressure (~ 100 psi) at which point a strong growth in apparent gas permeability is observed. The
differences between the dynamic (Ertekin) and static (Jones-Owens) slippage factors (Fig. 5a) are significant at high
pressures, but converge at low pressures. The apparent gas permeability growth predictions for both models also differ (Fig.
5b). For this high permeability case, with flowing pressure = 250 psia, the impact of gas-slippage on production forecasts
(Fig. 4) is shown to be minor, however.
SPE 144317 7

300

280

260

240

Gas Rate, Mscf/D


220

200

180

160

140

120

100
0 200 400 600 800 1000

Time, days

Case1 Case2 Case3 Case4

Fig. 4 — Simulation forecasts assuming high initial matrix permeability (0.005 md) for Cases 1 – 4.

140 b) 25
a)
120
20
Gas-Slippage Factor, psi

100

15
80
1+b/p

60
10

40
5
20

0 0
0 500 1000 1500 2000 2500 3000 3500 1 10 100 1000
Pressure, psi Pressure , psi
Ertekin Jones-Owens Ertekin Jones-Owens

Fig. 5 — Gas-slippage factor (a) and permeability multiplier (b) calculation using dynamic-slippage (Eq. 3) and static-slippage (Eq.
13) approaches for initial matrix permeability (0.005 md) assumption.

The highest forecasted production rates (Fig. 4) are for Case 2 (dynamic-slippage + desorption), followed closely by
Case 3 (static-slippage + desorption). Case 4 (no slippage, no desorption) yields the lowest production rates; the impact of
desorption in this example is minor, as evidenced by comparison of Case 4 and Case 1.
Turning our attention to rate-transient analysis, we now examine the impact of dynamic-slippage and desorption on
derived reservoir/hydraulic fracture properties, and establish whether or not the corrections to pseudovariables provided in
the Theory section are robust. We start with the comparison of semi-log derivative calculations for Case 4 and Case 2, with
and without correction for the pseudovariables (time and pressure) discussed above (Fig. 6). The derivative plots indicate the
expected sequence of flow regimes for vertical well completed with an infinite conductivity hydraulic fracture, namely linear
flow followed by pseudoradial then boundary-dominated flow. Note that a unit slope is not observed for boundary-
dominated flow because we have not used material balance pseudotime. We note that if corrections are not applied (Fig 6a)
to pseudovariables, there is a slight difference between derivative curves, which diminishes when the corrections are applied
(Fig. 6b).
8 SPE 144317

Semi-log Derivative Plot, No Corrections to Pseudovariables Semi-log Derivative Plot, With Corrections to Pseudovariables
a) b)

1.0E+07 1.0E+07

radial flow radial flow

d(∆m*(p)/q)/dlnta*
d(∆m(p)/q)/dlnta

1.0E+06 linear flow 1.0E+06 linear flow


boundary-dominated boundary-dominated
flow flow

1.0E+05 1.0E+05
1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05
Pseudotime, days Modified Pseudotime, days
Case 4 Case 2 Case 4 Case 2

Fig. 6 — Semi-log derivative calculations for Case 2 (with dynamic-slippage and desorption) and Case 4 (no gas-slippage, no
desorption) without (a) and with (b) corrections to pseudovariables.

Differences in the cases become more noticeable when the specific flow regimes are analyzed on their respective
specialty plots. Radial flow and linear flow plots are given in Figs. 7-8, respectively. The actual portion of the plot to which
a straight-line should be fit for analysis (as determined from derivative analysis) is indicated on the plots.
a) Radial Flow Plot, No Corrections to Pseudovariables b) Radial Flow Plot, With Corrections to Pseudovariables

1.5E+07 1.5E+07

radial radial
[m*(pi)-m*(pwf)]/qg
[m(pi)-m(pwf)]/qg

1.0E+07 flow 1.0E+07 flow

5.0E+06 5.0E+06

0.0E+00 0.0E+00
4 5 6 4 5 6
Superposition Time Superposition Time
Case 2 Case 4 Case 2 Case 4

Fig. 7 — Radial flow plot Case 2 (with dynamic-slippage and desorption) and Case 4 (no gas-slippage, no desorption) without (a) and
with (b) corrections to pseudovariables.

For radial flow (Fig. 7), the differences in the 2 cases is noticeable (Fig. 7a), translating into ~ 7% difference in the
computed matrix permeability, which is derived from a straight-line fit to the plot during the radial flow period. There is also
a slight difference in the calculated skins. After pseudovariable corrections are applied (Fig. 7b) the calculated permeabilities
are essentially the same for the two cases.
For linear flow (Fig. 8), the differences in the two cases are also noticeable (Fig. 8a), translating into ~ 11% difference in
k x f , which is derived from a straight-line fit to the plot during the linear flow period. Even after correction to the
pseudovariables, a small error remains (~ 4%), which we believe to be related to desorption effects. To test this hypothesis,
we ran one additional case (Case 5) which includes dynamic-slippage, but not desorption (Fig. 9). We note that after
removing the effects of desorption, the pseudovariable corrections for dynamic-slippage work very well.
It is clear that correction for desorption using the simple desorption compressibility approach (referenced to pore-volume
average pressure) is in error for these low permeability systems where pressure gradients near the fracture face are high.
From a practical point of view, however, if flowing pressure is large relative to Langmuir pressure (~ 535 psia in this case),
then the errors are expected to be smaller. For example, when we re-run Case 4 with a higher pwf (1000 psia versus 250 psia),
there is better agreement between Case 4 and Case 2 after corrections are made to the pseudovariables (not shown). The
error in k x f for Case 4 (after corrections applied) relative to Case 2 for example is now only ~ 2.5%.
SPE 144317 9

a) Linear Flow Plot, No Corrections to Pseudovariables b) Linear Flow Plot, With Corrections to Pseudovariables
1.5E+06 1.5E+06

1.0E+06 1.0E+06
[m(pi)-m(pwf)]/qg

[m(pi)-m(pwf)]/qg
linear flow linear flow

5.0E+05 5.0E+05

0.0E+00 0.0E+00
0 5 10 15 20 0 5 10 15 20
Superposition Time Superposition Time
Case 2 Case 4 Case 2 Case 4

Fig. 8 — Linear flow plot Case 2 (with dynamic-slippage and desorption) and Case 4 (no gas-slippage, no desorption) without (a) and
with (b) using corrections to pseudovariables.

a) Radial Flow Plot, With Corrections to Pseudovariables b) Linear Flow Plot, With Corrections to Pseudovariables
1.5E+06
1.5E+07

radial 1.0E+06
[m*(pi)-m*(pwf)]/qg
[m(pi)-m(pwf)]/qg

1.0E+07 flow linear flow

5.0E+05
5.0E+06

0.0E+00
0.0E+00 0 5 10 15 20
4 5 6
Superposition Time
Superposition Time
Case 4 Case 2 Case 5 Case 2 Case 4 Case 5

Fig. 9 — Radial flow plot (a) and linear flow plot (b) including Case 5 (dynamic-slippage, no desorption).

Lastly, the corrections to pseudovariables appear adequate for flowing material balance calculations (Fig. 10). For
example, analysis of Case 2 (dynamic-slippage + desorption) results in an OGIP and drainage area very close to the
simulation inputs (7100 MMscf and 160 acres, respectively).
Flowing Material Balance Plot, With Corrections to Pseudovariables

3.0E-04
Normalized Rate, scf/D/psi2/cp

2.0E-04

1.0E-04

0.0E+00
0 2000 4000 6000 8000
Normalized Cumulative Production, MMscf

Fig. 10 — Flowing material balance plot for Case 2.

Low Permeability Simulation Runs and Analysis. We repeat the analysis performed for the high permeability scenario in
the previous section to demonstrate the relatively larger effect that dynamic-slippage has on the analysis of low permeability
systems.
10 SPE 144317

100
90
80
70

Gas Rate, Mscf/D


60
50
40
30
20
10
0
0 200 400 600 800 1000
Time, days
Case1 Case2 Case3 Case4

Fig. 11 — Simulation forecasts assuming low initial matrix permeability (0.0005 md) for Cases 1 – 4.

a) 300 b) 25

250
20
Gas-Slippage Factor, psi

200
15
1+b/p

150
10
100

5
50

0 0
0 500 1000 1500 2000 2500 3000 3500 1 10 100 1000
Pressure, psi Pressure , psi
Ertekin Jones-Owens Ertekin Jones-Owens

Fig. 12 — Gas-slippage factor (a) and permeability multiplier (b) calculation using dynamic-slippage (Eq. 3) and static-slippage (Eq.
13) approaches for low initial matrix permeability (0.0005 md) assumption.

Figs. 11-12 are analogous to Figs. 4-5 for the high permeability case. Note the larger gas-slippage factors (Fig 12a) and
apparent gas permeability change (Fig 12b) for this case relative to the high-permeability scenario, which is expected. In the
low permeability scenario, significant permeability growth starts at higher pressure (compare Fig. 12b and Fig, 5b), and
hence the impact of gas-slippage is expected to be relatively higher. For example, at 10 days of production, there is > 17%
difference between cases 2 and 4 for the low-permeability scenario, but < 10% for the high permeability scenario.
Fig. 13 (analogous to Fig. 6), again demonstrates that slippage and desorption have a noticible impact on the derivative
calculations, but that the pseudovariable corrections (Fig. 13b) appear to work well. Notice that boundary-dominated flow is
not reached in this example.
Figs. 14-15 (analogous to Figs. 7-8) show radial (Fig. 14) and linear (Fig. 15) flow plots, and demonstrate the errors
associated with not correcting for slippage and desorption in the pseudovariables. A difference of ~ 11% in permeability and
16% in k x f between Cases 2 and 4 results if corrections are not applied. After corrections are applied the difference is
negligible and 1.7 %, respectively. As in the previous case, a small error results due to desorption, but correction for slippage
with no desorption appears accurate (Fig. 16). Further, as in the previous example, when we re-run Case 4 with a higher pwf
(1000 psia versus 250 psia), there is better agreement between Case 4 and 2 after corrections are made to the pseudovariables
(not shown). The error for Case 4 (after corrections applied) relative to Case 2 for example is < 1%.
SPE 144317 11

a) Semi-log Derivative Plot, No Corrections to Pseudovariables b) Semi-log Derivative Plot, With Corrections to Pseudovariables

1.0E+08 1.0E+08

radial flow radial flow

1.0E+07 1.0E+07

d(∆m(p)/q)/dlnt
d(∆m(p)/q)/dlnt

linear flow linear flow

1.0E+06 1.0E+06

1.0E+05 1.0E+05
1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05
Pseudotime, days Modified Pseudotime, days
Case 4 Case 2 Case 4 Case 2

Fig. 13 — Semi-log derivative calculations for Case 2 (with dynamic-slippage and desorption) and Case 4 (no gas-slippage, no
desorption without (a) and with (b) using corrections to pseudovariables.

a) Radial Flow Plot, No Correction to Pseudovariables b) Radial Flow Plot, With Correction to Pseudovariables

1.0E+08 1.0E+08

[m(pi)-m(pwf)]/qg
[m(pi)-m(pwf)]/qg

5.0E+07 5.0E+07
radial radial
flow flow

0.0E+00 0.0E+00
4 5 6 7 4 5 6 7
Superposition Time Superposition Time
Case 2 Case 4 Case 2 Case 4

Fig. 14 — Radial flow plot Case 2 (with dynamic-slippage and desorption) and Case 4 (no gas-slippage, no desorption) without (a)
and with (b) using corrections to pseudovariables.

a) Linear Flow Plot, No Corrections to Pseudovariables b) Linear Flow Plot, With Corrections to Pseudovariables
6.0E+06 6.0E+06

linear flow linear flow

4.0E+06
[m(pi)-m(pwf)]/qg

4.0E+06
[m(pi)-m(pwf)]/qg

2.0E+06 2.0E+06

0.0E+00 0.0E+00
0 10 20 30 0 10 20 30
Superposition Time Superposition Time
Case 2 Case 4 Case 2 Case 4

Fig. 15 — Linear flow plot Case 2 (with dynamic-slippage and desorption) and Case 4 (no gas-slippage, no desorption) without (a)
and with (b) using corrections to pseudovariables.
12 SPE 144317

a) Radial Flow Plot, With Correction to Pseudovariables b) Linear Flow Plot, With Corrections to Pseudovariables
6.0E+06
1.0E+08
linear flow
8.0E+07
4.0E+06

[m(pi)-m(pwf)]/qg
[m(pi)-m(pwf)]/qg

6.0E+07

radial
4.0E+07 2.0E+06
flow

2.0E+07

0.0E+00
0.0E+00 0 10 20 30
4 5 6 7
Superposition Time
Superposition Time
Case 2 Case 4 Case 5 Case 2 Case 4 Case 5

Fig. 16 — Radial flow plot (a) and linear flow plot (b) including Case 5 (dynamic-slippage, no desorption).

We now illustrate the impact of neglecting gas-slippage and desorption effects for rate-transient analysis of a real shale
gas well. The example is a multi-fractured horizontal well completed in 6-stages (30 perforation clusters) with slickwater
fractures in the Barnett Shale. Input data are provided in Table 3. This well exhibits long-term linear-flow characteristics,
identified from a ½ slope on the normalized pseudopressure versus time plot (Fig. 17) – the semi-log derivative (not shown)
also suggests a ½ slope. As in the simulated examples above, we analyze the linear flow period using the linear flow plot
(Fig. 18a), both neglecting the effects of slippage and desorption and then correcting for them by modifying pseudovariables.
Dynamic slippage (Fig. 18b) was again calculated using Eq. 3 and desorption was assumed to be instantaneous.
TABLE 3 — INPUT PARAMETERS FOR FIELD CASE .
MULTI-FRACTURED HORIZONTAL WELL COMPLETED IN
THE BARNETT SHALE
1.00E+07
Input Parameter Parameter Value
THICKNESS (ft) 280 Linear flow
3
BULK DENSITY (g/cm ) 2.47 1.00E+06
∆m(p)/qg

POROSITY (%) 8.4


GAS GRAVITY 0.677
% CH4 97.7 1.00E+05

% N2 1.5
% CO2 0.8
1.00E+04
ASSUMED ABSOLUTE 0.0001 1 10 100 1000 10000
PERMEABILITY (md)
Time, days
INITIAL RESERVOIR PRESSURE 2862
(psia)
33 Fig. 17 — Rate-normalized pseudopressure plot illustrating long-
INITIAL WATER SATURATION (%)
term linear flow (1/2 slope) for Barnett Shale well.
RESERVOIR TEMPERATURE (°F) 183
LANGMUIR VOLUME (scf/ton, in- 89
situ)
LANGMUIR PRESSURE (psia) 535.6
WELLBORE DIAMETER (in.) 6.25
SPE 144317 13

a) Linear Flow Plot b) 25

1.0E+06
20
linear flow
8.0E+05

15
[m(pi)-m(pwf)]/qg

6.0E+05

1+b/p
4.0E+05 10

2.0E+05 5

0.0E+00
0
0 50 100 150
1 10 100 1000
Superposition Time
Corrections Made No Corrections Made Pressure , psi

Fig. 18 — Linear flow analysis (a) of Barnett Shale well; dynamic-slippage and desorption are neglected and corrected for in the
analysis. (b) illustrates the growth in apparent gas permeability due to dynamic-slippage.

In this example, a significant difference (~20%) in the calculated k x f occurs between the corrected and uncorrected
cases; this large discrepancy is due to the very low assumed matrix permeability (0.0001 md), which in turn creates a
significant dynamic-slippage effect, particularly below 1000 psia. In this case, however, because flowing pressure is
significantly above desorption pressure, the effects of desorption are relatively small. The difference in calculated k x f
using pseudotime with and without desorption is < 5% in this example.

Discussion
In this work, we have demonstrated that if non-Darcy flow effects (tight gas and shale) and desorption (shale) occur for
unconventional reservoirs, these effects need to be accounted for to derive hydraulic fracture/reservoir properties from rate-
transient analysis. The effect of non-Darcy flow becomes more significant as the permeability (effective pore size) decreases,
due to the effects of slippage and diffusion. We have provided corrections to pseudovariables for both non-Darcy flow and
desorption that appear to work for cases of practical interest in shale gas reservoirs, but note that the correction for desorption
may still be in significant error if flowing pressure is below desorption pressure. We will continue to improve on these
corrections in the near future.
We note that the errors due to non-Darcy flow effects and desorption are in addition to those caused by improper
definition of pseudotime. In accompanying manuscripts (Nobakht and Clarkson, 2011a, Nobakht and Clarkson, 2011b) it
was noted that pseudotime calculations based upon initial pressure or pore-volume average pressure can lead to significant
errors in calculations related to linear flow, the dominant transient flow period for many shale gas reservoirs. Corrections to
pseudotime for constant rate- and constant-pressure cases were discussed in Nobakht and Clarkson, 2011a and Nobakht and
Clarkson, 2011b, respectively. A rigorous approach for production analysis of shale gas reservoirs would use the pseudotime
calculations suggested in those works, + the corrections for non-static (apparent gas) permeability and desorption suggested
in the current work. This will be the subject of a future paper.
Although we have used the dynamic-slippage approach suggested by Ertekin et al. (1986) as the method for calculating
apparent gas permeability as a function of pressure, temperature and gas composition in this paper, we note that other
approaches that account for multi-mechanistic (non-Darcy) flow have recently been used in the literature (ex. Javadpour,
2009 and Civan, 2010). Because our approach for correcting for non-Darcy effects in pseudovariable calculations and
reservoir simulation (through transmissibility multipliers) requires an apparent gas permeability calculation as a function of
pressure, the method is general enough to use any of the proposed non-Darcy flow calculations. For example, Fig. 19
compares the dynamic-slippage approach for calculating permeability multipliers used in the high and low-permeability
simulation cases (Fig. 19a) and the Barnett Shale (field) case (Fig. 19b) with that obtained using Javadpour’s approach. Note
that the permeability multiplier curve is very similar for the two approaches, but a pore radius is required as input for the
Javadpour’s calculations. To obtain a match to Ertekin’s prediction for the high and low permeability cases, 140 nm and 65
nm, respectively, were used as a pore radius for Javadpour’s calculations. For the field case, 37.5 nm was used.
14 SPE 144317

a) 200 b) 350
180
300
160
140 250
120
1 + b/p

200

1 + b/p
100
80 150
60
40 100
20
50
0
1 10 100 1000 0

Pressure, psi 1 10 100 1000

Pressure, psi
Javadpour (low k) Ertekin (high k)
Ertekin (low k) Javadpour (high k) Javadpour Ertekin

Fig. 19 — Comparison of permeability multiplier calculations using dynamic-slippage approach (Ertekin et al. 1986) and Javadpour
approach (2009) for simulated cases (a) and Barnett Shale (field) case (b).

Lastly, although multi-mechanistic flow is one possible cause of apparent gas permeability change with pressure during
tight gas/shale gas production, we note that there are other, potentially more important causes. For example, stress-
dependence of porosity and permeability may actually lead to a loss in absolute permeability with depletion, which has been
noted for some coalbed methane reservoirs, as well as tight gas reservoirs (Franquet et al. 2004). Very recently, Thompson et
al. (2010) modeled pressure-dependent permeability effects for overpressured shale reservoirs – they noted that a reduction of
pore pressure in these reservoir types may actually cause a significant reduction in fracture conductivity (modeled as effective
permeability changes). Ultimately, non-static porosity and permeability changes can be incorporated into pseudovariable
calculations for rate-transient analysis as we have done in this work, but investigation into the physical cause of such changes
must be performed along with rate-transient analysis. Thompson et al. (2010) suggested that intermittent pressure buildup
analysis must be performed along with rate-transient analysis to determine the causes; Gierhart et al. (2007) similarly used
time-lapse pressure-buildup analysis to quantify effective permeability to gas changes in coalbed methane reservoirs. It is
possible that the evolution of matrix pore structure (ex. reduction in pore throat size), particularly that of kerogen in shale,
may also provide an additional factor leading to changes in flow mechanisms with pressure, but to our knowledge, this has
not been investigated in detail.

Conclusions
It is expected that multi-mechanistic flow will occur for some tight gas/shale gas reservoirs, and that the importance of non-
Darcy flow increases with pressure depletion. In this paper we have used the dynamic-slippage approach as suggested by
Ertekin et al. (1986) to quantify apparent gas permeability changes with pressure, and establish the impact of these changes +
desorption on simulated gas production forecasts. Further, we have attempted to modify modern rate-transient methods to
include multi-mechanistic flow and desorption through modification of pseudovariables. The following conclusions may be
drawn:
1. For the cases investigated in this work, apparent gas permeability changes are most significant at pressures much less than
initial reservoir pressure; depending on flowing bottomhole pressure, the effects could be quite significant near wellbore.
2. Multi-mechanistic flow is more important for ultra-low permeability reservoirs, particularly for shale gas reservoirs with
nanodarcy matrix permeability.
3. Apparent gas permeability changes resulting from multimechanistic flow can be easily incorporated into commercial
simulators by using pressure-dependent transmissibility multipliers.
4. Failure to account for non-Darcy flow and desorption effects during rate-transient analysis can lead to significant errors in
derived hydraulic-fracture and reservoir properties.
5. Our method for accounting for multi-mechanistic flow in rate-transient analysis using modified pseudovariables appears
rigorous, but accounting for desorption by using a desorption compressibility term is only approximate.
6. The impact of desorption on analysis becomes less significant if flowing wellbore pressure is greater than Langmuir
pressure.

Acknowledgements
The authors would like to thank ConocoPhillips for their support of this research. Chris Clarkson would like to acknowledge
Encana for support of his Chair position in Unconventional Gas at the University of Calgary, Department of Geosciences.
SPE 144317 15

Nomenclature
Field Variables
A = drainage area, acres
Ax, Ay = gridblock cross-section area,
normal to direction given in subscript (Eq. 6), ft2
b = slippage factor, psi
ba = apparent slippage factor, psi
Bg = gas formation volume factor, RB/scf
cg = gas compressibility, psi-1
ct* = total compressibility including desorption, psi−1
*
ct = total compressibility, evaluated at average reservoir pressure, psi-1
D = diffusion coefficient, ft2/D
Gi = original gas in place, MMscf
h = formation thickness, feet
K1 = proportionality constant
Kn = Knudsen number, dimensionless
Kg = k∞ / ka , Eq. 7, dimensionless
ka = apparent permeability, md
k∞ = liquid-equivalent permeability, md
lc = characteristic length of the flow geometry, cm
m(p) = real gas pseudopressure, psi2/cp
m*(p) = corrected real gas pseudopressure, psi2/cp
p = pressure, psi
p = mean pressure, psi
pL = Langmuir pressure constant, psi
pR = reservoir pressure, psi
pwf = flowing bottomhole pressure, psi
qm = gas flow from microporosity, scf/D
qsc = gas well production/injection rate, scf/D
rc = capillary radius, cm
t = time, days
ta* = corrected pseudotime, hours or days
T = temperature, °R
Vb = bulk volume of the reservoir, ft3
VL = Langmuir volume constant, scf/ton
T = temperature, °R

Greek Variables
αc1 = unit conversion factor, 5.615
αc2 = unit conversion factor, 0.006328
βc = transmissibility conversion factor, 0.001127
γg = gas gravity, air = 1
μg = gas viscosity, cp
μ g = gas viscosity evaluated at average reservoir pressure, cp
φ = porosity, dimensionless, fraction
λ = mean free path, cm
ρ = density, g/cm3

References
Barree, R.D., Barree, V.L., and Craig, D.P. 2007. Holistic Fracture Diagnostics. Paper SPE 107877 SPE Rocky Mountain
Petroleum Technology Conference, Denver, Colorado, 16–18 April.
Bello, R.O. and Wattenbarger, R.A. 2008. Rate Transient Analysis in Naturally Fractured Shale Gas Reservoirs. Paper SPE
114591 presented at the CIPC/SPE Gas Technology Symposium 2008 Joint Conference, Calgary, Alberta Canada, 16-19
June.
Civan, F., 2010. Effective Correlation of Apparent Gas Permeability in Tight Porous Media. Transp. Porous Media 82: 375-
384.
16 SPE 144317

Clarkson, C.R. 2009. Case Study: Production Data and Pressure Transient Analysis of Horseshoe Canyon CBM Wells. The
Journal of Canadian Petroleum Technology 48 (10): 27-38.
Clarkson, C.R., and R.M. Bustin, 1999. The Effect of Pore Structure and Gas Pressure Upon the Transport Properties of
Coal: A Laboratory and Modeling Study. 2. Adsorption Rate Modeling. Fuel, 78: 1345–1362.
Clarkson, C.R. and McGovern, J.M. 2005. Optimization of Coalbed Methane Reservoir Exploration and Development
Strategies Through Integration of Simulation and Economics. SPEREE 8 (6): 502–519. SPE-88843-PA.
Clarkson, C.R., Bustin, R.M., and Seidle, J.P. 2007. Production-Data Analysis of Single-Phase (Gas) Coalbed-Methane
Wells. SPEREE 10 (3): 312–331. SPE-100313-PA.
Clarkson, C.R., Jordan, C.L., Ilk, D. and Blasingame, T.A. 2009. Production Data Analysis of Fractured and Horizontal CBM
Wells. Paper SPE 125929 presented at the 2009 SPE Eastern Regional Meeting held in Charleston, West Virginia, 23-25
September.
Clarkson, C.R., and Beierle, J.J. 2010. Integration of Microseismic and Other Post-Fracture Surveillance with Production
Analysis: A Tight Gas Study. Paper SPE 131786 presented at the SPE Unconventional Gas Conference held in Pittsburgh,
Pennsylvania, 23-25 February.
Ertekin, T., G.R. King, and F.C. Schwerer, 1986. Dynamic Gas Slippage: A Unique Dual-Mechanism Approach to the Flow
of Gas in Tight Formations. SPE Formation Evaluation, February 1986: 43-52.
Franquet, M., Ibrahim, M. Wattenbarger, R.A., and Maggard, J.B. 2004. Effect of Pressure-Dependent Permeability in Tight
Gas Reservoirs, Transient Radial Flow. CIPC Paper 2004-089 presented at the CIPC Conference, Calgary, Alberta
Canada, 8-10 June.
Gierhart, R.R., Clarkson, C.R., and Seidle, J.P. 2007. Spatial Variation of San Juan Basin Fruitland Coalbed Methane
Pressure Dependent Permeability: Magnitude and Functional Form. Paper IPTC 11333 presented at the International
Petroleum Technology Conference, Dubai, U.A.E., 4–6 December.
Javadpour, F., Fisher, D., and Unsworth, 2007. Nanoscale Gas Flow in Shale Gas Sediments. JCPT 46 (10): 55-61.
Javadpour, F., 2009. Nanopores and Apparent Permeability of Gas Flow in Mudrocks (Shales and Siltstones). JCPT 48 (8):
16-21.
Jones, F.O., and W.W. Owens, 1979. A Laboratory Study of Low Permeability Gas Sands. Paper SPE 7551 presented at the
1979 Society of Petroleum Engineers Symposium on Low-Permeability Gas Reservoirs, Denver, Colorado, 20-22 May.
Klinkenberg, L.J., 1941. The Permeability of Porous Media to Liquid and Gases. API Drilling and Production Practice
(1941), p. 200-213.
Loucks, R.G., R.M. Reed, S.C. Ruppel, and D.M. Jarvie, 2009. Morphology, Genesis, and Distribution of Nanometer-Scale
Pores in Siliceous Mudstones of the Mississippian Barnett Shale. Journal of Sedimentary Research 79: 848-861.
Nobakht, M. and Clarkson, C.R. 2011a. A New Analytical Method for Analyzing Production Data from Shale Gas
Reservoirs Exhibiting Linear Flow: Constant Pressure Production. Paper SPE 143989 presented at the North American
Unconventional Gas Conference, The Woodlands, Texas, Texas, 14–16 June.
Nobakht, M. and Clarkson, C.R. 2011b. A New Analytical Method for Analyzing Production Data from Shale Gas
Reservoirs Exhibiting Linear Flow: Constant Rate Production. Paper SPE 143990 presented at the North American
Unconventional Gas Conference, The Woodlands, Texas, Texas, 14–16 June.
Thararoop, P. 2010. Development of a Multi-Mechanistic, Dual Porosity, Dual Permeability Numerical Flow Model for
Coalbed Methane Reservoirs Accounting for Coal Shrinkage and Swelling Effects. Ph.D. Thesis, The Pennsylvania
State University, Department of Energy and Mineral Engineering, University Park, Pennsylvania, USA.
Thompson, J.M., Nobakht, M. and Anderson, D.M. 2010. Modeling Well Performance Data From Overpressured Shale
Reservoirs. Paper CSUG/SPE 137755 presented at the Canadian Unconventional Resources & International Petroleum
Conference, Calgary, Alberta, 19–21 October.
Wang, F.P., R.M. Reed, J.A. Jackson, and K.G. Jackson, 2009. Pore Networks and Fluid Flow in Gas Shales. Paper SPE
124253 presented at the 2009 Society of Petroleum Engineers Annual Technical Conference and Exhibition, New
Orleans, Louisiana, 4-7 October.

You might also like