You are on page 1of 13

Journal of Natural Gas Science and Engineering 22 (2015) 22e34

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage:

Production data analysis of tight gas condensate reservoirs

Hamid Behmanesh*, Hamidreza Hamdi, Christopher R. Clarkson
SPE, University of Calgary, Canada

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 22 August 2014
Received in revised form
31 October 2014
Accepted 3 November 2014
Available online 20 November 2014

The current focus on liquids-rich shale (LRS) plays in North America underscores the need to develop
reservoir engineering methods to analyze such reservoirs. Commercialization of LRS plays is now
possible due to new technology, such as multi-fractured horizontal wells (MFHW). Efcient production
from such reservoirs necessitates understanding of ow mechanisms, reservoir properties and the
controlling rock and uid parameters. Production-decline analysis is an important technique for analysis
of production data and obtaining estimates of recoverable reserves. Nevertheless, these techniques,
developed for conventional reservoirs, are not appropriate for ultra-low permeability reservoirs. There
are substantial differences in reservoir performance characteristics between conventional and ultra-low
permeability reservoirs. LRS reservoirs produce much leaner wellstreams compared to conventional
reservoirs due to very low permeabilities that result in very large drawdowns. Methods for analysis of
two-phase ow in conventional reservoirs, with underlying simplifying assumptions, are no longer
This paper discusses production data analysis of constant owing bottomhole pressure (FBHP) wells
producing from LRS (gas condensate) reservoirs. A theoretical basis is developed for a gas condensate
reservoir during the transient matrix linear ow (drawdown) period. The governing ow equation is
linearized using appropriately dened two-phase pseudopressure and pseudotime functions so that the
solutions for liquids can be applied. The derived backward model is employed to compute the linear ow
parameter, xfk.
Simulation results show that the liquid yield will be approximately constant for LRS wells during the
transient linear ow, from the early days of initial testing, if FBHP is almost constant. An analytical
formulation is used to prove this nding for 1D transient linear ow of LRS wells.
The proposed production data analysis (PDA) method is illustrated using simulated production data for
different uid models and relative permeability curves. Fine-grid compositional and black oil numerical
models are used for this purpose.
2014 Elsevier B.V. All rights reserved.

Production data analysis
Tight gas condensate reservoirs
Multi-fractured horizontal wells
Two-phase ow

1. Introduction
Unconventional resources have become a reasonably reliable
source of energy in North America. Horizontal drilling followed by
multi-stage hydraulic fracturing (multi-fractured horizontal wells
or MFHW) has become a widely employed practice for development of unconventional light oil (ULO) and unconventional gas
(UG) reservoirs. There is an increased importance placed on
reservoir engineering methods to analyze such reservoirs; specifically, production analysis techniques for ULO/UG must be adapted

* Corresponding author.
E-mail address: (H. Behmanesh).
1875-5100/ 2014 Elsevier B.V. All rights reserved.

to account for the unique aspects of MFHW production. Although

there have been a number of recent studies of this topic, particularly for shale gas, typically the developed PDA techniques rely on
simplifying assumptions about the reservoir rock and uid properties. In particular these studies generally assume single-phase
ow, which is not valid for saturated black oil and gas condensate
Hydraulically-fractured vertical and horizontal wells completed
in tight formations typically exhibit long periods of transient linear
ow. Large drawdowns are imposed on these very low permeability
formations to yield economic production. For dry gas and gas
condensate reservoirs, this can give rise to large variations in gas
physical properties and large gas-to-oil mobility ratio gradients
near the fracture face, respectively. Some techniques for analyzing

H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

transient linear ow in single-phase gas reservoirs have recently

been proposed (e.g. Nobakht and Clarkson, 2012). However, relatively little attention has been applied to the development of
analytical methods to address the two-phase ow complexities
associated with transient linear ow analysis of gas condensate
reservoirs. This paper addresses production data analysis of liquidsrich shale (LRS) reservoirs with a focus on gas condensate uid
systems. The results can be used in support of short- and long-term
production forecasts for liquid-rich shale.
2. Background and statement of problem
Well test analysis relies heavily upon solutions to ow equations
assuming single-phase ow of a liquid with small and constant
compressibility and viscosity. These solutions are often adapted for
rate-transient analysis (RTA) problems. If the wellbore pressure
falls below the bubble or dew point pressure, two-phase ow
conditions exist and the solutions for liquids are no longer valid.
However, liquid solutions for pressure- and rate-transient analysis
can still be used if (for example) pseudovariables are used to correct
for pressure-dependent uid properties and for multi-phase ow
(Fraim and Wattenbarger, 1987).
Analytical modeling of pressure/rate transients for multiphase
ow in porous media is challenging because the nonlinearities
associated with the uid must be contended with. Since the 1950's,
many investigators have proposed different approaches to tackle
this problem. Perrine (1956) and Martin (1959) suggested the
application of total mobility and the single-phase-ow concepts for
analyzing well responses when the pressure at the wellbore falls
below the bubble-point or the dew-point pressure. The premise of
their theory is that at some distance from the wellbore, the saturation gradient is small enough that can be ignored. Under such
circumstances, mobility and compressibility term in the singlephase ow expression may be replaced by total mobility (sum of
the mobility of each of the owing phases) and total compressibility (sum of the compressibility of each phase with each
compressibility term weighted by the saturation of that phase).
Later on, different investigators adapted various methods for twophase ow in radial models. Raghavan (1976) examined solutiongas-drive systems and proposed that a two-phase pseudopressure be used to determine formation permeability from well
test data. Aanonsen (1985) extended Raghavan's work and introduced the concept of the reservoir integral. Jones and Raghavan
(1988) and Camacho and Raghavan (1989) analyzed drawdown
and buildup responses in gas-condensate and solution-gas-drive
systems respectively by using Aanonsen's concept. Jones and
Raghavan (1988), Camacho and Raghavan (1989) and Hamdi
et al., 2013 used the ideas of Aanonsen (1985) to analyze drawdown and buildup responses in gas-condensate and solution-gasdrive systems. Some investigators utilized the powerful similarity method mathematical technique for analysis of these nonlinear problems. O'Sullivan and Pruess (1980) and O'Sullivan
(1981) proposed similarity solutions to highly nonlinear multiphase ow problems in geothermal well test analysis. The similarity method has also been applied in the study of multiphase ow
in solution gas-drive and gas condensate reservoirs (Be et al.,
1989; Vo et al., 1990).
In recent years, attention has shifted to low-permeability (unconventional) wells where transient linear ow is a dominant owregime and pressure gradients are large. Several studies (Ibrahim
and Wattenbarger, 2006, Poe, 2002, Miller et al., 2010 and
Nobakht and Clarkson, 2012) focused mainly on the ow of single
phase gas to obtain the linear ow parameter, xfk. Ibrahim and
Wattenbarger noted that the transient linear ow solution, derived
for liquids, required the use of pseudopressure and empirical


drawdown factor to account for changing gas properties with

pressure. Nobakht and Clarkson (2012) later used a corrected
pseudotime (as well as pseudopressure) to account for these effects. More recently, Qanbari and Clarkson (2013) and Chen and
Raghavan (2013) independently presented a similarity-based
methodology for innite-acting gas linear ow analysis for constant owing bottomhole pressure.
The analysis of two-phase ow, during the transient linear ow
period, provides unique challenges that are now receiving attention. The nonlinearities associated with two-phase ow (pressuredependent properties and high saturation gradients near the
wellbore) cannot be accounted for only by incorporating saturation
changes in the two-phase pseudopressure e more rigorous treatment is required to account for pressure-dependent properties of
the uids. Furthermore, the techniques that have been developed
for calculation of two-phase pseudopressure in radial models
cannot be condently used for linear models. This is because, for
radial ow, a steady-state region of signicant size develops,
whereas in linear ow, the steady state region progresses only a
small distance away from the wellbore.
In this work, we have developed a novel rate transient analysis
(RTA) technique for analyzing tight gas condensate reservoirs
during the transient linear ow period. The new method includes
evaluation of two-phase (gas condensate) pseudopressure, which
incorporates a new method for calculation of the saturationepressure relationship analytically, and two-phase pseudotime, which are used to linearize the diffusivity equation. The
reservoir uid is assumed to be a rich gas condensate uid. The
initial reservoir pressure is set close to the saturation pressure, and
the well is producing against a constant FBHP. The suggested
techniques presented in this work are valid for any multiphase ow
scheme provided that uid ow can be described by the diffusivity
equation based on formation volume factor formulations. This
manuscript is focused on the theoretical development of our multiphase analytical methods. We demonstrate the practical applicability of the presented method with eld cases. The signicance of
this work is that we have proven that the liquid-analog solution can
be applied (through use of carefully dened pseudovariables) for
low-permeability gas condensate reservoirs (under certain conditions), which are currently being exploited extensively by industry.
Several commonly observed behaviors of tight gas condensate
wells are also explained analytically.
The manuscript is organized as follows. First, we will discuss
two methods for linearizing the diffusivity equation as well as
backward solutions in the Model Development section. Next we
will discuss the setup of numerical simulation models in the
Simulation Model Setup section, followed by validation of our
new approaches using numerical simulation in the Model Verication Using Simulation section. Application of our new methods
for establishing the linear ow parameter (xfk) is demonstrated in
the Application of New Methods for Determination of xfk section. Finally, we will provide some brief conclusions obtained from
this work.
3. Model development
The emphasis of this study is on the transient (matrix) linear
ow regime associated with hydraulically-fractured tight gas and
shale gas condensate reservoirs. As observed from simulations,
when the pressure falls below the dew point pressure, the oil
saturation buildup near the fracture plane can cause a considerable
reduction in gas phase mobility (up to 70% according to our
simulation results). The corresponding diffusivity equation is highly
nonlinear because the uid properties are a strong function of
pressure and mobility is changing with saturation. Linearization of


H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

the diffusivity equation for non-linear ow behavior is possible

through the use of pseudovariables, i.e. pseudopressure and the
pseudotime. The pseudopressure transform is an exact mathematical transform whereas pseudotime is based on empirical
In this section, two methods for solving the partial differential
equation (PDE) corresponding to the tight gas/shale gas condensate
case are developed. The backward solutions for obtaining the linear
ow parameter, xfk, are given using these two approaches.
Further, important considerations for calculating pseudovariables,
which are required to enable the use of liquid solutions, are discussed. In a later section, our approximations are validated with
numerical simulation.
For the rst method of solving the PDE, the Boltzmann transform variable is applied. The corresponding pressure prole for a
single-phase liquid during innite acting linear ow for constant
rate and constant FBHP production are given in Eqs. (1) and (2),
respectively (Wattenbarger et al., 1998):

i vp
v h
rs krg mg Bgd kro =mo Bo

rs Sg Bgd So =Bo

0:00633$k vt


The initial and boundary conditions are:

p pi
p pwf
p pi


t >0
t >0

The same simplifying notations of Be et al. (1989) are used:

S So .
a krg mg Bgd Rs kro =mo Bo
a rs krg mg Bgd kro =mo Bo
b Sg Bgd Rs So =Bo
b rs Sg Bgd So =Bo
p x vx=vps x2a; a; b; b


h p

pD xD ; tD p 2 tD =p$exp  x2D 4tD
 . pi
 xD $erfc xD 2 tD


As derived in Appendix A, xfk is calculated according to

following equation:

 . p
pD xD ; tD erfc xD 2 tD



For the constant rate and constant pressure boundary condition

at the wellbore, the dimensionless pressures are dened as pD kh
(pi  p)/qBm and pD (pi  p)/(pi  pwf), respectively. The pressure
prole for constant pressure production is only a function of the xD/
tD variable. However, for constant rate production this is not the
case. This will be used later to explain some of the observations
made from numerical simulation results.
In the second method, it is hypothesized that the linearized
diffusivity equation for a dry gas reservoir is still valid if we use
two-phase pseudotime, ta,tp.



h$ Ppi  Ppwf $m

l=c* f


The detailed discussion on how to calculate pseudopressure is

given later.
For a single-phase gas reservoir, Eq. (5) can be simplied to Eq.
(6), which is exactly the same equation derived for dry gas (Ibrahim
and Wattenbarger, 2006). One should note that the appearance of a
constant 2 in the numerator of Eq. (6) is due to the fact that we
consider ow from only one side of the fracture.



2  315:4T

h$ fmct i $ Ppgi  Ppwf $m


3.1. Method 1: Boltzmann Transform Variable.

The combination of the Darcy and continuity equations yields
the governing PDE for ow of gas and oil components in a gas
condensate reservoir. The simplifying assumptions of constant rock
properties, thickness and temperature are used. It is further
assumed that Darcy's law is valid and capillarity and gravity effects
are ignored. Other effects such as gas slippage and turbulent ow
effect are also ignored e we will investigate these effects on the
outcome of rate-transient analysis in future work. However, we
assume that slippage effects are negligible because the relatively
large permeabilities (and hence pore sizes) assumed in this work
(the range of permeabilities studied in this work vary between
0.00001 and 0.1 md). Although turbulent ow could be expected to
occur in the fractures, the effect will manifest itself as a skin and
will not affect the derivation of linear ow parameter. We have also
assumed that the performance of the system can be modeled using
a single fracture with an innite conductivity.
As shown in Appendix A, the consequent governing PDEs (Formation Volume Factor formulation) for gas and oil components are
written as follows:

i vp
v h .
krg mg Bgd Rs kro =mo Bo

v .
Sg Bgd Rs So =Bo

0:00633$k vt

3.2. Method 2: Two-Phase Pseudopressure and Two-Phase

In the second method, it is assumed that, for the two-phase gas
condensate system, the linearized diffusivity equation for dry gas is
still valid if we use two-phase pseudopressure instead of dry gas
pseudopressure and a two-phase pseudotime instead of dry gas
Analogous to the dry gas formulation, a linearized form of the
diffusivity equation for two-phase ow conditions is given as:

v2 ppg

0:00633$k vta;tp

An empirical pseudotime transformation (Heidari and Gerami,

2011) is used here, and given as:







vb vPpg


k is calculated according to following equation:



h$ f$ Ppgi  Ppwf $mCP


H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

where m CP is the slope of 1/qgas versus ta. A derivation of Eq. (9) is

given in Appendix B.
Calculation of pseudopressure and pseudotime
for the backward
solution is required for determination of xf k. Assessment of these
parameters will be discussed further in the next section.

3.2.1. Pseudopressure Calculation and SaturationePressure

In both methods described above, pseudopressure is required to
linearize the diffusivity equation. A problem with pseudopressure
calculations for multiphase problems is that pressure must be
related to saturation to perform the calculations.
For radial ow problems, several investigators (Aanonsen, 1985;
Serra, 1988; Thompson and Reynolds, 1997) have shown that for the
constant-rate drawdown in a homogeneous, innite-acting gas
condensate or solution-gas-drive reservoir, the pressures, saturations, phase compositions, and hence the PVT properties and
relative permeabilities are unique functions of the Boltzmann variable. Furthermore, we see from Eq. (2) that for single-phase linear
ow, the pressure prole is also a function of this variable. By
assuming that the Boltzmann transformation is valid for two-phase
linear ow conditions under constant FBHP (skin is assumed zero),
a unique pressureesaturation (PeS) relationship can be found. This
has been veried against numerous compositional and black oil
simulations. This in turn implies that under a constant drawdown
condition, the pseudopressure drop is also invariant.
Following the Be et al. (1989) procedure, a saturationepressure
relationship for linear ow conditions is derived as follows:


method to obtain the saturation prole using the following



mg Bgd
1  rs$GOR mo Bo


A constant GOR during the transient linear ow period is

observed for all simulated LRS cases when producing against a
constant owing pressure. This observation is consistent with the
ndings of Whitson and Sunjerga (2012) and with our mathematical derivations for the linear ow regime. Based on numerical
simulations, Whitson and Sunjerga (2012) found that the liquid
yield (rp 1/GOR) remains approximately constant for long periods
of constant FBHP in LRS reservoirs. They noted this behavior to be
characteristic of LRS reservoirs for the case of innite-acting ow,
when the ow is 1D planar. Be et al. (1989) gave an analytical
solution that yields a constant GOR for constant rate innite-acting
radial reservoirs. The same conclusion can be made here based on
Eq. (12) and using the same approach as Be et al. to derive the GOR
behavior. This is always true where both the PDE and the associated
boundary and initial conditions can be expressed as a function of
the Boltzmann variable. This directly implies that for the linesource radial constant rate solutions, the PDE and the boundary
and initial conditions can be expressed in terms of the Boltzmann
variable, leading to a constant GOR. In the case of linear ow
(constant rate production), the inner boundary condition cannot be
expressed in terms of the Boltzmann variable and hence a constant
GOR cannot be obtained; indeed, it increases with time.

The limiting cases of Eq. (10) occur as y / 0 (limiting case I) or

in the late time (i.e. steady-state ow or equivalently close to the
wellbore) and as y / (limiting case II) or in the early time (or
equivalently at a far distance), where the simplied forms can be
derived for practical purposes. These limiting cases are given by:

3.2.2. Pseudotime Calculation

Use of the pseudopressure transformation alone does not account for all the nonlinearities involved with two-phase ow of oil
and gas. Interpreted results from simulated tests do not correlate
well with the liquid reference curve if the initial value of (c/l)*i is
used in Eq. (5). The empirical pseudotime transformation (Eq. (8)) is
used to linearize the temporal part of the diffusivity equation. For
boundary dominated ow, the integrand parameters are evaluated
at an average reservoir pressure. During the boundary dominated
ow regime, average reservoir pressure is declining by time. For
transient ow, however, we use the average pressure in the distance of investigation (DOI) as proposed by Anderson and Mattar
(2007). The average pressure in the DOI tends to be a constant
value for constant owing pressure as demonstrated in the work of
Nobakht and Clarkson (2012) for dry gas reservoirs. Likewise for gas
condensate reservoirs, as shown from our simulation results, the
same conclusion is drawn. This behavior is discussed further below.

ds=dp aa0  aa0 =aa_  aa

3.3. Average Pressure in the Distance of Investigation

ds aa0  aa0 dp=dy 2yab0  ab0

aa_  aadp=dy
2y ab_  ab_


Eq. (10) describes an unsteady pressure-saturation path. The

auxiliary equation for calculation of dp/dy (Eq. (11)) is not useful for
linear ow as wellbore pressure is independent of time.


dy A$a$ fk$m


ds=dp ab0  ab0 = ab_  ab_


Eq. (12) is for the long producing times and Eq. (13) is for early
producing times. For a specic uid system and a set of relative
permeability data, ds/dp can be easily calculated by these
Eq. (12) gives the saturation derivative for the steady state
condition where the saturation values (as a function of pressure)
can be obtained by integration. However, there is another, easier

Z 2p

Average pressure in the distance of investigation is calculated by

using the material balance equation for a tank of variable size. The
size of the tank is directly proportional to the distance of investigation in the linear ow regime. Distance of investigation is used in
well testing and is a measure of how fast the diffusion front moves
through porous media. The average pressure in the distance of
investigation can be estimated by the following equation using an
iterative scheme. The detailed derivation of Eq. (15) is given in
Appendix C.


1000$rgsc 5:615$rosc =GOR Bgdi $ mct i Ppgi  Ppwf q
$ 1
$ vb vPpg p
1000=5:615rgsc rosc rsi 0:194  62:55



H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

4. Simulation model setup

In order to validate the approaches described above, a series of
ne-grid black oil and compositional simulations were conducted.
Eclipse 100 and 300 (Schlumberger, 2010) are used for reservoir
simulations. The simulations were also used to study the performance of typical tight gas condensate reservoirs. The steep pressure and saturation gradients occurring near the hydraulic fracture
plane are captured using a geometric grid spacing scheme. Several
sensitivity analyses were performed to nd the optimum number
of grids and the geometric ratio. The distance between fractures is
discretized to an odd number of grids expanding away from the
fracture plane midway to the adjacent fracture and shrinking afterward. The mathematical formulation describing the relationship
between fractures distance and the geometric ratio, and size of rst
grid block is given by Eq. (16):

x0 $ an2 an1  a  1

Fig. 2. Schematic showing linear ow toward two fractures in multi-fractured horizontal well. The shaded rectangle indicates the element of symmetry. In our simulation
sensitivities, twice the area of shaded rectangular is modeled and rates from the halffracture model are scaled up to full well rates.


In Eq. (16), x0 is the grid block in which fracture is situated, d is

the fracture spacing, a is the geometric ratio and n is the number
of grid. A geometric ratio of 1.28 is used for grid size, with a large
distance d between the wells to ensure that the well is in transient ow for the entire producing period. In this work, an innite
conductivity fracture is assumed. The conductivity behavior of a
fracture is quantied by dimensionless fracture conductivity, FCD:


kf wf


FCD 50 is considered as the innite conductivity limit for

conventional reservoirs, however, as the matrix permeability of the
formations is very low, a higher FCD is selected in our cases. A single
grid is assigned to the fracture. As long as fracture conductivity
(fracture permeability times fracture width) is kept the same, there
will be no actual difference in the performance of different simulation cases. For evenly spaced fractures throughout the horizontal
section of the well, each fracture forms an element of symmetry.
The smallest element of symmetry is a quarter of the open area of
ow towards the fracture face from both sides. Fig. 1 illustrates a
sector model schematic used for the simulations. Fig. 2 shows a top
view of linear ow towards the fracture and the smallest element of
symmetry is shaded. In all simulations, twice the area of this
element is used and the rate performance for a fully gridded
reservoir is simply 2Nf-fold of the simulated model. Fig. 3 illustrates
the 1D planar fracture geometry gridding. We have assumed that
the drainage beyond the fracture's tips is negligible and used 1D
planar model for our mathematical and simulation modeling.
For numerical simulation sensitivities, two sets of relative
permeability curves, one with saturation exponents of 2.5 for both
oil and gas, and the other with saturation exponents of 3.5 and 5 for

Fig. 1. Schematic showing multi-fractured horizontal well geometry used for simulation. A subsection of this schematic, highlighted with dashed lines, is illustrated in
Fig. 2.

Fig. 3. Schematic illustrating model gridding.

the gas and the oil respectively, are used. Furthermore, a third set of
measured relative permeability data for a tight formation (i.e. a
porosity of 0.08 and a permeability of 0.18 md), in absence of the
initial water saturation, are used in the simulations. Relative
permeability curve 3 has been measured in the HerrioteWatt
University Gas-Condensate Research Lab. The pertinent relative
permeability data are given in Fig. 4. In addition, three different rich
gas condensate uids are used in this study. Table 2 provides the
key reservoir uid properties and composition of uid. Fig. 5 presents the calculated Constant Volume Depletion (CVD) data for the

Fig. 4. Relative permeability data used in simulation. Relative permeability set 1 is

actual measured relative permeability data obtained for a tight formation with
permeability 0.01 md.

H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34


Table 1
Reservoir Simulator required input data.



Matrix Permeability, k
Fracture permeability
Reservoir Thickness, h
Relative permeability
CVD of in-situ uid
Matrix porosity, f
Fracture half-length, xf
Wellbore diameter
Fracture conductivity
Fracture width
Fracture porosity

Fig. 4
Fig. 5



Table 2
Key reservoir PVT properties.

Initial pressure (psia)

Dew point pressure (psia)
Initial solution oil in gas (STB/MMscf)
Maximum CVD liquid dropout (%)
Reservoir temperature ()

Fluid 1

Fluid 2

Fluid 3




different uid systems used in this study. Fluid 1 and uid 2 are
synthetic uids and consist of 3 and 5 components respectively.
These uid samples are the Mixture 4 of Jones and Raghavan
(1988) and Rich Gas B uid sample from Heidari and Gerami
(2011). Fluid 3 is a real rich gas-condensate uid model and consists of 9 pseudo components with a maximum liquid-drop-out of
0.22 (from the CVD test) and a dew point of 3525 psia measured at
reservoir temperature of 200  F (Composition 1 of gas condensate
system from Shi's PhD thesis (2009)).
Porosity and absolute permeability are kept constant in the
simulation models. Other required information for the reservoir
simulator is given in Table 1. Gravitational and capillary effects are
neglected and the initial reservoir pressure is set to be only a few
psia above the saturation pressure for all simulated cases. This
initial boundary condition is the extreme case in terms of existing
nonlinearity as most of the investigated area is under two-phase
ow. Modication of our methods for undersaturated reservoirs
with different degrees of undersaturation is discussed elsewhere
(Behmanesh et al., 2014a,b).

Fig. 5. Constant volume depletion data for 3 different rich gas condensate uids used
in the numerical simulation.

Fig. 6. Simulated production performance from a liquid rich shale gas condensate
reservoir producing at constant FBHP (1000 psia) using a 1D planar fracture geometry.
For this simulation run, Sgi 1; uid composition 1 and relative permeability set 1
were used (pi 5600 psia, rsi 171 stb/MMscf).

Some sensitivity studies have been performed to understand the

effect of absolute permeability, relative permeability, uid type,
and production constraints on the reservoir performance. A constant GOR during linear transient period for all LRS producing
against constant pressure was observed (Fig. 6). This observation is
consistent with the ndings of Whitson and Sunjerga (2012) for the
radial ow and our mathematical derivations for the linear ow
5. Model verication using simulation
In this section, various model assumptions are tested, including
the calculation of pseudovariables and saturationepressure relationships. The accuracy of linear ow parameter xfk calculations
using Method 1 and Method 2 is also examined.
In order to verify our results, the gas and oil relative permeabilities and the uid ow properties of each simulation grid cell
were exported from the reservoir simulator and used in the evaluation of pseudopressure. Fig. 7 shows the saturationepressure

Fig. 7. Saturationepressure relationship obtained using uid 2 and relative permeability set 1. Reservoir saturations are plotted versus pressure at the end of rst and
second year. Saturation and pressure history of grid block 30 are also displayed. The
simulated well is owing against constant FBHP of 1000 psia. The unique saturationepressure relationship is valid during constant FBHP transient ow.


H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

Table 3
Numerical result of distance of investigation and average properties calculation.

Distance of
investigation, ft

Fluid 2-Relative permeability 1

1st Year
2nd Year
3rd Year
Fluid 3-Relative permeability 1
1st Year
2nd Year
3rd Year
Fluid 3-Relative permeability 3
1st Year
2nd Year
3rd Year

Average pressure,

Average condensate







relationship from numerical simulation results. A unique pressure

saturation relationship exists for all time steps. A xed average
pressure in the distance of investigation is calculated from the
pressure prole. This is also implied by the material balance
equation given by Eq. (15). The values of the distance of investigation, and the corresponding average pressure and the condensate
saturation for different set of relative permeability and uid system
are given in Table 3.
According to the simulation results, for the case of owing at
constant FBHP condition, the average pressure, as well as the
average condensate saturation within the DOI, is invariant with
time. Similar studies were conducted for a well owing with constant gas rate production and it was concluded that the average
pressure declines, while the average condensate saturation increases in the DOI over the time. The invariant average condensate
saturation in the DOI may be used to explain the constant GOR for
the constant FBHP cases. Conversely, as the condensate saturation
increases in the DOI, the GOR increases for constant rate production
during transient linear ow.
5.1. Saturation-Pressure in the Distance of Investigation
Comprehensive numerical experiments showed that the full
path of pressure and saturation can be estimated using the two
limiting cases described previously (limiting case I and limiting
case II) and interpolation. In this study, we used a cubic polynomial

Fig. 8. Saturationepressure relationship derived for uid 2, relative permeability set 1.

The numerical simulation-derived curve, interpolation using polynomial function, CCE
and CVD tests, steady state, and limiting case II are also shown.

Fig. 9. Saturationepressure relationship derived for uid 3, relative permeability set 1.

The numerical simulation-derived curve, interpolation using polynomial function, CCE
and CVD tests, steady state, and limiting case II are also shown.

of form a3x3 a2x2 a1x a0 for interpolation. This selection

provides reasonable results for many numerical simulations. The
four cubic polynomial coefcients (ai's) are obtained using (p, s)
points from the two limiting cases and their associated derivatives
(p, ds/dp). These points are selected close to where the limiting case
I starts to deviate from its initial trend, and where the limiting case
II deviates from the Constant Composition Experiment (CCE test).
This approach ensures that the slopes and trends of the two
limiting cases are reasonably preserved in the interpolation polynomial. Therefore, the polynomial is anchored to match the PeS
path in the early and late times (or equivalently the near and far
wellbore areas), and is used to predict the path for the intermediate
times (or areas). The use of a cubic polynomial for the PeS path
prediction yields negligible error in the calculation of pseudopressure with respect to reservoir integral pseudopressure. Fig. 8,
Fig. 9 and Fig. 10 show the relevant plots for different saturationpressure paths for different sets of uid models and relative
permeability curves. The CCE experiment, CVD test and the two
limiting cases are shown in each plot. The limiting cases I and II
adequately predict the saturation-pressure path at the pressure
close to the FBHP and the dew point pressure, respectively. For

Fig. 10. Saturationepressure relationship derived for uid 2, relative permeability set
3. The numerical simulation-derived curve, interpolation using polynomial function,
CCE and CVD tests, steady state, and limiting case II are also shown.

H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

Fig. 11. GOR and pressure prole for simulated unconventional (linear) gas condensate
reservoir. The steady state region advances very slowly even after 2 years of well
production against constant FBHP. Only 6% of the pressure-saturation path can be
correctly predicted from the steady state method.

pressures other than these regions, the corresponding cubic polynomial is used. The mere use of CCE or CVD experiments alone did
not give favorable results as these experiments underpredict or
overpredict the PeS path, respectively, depending on the selection
of uid system and relative permeability curve.
Although some authors (Jones and Raghavan, 1988; Camacho-V
and Raghavan, 1989) demonstrated that the steady state assumption can be used to describe the pressure-saturation path for conventional reservoirs, this was found to be a poor approximation for
tight formations. For example, Figs. 11 and 12 show the GOR and
pressure proles for a conventional (i.e. a radial homogeneous
reservoir with k 16 md) and a tight formation (i.e. a linear homogeneous reservoir with k 0.01 md). The relative permeability
data and the uid system of these two cases is the same. These plots
reveal that for the tight formation, the constant GOR (i.e. steady
state region) is limited to around 50 ft away from the wellbore after
2 years of production. This is indeed equivalent to a steady state
region after only 10 days of production in the conventional case.
The pressure drop within the steady state region of the conventional case contributes to almost 80% of total pressure drop


Fig. 13. The reciprocal of pseudotime integrand for different sets of uid and relative
permeability curves.

(pi  pwf) while this is only around 6% for the tight formation. This
simple example clearly shows that the steady state assumption
cannot fully characterize the reservoir performance in tight

5.2. Pseudotime in the Distance of Investigation

Calculation of pseudotime is simplied as follows due to the use
of average pressure in this distance of investigation for the constant
owing pressure case (average pressure is constant in the distance
of investigation):



vb vPpg p


The reciprocal of the pseudotime integrand, vb/vPpg, is plotted in

Fig. 13. This term simplies to (mgcg) for dry gas reservoirs. For the
two-phase ow condition, a correction factor, fCP, is introduced to
correct the xfk, which is calculated from the initial rock and uid
properties and Eq. (20) can alternatively be written in terms of fCP
and slope mCP:



k p 
fmct i $h Ppgi  Ppwf $mCP


where fCP is given as follows:


u mg ct i

vb vPpg 


Table 4
Rate transient analysis of different uid types and relative permeability curves. The
exact xfk 25.
Relative Permeability 1


k h$P 62:55

Fluid 1
Fluid 2
Fluid 3
xf k
Fig. 12. GOR and pressure prole for a simulated conventional (radial) gas condensate
reservoir. The steady state region advances much faster than linear model. We see that
the steady state method for prediction of saturationepressure relationship is valid in
this case.

Fluid 1
Fluid 2
Fluid 3


pwf $m

Permeability 3

$ l=c*i =f



Permeability 2


fmct i $Ppgi Ppwf $m0




H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

6. Application of new methods for determination of xfk

Eqs. (5) and (19) is the backward solution used for determination of the linear ow parameter, xfk. The procedure for analyzing
production data during transient linear ow for LRS gas condensate
wells subject to constant FBHP is adapted from Nobakht and
Clarkson (2012) as follows:
1. Plot the inverse gas rate versus the square-root of time in Cartesian coordinates. The data should form a straight line during
transient linear ow from which the slope (mCP) can be
2. Determine xfk using the initial rock and uid properties. The
solution is given as:



k p 
h fmct i $ Ppgi  Ppwf $mCP


3. From the material balance equation evaluate the average pressure in the region of investigation.
4. Evaluate the correction factor, fCP, and multiply xfk from step 2
by fCP.

Fig. 14. Illustration of the effect of corrected (for multiphase ow) pseudotime on the
slope of the square-root-of-time plot for uid 3, relative permeability set 1.

To demonstrate the applicability of the above procedure, we

selected ve combinations of uid compositions and relative
permeability curves from the sets previously discussed, and
compared our derived linear ow parameter with that used in
numerical simulation. For all the cases studied in this work the
exact value of xfk is 25 (ft md1/2).
As a rst attempt, the pseudopressure (i.e. the reservoir integral)
and the pseudotime transformations were evaluated based on the
actual pressure-saturation path and the average pressure from the
simulation results. For each simulation case the results of calculated
xfk from Eq. (5) (Method 1) and Eq. (9) (Method 2) are listed in
Table 4. (l/c)* is evaluated at initial reservoir pressure, pi, in Eq. (5)
The use of initial properties led to overestimation of xfk for all the
simulated cases. The calculated values clearly show that the liquid
analogy is achieved when the pseudopressure and the pseudotime
are used in Method 2. Method 1 does not achieve as good of a results as Method 2 because (l/c)* is assumed to be constant, which is
commonly assumed for radial ow.
In another attempt, and for practical application for each combination of uid and relative permeability, the pseudopressure and
pseudotime were calculated using our interpolation method to
derive the saturationepressure relationship, and the pseudotime
was evaluated at the average pressures in distance of investigation
(obtained from the material balance equation). In this attempt,
satisfactory results are also obtained and the corresponding relative
error analysis is tabulated in Table 5. The relative error is evaluated
based on the following equation:

The effect of the corrected (for multiphase ow) pseudotime

factor on the slope of the square-root-of-time plot (plot of 1/qg
versus t and ta) is illustrated in Fig. 14 for uid 2 and relative
permeability set 1. We see that the slope correction is signicant
and greater than tting error.

% relative error

jexact  calculatedj

7. Field case, MFHW completed in liquid-rich tight gas

condensate reservoir
This is an example of an MFHW completed in a tight gas
condensate reservoir in Western Canada. Details of well location,
formation, completions and reservoir properties are withheld to
preserve operator condentiality. The well uid production rates
and uid production ratios are given in Fig. 15. The uid production characteristics (Fig. 16) are similar to that expected for a wet
gas reservoir. Flowing pressures are approximately constant after
100 days of production, and there our analysis (which assumes
constant owing pressure), is focused on data after 100 days.
Fig. 16 shows the inverse gas rate vs. the square root of time plot


Table 5
Rate transient analysis of different uid types and relative permeability curves. The
exact xfk 25.
Relative Permeability 1
xf k p62:55
h$ fmct i $Ppgi Ppwf $m0
Fluid 1
Fluid 2
Fluid 3

Permeability 2

Permeability 3


Fig. 15. Fluid production rates and owing pressures for eld case.

H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

Fig. 16. Condensate and water/gas ratio (CGR/WGR) for eld case.

for this case. The data form a straight line (after 100 days), which
indicates that linear ow is the dominant ow regime. Using the
slope of the plot in Eq. (9) the total xfk is calculated (the sum of
xfk from each stage) is calculated to be 21.5 ft md0.5. And if
considering k 0.0005 md, the individual fracture half-length,
assuming equal contribution from fractures along the well, is
1000 ft. The resulting fracture half-length is consistent with that
used in numerical model history-matching, which will be presented in a future paper (Fig. 17).
8. Conclusions
In this study, the production performance of liquid rich tight
gas/shale gas condensate wells under constant FBHP in the linear
ow regime was studied. Different gas/condensate uid systems
and relative permeability data have been used. Numerous
compositional and black oil simulations were performed and
validity of the driven mathematical relationships was veried.
Some key observations and conclusions of this study are listed as
1. By using the pseudovariables, the diffusivity equation is linearized and the liquid analogy is applied. This leads to an excellent
prediction of xfk.

2. A unique pressureesaturation relationship for the linear systems under multiphase ow is proved and two limiting cases for
the early and the late time behavior are obtained.
3. The production performance of unconventional (tight formations) is different from the conventional reservoirs. It is found
that the unsteady pressure-saturation path for the tight formations cannot be fully characterized by the commonly-used
steady-state path which is used for conventional reservoirs.
However, a combination of a tted polynomial of degree 3 and
data obtained from the two limiting cases was found to
adequately approximate the unsteady pressure-saturation path.
4. A material balance approach was used to calculate an invariant
average pressure within the investigation distance. This average
pressure is critical in the determination of the two-phase
pseudotime variable.
5. It is proved analytically that GOR is constant for transient linear
ow with the constant owing bottomhole pressure constraint.
This implies the existence of a steady-state region around the
wellbore, although very limited in extent.
The authors would like to thank the sponsors of Tight Oil Consortium for funding this project. Chris Clarkson would like to
acknowledge Encana and Alberta Innovates Technology Futures for
support of his Chair in Unconventional Gas and Light Oil Research
in the Department of Geoscience, University of Calgary. Hamidreza
Hamdi thanks the Interactive Reservoir Modeling, Visualization and
Analytics Research Group at the University of Calgary for supporting his postdoctoral fellowship.

Fig. 17. Inverse gas rate vs. square-root-of-time plot for Field Example.



geometric ratio.
fracture half area, ft2.
dry gas formation volume factor, ft3/scf or RB/Mscf.
oil formation volume factor, RB/STB.
constant composition experiment
constant volume depletion
fracture spacing, ft.
distance of investigation
owing bottomhole pressure
dimensionless fracture conductivity.
gas oil ratio, Mscf/STB.
initial gas in place, MMscf.
formation permeability, md or nd.
fracture permeability, md.
gas relative permeability.
oil relative permeability.
length of square matrix block, ft.
fracture length, ft.
length of horizontal well, ft.
Liquid Rich Shale.
number of fractures per well.
number of cells inside the matrix blocks in x- and ydirection for fracture network grids.
grid block index.
initial moles in place, lbm.
number of cells outside the matrix blocks in x- and ydirection for fracture network grids.
produced moles, lbm.
oil gas ratio, STB/MMscf.
pressure, psia.
based pressure, psia
initial pressure, psia.


H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

Z 2p

average pressure, psia

owing bottomhole pressure (FBHP), psia.
Surface gas rate, scf/D, Mscf/D, MMscf/D.
Surface oil rate, STB/D.
producing oil-gas ratio or liquid yield, STB/MMscf.
solution oil-gas ratio, STB/Mscf.
initial solution oil-gas ratio, STB/Mscf.
solution gas-oil ratio (GOR), scf/STB, Mscf/STB.
initial solution gas-oil ratio (GOR), scf/STB, Mscf/STB.
rate transient analysis
time, hrs, days.
fracture width, ft.
Fracture half-length, ft.
two-phase Z-factor.
two-phase Z-factor at average pressure.
gas viscosity, cp.
oil viscosity, cp.


Appendix A. Derivation of the constant FBHP solutionmethod 1


The nonlinearities in Eq. A-1 given by coefcient a, can be

eliminated by introducing an integral transformation as follows:


The use of pseudopressure (Eq. A-2) in Eq. A-6 yields:

1 d2 Ppg
2 dy2


where (c/l)* is the generalized compressibility mobility ratio.

Expanding the terms gives:

c=l* a1 $db=dp a1 $ b0 bds=dp


If we assume (c/l)* is a constant, and can be calculated at a

reference pressure, then the solution for Eq. A-8 is:

Ppg y  Ppg pwf

p$ Ppi  Ppwf $

q Zy

c=l* $ exp  c=l* t2 dt



Note that for the single-phase ow of gas (c/l)* will be simplied

to (mgct). Eq. A-8 is an ordinary differential equation (ODE) with
corresponding boundary conditions. Use of the Boltzmann variable
causes the initial and boundary conditions in Eq. (3) to convert into
two boundary conditions. The resulting boundary conditions are:

p pwf
p pi

Focusing on the gas component, Eq. (3) may then be written as:

f vb

k vt

1 d
2 dy




For oil reservoirs, an analogous pseudopressure can be dened

(Be et al., 1989; Fevang and Whitson, 1996). To calculate the
pseudopressure, a relationship between saturation and pressure is
required, and is used in the evaluation of relative permeabilities
with pressure.
The Boltzmann variable of form y (fx)/2(0.00633 kt) is
now applied to simplify Eq. A-1. This is achieved by substituting the
space and time derivatives with the Boltzmann derivative through
application of the chain rule. From chain rule, the following
equality exists:

vm vm vy

vy vn


where m can be either p or b and n can be either x or t. The

Boltzmann derivatives with respect to space and time are given in
the following equations:

vx 2 0:00633k t


 p p
4 0:00633kt t


Therefore, by substituting the derivatives in terms of Boltzmann

variable, the analogous equations for the oil and the gas components will take the following forms:

1 d
2 dy


Moreover, the surface gas ow rate can be calculated as follows:

qgsc A$k$a 


where A 2xfh. All of the derived equations are based on this area
open to ow at the fracture face (i.e. to the wellbore). Merging Eqs.
A-10 and A-11, and after some mathematical manipulation, the
following equation is obtained:


 p q$ t
xf k$ Ppi  Ppwf $h f$ c=l


As observed from numerical simulation, when ow is essentially

1D linear (i.e. planar slab fracture) in LRS wells producing at constant FBHP, both gas and oil rates vary with t. This observation
mathematically implies 1/qg mt where m is the slope of
characteristic plot of 1/qg versus t.
Furthermore, xfk is calculated according to following

xf k


h$ Ppi  Ppwf $m

l=c f


Appendix B. Derivation of the constant FBHP solutionmethod 2.

The obtained partial-differential equation and its conditions (Eq.
(7)) are linear with respect to pseudofunctions. If we dene
dimensionless parameters as:

H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34




where, ni rgsc rosc rsi $Gand np 0 rgsc qgsc rosc qosc $dt. As
shown previously, GOR is constant during the transient linear ow
period and 1/qgsc mt m0 ta. From mathematical manipulation, the following can be derived:

Ppi  Pp
Ppi  Ppwf

np 2$ rgsc rosc =GOR $ t m

fmct i x2f

then Eq. (7), and the corresponding initial and boundary conditions
in dimensionless form, are as follows:

v2 ppD



PpD 1

xD 0

taD 0

PpD 1

xD 0

taD > 0

PpD 0


taD > 0


Furthermore, G, the gas-in-place in the investigated volume at

any time step is:

G 2xf $yinv: $h$f$Sgi Bgi


 p$ ta
h f$ Ppgi  Ppwf $xf k



Combining Eqs. C-2eC-5, the average pressure in the distance of

investigation can be estimated by the following equation using an
iterative scheme:


1000$rgsc 5:615$rosc =GOR Bgdi $ mct i Ppgi  Ppwf q
i$ 1 
$ vb vPpg p
1000=5:615rgsc rosc rsi 0:194  62:55
Z 2p Zi

The solution to Eq. (7) is the same as Eq. (2) except that PD
should be replaced by PpD and tD by taD. Having found the pseudopressure prole, the surface gas rate is then given by Eq. A-11.
Finally, Eq. B-2 is obtained after some manipulation and simplication (Wattenbarger et al., 1998):



For dry gas systems, this is simplied to:




Ppgi  Ppwf
$ zg mg cg i $
1  0:46$
pi Sgi
mct i


The average pressure in the distance of investigation for a

single-phase slightly compressible uid, during the constant pressure innite acting period, can be represented by Eq. C-8. This can
be simply proved by combining the material balance equation (i.e.
the compressibility equation) and the rate equation:

Appendix C. Determination of average pressure in the

distance of investigation.

p pi  0:46$ pi  pwf

Wattenbarger et al. (1998) proposed Eq. C-1 for determination of

distance of investigation in the linear models (see Kuchuk (2009)
and references cited therein for radius of investigation).

Eq. C-8 is used to give an initial guess for calculation of average

pressure in Eq. C-6.

yinv: 0:159$


kt fmct i


The formulation was obtained using a type curve. However, in

this study another formulation (Eq. C-2) is used, which is based on
the results of unit impulse theory as discussed in Behmanesh et al.

yinv: 0:194$


kt fmct i


The material balance equation for complex uid systems can be

modied to take the same form as the dry gas formulation. For gas
condensate systems, these modications include the two-phase Zfactor (Z2p), which is given by Eq. C-3 (Hagoort, 1988; Vo et al.,
1990). Average pressure can be evaluated according to following.

i$ 1 
Z 2p Zi



Aanonsen, S., 1985. Application of Pseudotime to Estimate Average Reservoir
Pressure. SPE 14256 presented at the SPE Annual Technical Conference and
Exhibition, Las Vegas, Nevada, USA. 22e26 September.
Anderson, D., Mattar, L., 2007. An improved pseudo-time for gas reservoirs with
signicant transient ow. J. Can. Petrol. Technol. 46 (7). SPE-PETSOC-07-07-05.
Behmanesh, H., Hamdi, H., Heidari, M., Clarkson, C.R., 2014a. Production data
analysis of overpressured liquid-rich shale reservoirs: effect of degree of
undersaturation. In: Paper SPE 168980 presented at the SPE Unconventional
Resources Conference, The Woodlands, Texas, USA, 1e3 April.
Behamensh, H., Tabatabaie, S.H., Heidari, M., Clarkson, C.R., 2014b. Modication of
the transient linear ow distance of investigation calculation for use in hydraulic fracture property determination. In: Paper SPE 168981 presented at the
SPE Unconventional Resources Conference, The Woodlands, Texas, USA, 1e3
Be, A., Skjaeveland, S., Whitson, C., 1989. Two-phase pressure test analysis. SPE
Form. Eval. 4 (4), 604e610. SPE-10224-PA.
Camacho-v, R.G., Raghavan, R., 1989. Performance of wells in solution-gas-drive
reservoirs. SPE Form. Eval. 4 (4), 611e620. SPE-16745-PA.


H. Behmanesh et al. / Journal of Natural Gas Science and Engineering 22 (2015) 22e34

Chen, C., Raghavan, R., 2013. On the liquid-ow analog to evaluate gas wells producing in shales. SPE Reserv. Eng. Eval. 16 (02), 209e215. SPE-165580-PA.
Fevang, ., Whitson, C.H., 1996. Modeling gas-condensate well deliverability. SPE
Res. Eng. 11 (4), 221e230. SPE-30714-PA.
Fraim, M.L., Wattenbarger, R.A., 1987. Gas reservoir decline-curve analysis using
type curves with real gas pseudopressure and normalized time. SPE Form. Eval.
2 (4), 671e682. SPE-14238-PA.
Hagoort, J., 1988. Fundamentals of Gas Reservoir Engineering. Elsevier Science,
Hamdi, H., Jamiolahmady, M., Corbett, P., 2013. Modeling the interfering effects of
Gas condensate and geological heterogeneities on transient pressure response.
SPEJ 18 (4), 656e669. SPE-143613-PA.
Heidari, M., Gerami, S., 2011. A new model for modern production-decline analysis
of gas/condensate reservoirs. J. Can. Petrol. Technol. 50 (7e8), 14e23. SPE149709-PA.
Ibrahim, M., Wattenbarger, R.A., 2006. Rate Dependence of Transient Linear Flow in
Tight Gas Wells.
Jones, J.R., Raghavan, R., 1988. Interpretation of owing well response in gascondensate wells (includes associated papers 19014 and 19216). SPE Form.
Eval. 3 (3), 578e594. SPE-14204-PA.
Kuchuk, F.J., 2009. Radius of Investigation for Reserve Estimation From Pressure
Transient Well Tests. SPE 120515.
Miller, M.A., Jenkins, C., Rai, R., 2010. Applying Innovative Production Modeling
Techniques to Quantify Fracture Characteristics, Reservoir Properties, and Well
Performance in Shale Gas Reservoirs. Paper SPE139097 presented at the SPE
Eastern Regional Meeting. Morgantown, West Virginia, 12e14 October. http://
Martin, John C., 1959. Simplied equations of ow in gas drive reservoirs and the
theoretical foundation of multiphase pressure buildup analyses. Trans. AIME
216, 309e311. SPE-1235-G.
Nobakht, M., Clarkson, C.R., 2012. A new analytical method for analyzing linear ow
in tight/shale gas reservoirs: constant-owing-pressure boundary condition.
SPE Res. Eval. Eng. 15 (3), 370e384. SPE-143989-PA.

O'Sullivan, M., Pruess, K., 1980. Analysis of injection testing of geothermal reservoirs. Trans. Geotherm. Resour. Counc. 4, 401e414.
O'Sullivan, M., 1981. A similarity method for geothermal well test analysis. Water
Resour. Res. 17 (2), 390e398.
Poe Jr., B.D., 2002. Effective Well and Reservoir Evaluation Without the Need for
Well Pressure History. Paper SPE 77691 presented at the SPE Annual Technical
Conference and Exhibition. San Antonio, Texas, 29 Septembere2 October.
Perrine, R.L., 1956. Analysis of Pressure-buildup Curves. Drilling and Production
Qanbari, F., Clarkson, C.R., 2013. A new method for production data analysis of tight
and shale gas reservoirs during transient linear ow period. J. Nat. Gas Sci. Eng.
14, 55e65.
Raghavan, R., 1976. Well test analysis: wells producing by solution gas drive. SPEJ 16
(4), 196e208. SPE-5588-PA.
Serra, K.V., 1988. Well-testing for Solution Gas Drive Reservoirs. Ph.D.thesis. University of Tulsa.
Shi, Ch, 2009. Flow Behavior of Gas Condensate Wells. Ph. D. thesis. Standford
University, Menlo Park, California.
Schlumberger, 2010. Eclipse 100 and 300 v.2010 Software.1.
Thompson, L., Reynolds, A., 1997. Pressure transient analysis for gas condensate
reservoirs. In Situ 21 (2), 101e144.
Vo, D.T., Jones, J.R., Camacho-V, R.G., Raghavan, R., 1990. A unied treatment of
material balance computations. In: Paper SPE 21567 presented at the CIM/SPE
International Technical Meeting, Calgary, 10e13 June.
Wattenbarger, R.A., El-Banbi, A.H., Villegas, M.E., Maggard, J.B., 1998. Production
analysis of linear ow into fractured tight gas wells. In: Paper SPE 39931 presented at the SPE Rocky Mountain Regional/Low-Permeability Reservoirs
Symposium, Denver, 5e8 April.
Whitson, C., Sunjerga, S., 2012. PVT in liquid-rich shale reservoirs. In: Paper SPE
155499 presented at the Annual Technical Conference and Exhibition, San
Antonio, Texas, USA. 8e10 October.