Sunteți pe pagina 1din 20

SPE 144028

Global Model for Fracture Falloff Analysis


M. Marongiu-Porcu, C. A. Ehlig-Economides / Texas A&M University, and M. J. Economides / University of Houston

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, 1416 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
Hydraulic fracturing is undeniably the crucial technology involved in the development of tight and/or unconventional gas
reservoirs. The fracture geometry and pumping execution, as well as the well architecture, can be designed to maximize the
well productivity, provided the reservoir permeability is known. Because it is difficult to estimate permeability from
conventional pressure transient tests, various authors have shown how fracture-injection/falloff tests designed for final
fracture design calibration (i.e. estimation of closure stress, leak-off coefficient and fracture fluid efficiency) can be used to
estimate reservoir permeability as well.
This paper presents a new comprehensive model for the analysis of fracture falloff, following a step-rate or constant rate
injection test. The derivative of the falloff pressure change with respect to the logarithm of superposition time shows a
progression of straight trends that include a newly introduced elastic closure-dominated flow enabling determination of the
closure stress (3/2 slope), post-closure formation-linear flow (1/2 slope) and infinite-acting pseudo-radial (0 slope). This
model provides a robust complete assessment tool that allows quantification of all fracture parameters (closure stress, fracture
extent, leak-off coefficient and fracture fluid efficiency) as well as reservoir permeability, provided that enough time is
allowed for the falloff to reach pseudo-radial flow regime. Both oil and gas reservoirs can be effectively evaluated.
While the approach is leveraging well-accepted pressure transient analysis methods, the inclusion of elastic closuredominated flow is new. This model can be used to design fracture-injection/falloff tests that would allow determination of all
the involved parameters, including reservoir permeability. Field data validate the model and demonstrate the value of this
approach.
Introduction
The fracture-injection/falloff test, often also called fracture-calibration test or minifrac, is frequently performed before the
main hydraulic fracturing treatment to estimate critical parameters required for the optimal tuning of the stimulation design,
and typically do not involve the use of any proppant material. Fracture-pressure falloff analysis was pioneered by Nolte
(1979, 1986 and 1988), with the introduction of a methodology for the determination of key parameters, such as the leak-off
coefficient, the fracture fluid efficiency and the fracture dimensions, provided that a suitable 2D model is used
(Khristianovitch and Zheltov, 1955; Perkins and Kern, 1961; Geertsma and De Klerk, 1969; Nordgren, 1972). Noltes
approach is based on a simple material balance scheme, which allocates the fluid injected for the calibration test as either lost
into the formation (leaking off through the fracture walls) or contributing to the fracture propagation within the reservoir
rock. The analysis of the recorded pressure falloff data is possible by means of the g-function, a special dimensionless
function that accounts for the evolution in time of the actual fracture wall surface exposed to fluid leak-off into the formation.
The Unified Fracture Design (UFD) approach indicates what the optimum fracture geometry (fracture half-length, fracture
width and dimensionless conductivity) should be for a given treatment proppant mass (Economides et al., 2002), provided
that a reliable value for the formation permeability is known as well as the approximate well drainage area. When the well
design incorporates transverse fractures created from a horizontal well in tight gas and, especially, in shale gas reservoirs,
formation permeability is important for establishing an optimal spacing between the transverse fractures (Song et al., 2011).
Pressure buildup and falloff tests are the standard analysis method used for reservoir permeability determination, but in
very low permeability reservoirs this is difficult or virtually impossible, because without stimulation the formation flow is
minimal. After-closure analysis (ACA) following a fracture calibration test offers a workable mechanism for permeability
estimation in very low permeability reservoirs. Several methods for estimating permeability and other calibration test
parameters are found in the literature.

SPE 144028

Gu et al. (1993) presented an ACA method based on the notion of impulse fracture, consisting of a small volume of
fluid injected in order to generate a short fracture and a shut-in period afterwards to record pressure falloff. This method
relies on the identification of a late time straight trend of the bottom hole recorded pressure versus the reciprocal of a shut in
time.
Nolte et al. (1997) provided a complex framework, based on the F-function, another dimensionless time function for afterclosure pressure analysis. They introduced a specialized plot (pressure versus squared values of dimensionless F-function)
from which reservoir permeability can be estimated from a late time negative unit slope indicating achievement of pseudoradial flow. Benelkadi and Tiab (2001) and Benelkadi et al. (2003) criticized Noltes approach and the difficulties associated
in the identification of after-closure linear flow regime and pseudo-radial, and provided a slightly modified approach, still
based on Noltes specialized plot, where the additional use of pressure derivative with respect to squared values of the
dimensionless F-function was found to be more suitable for proper characterization of linear flow and pseudo-radial flow
regimes and determination of reservoir permeability and extrapolated reservoir pressure.
Mayerhofer et al. (1995) provided a straight-line technique for determination of reservoir permeability and fracture face
resistance by representing the recorded fracture falloff data in yet another specialized plot. They modeled the total pressure
gradient from the fracture into the reservoir as the sum of two contributing terms: the pressure drop in the reservoir as effect
of an infinite conductivity fracture and the pressure drop across the fracture face. They used superposition to obtain a
transient pressure drop in the reservoir that accounted for variable leak-off rates through the fracture faces during injection
and shut-in, but the proposed techniques to calculate these leak-off rates require information not realistically available, such
as the total pressure difference between the fracture and the reservoir and the evolution of the leak-off process through the
varying fracture faces surface during injection. For this reason, Valk and Economides (1999) and Craig and Blasingame
(2006) proposed respectively two modified approaches, both based on the simplified assumption that the leak-off rate during
injection is constant.
All these ACA techniques depend on specialized plots designed to show a straight line for a portion of the data, from
which parameters are determined either from the slope of the line or from its endpoints. As such, there is a risk that apparent
straight lines may lead to erroneous or inconsistent results, particularly when the (frequent) absence of late time pseudo-radial
flow data is ignored or not recognized at all.
This paper starts from the use of the log-log diagnostic plot of the pressure falloff introduced by Mohamed et al. (2011).
We show how features identified in this plot provide immediate estimates for the closure stress and time, the formation
permeability, the created fracture half-length, the leak-off coefficient, and the fracture fluid efficiency, provided values for
Youngs modulus and the Poisson ratio are known. Then we introduce a new model combining before-closure and afterclosure behavior and show parametric studies using this model. Finally, we explain a new analysis approach and provide field
examples illustrating its application and the advantage of a global model over piecemeal approaches relying on a series of
independent specialized plots.
The Log-Log Diagnostic Plot Representation for a Fracture-Injection/Falloff Test
Bourdet et al. (1989) introduced the log-log diagnostic plot representation for pressure drawdown and pressure buildup tests,
where the pressure differences are calculated, respectively, as differences between the initial reservoir pressure and
instantaneous bottomhole flowing pressures (drawdown tests) or as differences between the bottomhole shut-in pressures and
bottomhole flowing pressure at shut-in (buildup tests). Furthermore, for drawdown tests the pressure derivative, p, is
computed numerically as the derivative of p with respect to the natural logarithm of the elapsed flowing time, while for
buildup tests the derivative is calculated with respect to the natural logarithm of the superposition time as

p =

dp
d ln

(1)

where is the superposition time function computed rigorously from the complete injection flow rate history (as described in
Lee et al., 2003).
For analysis of pressure buildup data acquired after a single constant production rate step, the superposition time function
is reduced to the simple form

t p + t
t

(2)

where tp is the production time. For variable production rates before shut-in, we use the Horner approximation (Horner, 1967)
in which tp is calculated as the cumulative hydrocarbon production divided by the last production rate. Blasingame and Lee
(1986) referred to this as material balance time.
Fracture-injection/falloff tests can be represented and analyzed using this methodology. Fig. 1 shows the schematic
sequence of events in the fracture-injection/falloff test. Injection of the same type of fluid to be used for the main treatment
pressurizes the formation until the rock breakdown is achieved. After breakdown the fracture propagates until the pumps are

SPE 144028

shut down. At this point the pressure falls off. An instantaneous decrease in pressure to the value labeled as ISIP
(Instantaneous Shut In Pressure) is due to friction losses. This pressure drop may be large if pressure is recorded at the
wellhead, but when bottomhole pressure is recorded the instantaneous pressure drop is mainly due to pressure losses in the
near wellbore area. As shown in Fig. 1, we systematically use the ISIP as the reference pressure for the calculation of the
falloff pressure difference, p. The closure pressure, pc, is marked in the figure, but this is not apparent in the data when
presented in this way.

Figure 1: Schematic sequence of events in a fracture-injection/falloff test

Fig. 2 shows an idealization of a fracture-injection/falloff test represented in a log-log diagnostic plot. The falloff pressure
derivative exhibits a characteristic progression of trends, firstly described in Mohamed et al. (2011). This succession of
trends includes:
- a newly introduced elastic closure-dominated flow (3/2 slope);
- the main fracture closure event (identified by the departure of the falloff pressure derivative from the 3/2 slope trend);
- an after-closure formation-linear flow ( slope) consistent with the after-closure pseudolinear flow regime described
by Craig and Blasingame (2006), Economides and Martin (2007) and by Baree et al. (2009);
- a late-time infinite-acting pseudo-radial flow (0 slope).
Fracture Pressure Falloff Model Construction
Our model combines before-closure and after-closure models in a piecewise function. The before-closure model couples
Carters leak-off model and Noltes assumption on fracture surface growth into a material balance scheme (exhaustively
presented by Valk and Economides, 1995; 1999) that is considered to be the most appropriate method for before-closure
pressure falloff modeling, since it directly relates injected volumes of fluids and injection times to fracture dimensions and
leak-off (fluid efficiency) development.
The after-closure model uses an enlarged equivalent wellbore radius to provide the effectively infinite conductivity
fracture behavior. We assume that at a far enough distance from the wellbore, the reservoir feels the transient pressure
behavior induced by leak-off during injection and fracture closure, followed by an effective zero-rate period starting
immediately after the closure time. Once consistent before and after-closure models are generated, a suitable spline algorithm
connects these two solutions to provide a continuous smooth model for the falloff pressure change and its derivative.
Before showing parametric studies, it is useful to explain in detail the parameters governing before and after-closure
behavior and how they are related in the combined model.

SPE 144028

Figure 2: Log-log diagnostic plot for an idealized fracture-injection/falloff test

Before Closure Theory


Valk and Economides (1995; 1999) presented a methodology for analysis and simulation of before-closure fracture pressure
falloff, based on Carters leak-off model, Noltes assumption on fracture surface growth, and elastic behavior of the closing
fracture walls under the formation in-situ stress. While tip-extension effects may occur in tight formations, we assume that
immediately after the shut-in the fracture ceases to propagate. The pressure falloff is governed by the gradual loss of
hydraulic fracture width w
(3)
p w = pc + S f w
where pc is the closure stress and Sf is the fracture stiffness, measured in psi/ft and representing the relationship between
fracture net pressure and fracture width according to the linear elasticity theory. Different analytical expressions have been
presented for Sf, all based on the assumption that there is no lateral flow in the fracture and the pressure is constant along the
fracture at any given shut-in time. Table 1 presents expressions for Sf for the vertical plane strain assumption (2D PKNgeometry, Perkins and Kern, 1961; Nordgren, 1972), for the horizontal plane strain assumption (2D KGD-geometry,
Geertsma and De Klerk, 1969) and for the simplest radial geometry.
The injection time, te, computed in an analogous way to the material balance time, is used in the dimensionless shut-in
time given by

t D =

t
te

(4)

Valk and Economides (1995; 1999) expressed the fracture width at a given time t after shut-in from a material balance
scheme accounting for leak-off in the formation and fracture propagation of the total injected fluid, resulting in

wte + t =

Vi
- 2S p 2C L te g (t D , )
Ae

(5)

Combining Eqs. 3 and 5 provides the final before-closure fracture pressure falloff model

)(

pw = pC + S f Vi / Ae - 2 S f S p - 2 S f C L te g (t D , )

(6)

where Vi is the volume of fluid injected in one wing of the fracture (i.e., half of the total injected fluid volume), Ae is the
equivalent surface area of one face of one wing, Sp is the spurt loss that indicates the fraction of fluid loss in formation at
the very early stages of the leak-off process, and CL is the leak-off coefficient.

SPE 144028

5
Table 1: Fracture stiffness and values for 2D fracture geometry models

Valk and Economides (1995) provided an analytical expression for the dimensionless g-function, g(t,) introduced by
Nolte (1979) for his well-known power law fracture surface growth assumption. This analytical expression is based on a
complicated mathematical function called Hypergeometric function, available in form of tables (Abramowitz and Stegun,
1972) and computing algorithms (Wolfram, 1991). A rigorous calculation of the necessary g-function values requires also
using the proper power law exponent , whose values for the familiar 2D fracture geometry models are presented in Table 1.
Equation 6 suggests that the bottomhole pressure decreases linearly with the g-function until the fracture finally closes,
after which the pressure falloff will depart from this linear trend. The linear nature of this model is better visualized when Eq.
6 is expressed as equation of a straight line of intercept bN and slope mN

pw = bN + mN g (t D , )

(7)

bN = pC + S f Vi / Ae , and

(8)

mN = -2S f C L te

(9)

where

Valk and Economides (1995; 1999) presented an exhaustive set of equations for the familiar 2D fracture geometry
models, based on the Shlyapobersky et al. (1998) assumption of negligible spurt loss, to calculate leak-off coefficient,
fracture extent, hydraulic average width (at end of pumping) and fracture fluid efficiency. All these equations are presented in
Table 2, and we can notice that bN and mN are necessary input parameters.
This technique normally requires the graph of the recorded wellbore pressure falloff data versus the values of the gfunction shown on the graph on the left in Fig. 3. In contrast, the analysis method presented in this paper doesnt require the
construction of the graph just mentioned, because of the intimate correspondence between the g-function and the
characteristic 3/2 slope trend in our log-log pressure derivative. In fact, if we begin considering the upper bond limiting form
of the g-function for fluid efficiency 100%

g (t D , = 1) =

3
3
4
(1 + t D )2 t D 2 ,

(10)

the presence of a time exponent of 3/2 reveals a priori that a pressure derivative calculated with respect to the natural
logarithm of the superposition time as in Eq. 1 will exhibit a 3/2 slope. When the pressure change is no longer dominated by
this behavior, the fracture is considered to be closed. In our method we pick closure time and stress as the point (pc, tc) on
the log-log diagnostic plot when the logarithmic derivative departs from the 3/2 slope.
A slight rearrangement of Eq. 10 gives

g (t D , = 1) =

4 3/ 2 2
t D 1 ,
3

(11)

where is defined as in Eq. 2, with tp replaced by the material balance injection time te. Substituting Eq. 11 in Eq. 7, and
taking the logarithmic derivative according to the basic derivative rule

p =
provides

dp
dp ,
=

d ln d

(12)

SPE 144028

p = 2m N t D5 / 2 1 1 / 2 .

(13)

Table 2: Fracture-injection/falloff analysis model based on the Shlyapobersky et al. (1998) assumption

Solving for mN and rearranging Eq. 7 provides the final expression for our two necessary parameters

mN =

p
2t 1 1 / 2

bN = p w m N

5/ 2
D

3
4 3 / 2 2
t D 1
3

(14)
(15)

where pw = ISIP-p.
In virtue of this result, we use tc, pc and pc at the closure time to calculate the required bN and mN, without
constructing the specialized plot pw vs g-function. This is schematically illustrated on the graph on the right in Fig. 3.
After Closure Theory
Although the fracture half-length increases during propagation, the after-closure response is dominated by the final fracture
geometry existing at the closure time. An approximate accounting for the leak-off rate could consider a two-rate injection
followed by shut-off, where the first rate is an assumed average leak-off rate during the fracture propagation, and the second
rate is an assumed average rate during closure. We have found that the Horner approximation (Horner, 1967) assuming an
average constant leak-off rate given by the ratio of the total leak-off volume at the closure time and the injection+closure time
can be successfully employed. Valk and Economides (1999) used this specific assumption to propose the following Laplace
transform of the dimensionless wellbore pressure

pwD = 1 e ( teD +tcD ) s

) qs

1
K0
s
2

(16)

where K0 is the modified Bessel function of the second kind of order zero. Refer to Valk and Economides (1999) for
definitions for dimensionless pressure, time and rate. Eq. 16 must be inverted numerically using, for example, the Stehfest
(1970) algorithm.

SPE 144028

Figure 3: Two approaches (traditional on the left, newly proposed on the right) for the determination of mN and bN

An even easier model that we find conveniently suitable for our purposes is the one for radial flow in an infinite-acting
reservoir with cylindrical source well and constant rate. The Laplace transform of the dimensionless pressure is (Lee et al.,
2003)
p wD =

K 0 rwD s
s

3/ 2

K1 rwD s

(17)

where K1 is the modified Bessel function of the second kind of order one. Refer to Lee et al., (2003) for definitions of
dimensionless pressure, time and rate. We account for the presence of the infinite conductivity fracture by using an
equivalent wellbore radius is rw = xf / 2 (Prats, 1961). Also Eq. 17 must be inverted numerically with the Stehfest (1970)
algorithm.
Finally, we model the after-closure behavior of the fracture pressure falloff using superposition (Lee et al., 2003) in time,
approximated by the injection analog to the previously described Horner time function.
Parametric Studies Using the Fracture Pressure Falloff Model
This section shows how to construct the piecewise global model for the before-closure and after-closure solutions described
in the previous sections, using a suitable spline algorithm to connect the two solutions and provide a continuous smooth
model for the falloff pressure change and its derivative. This approach selects a consistent distance (transitory gap) between
the before-closure and after-closure solutions. We generally found that one logarithmic cycle works fairly well in most
circumstances scrutinized so far. The work presented in this paper relies, on a convenient built-in spline function in the
Mathematica program (Wolfram, 1991). We will present a rigorous definitive spline methodology for this piecewise
modeling in a future publication.
The methodology in this section is suitable for the use of our global model in a design mode, while a slightly different
logic, depending on the number of input parameters available and the simplified 2D fracture geometry expected for the
created fracture, is used for the field data analysis presented in the next section.
The design mode parametric studies seek to define optimum injection parameters, as well as quantify the minimum
shut-in time that allows achieving pseudo-radial flow. The before and after-closure solutions must use the same fracture
geometry, in terms of fracture radius and width for a radial fracture geometry, or fracture width, height and half-length for
other 2D fracture geometry models (PKN or KGD). However, the fracture geometry is not an input. Rather, coupled with the
leak-off coefficient, it is a function of the selected injected volume and rate, fracture fluid efficiency, and reservoir rock
features such permeability, and the plane strain modulus, E= E / (1- 2), where E is Youngs modulus, and is the Poisson
ratio. We use these considerations to generate two parametric studies. For the first one we assess the impact of the fracture
fluid efficiency and consider two injected fluid volumes. The second parametric study assesses the impact of reservoir
permeability, for the same injection history and fracture fluid efficiency.
The Effect of Fracture Fluid Efficiency
We apply this analysis to a natural gas reservoir with pressure 2,400 psi, temperature 160 F and gas specific gravity of 0.7,
making use of the real-gas potential function m(p) (Al-Hussainy et al., 1966) to generate the log-log diagnostic plot. We
begin by generating the before-closure solution with Eq. 6 for an assumed radial fracture geometry. The plane strain modulus
E and the closure stress pc must be treated as independent input parameters, while the leak-off coefficient CL and the fracture

SPE 144028

wall surface area Ae (i.e., fracture radius Rf) are calculated in a consistent way, as functions of the considered fracture fluid
efficiency. For this purpose we solve a system of three equations for three unknowns: leak-off coefficient CL, fracture radius
Rf and average fracture width at the end of injection we .
The first equation is the definition of fracture fluid efficiency found in Table 2 for the radial fracture geometry.
The second equation comes from the same material balance scheme used to derive Eq. 5, which at the instant of shut-in
provides the volume of fluid that leaked off from one wing of the fracture during injection (for the given fracture fluid
efficiency and the injected volume) as

VLOi = Vi (1 ) = 2 AeC L t e g (t D = 0, )

(18)

where VLOi is the volume leaked off from one wing of the fracture during injection.
The final equation must provide an independent relationship between average fracture width at the end of injection and
fracture extent. The classic 2D fracture geometry models are of crucial help, providing simple analytical equations based on
linear elasticity, flow rheology and material balance. Valk and Economides (1995) provided an exhaustive description of the
published literature on the subject. For the selected radial fracture geometry and Newtonian fluid rheology (assumption
particularly suitable for any fracture injection performed in tight formation with KCl brines or fresh water) we use

q R
w = 2.24 i f
E

1/ 4

(19)

Where is the Newtonian viscosity of the injected fluid and qi is the fluid injection rate into one wing of the fracture (i.e.,
half of the total injection rate).
Furthermore, we need to define a closure time, in order to establish the limit for our before-closure solution. Knowing the
fracture fluid efficiency and the injection time, the fundamental equation presented by Nolte (1986) for fracture fluid
efficiency

=1

g (t D = 0, )
g (t D (tc ), )

(20)

can be rearranged to determine g (t D (tc ), ) , and thus the closure time. At this point, we are able to generate the beforeclosure solution with Eq. 6, then we use the first generated value of pw (ISIP) as the reference pressure for the calculation of
the falloff pressure difference and then we generate the pressure derivative with respect to the natural logarithm of the
superposition time as in Eq. 1.
Finally, we generate the after-closure solution for the infinite conductivity fracture of half-length consistent with the
fracture radius determined with the procedure described above and with the average constant leak-off rate given by the ratio
of the total leak-off volume at the closure time and the injection+closure time. The additional required input value for the
after-closure solution is the reservoir permeability. Table 3 presents all input parameters used to generate the following
parametric analyses.
Table 3: Input parameters for parametric analyses

Figures 4 and 5 show the results of our parametric assessment of fracture fluid efficiency for simulated injections of 500
gallons and 5,000 gallons of fluid, respectively. In each plot, the dashed curves represent the changes in m(p), while the solid
curves represent the logarithmic derivatives, calculated consistently with respect of the superposition time relative to the
actual injection time. The hollow circular marks represent the closure time, at which the before-closure solution (on the left
side of these marks) ends. The actual calculated after-closure solutions begins, for each set of curves, at one logarithmic cycle
distance from the closure time indicated by the hollow circular marks. The region in between is what results from the selected
spline procedure.
First, in both figures we notice that the effect of fracture fluid efficiency appears more pronounced for the before-closure
solution, where the green, red and blue curves are distinctively apart from each other. For the 500 gallons injection (Fig. 4)

SPE 144028

the resulting fracture radius are 51 ft, 60 ft and 69 ft, respectively for 50%, 70% and 90% fracture fluid efficiency.
Analogously, For the 5,000 gallons injection (Fig. 5) the resulting fracture radius are 143 ft, 166 ft and 186 ft, respectively for
50%, 70% and 90% fracture fluid efficiency. Clearly, larger fluid volume injected correspond to a longer fracture radius, but
the small relative difference among the three cases is such that after-closure transient effects are very similar, with the
consequence that the spacing among the different after-closure models is minimal.

Figure 4: Parametric study on fracture fluid efficiency for an injected fluid volume of 500 gallons

Figure 5: Parametric study on fracture fluid efficiency for an injected fluid volume of 5,000 gallons

Interestingly, we recognize a consistent generalized time shift of all curves between Fig. 4 and 5. For each value of
fracture fluid efficiency, the 5,000 gallons injection makes the closure times shifted by 1 logarithmic cycle (i.e. 10 times
larger) respect to the 500 gallons case. We observe this trend also for the time in which pseudo-radial flow begins, and while
in Fig. 4 all after-closure derivatives seem to achieve constant level in some 1,000 hrs, in Fig. 5 this takes some 10,000 hrs
(about 1 year).

10

SPE 144028

This behavior offers important insight on the selection of the fluid volume to use for fracture-injection/falloff test, if
reservoir permeability is one of the main objectives of the test. Clearly, the injected volume should be kept as small as
possible, in order to achieve transient pseudo-radial flow in the shortest possible time, as this will reveal reservoir
permeability. On the other hand, the use of greater injection fluid volume at rates similar to the ones considered for the main
fracture job design, may be preferred when the major test objective is to account for possible formation heterogeneities in
characterizing the leak-off process throughout a larger fracture wall surface area. The model presented in this paper provides
a useful tool for fracture-injection/falloff test design within the operational, economic and logistic constraints that the
fracture-injection/falloff test needs to satisfy.
The Effect of Reservoir Permeability
This analysis uses the same natural gas reservoir considered in the previous section. Since fracture fluid efficiency is still
treated as an input variable, we use again the same logic to determine consistent values of leak-off coefficient CL and the
fracture wall surface area Ae (i.e., fracture radius Rf).
Figure 6 shows the results of our parametric assessment of reservoir permeability for a simulated injection of 500 gallons
of fluid with a fracture fluid efficiency of 90%. Again, the dashed curves represent the changes in m(p), the solid curves
represent the logarithmic derivatives, and the hollow circular marks indicate the closure time.
As expected, reservoir permeability does not impact the before-closure solution, and the elastic-closure pressure falloff is
described in Fig. 6 by the couple of black curves for any value of reservoir permeability. This behavior is likely to be only
partially realistic, since it is well understood that reservoir permeability plays a big role in the characterization of the leak-off
process, together with fracture fluid composition and rheology. Thus, the leak-off coefficient used for the before-closure
solution should somehow reflect this relationship, but no rigorous modeling is available in the literature, apart from empirical
and laboratory experimental studies.
Figure 6 confirms and exacerbates the message delivered in the previous section. Achievement of pseudo-radial flow
requires unacceptable times, even for the small injection of 500 gallons considered, when reservoir permeability gets in the
order of magnitude of the microDarcy and lower. The case at 0.001 md achieves pseudo-radial flow in some 10,000 hours,
while the case at 0.0001 md takes about 100,000 hours (over 11 years).
We will see in the field examples section that while actual tight gas sand reservoirs analyzed with our technique appear in
good agreement with the theoretical trend represented by the blue curve (k 0.01 md), shale gas reservoirs present an afterclosure behavior that cannot entirely be reproduced by our model.

Figure 6: Parametric study on reservoir permeability for an injected fluid volume of 500 gallons and = 90%

Data Analysis Approach


To analyze field data, it is best to have direct measurements of the formation thickness, fracture height, hf, plane strain
modulus, porosity, formation fluid viscosity and compressibility, and for gas the formation temperature and the gas gravity.
To create the log-log diagnostic plot it is necessary to determine the ISIP and compute p as described in Fig. 1. In the loglog diagnostic plot of the pressure falloff data we look for 4 features: 3/2 slope, end of 3/2 slope indicating the closure time

SPE 144028

11

and 1/2 slope derivative trend related to the fracture extent, followed by constant derivative level related to the formation
permeability. With these inputs the interpretation consists of the following steps:
1.

2.

Initial assumption of the fracture geometry, with the two appropriate choices being PKN or radial. When the volume
injected is small or when the stress contrast between the productive and surrounding formations is small, the created
fracture geometry may be radial. When lithology suggests a strong fracture height containment, with sufficient injected
volume the fracture half-length will exceed its height, and the PKN model applies.
The point (pc, tc) at the end of the 3/2 slope trend in the derivative defines the closure time, and closure stress given
by the pressure at that time. We determine values for mN and bN from this point using Eqs 14 and 15. After this
determination, the next steps depend on the assumed fracture geometry.

PKN Geometry
3. Permeability is determined from the derivative level, p for oil, or m(p)for gas, using the equation

70.6qBo o
for oil
m' h
711qT
k=
for gas
m' h
k=

(21)

where for gas the fluid injection rate, q, is converted to the equivalent MSCF/d assuming piston-like displacement of the gas
by the injected fluid, the reservoir temperature T is in Rankine and m is the derivative level in the pseudo-radial flow region.
4. The initial reservoir pressure can also be estimated from the derivative level in the pseudo-radial flow region m and
corresponding values of p and t, using the following equation:

t + t slope=0
p (t slope=0 ) + ISIP
p* ~ pi = m ln e
t

slope
=
0

(22)

Fracture half-length is determined by selecting a point (t, p) in the after-closure derivative slope trend, using the
flow-regime analysis equation presented by Cinco-Ley and Samaniego (1981) for infinite conductivity fractures

4.064qB

xf =
m h kc
lf
t

40.9521qT
xf =

mlf h

where mlf =
5.

0 .5

for oil

k c
g t

(23)

0.5

for gas

2p' / t .

The fracture length, xf is used to estimate hf, which then enables estimation of the leak-off coefficient, CL, and the
average fracture width at the end of injection, we , and fracture fluid efficiency using the equations in Table 2.

Radial Geometry
3. In this case permeability cannot be estimated until the fracture radius is estimated. The fracture radius, Rf, CL,
4.
5.

we , and

fracture fluid efficiency are estimated using equations in Table 2.


If 2Rf < h, permeability is estimated assuming h = 2Rf. If 2Rf > h, permeability is estimated using the actual reservoir
thickness. Eq. 21 applies for both cases.
The initial reservoir pressure is estimated with Eq. 22.

Finally, we generate the piecewise global model match for the data.
Field Examples
We present 3 field examples. The first two examples are from tight gas wells in the Cotton Valley sandstone and the
Mesaverde sandstone, and both were presented previously by Mohamed et al. (2011). The final example is from a shale gas
well in the Haynesville shale.

12

SPE 144028

Cotton Valley Tight Gas Well


The Cotton Valley tight gas formation is located in East Texas above the Haynesville Shale (McCain et al., 1993). The
Cotton Valley trend can be found also in Northwest Louisiana. The formation rock contains shale, sandstone, and clay
deposits that produce natural gas. The Cotton Valley formation is typified by low porosities (5 to 10 %) and permeability
values in the micro-darcy range, and may generally be described as medium-hard, gray sandstone. The Cotton Valley
formation can be found at depths between 7,500 and 10,000 feet.
A fracture-injection/falloff test was performed in a vertical gas well completed in the Taylor sand interval in the Cotton
Valley formation. The entire gross pay interval (10,270-10,302 ft TVD) was perforated, but the petrophysical analysis
indicated a net pay thickness of only 15 ft. The test was conducted by pumping 36.25 bbls of 3% KCl water in a total of 19
minutes. The bottomhole pressure falloff was monitored for 6 days before retrieving the downhole gauges. Refer to
Mohamed et al. (2011) for detailed information about the actual injection schedule, and Table 4 provides reservoir and fluid
properties used for the analysis. Additionally, we use a plane strain modulus of 6.106 psi for this analysis.
Table 4: Input parameters for the Cotton Valley fracture-injection/falloff test

Figure 7 shows the log-log diagnostic plot for the pressure falloff data in terms of the real-gas potential function m(p) for
an ISIP of 7,287 psi. All the anticipated features are easily identifiable from this figure. Lithology and petrophysics suggest
the PKN fracture geometry model, due to the fracture height containment that is expected from the confining shales. In
general, this kind of assumption could be verified by a temperature-log survey, but the considerable shut-in times required for
tight formations like the one under analysis makes this field measurement unsuitable.
The 3/2 slope trend is visible for more than half a logarithmic cycle, and a clear departure from this trend is identified at
t = 1 hr and m(p) = 1.35.109 psi2/cp (p = 2,985 psi), giving a closure stress of 6,370 psi. Mohamed, et al. (2011) showed
that the conventional analysis arrives at the same result. Using the values for t, p, and p at the closure time, we use Eqs.
14 and 15 to calculate the parameters mN and bN, respectively 302 and 7,670 psi.
We use Eq. 21 to estimate permeability as 0.0095 md from for the derivative level indicated in Fig. 7, and we use Eq. 22
to estimate the reservoir pressure as 4,830 psi. Next, we use Eq. 23 to estimate the fracture half-length using the point t =
3.38 hr and m(p) = 3..109 psi2/cp-cycle = giving mlf = 3.23.109 and xf = 80 ft. This analysis is in perfect agreement with
independent estimations performed by Mohamed et al. (2011) using the Gu et al. (1993) and Benelkadi et al. (2003)
methods.
Furthermore, using the fracture half-length, we estimate the fracture height, hf = 62 ft, and, in turn, the leakoff coefficient,
CL = 0.000614 ft/min0.5, we = 0.16 inches and e = 67%.
Finally, Fig. 8 shows the piecewise global model match for the data generated for these results. We see that the global
match with pressure change and derivative is excellent before closure and through the first portion of the after closure
response. However, after about 5 hours the data show more complex behavior than that exhibited by the model and arrive
earlier at the level of the derivative interpreted as pseudo-radial flow.

SPE 144028

13

Figure 7: Log-log diagnostic plot for data from the Cotton Valley fracture-injection/falloff test

Figure 8: Global model match for the Cotton Valley fracture-injection/falloff test

Mesaverde Tight Gas Well


The Mesaverde formation occurs in various structural basins in the western United States such as Piceance Basin, Powder
River Basin, Uintah Basin, Washakie Basin, and Wind River Basin. Pierce (1997) described the Mesaverde formation as
interbedded light gray sandstone and gray shale in upper part; lower part massive, light-buff, ledge-forming sandstone
containing thin lenticular coal beds.
Craig and Blasingame (2006) previously analyzed the fracture-injection/falloff test on the tight gas well 543-33. A
fracture-injection and falloff sequence was performed in an isolated sandstone layer, consisting of the injection of 17.69 bbl
of 1% KCl treated water at an average injection rate of 3.3 bbl/min for 5.3 minutes and a pressure falloff recorded for 16.2 hr.

14

SPE 144028

Immediately following the falloff, the well was flowed at 100 Mscf/D for 141.7 hrs, then lowered at 98 Mscf/D for the next
24.3 hrs, then lowered at 60 Mscf/D for the next 0.6 hrs and finally lowered at 50 Mscf/D for the final 0.1 hrs before shutting
the well in for a pressure buildup test lasting 14.95 days. Any additional information about the well and the sequence of
operations can be retrieved from Craig and Blasingame (2006), and Table 5 provides reservoir and fluid properties used for
the analysis.
Table 5: Input parameters for the Mesaverde well 543-33 fracture-injection/falloff test

Craig and Blasingame (2006) used their type-curve method to match this buildup, which provided a reservoir permeability
of 0.012 md, a fracture half-length of 121 ft and a fracture conductivity of 18 md.ft. Mohamed et al. (2011) confirmed these
results with a traditional log-log diagnostic plot interpretation, emphasizing the fact that the buildup time was not long
enough to identify pseudo-radial flow, and that the assumed flat level for the derivative revealing the reservoir permeability
value appears to be completely arbitrary.
As in the previous example, lithology and petrophysics suggest the PKN fracture geometry model, due to the fracture
height containment that is expected from confining shales. Figure 9 shows the comparative log-log diagnostic plot showing
both the pressure falloff data and the pressure buildup data in terms of real-gas potential function m(p). As in the previous
example, the anticipated features for the pressure falloff derivative are easily identifiable from this figure, except the missing
late time flat derivative revealing of the pseudo-radial flow regime.

Figure 9: Log-log diagnostic plot for data from the Mesaverde fracture-injection/falloff test and buildup

Similar to the previous example, the 3/2 slope trend is visible, and a clear departure from this trend is identified at t =
0.46 hr and m(p) = 3.53.106 psi2/cp (p = 856 psi), giving a closure stress of 2,790 psi for an ISIP of 3125 psi. Again,

SPE 144028

15

conventional analysis with the G[dp/dG] methodology (Nolte, 1979) arrives at the same result. Using the values for t, p,
and p at the closure time, we use Eqs. 14 and 15 to calculate the parameters mN and bN, respectively 63 and 3,091 psi.
The lack of the radial flow response makes this example more challenging. The absence of pseudo-radial flow regime
feature in the falloff derivative prevents estimation of the extrapolated initial reservoir pressure. However, for any assumed
radial flow there is a unique fracture half length that can satisfy the slope trend seen after closure. For illustration purposes,
we assume as before a permeability value of 0.012 md. For t = 9 hrs and m(p)=2..107 psi2/cp, we determine mlf = 1.33.107,
and finally a value of fracture half-length of 124 ft. Using this value for the fracture half-length, we estimate the fracture
height, hf = 65 ft, and, in turn, the leakoff coefficient, CL = 0.00029 ft/min0.5, we = 0.05 inches and e = 70%.
Mohamed et al. (2011) pointed out that once the pressure falloff derivative achieves the after-closure pseudolinear flow, it
overlays quite compellingly upon the pressure buildup derivative (with its typical slope straight trend firstly illustrated by
Cinco-Ley and Samaniego, 1981). This provided a crucial empirical proof of the validity of the analysis method for fractureinjection/falloff test discussed in this paper, based on the use of the log-log diagnostic plot.
Figure 10 shows the piecewise global model match for the data generated for these results. Interestingly, an alternative
match is shown, performed assuming a reservoir permeability three time smaller than the one initially used (i.e., 0.004 md).
For this value of reservoir permeability, we obtain the following results: xf = 173 ft, hf = 54 ft, CL = 0.00024 ft/min0.5, we =
0.04 inches and e = 70%. We have performed this alternative analysis to show that in absence of pseudo-radial flow, the
reservoir permeability estimation is highly ambiguous, and consequently the entire analysis. In fact, this analysis is no less
realistic than the first one, since both rely on assumed an value for reservoir permeability.

Figure 10: Global model match for the Mesaverde fracture-injection/falloff test

Haynesville Shale Gas Well


The Haynesville Shale natural gas formation is located in Northwest Louisiana, East Texas, and extends into Arkansas.
Haynesville Shale is a very promising formation due to its abnormal high pressure and large thickness. The Haynesville shale
can be found at depths between 10,000 and 14,000 feet.
A fracture-injection/falloff test was conducted on the toe of a cased horizontal drain at an average true vertical depth of
12,300 ft. A set of three perforation clusters with a spacing of 89 ft was opened via TCP guns; with a perforation density of
12-SPF, the three clusters of respectively 4 ft, 4 ft and 2 ft resulted in a total of 120 (48+48+24) holes theoretically open to
flow. Actual perforation efficiency was not estimated by the operator.
The injection was performed by pumping 20 bbls of fresh water, at a constant rate of 3 bbls/min in a time interval of 6.6
minutes. The bottomhole pressure decline was monitored for 67 hours. Table 6 shows reservoir and fluid properties used for
the analysis.
Because there were 3 perforation clusters, this example is complicated by the possibility that more than one fracture may
have been created. The before-closure modeling has been performed for two simplified schematic configurations:

16

SPE 144028

a)

Only one radial fracture has been created by the total diversion of the injected fluid throughout only one of the
three clusters of perforations;
b) The injected fluid has been equally diverted among the three clusters, corresponding in three identical radial
fractures with the same fracture fluid efficiency and same leak-off coefficient.
Table 6: Input parameters for Haynesville Shale fracture-injection/falloff test

The relatively small injected volume may not have been sufficient for the fracture to grow in height to the shale thickness.
For this reason, we assume for this example radial fracture geometry, and we will confirm the validity of this assumption
once the fracture radius is determined.
Figure 11 shows the log-log diagnostic plot for the pressure falloff data in terms of real-gas potential function m(p). As in
the previous examples, the before-closure 3/2 slope and the closure stress are easily identified. Unlike the previous examples,
the after-closure behavior does not show a slope trend. Instead, soon after closure the derivative becomes level. This case
is actually quite challenging because lack of the slope trend after closure means that there is no definitive indication of the
fracture half-length or radius. Only by combining before and after closure analysis can we estimate the permeability and
fracture geometry.

Figure 11: Log-log diagnostic plot for data from the Haynesville Shale fracture-injection/falloff test

We identify the closure event at the time of the departure of the derivative from the characteristic 3/2 slope, at about 8.4 hr
of elapsed shut-in time, and m(p) = 5.4.109 psi2/cp (p = 14,286 psi), giving a closure stress of 11,570 psi for an ISIP of
14,668 psi. Also for this case, we found a very good agreement with results of independent analyses of the closure time and
stress with the G[dp/dG] methodology (Nolte, 1979). We use Eq. 22 to estimate the reservoir pressure as 11,340 psi. Using
the values for t, p, and p labeled in Fig. 11, we use Eqs. 14 and 15 to calculate the parameters mN and bN, as 21 and

SPE 144028

17

12,010 psi respectively. Then, we use these parameters to estimate fracture radius, leak-off coefficient, average end-of-job
fracture width and fracture fluid efficiency, from the equations in Table 2 for the assumed radial fracture geometry.
For the single fracture configuration we obtain Rf = 66 ft, CL = 0.0000915 ft/min0.5, we = 0.08 inches and e = 92%.
For the 3 equal fractures configuration we obtain Rf = 46 ft, CL = 0.0000635 ft/min0.5,

we = 0.06 inches and e = 92%.

The level of the derivative during pseudo-radial flow provides an estimate for kh = 2kRf = 0.193. We estimate the shale
permeability as k = 0.00415 md assuming one created fracture and k = 0.0022 md for three created fractures using the
previously-determined value for kRf divided by 3 to account for only 1/3 of the injected volume for each fracture.
These two permeability values of are in good agreement with the results obtained by Song et.al, (2011), which performed
a long-term production data analysis for the same well, and determined a representative reservoir permeability of 0.0031 md.
Figure 12 shows the agreement achieved for this and several other Haynesville Shale wells. In three of the wells studied, the
radial flow trend is missing in the falloff response. In that case, taking the derivative level just above the derivative value
observed at closure provides a maximum possible permeability value.

Figure 12: Comparison between permeability values determined from fracture-injection/falloff


test and production data analysis for Haynesville Shale wells

Both fracture geometries and leak-off coefficients for the two assumed configurations provide identical before-closure
model match for m(p) and derivatives shown in the log-log diagnostic plot in Fig. 13. As for the Cotton Valley example, the
data show more complex behavior than that exhibited by the model and arrive earlier at the level of the derivative interpreted
as pseudo-radial flow than what the after-closure model indicates. Please notice that we have intentionally avoided any
attempt of joining the two solutions to show the exact entity of the macroscopic discontinuity between before-closure and
after-closure m(p) and derivative, which is also practically identical for both fracture geometries of 66 ft and 46 ft
corresponding to one or three created fractures. Comparing the position of the after-closure derivative curve with the data
derivative suggests that the fracture half-length indicated by the data is much shorter than the one determined with our
before-closure model. Perhaps the explanation relates to the transition that must occur after closure when the derivative value
at the closure time approaches or exceeds the constant level of the derivative associated with pseudo-radial flow.
The behavior shown in this example is not isolated. Several other tests in the Haynesville shale conducted in a similar
fashion also exhibited after closure behavior that quickly assumed apparent pseudo-radial flow without the slope trend
expected for the fracture implied by the before closure analysis. Furthermore, production data analysis in Song et al. (2011)
indicates much smaller fracture extents than what anticipated based on known injected-fluid volumes and fracture fluid
efficiencies.
The apparent discrepancies in these analyses cannot be resolved with the conceptual models we are currently using. While
the fracture-injection/falloff test model seems to work well for conventional reservoirs and for tight sandstone, it is not able

18

SPE 144028

to reproduce after-closure fracture falloff behavior observed in shale. However, we must reiterate that conducting the
calibration injection through multiple perforation clusters is highly unfavorable from the analysis perspective.

Figure 13: Before-closure model match and after-closure inconsistency for the Haynesville Shale fracture-injection/falloff test

Conclusions
This paper offers a unified approach for the fracture-injection/falloff test that avoids the need for assorted specialized plots.
We provide an analysis procedure based on a single log-log diagnostic plot of the falloff pressure difference and its
logarithmic derivative, in the same manner as what is commonly done in pressure transient analysis. Then we provide a
global model for the observed response that combines a before-closure model coupling Carters leak-off model and Noltes
assumption on fracture surface growth into a material balance scheme with an after-closure model using an enlarged
equivalent wellbore radius to provide effectively infinite conductivity fracture behavior.
This model provides a complete assessment tool that rigorously matches the complete fracture pressure falloff response,
and in turn allows quantification of all fracture parameters (closure stress, fracture length, fracture height, leak-off coefficient
and fracture fluid efficiency) as well as reservoir permeability, provided that enough time is allowed for the falloff to
establish pseudo-radial flow.
The additional value of the global model is that it can be used to design and optimize fracture-injection/falloff tests in low
permeability reservoirs, where the optimal injection rate, injected fluid volume and injection time become the final output of
the design procedure. In this case, the goal is to design a fracture-injection/falloff test that enables estimation of both the main
fracturing treatment calibration parameters (closure stress, leak-off coefficient and fracture fluid efficiency) as well as
reservoir permeability, where optimized injected fluid volumes and reduced fracture dimensions allow development of the
pseudo-radial flow transient in a practical length of time.
Nomenclature
CL= leak-off coefficient, ft/min0.5
ct = total compressibility, psia-1
E= Youngs modulus, psi
E= plane strain modulus, psi
F= F-function, dimensionless
g= g-function, dimensionless
h = net pay thickness, ft
hf = hydraulic fracture height, ft
ISIP = instantaneous shut-in pressure, psi
k = permeability, md
kf w = fracture conductivity, md-ft
m= constant derivative level in a log-log plot

SPE 144028

19

p = pressure, psia
p* = extrapolated pressure, psia
pc =fracture closure stress, psia
pi = initial pressure, psia
pr = reservoir (initial) pressure, psia
pw = wellbore pressure, psia
q = flow rate, STB/D or MSCF/D
qi = injection rate into one wing of the fracture, bbl/min
qi TOT = total injection rate, bbl/min
rw = wellbore radius, ft
Sp = spurt loss, in
t = time, hr
tc= fracture closure time, hr
tp = time of pumping or production, hr
Vi = volume of fluid injected in one wing of the fracture, ft3
Vi TOT = total volume of fluid injected, ft3
VLOi = volume leaked off from one wing of the fracture during injection, ft3
we = width at end of injection, in
xf = fracture half-length, ft
p = pressure change, psia
p = pressure change derivative, psia
t = shut-in time, hr
= fracture fluid efficiency, %
= viscosity, cp
v = Poissons ratio
= porosity, %
= superposition time, dimensionless
Acknowledgements
The authors would like to thank BP for sharing data used for the Cotton Valley field example, and Shell-SWEPI and EnCana
Oil and Gas for the Haynesville Shale example.
References
Abramowitz, M. and Stegun, I.A.: Handbook of Mathematical Fucntions. Dover, New York, 1972.
Al-Hussainy, R., Ramey, H.J. Jr., and Crawford, P.B.: The Flow of Real Gases Through Porous Media. JPT (May) 624; Trans., AIME,
237, 1966.
Barree, R.D., Barree, V.L., and Craig, D.P.: Holistic Fracture Diagnostics: Consistent Interpretation of Prefrac Injection Test Using
Multiple Analysis Methods. SPE Production & Operations pp.396-406, August 2009.
Benelkadi, S., and Tiab, D.: Reservoir Permeability Determination Using After Closure Analysis of Calibration Test. SPE paper 70062,
2001.
Benelkadi, S., Belhaouas, R., and Sahar, M.: Use of After Closure Analysis to Improve Hydraulic Fracture Designs, Application on
Algerias In-Adaoui Gas Field. SPE paper 80936, 2003.
Blasingame, T.A., and Lee, W.J.: Variable-Rate Reservoir Limits Testing. SPE paper 15028, 1986.
Bourdet, D., Ayoub, J.A., and Pirard, Y.M.: Use of pressure derivative in well-test interpretation. SPEFE (June), 293-302, 1989.
Cinco-Ley, H., and Samaniego-V., F.: Transient Pressure Analysis for Fractured Wells. JPT (September 1981), 1749-1766.
Craig, D.P., and Blasingame, T.A.: Application of a New Fracture-Injection/Falloff Model Accounting for Propagating, Dilated and
Closing Hydraulic Fractures. SPE paper 100578, 2006.
Economides, M.J., and Martin T.: Modern Fracturing Enhancing Natural Gas Production. ET Publishing, 2007.
Economides, M.J., Oligney, R.E., and Valk, P.: Unified Fracture Design. Orsa Press, 2002.
Geertsma, J. and De Klerk, F.: A Rapid Method of Predicting Width and Extent of Hydraulically Induced Fractures. JPT (December
1969) pp 1571-1581.

20

SPE 144028

Gu, H., Elbel, J.L., Nolte, K.G.: Formation Permeability Determination Using Impulse-Fracture Injection. SPE paper 25425, 1993.
Horner, D.R.: Pressure Buildups in Wells. Pressure Analysis Methods, Reprint Series, SPE, 9, 25-43, 1967.
Khristianovitch, S. A. and Zheltov, Y. P.: Formation of Vertical Fractures by Means of Highly Viscous Liquid. Proceedings Fourth
World Pet.Cong. (1955) Section II, pp 579-586.
Lee, W.J., Rollins, J.B., and Spivy, J.P.: Pressure Transient Testing. Textbook Series, SPE, Richardson, TX, 2003.
Mayerhofer, M.J., Ehlig-Economides, C.A., and. Economides, M.J.: "Pressure Transient Analysis of Fracture Calibration Tests". SPE JPT,
vol. 47, pp. 229-234, 1995
McCain, W.D, Voneiff, G.W., Hunt, E.R., and Semmelbeck, M.E.: A Tight Gas Field Study: Carthage (Cotton Valley) Field. SPE paper
26141, 1993.

Mohamed, I.M., Nasralla, R.A., Sayed, M.A., Marongiu-Porcu, M., and Ehlig-Economides, C.A.: Evaluation of After-Closure
Analysis Techniques for Tight and Shale Gas Formations. SPE paper 140136, 2011.
Nolte, K.G.: Determination of Fracture Parameters from Fracturing Pressure Decline. SPE paper 8341, 1979.
Nolte, K.G.: A General Analysis of Fracturing Pressure Decline with Application to Three Models. SPEPE (December) 571-83, 1986.
Nolte, K.G.: Principles for Fracture Design Based on Pressure Analysis. SPEPE (February) 22; Trans., AIME, 285, 1988.
Nolte, K.G., Maniere, J.L., and Owens, K.A.: "After-Closure Analysis of Fracture Calibration Tests". SPE paper 38676, 1997.
Nordgren, R. P.: Propagation of Vertical Hydraulic Fracture. SPEJ (August 1972) pp 306-314.
Pierce, W.G.: Geologic map of the Cody 1 degree x 2 degrees quadrangle, northwestern Wyoming. U.S. Geological Survey,
Miscellaneous Geologic Investigations Map I-2500, 1997.
Perkins, T. K. and Kern, L. R.: Widths of Hydraulic Fracture. SPE JPT (September 1961) pp 937-949.
Prats, M.: Effect of Vertical Fractures on Reservoir Behavior Incompressible Fluid Case. SPEJ, 1, (1961) pp 105-118.
Shlyapobersky, J., Walhaug, W.W., Sheffield, R.E., Huckabee, P.T.: Field Determination of Fracturing Parameters for Overpressure
Calibrated Design of Hydraulic Fracturing. SPE paper 18195, 1998.

Stehfest, H.: Numerical Inversion of Laplace Transforms. Commun. ACM 13 (1), 47-49, 1970.
Song, B., Economides, M.J., and Ehlig-Economides, C.A.: Design of Multiple Transverse Fracture Horizontal Wells in Shale Gas
Reservoirs. SPE paper 140555, 2011.
Song, B., and Ehlig-Economides, C.A.: Rate-Normalized Pressure Analysis for Determination of Shale Gas Well Performance. SPE
paper 144031, 2011.
Valk, P.P., and Economides, M.J: Hydraulic Fracture Mechanics. John Wiley & Sons, Chichester, England, 1995.
Valk, P.P., and Economides, M.J: Fluid-Leakoff Delineation in High-Permeability Fracturing. SPEPF, 14 (2), pp117-130, May 1999.
Wolfram, S.: Mathematica: A System for Doing Mathematics by Computer, 2nd ed., Addison-Wesley, NY, 1991.

SPE Metric Conversion Factors


bpm
x 2.649 788
E 03 = m3/s
F
(F 32)/1.8*
= C
ft
x 3.048*
E 01 = m
gal
x 3.785 412
E 03 = m3
inch
x 2.54*
E 02 = m
lbs
x 4.535 924
E 01 = kg
ppg
x 1.198 264
E + 02 = kg/m3
psi
x 6.894 757
E + 00 = kPa
* - conversion factor is exact

S-ar putea să vă placă și