SENSITIVITY OF FRACTURED HORIZONTAL WELL PRODUCTIVITY TO

RESERVOIR PROPERTIES IN SHALE-GAS PLAYS

By:
Juan Carlos Carrat

A thesis submitted to the Faculty and the Board of Trustees of the Colorado
School of Mines in partial fulfillment of the requirements for the degree of Master of
Science (Petroleum Engineering).

Golden, Colorado
Date: ____________________

Signed: __________________________
Juan Carlos Carrat

Approved: __________________________
Dr. Erdal Ozkan
Thesis Advisor

Golden, Colorado
Date: ____________________

Signed:__________________________
Dr. William Fleckenstein
Adjunct Professor and Interim Head
Department of Petroleum Engineering

ii

ABSTRACT

This research presents an investigation of the sensitivity of fractured horizontal

well performances to reservoir properties in ultra-tight, naturally fractured shale-gas
reservoirs under depletion. The method of the research is analytical. Following a
general review of the existing literature on well performances in unconventional tight
reservoirs, the analytical trilinear flow model developed by Ozkan et al. (2011) is
introduced. The original dimensionless trilinear flow solution in Laplace domain is
converted to dimensional form to be used in productivity index calculations.
The productivity of fractured horizontal wells in shale-gas plays is discussed in
terms of transient productivity index defined by Araya and Ozkan (2002). For the
purposes of this research, the stimulated reservoir volume (hydraulically fracturing the
horizontal well in order to create pathways from the induced fractures towards the
existing naturally fractured reservoir, creating either a complex network of fractures
through hydraulic fracturing, or rejuvenating the existing healed fractures to create an
effective stimulation zone around the well) assumption is implemented. This
assumption, commonly used for shale-gas wells, limits the drainage of the well to the
stimulated volume around the well and strongly influences the average pressure to be
used in productivity computations. Derivation of the average pressure relationship for
the dual-porosity (naturally fractured) stimulated reservoir volume is presented and the
expression for the average pressure is combined with the transient trilinear flow solution
to obtain a relation for the transient productivity index. As a means of investigating the

iii

sensitivity of productivity to a selected set of reservoir properties, the derivatives of the

productivity index with respect to these properties are derived.
The results of the research are presented in three forms: (i) productivity index
versus time for the low, base, and high values of the selected property and the
productivity index versus the change in property value at selected times, (ii) tornado
charts indicating the sensitivity of the productivity index to the selected properties at
different times, and (iii) change in productivity index (derivative of productivity index)
with the variation of the selected properties. The results indicate that, unlike the
common expectation, the productivity of the fractured horizontal wells in shale-gas plays
is not very sensitive to the permeability of the natural fractures in the stimulated
reservoir volume and/or the distance between hydraulic fractures. The permeability of
the shale matrix and the density of the natural fractures in the stimulated reservoir
volume, on the other hand, influence productivity by 600% and 300% respectively. The
hydraulic fracture permeability and half-length appear to moderately influence
productivity on a 5% to 20% range. These results should be useful to guide well
completion in shale-gas plays, and the sensitivity analysis provided in this thesis should
be useful for engineering analysis of projects.

iv

TABLE OF CONTENTS

ABSTRACT ................................................................................................................... iii

LIST OF FIGURES ......................................................................................................... vii
LIST OF TABLES ............................................................................................................ x
ACKNOWLEDGMENTS ..................................................................................................xi
CHAPTER 1 INTRODUCTION ..................................................................................... 13
1.1

Motivation ................................................................................................ 14

1.2

Research Objectives ................................................................................ 15

1.3

Methodology ............................................................................................ 16

CHAPTER 2 BACKGROUND AND LITERATURE REVIEW ........................................ 17

2.1

Development of Shale-Gas Plays ............................................................ 17

2.2

Hydraulic Fracture Design ....................................................................... 20

2.3

Flow in Fractured Shale ........................................................................... 23

2.4

Knudsen Diffusion and Desorption in Shale Matrix .................................. 24

2.5

Modeling Production from Fractured Horizontal Wells ............................. 30

2.6

Pressure Dependent Natural Fracture Permeability ................................ 40

CHAPTER 3 TRILINEAR FLOW MODEL ..................................................................... 41

3.1

Physical Definition and the Assumptions of the Trilinear-Flow Model ...... 41

3.2

Trilinear-Flow Solution ............................................................................. 42

3.3

Dimensional Form of the Trilinear-Flow Solution ..................................... 47

CHAPTER 4 PRODUCTIVITY INDEX .......................................................................... 51

4.1

Productivity Index .................................................................................... 51

4.2

Derivative of the productivity index with respect to key system

properties ................................................................................................. 53
4.2.1 Derivative of the productivity ( J ) with respect to natural k f .......... 54
4.2.2 Derivative of the productivity ( J ) with respect to km ..................... 55
4.2.3 Derivative of the productivity ( J ) with respect to f ..................... 55
4.2.4 Derivative of the productivity ( J ) with respect to Lh ...................... 55
4.2.5 Derivative of the productivity ( J ) with respect to d F 2 ye ............ 56
4.2.6 Derivative of the productivity ( J ) with respect to k F ..................... 56
4.2.7 Derivative of the productivity ( J ) with respect to xF xe ............... 57

CHAPTER 5 SENSITIVITY OF PRODUCTIVITY TO KEY SYSTEM PROPERTIES .... 58

5.1

Productivity Index as a Function of Time ................................................. 60

5.2

Tornado Charts ........................................................................................ 70

5.3

Rate of Change of Productivity ................................................................ 74

CHAPTER 6 CONCLUSIONS ...................................................................................... 81

NOMENCLATURE ........................................................................................................ 84
REFERENCES .............................................................................................................. 88
APPENDIX A ................................................................................................................. 92

vi

LIST OF FIGURES
Figure 2.1

Horizontal rig activity in Major Shale Basins in the USA. (Viswanathan et

al., 2011) .................................................................................................. 18

Figure 2.2

Shale-gas plays in the USA (SLB, 2005) ................................................. 19

Figure 2.3

Types of Fractures & Events from Microseismic Mapping (Cipolla et al.,

2009)........................................................................................................ 21

Figure 2.4

Fracture Conductivity impact on Productivity (Cipolla et al., 2009b) ........ 22

Figure 2.5

Gas shale storage and flow capacity diagram showing pore type, flow
type, dominant particle motion within a given flow regime. Ion-milled SEM
image of Devonian gas shale. (Sodergeld et al., 2010) ........................... 25

Figure 2.6

Comparison of gas flow (a) in micropores where the flow is no slip and (b)
in nanopores where flow is slip (Javadpour et al., 2007) ......................... 25

Figure 2.7

Schematic of gas-molecule locations in a small part of a kerogen grain

pore system of a mudrock (Javadpour et al., 2009) ................................. 26

Figure 2.8

Comparison between pores in conventional (a) and unconventional shale

gas (b) reservoirs. (Javadpour et al., 2007) ............................................. 27

Figure 2.9

Gas evolution and production in shale gas sediments at different length

shales: (a) Macroscale: gas flow from the reservoir to the wellbore, (b)
Mesoscale: gas flow in large pores, (c) Microscale: gas flow in nanopores,
(d) Nanoscale: gas desorption from the nanopore walls, and (e) Molecular
scale: diffusion of gas molecules in the kerogen/clay. (Javadpour et al.,
2007)........................................................................................................ 28

Figure 2.10 Error caused by neglecting the contribution of slip flow on the productivity
of a fractured horizontal well (Ozkan et al., 2010) .................................... 28
Figure 2.11 Gas displacing water saturation profiles. Comparison of the Knudsen
diffusion approach (solid lines) and Darcy's apparent permeability
approach (Dashed Lines). (Andrade et al., 2011) .................................... 29
Figure 2.12 Schematic of the Quad Porosity; Green Circles: Pores in Inorganic, Gray
Circles: Kerogen (organic matter), tiny dots inside the gray circles
represent the pores within the organics, Red lines: Network of natural and
small induced fractures, Blue Bar: Major hydraulic fracture (Andrade et al.,
2011)........................................................................................................ 30

vii

Figure 2.13 Sketch of a fractured horizontal well in a tight-gas reservoir and the
convergence of flow around the well. (Ozkan et al., 2011) ...................... 31
Figure 2.14 Effective fracture concept for multiply fractured horizontal well in an
unconventional tight reservoir (Brown et al., 2011) .................................. 33
Figure 2.15 Trilinear flow model (Ozkan et al., 2011) ................................................. 34
Figure 2.16 Trilinear model match of the field data (Brown et al., 2009) ..................... 34
Figure 2.17 Trilinear Model-Effect of Outer Reservoir A-kf=2000 md and B-kf=20000
md (Ozkan et al., 2011) ........................................................................... 35
Figure 2.18 Trilinear Model-Effect of Outer Reservoir A-f=0.8 f/ft and B-f=2 f/ft
(Ozkan et al., 2011) ................................................................................. 36
Figure 2.19 Trilinear Model-Effect of Inner Reservoir Matrix Permeability km=1E-8 md
until km=1E-1 (Ozkan et al., 2011) .......................................................... 37
Figure 2.20 Trilinear Model-Effect of Inner Reservoir Matrix Permeability between
unconventional reservoir vs. homogeneous tight conventional (Brown et
al., 2011) .................................................................................................. 38
Figure 2.21 Effect of natural fracture density (Ozkan et al., 2011) .............................. 39
Figure 2.22 Effect of hydraulic fracture conductivity on the productivity of multiplyfractured-horizontal-wells (Ozkan et al., 2011) ........................................ 40
Figure 3.1

Trilinear-Flow Model Schematic (Brown et al., 2009)............................... 42

Figure 5.1

Effect of horizontal well length on productivity index ................................ 61

Figure 5.2

Effect of distance between hydraulic fractures on productivity index. ...... 61

Figure 5.3

Effect of matrix permeability on productivity index ................................... 62

Figure 5.4

Effect of natural fracture permeability on productivity index. .................... 62

Figure 5.5

Effect of natural fracture density on productivity index. ............................ 63

Figure 5.6

Effect of hydraulic fracture permeability on productivity index. ................ 63

Figure 5.7

Effect of hydraulic fracture half-length on productivity index. ................... 64

Figure 5.8

Productivity index as a function of horizontal well length. ........................ 66

Figure 5.9

Productivity index as a function of distance between hydraulic fractures. 66

Figure 5.10 Productivity index as a function of matrix permeability............................. 67

viii

Figure 5.11 Productivity index as a function of natural fracture permeability. ............. 67

Figure 5.12 Productivity index as a function of natural fracture density. ..................... 68
Figure 5.13 Productivity index as a function of hydraulic fracture permeability. .......... 68
Figure 5.14 Productivity index as a function of hydraulic fracture half-length. ............ 69
Figure 5.15 Sensitivity of productivity to various parameters; t = 1 hr ......................... 72
Figure 5.16 Sensitivity of productivity to various parameters; t = 10 hr ....................... 72
Figure 5.17 Sensitivity of productivity to various parameters; t = 100 hr ..................... 73
Figure 5.18 Sensitivity of productivity to various parameters; t = 1000 hr ................... 73
Figure 5.19 The Rate of change of productivity for different horizontal well lengths over
the life of the well. .................................................................................... 74
Figure 5.20 The Rate of change of productivity for low, base, and high values of the
distance between hydraulic fractures as a function of time. ..................... 75
Figure 5.21 The Rate of change of productivity for low, base, and high matrix
permeabilities over the life of the well. ..................................................... 75
Figure 5.22 The Rate of change of productivity for low, base, and high natural fracture
permeabilities as a function of time. ......................................................... 76
Figure 5.23 The Rate of change of productivity for low, base, and high densities of
natural fractures as a function of time. ..................................................... 76
Figure 5.24 The Rate of change of productivity for different values of hydraulic fracture
permeability during the life of the well. ..................................................... 77
Figure 5.25 The Rate of change of productivity for different values of hydraulic fracture
half-length as a function of time. .............................................................. 77
Figure 5.26 Comparison of the rate of change of productivity for the low values of the
selected properties as a function of time.................................................. 79
Figure 5.27 Comparison of the rate of change of productivity for the base values of the
selected properties as a function of time.................................................. 80
Figure 5.28 Comparison of the rate of change of productivity for the high values of the
selected properties as a function of time.................................................. 80

ix

LIST OF TABLES
Table 5.1 Base-case data for the well, hydraulic fracture, and reservoir properties ...... 59
Table 5.2 Range of the parameters used in the sensitivity study .................................. 60
Table 5.3 Data used in the sensitivity analysis; t = 1 hr. ................................................ 70
Table 5.4 Data used in the sensitivity analysis; t = 10 hr. .............................................. 71
Table 5.5 Data used in the sensitivity analysis; t = 100 hr. ............................................ 71
Table 5.6 Data used in the sensitivity analysis; t = 1000 hr. .......................................... 71

ACKNOWLEDGMENTS
I would like to express my sincere gratitude to my advisor Prof. Erdal Ozkan for
the continuous support during the process of developing my Masters Thesis. It would
be unfair not to mention his patience, motivation, enthusiasm, and understanding as
well as his knowledge that helped me through the learning stages of my thesis. I am
sure it would have not been possible without his help.
Special thanks to the petroleum engineering department faculty, specifically Dr.
Jennifer Miskimins, Dr. Hossein Kazemi and Dr. Ramona Graves for their consistent
willingness to share their knowledge and answer my continuous concerns. There infinite
support and continuous encouragement were key to fulfilling this document.
I cannot leave behind the petroleum engineering department Godmother
Madrina, Denise Winnbower, all my love and thanks goes for this woman who day in
and day out is always helping students reach their goals.
To my colleagues, fellow graduate students I send special thanks, specifically to
Najeeb Alharthy, John Akinboweya, Margaret Brown, Benjamin Ramirez, Mojtaba Kiani,
Felipe Silva, Jeff Brown, Jared Atkinson, Andrea Vega, Antonio Velasquez, Franklin
Useche and Carelia Rojas, among others. Your technical support and encouragement
were key for fulfilling this goal.
Many people and organizations helped me pursue and complete this personal
goal. Specifically, Bill Gumma and Luis Porras from Vinccler Oil & Gas S.A as well as
Dr. Mark Hart, Guy Goudy, Jim Sorensen and Aaron Goss from Austin Exploration.

xi

Finally, I would like to address the most important thanks to my family, specially
my parents, who have always encouraged me taking challenging paths, without their
support, guidance and humble wisdom, fulfilling a Masters degree at Colorado School
of Mines would non existant.
.

xii

CHAPTER 1
INTRODUCTION
During the last century, societies have relied on energy to develop and maintain
their economic and social strength. Energy is a necessity when talking about growth. Oil
and gas are energy-efficient, non-renewable resources. In recent decades, exploration
and production practices for oil and natural gas have focused on efficient ways of
recovery.
Due to the high-energy demand and high depletion of conventional reservoirs,
much interest has focused on unconventional reservoirs, most of which are low
permeability formations. These types of reservoirs can be defined as oil and gas
accumulations that cannot be produced at economic flow rates without the help of
massive stimulation treatments or special recovery processes and technologies, such
as hydraulic fracturing and steam injection. Coal-bed methane, tight-gas sands and
shale-gas reservoirs belong in this category (Holditch, 2001).
According to the US Department of Energys Energy Information Association, the
gross production from shale plays in the US was nearly doubled (from 1,180 Bscf to
more than 2,000 Bscf) from 2007 to 2008 (EIA 2009) and it continues growing. Most of
the production has been supplied from large shale plays, such as the Barnett and
Devonian Shale, while the continuing increase is due to the development of newer plays
including Woodford, Haynesville, Marcellus, Eagle Ford, and Horn River. Shale gas has
been providing the largest source of growth in the United States natural gas supply and
that contribution is expected to grow 113 percent from 2011 to 2040 (EIA 2011).

13

Effective drainage of shale-gas reservoirs is a challenge for reservoir engineers.

It is agreed that one of the major factors determining the productivity characteristics of
ultra-low permeability reservoirs is the existence of a natural fracture network. A
common and effective practice in these reservoirs is drilling horizontal wells and then
hydraulically fracturing in order to create pathways from the induced fractures towards
the existing naturally fractured reservoir, creating either a complex network of fractures
through hydraulic fracturing, or rejuvenating the existing healed fractures to create an
effective stimulation zone around the well (stimulated reservoir volume SRV). The
combination of an ultra-tight shale matrix with hydraulic fractures, natural fractures, and
a horizontal well is a complicated flow system a reservoir engineer needs to understand
and model in order to predict and optimize the production performance. Most of the
production optimization practices in these systems have been by a trial and error
approach or by following old procedures that were effective in similar reservoirs.
Although these approaches may be attractive in early and fast development phases,
they do not ensure optimum productivity and, in some cases, end up in failure.
This thesis research is intended to develop a better understanding of the large
number of parameters influencing the productivity of fractured horizontal wells in shalegas reservoirs. The ultimate goal is to improve the optimization of production from
shale-gas reservoirs.
1.1

Motivation
Among many activities of petroleum reservoir engineering, the most important

responsibility is optimizing hydrocarbon production and reservoir drainage accounting

for project economics. Although improving well completions and reservoir management
14

practices based on past experiences are reasonable approaches, they are not very
useful during early phase of development of a new play nor do they provide a predictive
capacity. For this reason, the creation of reservoir models for static and dynamic
subsurface conditions is an effective tool to simulate fluid flow towards wellbores and
determine the conditions for best productivity. For shale-gas plays, in addition to
inherent challenges of modeling subsurface flows, models should account for an
average pore size in the micrometer range (Sondergeld et al, 2010) and incorporate
fluid flow dominated by diffusion.
The development of unconventional shale-gas plays in the US is rightfully
attributed to improvements in horizontal drilling and staged fracturing techniques and
successful execution of new technologies to improve project economics. The fast
development pace of the resource plays was amenable to a trial and error approach to
improve the short-term performances by experimenting with different drilling and
completions techniques. As the shale-gas industry matures in the US and long-term
sustainable natural-gas production is deemed important for the energy security of the
country, better predictive approaches are required to optimize performance of shale-gas
wells. Such optimization requires better understanding of the sensitivities of well
performance to well, hydraulic fracture, and reservoir properties. This understanding
sets the motivation for the research presented in this thesis.
1.2

Research Objectives
The objective of this research is to improve understanding of the parameters that

affect recovery from shale-gas plays. These parameters include horizontal well
properties, such as length, hydraulic fracture properties, such as conductivity, half15

length, and distance between hydraulic fractures, and reservoir properties, such as
permeability of the matrix and natural fractures and natural fracture density (or the size
of the matrix blocks). The results are expected to improve optimization of hydraulic
fracturing treatments, horizontal well lengths, and well spacing in shale-gas plays.
1.3

Methodology
In this research, an analytical approach will be used to represent the

performances of fractured horizontal wells in shale-gas plays. The trilinear model

formulation (Ozkan et al., 2011, and Brown et al., 2011) will be used to derive closedform analytical expressions for the change of the productivity of fractured horizontal
wells in shale-gas reservoirs as a function of horizontal well, hydraulic fracture, natural
fracture, and matrix properties. Analytical derivatives of the productivity index with
respect to the parameters of interest are provided. The results are presented as a
function of time to delineate the short- and long-term effects of these parameters. A set
of base-case data will be selected and the results will be generated for a range of the
selected properties around the base case. Tornado charts will also be used to
demonstrate the sensitivity of the productivity to the variations of the selected set of
properties.

16

CHAPTER 2
BACKGROUND AND LITERATURE REVIEW
Due to the demonstrated potential and growing social and political acceptance of
shale gas as a long-term energy source, research and development efforts as well as
technology implementations in the field have been vibrant. In this chapter, a summary of
the most relevant developments will be provided with a review of the literature.
2.1

Development of Shale-Gas Plays

Economic gas and oil production from low permeability reservoirs has been a

challenge for the oil and gas industry. Most of the high permeability reservoirs have
been depleted leaving behind much lower permeability formations, which were deemed
unviable due to technical complexities and unfavorable economics. In the last decade,
encouraged by the improvements in technical and economic conditions, particular
attention has been given to tight gas reservoirs with permeabilities in the range of microdarcies, such as shale-gas reservoirs.
Shale-gas reservoirs has become a main interest in the United States due to the
success of the Barnett, Woodford, and Fayetteville Shales. The popular technique for
producing from these reservoirs has been the stimulation of horizontal wells. Even with
the drop of natural gas price in 2009, horizontal rig activity (Figure 2.1) has been
relatively consistent in shale plays (Viswanathan et al., 2011). However, beyond the
early production performances, there is a limited understanding of what properties or
factors control the long-term productivity of the shale gas wells.

17

Figure 2.1 Horizontal rig activity in Major Shale Basins in the USA. (Viswanathan et al.,
2011)
Shale-gas reservoirs are organic-rich shale formations that store gas in limited
rock pore space. Some amount of the gas in place may be absorbed on the organic
material (kerogen). Due to the basin-scale accumulation of shale-gas resources,
estimated gas-in-place volumes are large. For example, the estimated 500-1000 Tcf
(Arthur et al., 2008) gas potential in the US represents a jump from 150 Tcf predicted
not more than a decade ago. These estimates have changed due to the economic
success of the Barnett Shale and many that followed (Woodford, Fayetteville, Marcellus,
Eagle Ford, and Haynesville). Figure 2.2 shows the major shale-gas plays in North
America.

18

Figure 2.2 Shale-gas plays in the USA (SLB, 2005)

Shale formations, which are the most common of sedimentary rocks, may be
classified as source rock as well as the reservoir rock. Made up of clay size weathering
debris, shale formations are fine-grained laminated clastic sedimentary rocks that are
soft and fissile (Sondhi et al., 2011). Typically they have a thickness of 50 to 600 ft,
porosities between 2-8%, permeabilities from 10 to 100 nano-darcy, organic content of
1-14%. They are encountered in depths ranging from 1000 to 13000 ft (Cipolla et al.,
2009).
One way to characterize shale is to determine if it is silica-rich, carbonate-rich or
clay-rich. The geological setting of the rock may be inferred from its mineralogy and
used to assess formation fraccability and effective completion design, which in some
cases could shadow the potential of hydrocarbon prone shale (Sondhi et al., 2011).
The success in developing shale-gas plays in the last decay has been attributed
to improvements in completion and stimulation techniques on horizontal wells as well as

19

increased gas price. Unfortunately, an understanding of the factors influencing

productivity has not been achieved at the same pace.
2.2

Hydraulic Fracture Design

Developments in the area of hydraulic fracturing were a key proponent to the

success of shale-gas plays. Design of fractures has changed from simple (bi-wing)
fractures to more complex fracture networks to accomplish the near-wellbore stimulation
effect (stimulated reservoir volume, SRV) needed in shale plays (Figure 2.3).
Combination of reservoir modeling with microseismic mapping (Figure 2.3) has
cultivated understanding and improved fracture design techniques. Integrating the latter
with core data, better understanding of the performance of the combined system
consisting of the well, hydraulic fracture, and reservoir (natural fractures and shale
matrix) can be achieved. From modeling studies, it was proven that creating a highconductivity hydraulic fracture would affect stress on the fracture network and matrix
resulting in improved productivity (Cipolla et al., 2009a). This also indicates the
necessity for proper understanding of the matrix environment when designing horizontal
wells and fracture treatments. Such an understanding is not possible only from
microseismic mapping. Mayerhofer et al. (2006), conducted similar analyses by
integrating Microseismic fracture mapping and modeling flow in fracture networks in the
Barnett shale. They evaluated the influences of the network size, network density,
network conductivity, and high near-well fracture conductivity among others. Their
primary conclusion was that in any completion strategy the main goal is to generate the
largest possible fracture network with the highest possible fracture density to maximize
fracture surface area.
20

Gas Shales: Network Development

Waterfrac Examples From Microseismicity
1500

All Microseisms
1000

~2500 ft

~2700 ft

1000

500

500

-500
Simple Fracture

Northing (ft)

Northing (ft)

Observation
Well
Complex Fracture

Frac
Well

~1200 ft

-500

~600 ft

-1000

Possible
Aligned
Features

-1000

Frac Well

-1500

-1500
-2000
-1000

-500

Complex Fracture
With Fissure Opening

500

Easting (ft)

1000

Observation Well

1500

Complex Fracture
Network

-2000
-1000

From Fisher et al, SPE 77411

-500

500

1000

1500

Easting (ft)

Figure 2.3 Types of Fractures & Events from Microseismic Mapping (Cipolla et al.,
2009)
Neuhaus et al. (2012) took a similar approach in the Marcellus shale by using
microseismic technology and integrating it with well logs, well cores, and other reservoir
data to observe how the Microseismic events reflect fracture propagation via dip and
strike analysis. With this information, they proposed to couple reservoir models with
fracture simulators to create a robust tool for analysis of the field performance.
Viswanathan et al. (2011), combined data from subsurface conditions and the
effect of stimulation techniques on production into a set of methodologies (normalized
production comparison, productivity index analysis, numerical simulations and neural
network trained self-organizing maps). They used the results to assess the optimal
stimulation technique including the selection of the proppant and fracture fluids as well
as the effective cluster spacing.

21

Based on reservoir simulation results and production profiles evaluated, Cipolla,

et al. (2009b) noted the conductivity of the hydraulic fracture as one of the critical
parameters, which controls horizontal-well performances in unconventional reservoirs.
Figure 2.4 shows that, as fracture conductivity is increased, productivity increases. In
this analysis, all properties were kept constant; the only value that was changed was the
hydraulic fracture conductivity, which started at 2 md-ft (equal; to the matrix
permeability) and increased to 500 mD-ft. The results in Figure 2.4 show the importance
of generating a path that interconnects the wellbore with the matrix network. The blue
dotted curve, denoted as Well A in Figure 2.4, is an actual field data matching the
simulation with a fracture conductivity of 2 mD-ft.

Figure 2.4 Fracture Conductivity impact on Productivity (Cipolla et al., 2009b)

Other parameters, such as fracture spacing, effectiveness of propped fracture
length, matrix conductivity, natural fracture spacing, and formation permeability have
also been the subject of research, Fisher et al. (2004),), Ketter et al. (2006), and, Cipolla
et al. (2008). These studies have emphasized the importance of understanding matrix

22

permeability and un-propped fracture conductivity, as well as hydraulic fracture

conductivity, when optimizing stimulation treatments in shale-gas reservoirs.
Publications by Fisher et al. (2004) and Ketter et al. (2006) have noted that inefficient
fracture initiation is the most repetitive problem when completing horizontal wells in the
Barnett shale. An analysis on 256 wells indicated multiple areas that could be improved.
Among the most important was the use of image logs in order to predict horizontal
stress anisotropy. This information could be used to reduce stress-shadowing effects
due to competition between fractures by selectively picking perforation clusters, and to
choose the appropriate cross-link gel for good proppant placement in order to generate
large fracture width.
2.3

Flow in Fractured Shale

In order to understand and model flow in fractured shale, the concept of dual

porosity has been extensively used (eg., Medeiros et al., 2008, Ozkan et al., 2011,
Brown et al., 2011, and Bello and Wattanberger, 2008 & 2010). This concept assumes
that the matrix has large storativity but negligible conductivity while the fracture network
has negligible storativity but high conductivity. In dual porosity idealization, matrix and
fracture media are represented as overlapping continua coupled by transfer functions
representing the fluid transfer from the matrix to fracture network. Warren & Root (1963)
were one of the first to introduce dual-porosity formulation to petroleum engineering
literature. They developed a mathematical formulation of single-phase flow in dual
porosity systems assuming pseudo steady state flow from matrix to fractures. The
Warren and Root model was later extended by Kazemi et al. (1969), to unsteady state
fluid transfer between matrix and fractures. Kazemis model was capable of solving
23

matrix/fracture fluid transfer using multiphase transfer functions. For tight shale-gas
plays, transient fluid transfer assumption between the matrix and fractures has been
shown to be more appropriate (Brown et al., 2011).
2.4

Knudsen Diffusion and Desorption in Shale Matrix

Geologists categorize shale as mud rocks (Folk et al., 1974). Mud rocks are

sedimentary rocks, which have a grain size distribution lower than 62 m while pore
body and throat distribution are typically on a smaller range making it very difficult to
measure their physical properties (Sodergeld et al., 2010). Therefore, to model fluid flow
in shale, understanding fluid flow in pores with sizes ranging from micro meters to nano
meters is essential. This requires understanding the molecular size of the working fluids
that can access the small rock conduits. However, core measurement procedures for
unconventional reservoirs have not yet been established as they have for conventional
reservoirs (Recommended Practices 40, API 1998).
As shown in Figure 2.5 (Sodergeld et al., 2010), at nano meter pore scale,
Knudsen diffusion and slip flow regimes are expected to be dominant. In the case of
conventional Darcy flow, no-slip condition prevails in the flow channels (Figure 2.6a).
When the size of the pore channels becomes small and the pressure drops sufficiently
for the mean free path of the fluid molecules becomes larger that the pore diameter
(Figure 2.6b), then slip flow regimes develop. Darcys Law has been used to model fluid
flow in conventional reservoirs, initially; it was also applied to unconventional reservoirs
to model fluid flow in shale rocks. However, this approach consistently underestimated
production from shale-gas reservoirs because of flow mechanisms not accounted for by
Darcys law.
24

Figure 2.5 Gas shale storage and flow capacity diagram showing pore type, flow type,
dominant particle motion within a given flow regime. Ion-milled SEM image of Devonian
gas shale. (Sodergeld et al., 2010)

Figure 2.6 Comparison of gas flow (a) in micropores where the flow is no slip and (b) in
nanopores where flow is slip (Javadpour et al., 2007)
Javadpour et al. (2007) and Javadpour et al. (2009) discussed that flow in
nanopores of a shale matrix is described by Knudsen diffusion and slip flow while Darcy
flow covers flow in micropores, desorption occurs at kerogens surface, and finally
diffusion takes place in solid kerogen. As shown in Figure 2.7, they explained that gas
25

fills pores on shale rocks as compressed gas, remains at the kerogen surface as
adsorbed gas, and is dispersed inside kerogen as dissolved gas.

Figure 2.7 Schematic of gas-molecule locations in a small part of a kerogen grain pore
system of a mudrock (Javadpour et al., 2009)
Shale rocks, however, do not display a uniform pore size; instead, they have a
variety of pore sizes, ranging from micro size pores to nano size pores. In conventional
reservoirs, because larger pores dominate flow, the smaller pores stay below the
porosity cut-off (Figure 2.8a). In shale, on the other hand, most of the pore volume is in
the form of nano-pores and cannot be neglected (Figure 2.8b). Therefore, in dealing
with shale-gas reservoirs, multiple flow mechanisms prevailing in different size of pores
should be taken into consideration. Furthermore, existence of organic material
(kerogen) also brings the effect of desorption into consideration.

26

Figure 2.8 Comparison between pores in conventional (a) and unconventional shale gas
(b) reservoirs. (Javadpour et al., 2007)
According to Javadpour et al. (2007) and Javadpour et al. (2009), the order of
events for gas production from shale starts by generating an equilibrium disruption via a
drilled well or an inwduced fracture. These generate a pressure drop for the
compressed gas in the pores to be produced, followed by the desorption of gas
molecules on the kerogen surface due to decreased pressure in the kerogen pores. Due
to desorption, a concentration gradient is generated between the bulk of the kerogen
and its pore surface that triggers gas diffusion in kerogen. This mechanism is illustrated
in Figure 2.9 and represented by the following advection-diffusion-desorption equation:
c
U .c D : 2c c 0
t

(2.1)

In Eq. 2.1, the first term represents change of concentration with time, the second term
is referred to as advection, the third is Knudsen diffusion, and the last term refers to
desorption.

27

Figure 2.9 Gas evolution and production in shale gas sediments at different length
shales: (a) Macroscale: gas flow from the reservoir to the wellbore, (b) Mesoscale: gas
flow in large pores, (c) Microscale: gas flow in nanopores, (d) Nanoscale: gas
desorption from the nanopore walls, and (e) Molecular scale: diffusion of gas molecules
in the kerogen/clay. (Javadpour et al., 2007)
Ozkan et al., (2010), used this concept of multiple mechanisms contributing to
flow to develop a dual mechanism, dual porosity formulation for naturally fractured
reservoirs, which considers Darcy and Slip flow regimes concurrently. The conclusion
revealed that neglecting slip flow in matrix results in significant underestimation of well

400
350

Error for km = 1E-5 md

Error for km = 1E-8 md

100 x (Js-Jd)/Js, %

ERROR IN PRODUCTIVITY ESTIMATES

productivities (Figure 2.10).

300
250
200
150
100
50
0
1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

ELAPSED TIME, Days

Figure 2.10 Error caused by neglecting the contribution of slip flow on the productivity of
a fractured horizontal well (Ozkan et al., 2010)

28

Andrade et al. (2011), incorporated shale-gas reservoir flow mechanisms into a

1D simulator. They ran the model by using conventional Darcys law versus Knudsen
diffusion approach. The experiment was performed using pure methane gas injected
into a porous medium of nano-scale permeability (100% water saturated). Figure 2.11
shows that the application of the apparent permeability correction due to the Knudsen
diffusion yields an enhancement in the ability of gas to flow.

Figure 2.11 Gas displacing water saturation profiles. Comparison of the Knudsen
diffusion approach (solid lines) and Darcy's apparent permeability approach (Dashed
Lines). (Andrade et al., 2011)
Andrade et al, (2011) proposed a quad porosity reservoir model (Figure 2.12)
considering that the reservoir is comprised of four pore spaces: pores in the inorganic
and organic regions of the formation, natural fractures and hydraulic fractures.

29

Figure 2.12 Schematic of the Quad Porosity; Green Circles: Pores in Inorganic, Gray
Circles: Kerogen (organic matter), tiny dots inside the gray circles represent the pores
within the organics, Red lines: Network of natural and small induced fractures, Blue Bar:
Major hydraulic fracture (Andrade et al., 2011)
2.5

Modeling Production from Fractured Horizontal Wells

Hydraulically fractured horizontal wells have proven to effectively produce oil and

gas from tight, unconventional reservoirs. In tight-gas reservoirs, hydraulic fractures

were shown to reduce well drawdown, increase productivity of horizontal wells by
increasing the surface area in contact with formation, and provide high productivity
paths to wellbore. Project economics in tight formations, however, depend strongly on
well spacing and number of hydraulic fractures required to drain a reservoir efficiently.
Chen and Raghavan (1997) and Raghavan et al. (2007) developed a source-function
solution to evaluate the pressure behavior and production performance of fractured
horizontal wells in tight-gas reservoirs. Their main conclusion was summarized in the
following statement: The objective of fracturing a horizontal well is to create a system

30

whose long-term performance will be identical to that of a single fracture of length equal
to the spacing between the outermost fractures. (Raghavan et al., 1997). With this
interpretation, performance of fractured horizontal wells can be correlated in terms of an
effective fracture conductivity and effective fracture half-length. The conductivity of this
effective fracture depends on the permeability of the reservoir and the number, distance
between, and conductivities of the individual hydraulic fractures. The effective fracture is
envisioned to produce from the region beyond the tips of the fractures as shown in
Figure 2.13.

wf

CF

C F ,eff

wF ,eff 2 x F

xF

x F ,eff D / 2

Figure 2.13 Sketch of a fractured horizontal well in a tight-gas reservoir and the
convergence of flow around the well. (Ozkan et al., 2011)
Medeiros et al. (2007) built a semi-analytical model using Greens function
solution for the diffusivity equation, which had the capability of incorporating local
heterogeneities. They used this model to investigate the performances of horizontal
wells with multiple induced finite-conductivity fractures (transverse and longitudinal).
31

Dual-porosity idealization was used to incorporate a naturally fractured zone (a

stimulated reservoir volume) around the well and hydraulic fractures into the model.
Using their semi-analytical model, Medeiros et al. (2007) reached several
important conclusions. One of the most important conclusions was if hydraulic fractures
affect the stress distribution to create and rejuvenate natural fractures around the well,
then the productivity of the system is significantly increased; in this case, the main
production contribution comes from the natural fractures. This conclusion points to the
importance of focusing on stress distribution around the well in fracture design. Another
important conclusion was extended transient flow periods control the productivity of
these systems.
Ozkan et al. (2011) and Brown et al. (2011) presented an analytical trilinear flow
solution to simulate the pressure transient and production behavior of fractured
horizontal wells in unconventional reservoirs. Their model is based on the perception
that, similar to the work of Chen and Raghavan (1997) and Raghavan et al. (2007), the
fractured horizontal well can be represented by an equivalent fracture. In the case of
shale-gas reservoirs, because flow is mostly assumed to take place in the stimulated
reservoir volume covering the inner reservoir between hydraulic fractures, the length of
the equivalent fracture is equal to the aggregate length of individual fractures, as shown
in Figure 2.14, and the conductivity of the equivalent fracture is equal to average
conductivity of the fractures. The drainage volume around the equivalent fracture can be
seen as parallelepiped with the length of the equivalent fracture and the width equal to
the average hydraulic fracture spacing as seen on Figure 2.14.

32

ye,eff = d/2
wF

CF

d D / nF 1

2xF

2xF,eff = 2nxF

xF

C F ,eff C F

wF ,eff wF
Figure 2.14 Effective fracture concept for multiply fractured horizontal well in an
unconventional tight reservoir (Brown et al., 2011)
The Trilinear analytical model couples three environments (Figure 2.15), a tight,
homogeneous reservoir beyond the tips of the hydraulic fractures (outer reservoir), a
naturally fractured reservoir between hydraulic fractures (inner reservoir), and hydraulic
fractures distributed along the length of the horizontal well. Flow convergence is
essentially linear in hydraulic fractures and between fractures. Outer reservoir response
will also be linear initially. Therefore, coupling three contiguous linear flow regions
constitutes the essence of the Trilinear analytical flow model (Figure 2.15).

33

ye = dF/2

HYDRAULIC
FRACTURE
kF, wF, F, ctf

OUTER RESERVOIR
ko, o, cto

xe

xF

INNER RESERVOIR
NATURALLY
FRACTURED
kf, f, ctf,
km, m, ctm

HORIZONTAL WELL

Figure 2.15 Trilinear flow model (Ozkan et al., 2011)

Brown et al., (2011) showed that the Trilinear model matches with numerical the
numerical models and follow the trend of the field data (Figure 2.16). The results in
Figure 16 were obtained by implementing regression calculations since most of the
reservoir parameters were unknown. Nevertheless both pressure and derivative trilinear
model curves match the pressure and derivative field data curves reasonably well.

Figure 2.16 Trilinear model match of the field data (Brown et al., 2009)

34

Ozkan et al., (2011) also investigated the effect of the stimulated reservoir
volume consisting of the hydraulic and natural fractures on the production behavior of
horizontal wells. Similar to Mayerhofer et al. (2008), they noted that the production
beyond the stimulated reservoir volume is negligible. Taken from Ozkan et al. (2011),
Figure 2.17 shows the well performance for different sizes of the outer reservoir for two
values of natural fracture permeability in the inner zone. When the natural fracture
permeability is 2,000 md, the effect of outer reservoir is not seen until 10,000,000 hours
(around 100 years). For the natural fracture permeability of 20,000 md, the effect of the
outer boundary is noticed at approximately 100,000 hours (11 years), which is still
insignificant. Similar observations are noted for the effect of the natural fracture density
shown in Figure 2.18.

Figure 2.17 Trilinear Model-Effect of Outer Reservoir A-kf=2000 md and B-kf=20000 md

(Ozkan et al., 2011)

35

Ozkan et al. (2011) have also shown that the productivity of horizontal wells in
shale is not sensitive to natural fracture permeability (Figure 2.17) but influenced
significantly by the density of natural fractures (Figure 2.18). Similarly, will be shown
further in the sequence figures (Figure 2.21), taken from Ozkan et al. (2011), shows that
increasing hydraulic fracture conductivity will increase productivity, however the benefit
decreases as conductivity increases.

Figure 2.18 Trilinear Model-Effect of Outer Reservoir A-f=0.8 f/ft and B-f=2 f/ft (Ozkan
et al., 2011)
Unconventional reservoirs range their matrix permeabilities in nano darcies
(Javadpour et al., 2007, Sondergel et al., 2010, Orangi et al., 2011) and effect of matrix
permeability (Ozkan, et al., 2011) is seen at intermediate times by reducing the
pseudopressure drop.
Another key reservoir property analyzed was the matrix permeability (Figure
2.19). As matrix permeability is high, the effect is seen at intermediate times by reducing
36

the pseudopressure drop. At late times all responses merge giving the impression that
matrix permeability doesnt play a significant role. However when looking at late time
productivities, matrix permeability does influence when matrix permeability is in the
nano darcy range.

Figure 2.19 Trilinear Model-Effect of Inner Reservoir Matrix Permeability km=1E-8 md

until km=1E-1 (Ozkan et al., 2011)
Brown et al., (2011) compares conventional tight reservoirs with unconventional
reservoirs showing that (Figure 2.20) shows productivity drop is very similar, in fact,
showing that the unconventional case, behaves best. This is a clear proof of how
important is to have a proper understanding of the potential completion designs to
enhance productivity via the stimulated reservoir zone.

37

Productivity Index, J/nf, (Mscf - cp)/(d - psi )

1.E-03

1.E-04

Naturally Fractured
Homogeneous, k = 0.1 md

1.E-05
1.E-03

1.E-02

1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

Time, t, hr

B - Productivity Index vs Time

Figure 2.20 Trilinear Model-Effect of Inner Reservoir Matrix Permeability between

unconventional reservoir vs. homogeneous tight conventional (Brown et al., 2011)
Finally, we discuss the findings on natural fracture permeability and, as simple as
possible, if high permeability is seen, better productivity. This only works, if large
volumes or pools of fluids are interconnected within the natural fractures. It is known
that most of the fluid potential will come from the matrix network. As long as more
conduits are interconnected, there is a higher chance of fluid passage. Ozkan et al.,
(2011) shows how the natural fracture density increases productivity (Figure 2.21).
This is reasoned by understanding that high permeabilities contribute to flow
within natural fractures, not within matrices. If no fluid transfer happens between
matrices and fractures no additional productivity will take place.

38

As a higher density of natural fractures is achieved, a higher drainage is

achieved. A successful design is dependent on hydraulic fractures and rock and fluid
properties.

Figure 2.21 Effect of natural fracture density (Ozkan et al., 2011)

Similarly, the stimulated reservoir will vary in performance depending on how the
hydraulic fractures are designed as well as the spacing between them. Figure 2.22
shows that increasing hydraulic fracture conductivity will increase productivity, however
the benefit decreases as conductivity increases. From a fundamental viewpoint, higher
productivity is achieved by making the dimensionless conductivity, larger. This concept
is relatively straightforward for the design of hydraulic fractures in conventional tight
formations (such as tight gas sands) where higher conductivities can be achieved for
low reservoir permeability. A large dimensionless conductivity indicates that the
hydraulic fracture can transmit all the fluid provided by the reservoir.

39

Figure 2.22 Effect of hydraulic fracture conductivity on the productivity of multiplyfractured-horizontal-wells (Ozkan et al., 2011)
Increasing hydraulic fracture conductivity will increase productivity and, from a
fundamental viewpoint, higher productivity is achieved by making the dimensionless
conductivity larger. A large dimensionless conductivity indicates that the hydraulic
fracture can transmit all the fluid provided by the reservoir.
At late times, hydraulic fracture permeability merges together for
pseudopressure; as per productivity they all merge at about the same trend.
2.6

Pressure Dependent Natural Fracture Permeability

Because natural fracture are imported for efficient drainage of the stimulated

reservoir volume, and they open and close under stress and strain, there has been
interest in investigating the effect of stress dependent natural permeability on horizontal
well productivity in shale-gas plays. Cho et. al. (2012) developed an experimental study
of pressure-dependent natural fracture permeability in tight unconventional reservoirs.
They show that the effect of pressure-dependent, natural-fracture permeability on shalegas-well production is a function of the matrix permeability. If the matrix permeability
was too tight, then the retained permeability of the natural fractures could still be
sufficient for the available volume of the fluid.
40

CHAPTER 3
TRILINEAR FLOW MODEL
In this research, the trilinear flow model developed by Ozkan et al. (2011), is
used to represent the performances of fractured horizontal wells in shale-gas plays.
This chapter describes the general features and the assumptions of the trilinear model
and presents the basic equations and definitions of the terms used. Because the
original derivation of the trilinear flow solution by Ozkan et al. (2011), was in
dimensionless form in the Laplace transform domain, for the purposes of this research,
it is converted to dimensional form by using the Laplace transform properties. This
analytical conversion is required because some of the properties of interest, which are
critical for the discussion of productivity, are used in the definition of dimensionless time,
and upon application of Laplace transformation to dimensionless time, these properties
become implicit in the Laplace transform parameter.
3.1

Physical Definition and the Assumptions of the Trilinear-Flow Model

The trilinear model has been developed by coupling flows in three interconnected

subsurface environments contributing to the production of a fractured-horizontal-well

surrounded by a stimulated reservoir volume (Figure 3.1). The most important
assumption used in the derivation of the model is that the flow in each environment is
linear; that is, linear flow from the outside reservoir towards the naturally fractured
stimulated reservoir volume, from the naturally fractured reservoir towards the
hydraulically induced fracture, and finally from the hydraulic fracture towards the
wellbore. The model incorporates the radial convergence of flow toward the wellbore
41

intersection in the hydraulic fracture by using a flow choking term (Ozkan et al., 2011).
In the derivation of the model, the lines of symmetry formed by the no-flow boundaries
are used (Figure 3.1). Because the detailed derivation and verification of the trilinear
model has been given by Brown et al (2011) and Ozkan et al. (2011), no further details
of the development and derivation of the trilinear flow model will be given here. Only
some definitions that are essential for the purposes of this research will be provided in
the following sections.

Figure 3.1 Trilinear-Flow Model Schematic (Brown et al., 2009)

3.2

Trilinear-Flow Solution
The dimensionless solution for the wellbore pressure obtained from the trilinear

model is given by

P wD sD

sDCFD tanh

(3.2.1)

42

The definitions of the terms used in Eq.3.2.1 are as follows:

2 F
s
D
CFD FD

(3.2.2)

F O tanh O yeD D
2

O
CRD yeD

(3.2.3)

uD

(3.2.4)

u D sD f s D

(3.2.5)

where, f ( s) is the transient dual-porosity matrix-to-fracture transfer function (Kazemi,

1969, deSwaan-O, 1976, Serra et al., 1983) given by

f sD 1

3 sD
tanh
3sD

(3.2.6)

With

m ctm hmt m ctm hm

f ctf h ft f ctf h f

(3.2.7)

And

43

 xF2
12 2
hm

sD

OD

FD

km hmt

k f h ft

xF2
12 2
hm

km hm

k
h
f f

(3.2.8)

sD

tanh
xeD 1
OD

(3.2.9)

F
f

(3.2.10)

Where

Ck F
ct F

Ck f

(3.2.11)

ct f

(3.2.12)

Ck f

ct f

OD

(3.2.13)

O
f

(3.2.14)

And

CkO
ct O

(3.2.15)

44

In Eqs. 3.2.11, 3.2.12, and 3.2.14,

C 2.637 X 104

(3.2.16)

and sD is the Laplace transform parameter with respect to dimensionless time,

t D The definitions of the dimensionless variables used in Eq. 3.2.1 are given by

pD

tD

k f h ft
141.2q

k f h ft
141.2q

pi p

(3.2.17)

f
(3.2.18)

xF2

Where

xF

; x or y

wD

wF
xF

CFD

k F wF

k f xF

(3.2.19)

(3.2.20)

k F wF
h
k f ft xF
h

(3.2.21)

And

CRD

h
k f ft xF
h
k f xF

kO ye
kO ye

(3.2.22)

In Eqs. 3.2.20 and 3.2.21, k f is the bulk natural fracture permeability defined by
45

V
h
k f k f f k f ft
h
Vb

(3.2.23)

Also, it must be noted that the solution given by Eq. 3.2.1 uses the symmetry of
the system and considers only one of the hydraulic fractures with its corresponding
drainage volume. Because the hydraulic fractures are identical and equally spaced, the
production rate

of each fracture is

qt / nF

the number of hydraulic fractures; that is,

where

qt is the total flow rate and nF is

qt q nF .

Note that the trilinear flow solution given above assumes the flow of a constant
compressibility fluid (oil). Because the flow of gas in shale is considered in this study,
we assume that the following pseudopressure transformation (Al-Hussainy and Ramey,
1966, and Al-Hussainy et al., 1966) can be used to obtain a liquid-equivalent form of the
gas diffusion equation:

m p 2

Pb

p'
dp '
Z

(3.2.24)

Under this transformation, gas flow solution in terms of dimensionless

pseudopressure defined by

mD

kh
m pi m pwf

1422qscT

(3.2.25)

becomes equivalent to liquid solution flow solution in terms of dimensionless pressure

(Eq. 3.2.1 and Eq. 3.2.17). In the rest of this thesis, we will present the mathematical
expressions in terms of dimensionless pressure but the results will be obtained in terms
of pseudopressure by applying Eqs. 3.2.24 and 3.2.25. Furthermore, the discussion of
46

the limitations of the pseudopressure approach in general and for applications to shalegas systems will be left for other studies.
3.3

Dimensional Form of the Trilinear-Flow Solution

The solution given in Eq. 3.2.1 is in the Laplace domain and

sD

represents the

Laplace transform-parameter with respect to t D . Because k f is involved in the definition

of dimensionless time,

t D , k f is implicit in sD . Therefore, conversion of Eq. 3.2.1 to

dimensional form cannot be accomplished by a straightforward substitution of the

definitions of dimensionless variables. Below, the appropriate procedure is outlined:
We first invoke the definition of dimensionless pressure to obtain:
p wf sD

1422q

k f h ft sDCFD F tanh

(3.2.26)

Then, we note the following Laplace transform property: If

F sD L F tD L F ct

(3.2.27)

where,

f
(3.2.28)

xF2

then

47

1 sD
F L F t F s
c c

(3.2.29)

where s and sD correspond to the Laplace transform parameters with respect to t and

tD, respectively. Using Eqs. 3.2.26 through 3.2.29, we can write

1
s
L pwf t p wf
c
c
141.2q

k f h ft c s / c C tanh
FD

L pwf t

(3.2.30)

141.2q

k f h ft sC tanh
FD

pwf s

the other definitions of the terms used in Eq. 3.2.27 are given below:

2 F
s
1 2c F
s

CFD c FD c CFD FD

1 2 F
s

c CFD FD

F c F c O yeD D
2

O
CRD yeD

(3.2.31)

(3.2.32)

s s 1 O
f
u
c c c CRD yeD

(3.2.33)

where

48

 O c O c

s
cOD

s
tanh
xeD 1
cOD

(3.2.34)

s
u sf s f s
c

(3.2.35)

3 s
3 s
c

s
f s f 1
tanh
tanh
1

3s
3s
c
c

(3.2.36)

and

in Eq. 3.2.36,

c 12

f xF2 km hmt

f km hm

12 2

xF2 hm2 k f h ft
hm

k f hf

(3.2.37)

let us expand and simplify the definitions given in Eqs. 3.2.31 through 3.2.33

xF2 2 F
s 2 xF2 F sxF2 2 xF2 F sxF2
F

f CFD FD f CFD F f CFD F

F
xF

xF2

1
w

c F F O tanh
O ye
f
2

xF
xF2

O
w

tanh

xF O tanh O ye
2
2 e
xF
2
2

xF

O
2
F

(3.2.38)

1 O

1 O
u
u

2
xF c CRD ye
f CRD ye

where

49

(3.2.39)

(3.2.40)

 O c O c
O

f
xF

s
cOD

s
tanh
xeD 1
cOD

tanh
xe xF
O
O

(3.2.41)

in the rest of this work, we will use Eq. 3.2.30 and Eqs. 3.2.31 through 3.2.41 without
the tilde sign.

50

CHAPTER 4
PRODUCTIVITY INDEX
The use of productivity index is a convenient means of investigating the effect of
various rock, fluid, well, and completion properties on the performance of a fractured
horizontal well in a shale-gas play. In conventional plays, the productivity index is
usually defined during stabilized (boundary-dominated) flow periods. For long horizontal
wells with multiple fractures in tight formations, transient flow periods are elongated and
most wells reach their economic cut-off rates either during transient flow or shortly after
reaching the boundary dominated flow period. Therefore, the conventional definition of
productivity index based on stabilized flow is not an appropriate measure of the
performance of these fractured horizontal wells in shale-gas plays.
In this chapter, a generalized definition of productivity index comprising both
transient and boundary-dominated flow periods is provided. The definition of the
productivity index is combined with the trilinear flow solution and applied to shale-gas
wells with a stimulated reservoir volume. To be used later in the discussion of the
sensitivity of productivity to various well, fracture, and reservoir properties, derivatives of
the productivity index with respect to selected properties are also derived.
4.1

Productivity Index
Productivity index is defined as production rate per unit pressure drawdown with

respect to the average pressure in a closed reservoir. The mathematical expression of

the productivity index is given by

51

 qt
P P
wf

J
qt

m P m Pwf

for oil

(4.1.1)

for gas

where qt is the total production rate from the well and p is the average pressure in the
drainage area. As noted in Chapter 3, for simplicity, in this work we present the
formulations of productivity index in terms of pressure (liquid flow), but, in the
computation of the results for gas flow, we use the pseudopressure formulation.
To be able to use the definition of productivity index with the trilinear flow
solution, following Araya and Ozkan (2002) and Medeiros et al. (2008), we write Eq.
(4.1.1) as follows:

qt
qt
qt
=
= 1
P Pwf Pi Pwf Pi P L p wf s P

(4.1.2)

In Eq. 4.1.2, p wf s is the trilinear flow solution (in Laplace domain) given in

dimensionless form by Eq. 3.2.27 in Chapter 3 and Pi P is obtained from material

balance on the closed volume of the reservoir as follows:
Consider,

V t
1 V
cV
V p
t p

(4.1.3)

applying Eq. 4.1.3 for a dual-porosity system, we have

cmVm c f V f

1
t
Vm V f
t
p

(4.1.4)

52

which yields

c V

m m

c f V f

pt 1t V

V f

5.615qt
24

(4.1.5)

We can rearrange Eq. 4.1.5 to write

0.234qt
0.234qt
p

V
t

Vm
V c V

f c f f V
Ah f c f f 1 m m m
m cm
V
V
V f c f V f

0.234qt

h c h

Ah f c f f 1 m m mt
h f c f h ft

0.234qt

Ah f c f h ft 1

(4.1.6)

Integrating Eq. 4.1.6 yields

Pi P

0.234qt
t
Ah f c f h ft 1

(4.1.7)

Note that the derivation of the material balance equation given above uses the
usual assumption that the production beyond the stimulated reservoir volume is
negligible; that is, the extent of the drainage volume perpendicular to horizontal well axis
in each direction is equal to the half-length of the hydraulic fractures. If the production
beyond the stimulated reservoir volume needs to be accounted for, then the additional
volume of the reservoir beyond the stimulated reservoir volume should be entered into
computations as a homogeneous medium having the properties (compressibility) of the
matrix.
4.2

Derivative of the productivity index with respect to key system properties

In Chapter 5, the variation of the fractured horizontal well productivity with the

change in some key reservoir properties will be discussed. One of the approaches used
53

in these discussions is the examination of the derivative of the productivity index with
respect to the selected system properties. In this thesis, the following well, fracture, and
reservoir properties, which are commonly encountered in the discussion of horizontal
well productivity in shale-gas reservoirs, will be used:
(i) Horizontal well length, Lh
(ii) Matrix permeability, km
(iii) Natural fracture permeability, kf
(iv) Density of natural fractures, f
(v) Distance between hydraulic fractures, dF
(vi) Hydraulic fracture permeability, kF
(vii) Hydraulic fracture half-length, xF
The selection of the above list of the properties should be deemed arbitrary, as
the importance of any system property will depend on the particular purpose of the
investigation. However, the procedure demonstrated here will be applicable to any other
property chosen for investigation. Below, the derivatives of J with respect to the
properties listed above are presented. The derivation of these relationships is rather
long and tedious and is given in Appendix A.
4.2.1 Derivative of the productivity ( J ) with respect to natural k f
s
2
p wf s
F
n
h

J
F ft

J 2 L F tanh F F F tanh 2 F
2

k f 141.2qt 2

1
w 2
w

F ye F xF ye O
xF
2
2

54

(4.2.1)

4.2.2 Derivative of the productivity ( J ) with respect to km

s p 2 s

wf

F
tanh F F 1 tanh

F
F2
1
w

4n k xF x ye 2
F O

f m
F O

nF k f h ft
J

2 1
2

J L

2
2
3
s
km 141.2qt
f s 1

s f s 1 s

sn f
xF 2
1

O
O
e
F
CRD ye
O xF F

2O

CRD km ye

(4.2.2)

4.2.3 Derivative of the productivity ( J ) with respect to f

s p 2 s

wf

CFD F f

tanh F F 1 tanh 2 F

nF k f
J
2C y

J 2 L1
2
RD e
1

f 141.2qt
F

3 s (4.2.3)
f O h ft

2

s 1 tanh


hmt

F2
w
F

xF O x ye 2

F
O
O

1 2 p
J
qt
f

4.2.4 Derivative of the productivity ( J ) with respect to Lh

J
1
1
p

J 2 L1 pwf s J 2
Lh qt Lh 2 ye
qt
Lh 2 ye
55

(4.2.4)

4.2.5 Derivative of the productivity ( J ) with respect to d F 2 ye

nF k f h ft
2
J

J 2 L1 s p wf s
d F 141.2qt2 B
1

xF2 O F2 tanh F F 1 tanh 2

2 xF F

(4.2.5)

1
1
p
J 2 L1 p wf s J 2
2qt ye
qt
Lh 2 ye

4.2.6 Derivative of the productivity ( J ) with respect to k F

s p 2 s
wf

CFD
F tanh F

k
F
k f h ft
J
1 2 1

J L
2
2 F sxF2
1
k F 141.2q q

2k F F CFD F

tanh F F 1 tanh 2

56

(4.2.6)

4.2.7 Derivative of the productivity ( J ) with respect to xF xe

nF k f h ft
J

J2
2
xF 141.2qt

1
F CFD tanh F

xF

2
2
F
s p wf ( s ) CFD tanh F F 1 tanh

L1
sxF
2 F

xF CFD F F F

2 f CRD CFD F O xF ye

f s xF2 2 xe

O 1

O
f

F x F y wF

F O x e 2
O
F
O

1 p xe
q J x x
e
F
t

57

(4.2.7)

CHAPTER 5
SENSITIVITY OF PRODUCTIVITY TO KEY SYSTEM PROPERTIES
In this chapter, the main results about the sensitivity of productivity to some key
properties of the well, hydraulic fractures, and the reservoir are presented. The basecase data shown in Table 5.1 is used to study the sensitivity of fractured horizontal wells
to the following properties of the well, hydraulic fractures, and the reservoir (highlighted
in red in Table 5.1):
(i) Horizontal well length, Lh
(ii) Matrix permeability, km
(iii) Natural fracture permeability, kf
(iv) Density of natural fractures, f
(v) Distance between hydraulic fractures, dF
(vi) Hydraulic fracture permeability, kF
(vii) Hydraulic fracture half-length, xF
The ranges of the properties used in this work are shown in Table 5.2.
It must be emphasized that no claim is made about the representativeness of the
base-case data and the ranges of the sensitivity properties shown in Tables 5.1 and 5.2,
respectively. In this study, they are rather an arbitrary compilation of the data we have
gathered. However, the qualitative interpretation of the sensitivities demonstrated by the
results presented in this section is not hampered by our choice of data. Similarly, the list
of the properties used in the sensitivity discussions is not all-inclusive; some common
properties usually associated with productivity issues have been selected. The

58

approach used here can be extended to the study of the effect of any other property. In
practice, the approach used here should be applied after bracketing the data ranges for
the particular play of interest. The sensitivity parameters and their ranges (the high,
base, and low values shown in Table 5.2) will be a choice of the analyst, which might be
influenced by many different considerations.
Table 5.1 Base-case data for the well, hydraulic fracture, and reservoir properties

59

Table 5.2 Range of the parameters used in the sensitivity study

Three approaches are used in this chapter to discuss the sensitivity of

productivity to the selected properties:
(I) Consider the base case data shown in Table 1 and present the productivity
index as a function of time for the ranges of the properties shown in Table 2. In
addition, present the productivity index as a function of each property at
selected times over the life of the well.
(II) Consider the low, base, and high values of the properties in Table 2 and present
tornado charts of sensitivity at selected times over the life of the well.
(III) Consider the derivative of productivity index as a function of the selected
properties and present the change in productivity with respect to the selected
parameters as a function of time for low, base, and high values of the property
shown in Table 2.
5.1

Productivity Index as a Function of Time

Figures 5.1 through 5.7 show the productivity index as a function of time for the

low, base, and high values of the parameters listed in Table 5.2. In each figure, the
effect of changing one property while keeping everything else at the base case value is
investigated. In general, the productivity index decreases as a function of time at early

60

times and then stabilizes at late times indicating the establishment of boundarydominated flow.

Figure 5.1 Effect of horizontal well length on productivity index

Figure 5.2 Effect of distance between hydraulic fractures on productivity index.

61

Figure 5.3 Effect of matrix permeability on productivity index

Figure 5.4 Effect of natural fracture permeability on productivity index.

62

Figure 5.5 Effect of natural fracture density on productivity index.

Figure 5.6 Effect of hydraulic fracture permeability on productivity index.

63

Figure 5.7 Effect of hydraulic fracture half-length on productivity index.

The results shown in Figures 5.1 and 5.7 are complementary and display the
interaction among horizontal well length, number of hydraulic fractures, and the distance
between hydraulic fractures. In Figure 5.1, increasing the length of the horizontal well,
keeping the distance between hydraulic fractures the same (250 ft), increases the
number of hydraulic fractures (10, 18, and 36 fractures for 2,500 ft, 4,500 ft and 9,000 ft
horizontal well lengths, respectively), but the drainage area for each fracture stays the
same. Because the total flow rate is fixed and equally distributed to each hydraulic
fracture, pressure drop in the total system decreases with the increase in horizontal well
length (total drainage area or stimulated reservoir volume). Therefore, the change
(increase) in productivity is governed by the lowered pressure drawdown due to
reduced production per unit length of the horizontal well.
In Figure 5.2, on the other hand, increasing the distance between hydraulic
fractures, keeping the horizontal well length fixed (4,500 ft), reduces the number of
64

hydraulic fractures (45, 18, and 10 fractures for 100 ft, 250 ft, and 450 ft distance
between hydraulic fractures, respectively). As the drainage area per fracture increases
at a faster rate than that for the production rate per fracture (total rate is fixed), the
productivity decreases as the distance between fractures increases. Comparison of the
results in Figures 5.1 and 5.2 indicate that despite the similar trends of the variations of
productivity, the cause of the productivity change is different. Moreover, the change of
one property may cause changes in the influence of the others. Thus, instead of
considering the individual effects of the horizontal well length, distance between
fractures, and the number of fractures, their conglomerated effect should be factored
into the design of the well construction
Similarly, multiple interpretations are possible from some figures. For example,
because the formation thickness and the individual natural fracture thickness are fixed,
increasing the density of natural fractures in Figure 5.5 is equivalent to changing the
thickness of matrix slabs, or increasing the horizontal well length (Figure 5.1) or
hydraulic fracture half-length (Figure 5.7) is similar to increasing the drainage area
(stimulated reservoir volume).
To assess the influence of each property on the productivity index, the results
shown in Figures 5.1 through 5.7 are replotted in Figures 5.8 through 5.14. It is
indicated by Figure 5.8 that the productivity index linearly increases by the increase of
horizontal well length and the increase is larger at earlier times. As explained above in
the discussion of Figures 5.1 and 5.2, the distance between hydraulic fractures (Figure
5.9) has a similar (but reverse) effect on productivity to that of horizontal well length.

65

Figure 5.8 Productivity index as a function of horizontal well length.

Figure 5.9 Productivity index as a function of distance between hydraulic fractures.

66

Figure 5.10 Productivity index as a function of matrix permeability.

Figure 5.11 Productivity index as a function of natural fracture permeability.

67

Figure 5.12 Productivity index as a function of natural fracture density.

Figure 5.13 Productivity index as a function of hydraulic fracture permeability.

68

Figure 5.14 Productivity index as a function of hydraulic fracture half-length.

Except for the distance between hydraulic fractures (Figure 5.9), Figures 5.10
through 5.14, show that the productivity is higher for higher matrix, natural fracture, and
hydraulic fracture permeabilities, higher natural fracture densities, and larger fracture
half-lengths. The rate of increase of productivity index becomes larger as the matrix
permeability (Figure 5.10) and natural fracture density (Figure 5.12) increases (the
increase is less apparent for natural fracture density at early times). Furthermore, the
difference between the early- and late-time productivities is insignificant for higher
matrix permeabilities and natural fracture densities (Figures 5.10and 5.12). The
increase in productivity becomes smaller as the natural fracture permeability (Figure
5.10), hydraulic fracture permeability (Figure 5.13), and hydraulic fracture half-length
(Figure 5.14) increases. For all these cases, the rate of increase in productivity is
smaller at late times.

69

Among all the cases shown in Figures 5.8 through 5.14, the productivity index
appears least sensitive to the increase in natural fracture permeability (Figure 5.11).
This observation may appear to be contrary to common expectation, but it can be
explained by the fact that even the natural fractures with relatively small permeabilities
constitute infinite-conductivity properties compared with the very tight matrix
permeability. Finally, as expected, the productivity decreases as the distance between
hydraulic fractures increases (Figure 5.9).
5.2

Tornado Charts
Tornado charts are useful to investigate the sensitivity of the productivity index to

the changes of the well, hydraulic fracture, and reservoir properties in the vicinity of their
base case values. The low, base, and high values of the properties given in Table 5.2
have been considered here. The data shown in Tables 5.3 through 5.6 have been used
to construct the tornado charts.
Table 5.3 Data used in the sensitivity analysis; t = 1 hr.

70

Table 5.4 Data used in the sensitivity analysis; t = 10 hr.

Table 5.5 Data used in the sensitivity analysis; t = 100 hr.

Table 5.6 Data used in the sensitivity analysis; t = 1000 hr.

The tornado charts shown in Figures 5.15 through 5.18 indicate that, among the
reservoir properties, the productivity index is most sensitive to matrix permeability. On

71

the other hand, the sensitivity of productivity to natural-fracture permeability is

insignificant while the sensitivity to the density of natural fractures is high.

Figure 5.15 Sensitivity of productivity to various parameters; t = 1 hr

Figure 5.16 Sensitivity of productivity to various parameters; t = 10 hr

72

Figure 5.17 Sensitivity of productivity to various parameters; t = 100 hr

Figure 5.18 Sensitivity of productivity to various parameters; t = 1000 hr

Among the properties of the well and hydraulic fractures, the well length has the
highest influence on productivity; hydraulic fracture permeability has moderate effect,
and the distance between hydraulic fractures and the hydraulic fracture half-length have
relatively insignificant effect.
73

5.3

Rate of Change of Productivity

One of the important considerations in well construction and stimulation is to

improve the properties, which are the most influential on productivity. One way of
determining these critical properties is to consider the rate of change of productivity with
respect to the key properties as shown in Figures 5.19 through 5.25.
It must be noted that results shown here display a different aspect of the
sensitivities of the productivity to selected properties. The rate of change of productivity
with respect to a property indicates the change in productivity per unit change in the
property. In the previous sections, we have examined the total change productivity
within the possible range of the variation of each property. For example, in Figure 5.19,
the rate of change of productivity with respect to unit change (1 ft) of horizontal well
length is very small. In the previous sections, however, we have observed that within
the range of the possible horizontal well lengths, the change in productivity is
reasonably large. In other words, if the well length can be increased by a large amount,
then the total change in productivity may be still considerably large. This interpretation is
applicable to all the results shown in this section.

Figure 5.19 The Rate of change of productivity for different horizontal well lengths over
the life of the well.

74

Figure 5.20 The Rate of change of productivity for low, base, and high values of the
distance between hydraulic fractures as a function of time.

Figure 5.21 The Rate of change of productivity for low, base, and high matrix
permeabilities over the life of the well.

75

Figure 5.22 The Rate of change of productivity for low, base, and high natural fracture
permeabilities as a function of time.

Figure 5.23 The Rate of change of productivity for low, base, and high densities of
natural fractures as a function of time.

76

Figure 5.24 The Rate of change of productivity for different values of hydraulic fracture
permeability during the life of the well.

Figure 5.25 The Rate of change of productivity for different values of hydraulic fracture
half-length as a function of time.
The rate of change of productivity with horizontal well length (Figure 5.19), matrix
permeability (Figure 5.21), natural fracture permeability (Figure 5.22), and hydraulic
fracture half-length (Figure 5.25) varies in time during early times but stabilizes at late
77

times. As the distance between fractures, matrix and natural fracture permeabilities, and
fracture half-length increase, the rate of change of productivity becomes smaller. This
indicates that if the distance between fractures cannot be decreased, then increasing
the number of fractures may not contribute to productivity significantly. Similarly, after a
threshold value, further increase in fracture half-length may not provide economically
meaningful improvement in productivity. Also, if the matrix and natural fracture
permeabilities are in the higher range, then it is more difficult to improve productivity by
further increasing the matrix and natural fracture permeabilities (in practice, only natural
fracture permeability may be changed by hydraulic fracturing but the matrix permeability
cannot be artificially changed).
Interpretation of the results for the natural fracture density and hydraulic fracture
permeability is more complex. Focusing only at large times, it can be seen that the
potential to increase the productivity is more significant at larger natural fracture
densities. However, the rate of change appears to be approaching a terminal value as
the natural fracture density increases. For the lower values of hydraulic fracture
permeability, there is a slightly better chance of improving productivity by increasing
hydraulic fracture permeability, but the increase is relatively insignificant.
Finally, in Figures 5.26 through 5.28, we summarize all the results of this section
for the low, base, and high values of the properties. As observed in the previous
sections, Figures 5.26 through 5.28 indicate that, among the properties which can be
controlled by well construction and stimulation design, the rate of change (sensitivity) of
the productivity is the highest to the density of natural fractures and the distance
between the hydraulic fractures. Increasing the natural fracture permeability and the

78

horizontal well length are not good choices for improving productivity (economically,
increasing horizontal well length can still be a good option). Also, increasing hydraulic
fracture conductivity may not provide an economically justifiable improvement in
productivity.

Figure 5.26 Comparison of the rate of change of productivity for the low values of the
selected properties as a function of time.

79

Figure 5.27 Comparison of the rate of change of productivity for the base values of the
selected properties as a function of time.

Figure 5.28 Comparison of the rate of change of productivity for the high values of the
selected properties as a function of time

80

CHAPTER 6
CONCLUSIONS
The results obtained in this research should be a useful guide for well completion
design in shale-gas plays and the sensitivity analysis provided should be useful for
engineering parameter optimization.
However, extra attention should be taken when applying the model by following
its assumptions (linear flow, uniform spacing of hydraulic fractures) so if those
conditions are not met, they are addressed accordingly.
Also, results shown are based from the low, base and high cases used from table
1 and values used are specific to this research. For any particular shale-gas play, the
knowledge of the field will mold what values and ranges are needed to use in order to
draw the right sensitivities.
Based on the work presented in this thesis, the following conclusions can be drawn:
1.

Unlike the common expectation, the productivity of the fractured horizontal

wells in shale-gas plays is not very sensitive to the permeability of the
natural fractures in the stimulated reservoir volume and the distance
between hydraulic fractures. The natural fracture permeability sensitivity to
productivity is close to none, while the distance between hydraulic fractures
is in the 7% influence, inverse to increase of the parameter trend.

2.

The permeability of the shale matrix and the density of the natural fractures
in the stimulated reservoir volume, on the other hand, influence productivity
by 600% and 300% respectively.

81

3.

The hydraulic fracture permeability and half-length appear to have

moderate influence of productivity with an approximate 7% on the high
case influence while a 15% on the low case. For fracture half-length the
span goes 7% on high and low case.

4.

The rate of change of productivity with horizontal well length, matrix

permeability, natural fracture permeability, and hydraulic fracture halflength varies in time during early times, but stabilizes at late times.

5.

As the distance between fractures, matrix and natural fracture

permeabilities, and fracture half-length increase, the rate of change of
productivity decreases.

6.

For considerable gain in productivity by increasing the number of hydraulic

fractures, the distance between hydraulic fractures should be small.

7.

After a threshold value, further increase in fracture half-length may not

provide economically meaningful improvement in productivity.

8.

If the natural fracture permeabilities are in the higher range, then it is more
difficult to improve productivity by further increasing the natural fracture
permeabilities.

9.

Potential to increase productivity is more significant at larger natural

fracture densities. However, the rate of change approaches a terminal
value as the natural fracture density increases.

10.

For the lower values of hydraulic fracture permeability, there is a slightly

better chance of improving productivity by increasing hydraulic fracture
permeability, but the increase is relatively insignificant.

82

11.

Among the properties which can be controlled by well construction and

stimulation design, the rate of change (sensitivity) of the productivity is the
highest to the density of natural fractures and the distance between the
hydraulic fractures.

12.

Increasing the natural fracture permeability and the horizontal well length
are not good choices for improving productivity (economically, increasing
horizontal well length can still be a good option).

13.

Also, increasing hydraulic fracture conductivity may not provide an

economically justifiable improvement in productivity.

Finally, it is recommended that the results presented in this work are combined
with an economic analysis to assess the impact of the variation of each property on
project economics.
.

83

NOMENCLATURE

Drainage area [ft2]

Bo

Formation volume factor [rbbl/STB]

Wellbore-storage coefficient [rbbl/psi]

CA

Shape factor

CD

Wellbore-storage coefficient, dimensionless

C FD

Hydraulic fracture conductivity, dimensionless

C RD

Reservoir Conductivity, dimensionless

c
c~

Compressibility [psi-1]

Distance between outermost fractures [ft]

Distance between two adjacent fractures [ft]

Reservoir thickness [ft]

hf

Thickness of natural fractures [ft]

h ft

Total thickness of natural fractures [ft]

hm

Thickness of matrix slabs [ft]

h ft

Total thickness of matrix slabs [ft]

Transient productivity index [stb/d/psi], [Mscf/d/psi2/cp]

Permeability tensor [md]

Permeability [md]

kI

Permeability of the inner reservoir [md]

kf

Natural fracture intrinsic permeability [md]

~
kf

Natural fracture bulk permeability [md]

kF

Hydraulic fracture permeability [md]

kO

Permeability of the outer reservoir [md]

Bulk compressibility [psi-1]

84

km

Matrix intrinsic permeability [md]

Length [ft]

Reference length [ft]

2
Pseudopressure, [ psi cp ]

~
m

2
Average reservoir pseudopressure, [ psi cp ]

nF

Number of hydraulic fractures

nf

Number of natural fractures

nm

Number of matrix blocks

p
~p

Pressure [psia]

Volumetric rate [rbbl/day; Mscf/day]

rw

Wellbore radius [ft]

rwj

Fracture effective wellbore radius [ft]

rwt

Total system effective wellbore radius [ft]

Laplace parameter

sc

Horizontal well flow choking skin factor

sD

Laplace parameter after time scaling property

Reservoir temperature [R]

Time [hours]

Velocity vector [ft/hours]

wF

Hydraulic fracture width [ft]

Point coordinate in x direction [ft]

xe

Reservoir size, x-direction [ft]

xF

Hydraulic fracture half-length [ft]

Point coordinate in y direction [ft]

ye

Reservoir size, y-direction [ft]

Point coordinate in z direction [ft]

Average reservoir pressure [psia]

85

Gradient operator

Three-dimensional Laplace operator

GREEK

Parameter defined in trilinear flow model

Parameter defined in trilinear flow model

Difference operator

Diffusivity [ft2/hr]

Transmissivity ratio

Viscosity [cp]

Pi constant

Fluid density [lbm/ft3]

Shape factor [ft-2]

Porosity

Bulk porosity

Intrinsic property

Bulk property

Storativity ratio

SUBSCRIPTS

Dimensionless

External boundary

Natural fracture

Hydraulic fracture

wf

Flowing wellbore

Horizontal

Initial

Inner Reservoir
86

Matrix

Outer Reservoir

Producing

Reservoir

sf

Sand face

Total

Internal boundary (wellbore)

x, y, z 3-D Cartesian-directions

Type of medium: m (matrix) or f (fracture)

87

REFERENCES
Araya, A. and Ozkan, E. 2002. An Account of Decline-Type-Curve Analysis of Vertical,
Fractured, and Horizontal Well Production Data, Paper SPE 77690 presented at
the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, Sept.
29-Oct. 2, 2002.
Al-Hussainy, R., and Ramey, H. J. 1966. Application of Real Gas Theory to Well Testing
and Deliverability Forecasting. JPT 18 (5): 637-642. SPE 1243-PA.
Al-Hussainy, R., Ramey, H. J., and Crawford, P. B. 1966. The Flow of Real Gases
through Porous Media. JPT 18 (5): 624-636. SPE 1243-A.
Andrade, J., Civan, F., Devegowda, D., and Sigal, R. 2010. Accurate Simulation of
Shale-Gas Reservoirs. Paper SPE 135564 presented at the SPE Annual
Technical Conference and Exhibition held in Florence, Italy, 19-22 September
2010.
Andrade, J., Civan, F., Devegowda, D., and Sigal, R. 2011. Design and Examination of
Requirements for a Rigorous Shale-Gas Reservoir Simulator Compared to
Current Shale-Gas Simulators. Paper SPE 144401 presented at the SPE North
American Unconventional Gas Conference and Exhibition held in The
Woodlands, Texas, USA, 14-16 June 2011.
Arthur, J.D., 2008. Hydraulic Fracturing for Natural Gas Well of the Marcellus Shale.
presented at the SPE North American Unconventional Gas Conference and
Exhibition held in The Woodlands, Texas, USA, 14-16 June 2011.Barenblatt,
G.I., Zheltov, I.P., and Kochina, I.N. 1960. Basic Concepts in the theory of
seepage of homogeneous liquids in fissured rocks. J. of Applied Mathematics
and Mechanics 24 (5): 1286-1303. DOI: 10.1016/0021-8928(60)90107-6.
Bello, R. O., Wattenbarger, R. 2010. Multi-stage Hydraulically Fractured Shale Gas
Rate Transient Analysis. Paper SPE 126754 presented at SPE North Africa
Technical Conference and Exhibition, Cairo, Egypt, 14-17 Feb.
Brown, M., 2009. Analytical Trilinear Pressure Transient Model for Multiply Fractured
Horizontal Wells in Tight Unconventional Reservoirs. Masters Thesis
Dissertation. Colorado School of Mines. Golden, CO.
Brown, M., Ozkan, E., Ragahavan, R., and Kazemi, H. 2011. Practical Solutions for
Pressure Transient Responses of Fractured Horizontal Wells in Unconventional
Reservoirs. SPEREE, Volume 14, Number 6 (December) pages 663-676.
Chen, C., and Raghavan, R.S. 1997. A multiply-fractured Horizontal Well in a
Rectangular Drainage Region. Paper SPE 37072 presented at the SPE Journal,
Volume 2, December 1997, pages 455-465.
88

Cho, Y., Apaydin, O.G., Ozkan, E. 2012. Pressure-Dependent Natural-Fracture

Permeability in Shale and its Effect on Shale-Gas Production. Paper SPE
159801 presented at the SPE Annual Technical Conference and Exhibition held
in San Antonio, Texas, USA, 8-10 October 2012
Cipolla, C.L., Warpinski, N.R., Mayerhofer, Lolon, E.P., and Vincent, M.C. 2008. The
relationship between Fracture Complexity, Reservoir Properties, and Fracture
Treatment Design. Paper SPE 115769 presented at SPE Annual Technical and
Exhibition held in Denver, Colorado, USA, 21-24 September.
Cipolla, C.L., Lolon, E.P., Mayerhofer, M.J.,Warpinski, N.R. 2009. Fracture Design
Considerations in Horizontal Wells Drilled in Unconventional Gas Reservoirs.
Paper SPE 119366 presented at SPE Hydraulic Fracturing Technology
Conference, The Woodlands, Texas, 19-21 January.
Cipolla, C.L., Lolon, E.P., Erdle, J.C., Rubin, B. 2009. Reservoir Modeling in Shale-Gas
Reservoirs. Paper SPE 125530 presented at SPE Eastern Regional Meeting,
Charleston, West Virginia, 23-25 September.
deSwaan-O, A. 1976. Analytic Solutions for Determining Naturally Fractured Reservoir
Properties by Well Testing. SPEJ 16 (3): 117-122. SPE 5346-PA.
Fisher, M.K., Heinze, J.R., Harris, C.D., Wright, C.A., and Dunn, K.P. 2004. Optimizing
Horizontal Completion Techniques in the Barnett Shale using Microseismic
Fracture Mapping. Paper SPE 90051 presented at the SPE Annual Technical
Conference and Exhibition held in Houston, Texas, USA, 26-29 September.
Holditch, S. A. (2001, March 12). The Increasing Role of Unconventional Reservoirs in
the Future of the Oil and Gas Business.
Javadpour, F. 2009. Nanopores and Apparent permeability of Gas Flow in Mudrocks
(Shales and Siltstones). Journal of Canadian Petroleum Technology, Vol. 48, No
48, pp. 16-21, August 2009.
Javadpour,F., Fisher, D.,Unsworth, M. 2007. Nanoscale Gas flow in Shale Gas
Sediments. Journal of Canadian Petroleum Technology, Vol. 46, No 10, pp. 5561, October.
Kazemi, H. 1969. Pressure Transient Analysis of Naturally Fractured Reservoirs with
Uniform Fracture Distribution. SPEJ 9 (4) 451-462. SPE 2516-A.
Ketter, A.A, Daniels, J.L, Heinze, J.R., Waters, G. 2006. A Field Study Optimizing
Completion Strategies for Fracture Initiation in Barnett Shale Horizontal Wells.
Paper SPE 103232 presented at SPE presented at SPE Annual Technical
Conference and Exhibition held in San Antonio, TX, USA, 24 27 Sept.
Ilk, D., Jenkins. C.D., Blasingame, T.A. 2011. Production Analysis in Unconventional
Reservoirs-Diagnostics, Challenges, and Methodologies. Paper SPE 144376

89

presented at SPE North American Unconventional Gas Conference and

Exhibition held in The Woodlands, Texas, 14-16 June.
Mayerhofer, M.J., Lolon, E.P., Youngblood,J.E., and Heinze, J.R. 2006. Integration of
Microseismic Fracture Mapping Results with Numerical Fracture Network
Production Modeling in the Barnett Shale. Paper SPE 102103 presented at
presented at SPE Annual Technical Conference and Exhibition held in San
Antonio, TX, USA, 24 27 Sept.
Mattar, L., Gault, B., Morad, K., Clarkson, C.R, Freeman, C.M., Ilk, D. Blasingame, T.A.
2008. Production Analysis and Forecasting of Shale Gas Reservoirs: Case
History-Based Approach. Paper SPE 119897 presented at SPE Shale Gas
Production Conference, Fort Worth, Texas, 16-18 November.
Medeiros, F. Jr., 2007. Semi-Analytical Pressure-Transient Model for Complex WellReservoir Systems. PhD Dissertation. Colorado School of Mines. Golden, CO.
Medeiros, F., Kurtoglu, B., Ozkan, E., and Kazemi, H. 2007. Analysis of Production
Data From Hydraulically Fractured Horizontal Wells in Tight, Heterogeneous
Formations. Paper SPE 110848-MS presented at presented at SPE Annual
Technical Conference and Exhibition, Anaheim, CA, 11 14 Nov.
Medeiros, F., Kurtoglu, B., Ozkan, E., and Kazemi, H. 2007. Pressure-Transient
Performances of Hydraulically Fractured Horizontal Wells in Locally and Globally
Naturally Fractured Formations. Paper IPTC 11781-MS presented at
International Petroleum Technology Conference, Dubai, U.A.E., 4 6 Dec.
Medeiros, F., Jr. Ozkan, E., and Kazemi, H. 2006. A Semianalytical, Pressure-Transient
Model for Horizontal and Multilateral Wells in Composite, Layered, and
Compartmentalized Reservoirs. Paper SPE 102384-MS presented at SPE
Annual Technical Conference and Exhibition, San Antonio, TX, 24-27 Sept.
Medeiros, F., Jr., Ozkan, E., and Kazemi, H. 2008. Productivity and Drainage Area of
Fractured Horizontal Wells in Tight Gas Reservoirs. SPEREE 11 (5): 902-911.
SPE 118110-PA.
Neuhaus, C.W., Stroud-Williams, S., Remington, C., Barker, W., Blair, K., Neshyba, G.,
McCay, T., 2012. Integrated Microseismic Monitoring for Field Optimization in
the Marcellus Shale-A Case Study. Paper SPE 161965 presented at Canadian
Unconventional Resources Conference held in Calgary, Alberta, Canada, 30
Oct.-1st Nov.
Ozkan, E., Brown, M., Raghavan, R., and Kazemi, H. 2011. Comparison of FracturedHorizontal-Well Performance in Tight Sand and Shale Reservoirs. SPEREE
Volume 14, Number 2. (April) Pages 248-259

90

Ozkan, E., Ohaeri, U., and Raghavan, R. 1987. Unsteady Flow to a Well Produced at a
Constant Pressure in a Fractured Reservoir. Paper SPE 9902 presented at the
SPE California Regional Meeting held in Bakersfield, 25-26 March.
Ozkan, E., Raghavan, R., and Aypadin, O. 2010. Modeling of Fluid Transfer from Shale
Matrix to Fracture Network. Paper SPE 134830 presented at the SPE Annual
Technical Conference and Exhibition held in Florence, Italy, 19-22 Sept.
Raghavan, R.S., Chen, C., and Agarwal, B. 1997. An Analysis of Horizontal Wells
Intercepted by Multiple Fractures. Paper SPE 27652 presented at the SPE
Journal, Volume 2, September 1997, pages 235-245.
Serra, K., Reynolds, A. C., and Raghavan, R. 1983. New Pressure Transient Analysis
Methods for Naturally Fractured Reservoirs. JPT 35 (12): 2271-2283. SPE
10780-PA.
Soliman, M. Y., Hunt, J. L., and El Rabaa, A. M. 1990. Fracturing Aspects of Horizontal
Wells. JPT 42 (8): 966-973. SPE 18542-PA.
Sondergeld, C.H., Newsham, K.E., Comisky, J.T., Rice, M.C., Rai, C.S., 2010.
Petrophysical Considerations in Evaluating and Producing Shale Gas Resources.
Paper SPE 131768 presented at SPE Unconventional Gas Conference held in
Pittsburgh, Pennsylvania, USA 23-25 Feb.
Sondhi, N., 2011. Petrophysical Characterization of Eagle Ford Shale. Masters Thesis
Dissertation. University of Oklahoma. Norman, OK.
Stehfest, H. 1970. Numerical Inversion of Laplace Transforms. Communications. ACM
13 (1). 4749.
Van Everdingen, A. F., and Hurst, W. 1949. The Application of Laplace Transformation
to Flow Problems in Reservoirs. Petroleum Transactions, AIME 186: 305-324.
SPE 949305-G.
Viswanathan, A., Altman, R., Oussoltsev, D., Kanneganti, K., Xu, J., Grant, D., Indriati,
S., Pea, A., Loayza, M., Kirkman, B., Rhein, T. 2011. Completion Evaluation of
the Eagle Ford Formation with Heterogeneous Proppant Placement. Paper SPE
149390 presented at SPE Canadian Unconventional Resources Conference held
in Calgary, Alberta, Canada 15-17 Nov.
Warren, J. E., and Root, P. J. 1963. The Behavior of Naturally Fractured Reservoirs.
SPEJ 3 (3): 245 255. SPE 426-PA.

91

APPENDIX A
The derivative of the productivity index (Eq. A.1) with respect to a well, reservoir,
or fracture property, , is given by

qt
J

L pwf s p

qt
1

L pwf s p
2

L pwf s p

J
1
J 2 L1 pwf s
p

qt

(A.1)

4.2.1 Derivative of J with respect to natural fracture permeability, k f

From Eq. A.1, we have

J
1
J 2 L1
pwf s
p
kf
qt
kf
kf

(A.2)

Using Eq. 3.2.23, we obtain

141.2qB
1
pwf s
kf
hft s
kf kf CFD F tanh

141.2qB

2
kf hft sCFD F tanh

kf CFD F tanh F

kf

(A.3)

hft s

pwf2 s
kf CFD F tanh
141.2qB
kf

consider

92


kf CFD F tanh
kf
kf F tanh

F CFD F tanh

Ck

FD
f

(A.4)

k
tanh

kf CFD tanh

kf CFD F

kf

we have

CFD
k w
kF w F
C
1
F F
FD

kf
kf
hft kf kf
h
kf2 ft xF
h xF

h

F
kf

(A.5)

F
1
2F sxF2

2 F kf
2 F kf CFD F
1

2 F 2F CFD
2

2 F CFD kf CFD kf

2 F
2 F

2 F CFD kf kf CFD

(A.6)

where

xF O tanh O y e
kf
2

F
w

xF tanh O y e
O O
tanh O y e
kf
2 kf
kf
2

kf

93

(A.7)


kf

O
1
1 O

u

2 O kf
2 O kf f CRD y e

1

1 O

1
u f
2
2 O f CRD y e
kf f kf

2f O f

kf kf

O

u f

CRD y e
kf kf

O u

CRD y e kf

Ckf f

f
tf

kf

O
1

CRD y e CRD y e

1 O

2O
CRD y e kf CRD y e

kf

O u

CRD y e kf

(A.8)

(A.9)

O O 1

kf
y e kf CRD
CRD
kf

(A.10)

O 1 O 1 CRD

0
CRD y e O kf CRD kf

because
1 O

O kf

1
O x F

1
O x F

O kf

xF

tanh
xe xF
O
O

s

s

tanh
xe xF f
O
O
kf
s

s
1
tanh
xe xF f
O
O
kf kf
s

(A.11)

and

94


k
1 CRD
1 f

CRD kf CRD kf

hft
h xF 1

kO y e kf

(A.12)

also

f s
u

s
s
kf
kf
kf

3s

tanh
1

3s

3s
s tanh
k

3s

tanh
0

3s
3s kf

(A.13)

because

kf

3s

3s 2 kf
1

3s kf

3s
kf
3s 2 kf

f km hm
12 2
0
h
k
h

f
f

(A.14)

and

3s

tanh
0

kf

(A.15)

thus

kf

2kff O CRD y e

(A.16)

consider

95

w
w
w

tanh O y e 1 tanh2 O y e
O y e

kf
2
2 kf
2

w
w

1 tanh2 O y e y e
O
2
2 kf

ye 2

O

u 1 tanh2 O y e

2
2kff O CRD y e

(A.17)

thus
F
xF

kf 2kff O

u tanh O y e

CRD y e

xF y e

w
2 O

u 1 tanh2 O y e

2kff
2

CRD y e

xF O

tanh O y e
2kf
2

xF O y e
w
2

1 tanh O y e
2kf
2

(A.18)

w
w

xF O y e F2 y e

2
2

2kf
2kf
2 xF k f

rearranging

F
1

kf
2xF kf

w
w 2

2
xF F xF y e 2 O y e 2 F

(A.19)

and thus
F
kf

1
2kf F CFD

1
F
xF

w 2
w

y e 2 F xF y e 2 O

Also, we have

96

(A.20)


tanh
kf

F 1 tanh2

F
F

(A.21)

kf

Thus, we have

kf CFD F tanh
kf
F 1 tanh2

1
2 F

F kf CFD tanh

F
F

F tanh

kf

F F 1 tanh2

(A.22)

1
w 2
w

F
y

x
y

O
e
F
F e
xF
2
2

finally,
2
hft s

pwf s
pwf s
kf
141.2qB

1
2 F

1
F
xF

tanh

F F 1 tanh2

(A.23)

w 2
w

y e 2 F xF y e 2 O

and because

p 0
kf

(A.24)

we obtain

97

nF hft
J
2 1 s

J
L

p
s

wf
kf 141.2qt2B 2
F

F tanh

1
F
xF

F F F tanh2

(A.25)

w 2
w

y e 2 F xF y e 2 O

4.2.2 Derivative of J with respect to matrix permeability, km

From Eq. 3.1, we have

J
1

( ) J 2 L1
pwf s
p
km
qt

km
km

(A.26)

From Eq. 3.2.23, we obtain

141.2qB
1
pwf s
km
kf hft sCFD km F tanh F

2
k h sC

f ft FD pwf s
F tanh F

141.2qB
km

(A.27)

consider

F tanh
km

tanh

F tanh

F 1 tanh2

tanh
km

F
km

(A.28)

we have
F
km

F
F
1
2F sxF2
1

2 F km 2 F km CFD F
F CFD km
1

where

98

(A.29)


F

w

xF O tanh O y e
k m k m
2

w O

xF tanh O y e
xF O
tanh O y e
2 k m
k m
2

w O

xF tanh O y e

(A.30)

w
w O

xF O 1 tanh2 O y e y e
2
2 k m

F
F2
xF O
O
xF O

w O
ye

2 k m

In Eq. A.30

O
k m

1
2f
1
2f

O
1

2 O k m 2f O k m
1

CRD y e

1 O u

O y e k m CRD k m
1

O y e

1 O

CRD k m

1
O

2CRDf O y e O
1
u

2f O k m

k m

O CRD u

2

C

RD
m
m

1 CRD

CRD k m

99

(A.31)

1 O

O km

O km

xF

tanh
xe xF
O
O

s

f
s

s
xe xF
tanh

F
O
m
O
1

s
s

f
x k tanh xe xF
O

O
m
F
f

s
xe xF 3 2 O
tanh

O
m
O

1 F

s
s
s

2
O
f

tanh
x

x
x

e
F
e
F
32
x

O

O
2O k m

F

x
1 O sf
2 xe xF xe xF F O2 O

2O O O
fO km
sf
xF 2 O
1 O

e
F
e
F
2O O O2 xF
fO O km

O
sf
xF 2

O
O
e
F
2k m O
k m
O xF f

Similarly, by expanding the partial derivative term with respect to matrix

permeability, equation (A.32) becomes the following:

100

(A.32)

 3s
u f s

s
s
tanh
1

k m k m
k m
3s

3s
s tanh
k

tanh

s

3s

3s

3s
tanh

3s k m

3s

3s k m

3s

2
1 tanh
3s

k m

3s 1
tanh

3s 2 k m
s

s
1

2
3s 1 tanh 3s 2 3 2 k

3s 1
tanh

3s 2 k m
s

3s 3s
1

2
3s 1 tanh 2 3 2 k

3s 1
tanh

3s 2
s

s
3

s
1

k m

tanh

3s

3s
tanh

s 3s

2k m 3s
3s
3s
2
3s 3s tanh

u f s
s
s

k m k m
2k m

2
3s

f s 1
f s 1

thus

101

(A.33)

 O
km

3s 2

f s 1
s
f
s

sf

2O
xF
1
4f km O

x xF

CRD y e O O xF f O e
CRD km y e

(A.34)

And

F F
F2

xF O
k m O
xF O

w
y e

3s 2

f s 1
s
f
s

sf
2 O
xF 2
1
4f km O

F
CRD y e O O xF F O e
CRD km y e

(A.35)

and from (A.29)

F
k m

F CFD

F2
F xF O
xF O
O

w
y e

3s 2
f s 1
s f s 1 s

sf
2 O
xF 2
1
4f km O

x xF

CRD y e O O xF F O e
CRD km y e

Then

102

(A.36)

F tanh
km

tanh 1 tanh2

F
F
F

2
F

1
w

xF F y e

xF O
2
4f k m F CFD O

(A.37)
2

3s 2

f s 1
s f s 1 s

1 sf xF 2 x x 2O
F
CRD y e O O xF F O e

CRD km y e

finally,
2
kf hft s

pwf s
pwf s
km
141.2qB

tanh

F F 1 tanh2 F

1
4f km

F
F2
xF

xF O
F O

w
y e
2

(A.38)

3s 2
f s 1
s f s 1 s

sf
2 O
xF 2
1

O
O
e
F
CRD y e
CRD k m y e
O xF F

because

p 0
km

(A.39)

we obtain

103

nF kf hft
2
J

J 2L1 s p pwf s
2
km 141.2qt B

tanh

F 1 tanh2

2
w
F xF F y e
xF O
2
4nf km F O

2
3s 2
f s 1
s f s 1 s

snf

2O
x
F O2 xe xF
O

CRD y e
CRD km y e
O xF F

(A.40)
4.2.3 Derivative of J with respect to natural fracture density, f
From Eq. A.1, we have

J
1

J 2 L1
pwf s
p
f
qt

f
f

1
1
J 2 L1
pwf s hft p

qt
f hft

p hft
1 2 1
J L
pwf s

qt

f
hft f

p
J
1
J 2 L1
pwf s

f
qt

f
f

(A.41)

where we have used

hft
h
f hhf hhf ft
f f
f

(A.42)

From Eq. 3.2.23, we obtain

104

141.2qB
1
pwf s
f
kf s
f hftCFD F tanh

141.2qB
1

h C
kf s
tanh
ft FD F

hftCFD F tanh F

(A.43)

2
kf s

pwf s
hftCFD F tanh
141.2qB
kf

Consider

hftCFD F tanh
f

tanh

hft F tanh

F
F

F CFD F tanh

FD

ft
f

(A.44)

hftCFD
F 1 tanh2

where

kF w

k f xF f
CFD
1

f hft
hft

CFD

f
f

hkF w

kf hft xF

hftCFD

hft
C
FD
f
f

(A.45)

And

105

F
1
2F sxF2

2 F f
2 F f CFD F
1

1 1 F F CFD
2

F CFD f CFD f

F
CFD

(A.46)

1 F
1

F F f f

In Eq. A.38

f f

xF O tanh O y e
2

w O

xF tanh O y e
2 f

xF

w
w O

O 1 tanh2 O y e y e
2
2 f

F2
F xF O
xF O
O

(A.47)

y e

where

O
f

1 O u

O y e f CRD f

O y e

2f
2f

O
f

O
1
O

2 O f
2f O f CRD y e

1 O

CRD f

O CRD
2
CRD f

CRD
1
u

2
2CRD
f O y e f
2f O f
1

In Eq. A.48,

106

(A.48)

CRD
f

h
kf ft xF
k f xF

h hft CRD hft CRD

f km y e f hft km y e f
hft f
f

f s
u

s
s
f
f
f

3s
s tanh

(A.49)

3s

tanh
1

3s

3s

3s

tanh

3s f

3s 1 3s 1 1
s

1 tanh2

3
s

f
f

3s 1 3s
1 1 hm
s

h
1 tanh2

m

f hm hm f
3s
2

3s 2
s
1 tanh2

3s
hm

3s hmt

3s 2
s
1 tanh2

3s
hmt

3s
h

ft
f

3s 2
s
1 tanh2

3s
hmt

3s hft

hs
u
2 ft
f
hmt f

(A.50)

3s
2
1 tanh

thus

O
f

3s
h
O

ft s 1 tanh2

f O f 2CRD y e hmt

107

(A.51)

3s
hft
F
1
2
O

s
1

tanh

f
f O f 2CRD y e hmt

F2
w
F

xF O
y e
2
xF O
O

F
f

(A.52)

F
1

CFD F f CFDf O F f

3s
hft
2
O

s
1

tanh

2CRD y e hmt

F2
w
F

xF O
y e

O
F
O

(A.53)

Therefore

hftCFD F tanh
f

tanh

1
CFD F f

F F 1 tanh2 F

1
F
f O

(A.54)

O
h

3s

ft s 1 tanh2

2CRD y e hmt

F2
w
F

xF O
y e
2
xF O
O

108

2
kf s

1
pwf s
pwf s

f
141.2qB
CFD F f

tanh

1
F
f O

F 1 tanh2

3s
hft
O
2

s 1 tanh

2
C
y
h

RD
e
mt

F2
F

xF O

xF O
O

w
ye

s p 2 s
nF kf
J
wf

J 2L1
2
f 141.2qt B
CFD F f

tanh

F F 1 tanh2 F

1
F
f O

F xF
O

(A.55)

3s
hft
O
2

s
1

tanh

2
C
y
h

RD
e
mt

w
O
ye
2
xF O

2
F

1 2 p
J
qt
f

4.2.4 Derivative of J with respect to horizontal well length, Lh

From Eq. A.1, we have

109

(A.56)



J
1
= - J2 L-1
p wf s p
Lh qt Lh
Lh
=-

1 2 -1
1
J L
p wf s - Ap

qt Lh

L
h A

=-

1
1 2 -1
A
J L
p wf s + p

qt Lh
A Lh

=-

1
1 2 -1

2x e Lh + 2y e
J L
p wf s + p
qt Lh

A Lh

=-

1 2 -1
p
J L
p wf s +

qt Lh
Lh + 2y e

141.2qB 2 -1

=
JL
k f hft qtLh
sCFD F tanh F
1
p
- J2
qt Lh + 2y e

1
1
p
J2L-1 p wf s - J2
qt Lh + 2y e
qt Lh + 2y e

In Eq. A.57, we have used Eq. 3.2.23 to obtain

110

(A.57)

141.2B
1
pwf s
q
Lh
kf hft sCFD Lh F tanh F

1
q

tanh Lh
F
F

141.2B
1

q
kf hft sCFD F tanh F

qt

tanh
F

2y e
F Lh Lh 2y e

F tanh
Lh

qt
141.2B
1
2y

e
kf hft sCFD F tanh F
Lh Lh 2y e

qt
2y e
141.2B

kf hft sCFD F tanh F L 2y 2

e

141.2B
q

kf hft sCFD Lh 2y e F tanh F

(A.58)

where

F tanh
Lh

F tanh

F F 1 tanh2

F
F
0
Lh

(A.59)

as a result of the following relations:

F
Lh

F
F
1
2F sxF2
1

2 F Lh 2 F Lh CFD F CFD F Lh
1

111

(A.60)

xF O tanh O y e
2

Lh Lh

w O

xF tanh O y e
2 Lh

xF

w
w O

O 1 tanh2 O y e y e
2
2 Lh

F2
F

xF O
xF O
O
O
Lh

1
2f
1
2f

O
Lh

(A.61)

w O
ye

2 Lh

O
1
O

2 O Lh 2f O Lh CRD y e

1 O u

O y e Lh CRD Lh
1

O y e

1 O

CRD Lh
1

2C

2
RD f

O CRD
2
CRD Lh

Lh

(A.62)

CRD
1
u

O y e Lh 2f O Lh

CRD
0
Lh

(A.63)

u
0
Lh

(A.64)

Thus

O
Lh

(A.65)

F
0
Lh

(A.66)

112

 F
Lh

(A.67)

We have also imposed the condition that the distance between hydraulic
fractures and the total flow rate qt from the well stays the same as the well length
changes. Therefore, the production rate for each fracture is given by q= qt/nF where nF
is the number of hydraulic fractures, which depends on the well length by nF
=[Lh/(2ye)]+1.
4.2.5. Derivative of J with respect to distance between hydraulic fractures,

d F 2 ye
From Eq. A.1, we have

J
1 J
1

J 2 L1
pwf s
p
dF 2 y e
qt
y e

y e

1 2 1
1

J L
pwf s A p

2qt
y e A
y e

p
1 2 1
2xe Lh 2y e

J L
pwf s
2qt
y e
A y e

1 2 1
2 p

J L
pwf s

2qt
y e
Lh 2y e

113

(A.68)

From Eq. 3.2.23, we obtain

141.2B
1
pwf s
q
y e
kf hft sCFD y e F tanh F

1
q

tanh y e
F
F

141.2B
1

q
kf hft sCFD F tanh F

F tanh
y e

qt
2y e

tanh y e Lh 2y e
F
F

kf hft sCFD
2

pwf s
F tanh
141.2qB
y e

141.2B
q

kf hft sCFD y e F tanh F

k h sC
2

f ft FD pwf s
F tanh
141.2qB
y e

141.2qB

kf hft y e sCFD F tanh F

k h sC
2

f ft FD pwf s
F tanh
141.2qB
y e

(A.69)

ye

y e

1
F
pwf s

consider

F tanh
y e

F tanh

F F 1 tanh2

F
F

(A.70)

where
F
y e

F
F
1
2F sxF2
1

2 F y e 2 F y e CFD F CFD F y e
1

In Eq. 3.38

114

(A.71)


F

w

xF O tanh O y e
y e y e
2

w O

xF tanh O y e
xF O
tanh O y e
2 y e
y e
2

w O

xF tanh O y e
2 y e

w
w O

xF O 1 tanh2 O y e y e
O
2
2 y e

F2
F2
w O
F

xF O
xF O
y e
2 y e
xF
xF O
O

(A.72)

where

O
y e

O
1
O

2 O y e 2f O y e CRD y e

O CRD
O
u

2
2
2f O CRD y e y e CRD y e y e
1

(A.73)

Because

CRD
C
k f xF k f x F

RD

2
y e
y e km y e km y e
ye

(A.74)

and

3s
f s
u

s
s
tanh
0
1

y e
y e
y e
3s

Thus

115

(A.75)

F
2
xF O F
y e
xF
F
y e

(A.76)

F
1
2F sxF2

2 F y e 2 F y e CFD F
1

1
CFD

F2
1
xF2O F2
xF O

xF xF CFD F
F

(A.77)

and

F tanh
y e

tanh

1
xF CFD F

F 1 tanh

x
2
F

F2
(A.78)

therefore

k f hft
2

p wf s = sp wf s
ye
141.2qB

xF2 O - F2

1
xF

tanh

(A.79)

F 1- tanh2

1
F + p wf s
ye

and thus

nF kf hft
2
J

J 2L1 s pwf s
dF 141.2qt2B
1

xF2O F2 tanh F F 1 tanh2

2 xF F

1
1
p
J 2L1 pwf s J 2
2qt y e
qt
Lh 2y e

116

(A.80)

4.2.6. Derivative of J with respect to hydraulic fracture permeability, k F

From Eq. A.1, we have

J
1

1 J 2 L1
pwf s
p
kF
qt

kF
kF

1
1 J 2L1
pwf s
qt
kF

(A.81)

From Eq. 3.2.23, we obtain

141.2qB
1
pwf s
kF
kf hft s
kF CFD F tanh

141.2qB
1

kf hft s
kF CFD F tanh

CFD F tanh F

kF

(A.82)

kf hft s

2
pwf
s CFD F tanh
141.2qB
kF

consider

CFD F tanh
kF

tanh

tanh

F F tanh

F 1 tanh2

C
FD F tanh
kF

Ck

F
F
kF

F 1 tanh2

FD

F
F
kF

where

117

(A.83)

kF

F
1
2F sxF2

2 F kF 2 F kF CFD F
1

2 F 2F CFD sxF2 F
2
2

F kF
2 F CFD kF CFD kF
1

(A.84)

2 F
2 F
sxF2

2 F CFD kF kFCFD kFF

1

kF

In Eq. A.84

F

w

xF O y e 0
kF kF
2

(A.85)

Thus
F

kF

1
2kF

2F sxF2

F CFD F

(A.86)

And

CFD F tanh
kF

1
2kF

C
F FD F tanh
kF

2F sxF2

F CFD F

tanh

F 1 tanh2

(A.87)

Therefore

118

kf hft s

pwf s
kF
141.2qB

CFD
F tanh F
k
F

2
2F sxF2
1
pwf s

C
F
2
k

FD

F
F

2
tanh F F 1 tanh

(A.89)

and thus

nF kf hft
2
J

J 2L1 s pwf s
kF 141.2qt2B
C
FD F tanh
kF

tanh

2
sx 2
F F
2kF F CFD F
1

F 1 tanh2

(A.90)

4.2.7. Derivative of J with respect to hydraulic fracture half-length, xF xe .

From Eq. A.1, we have

J
1

J 2 L1
pwf s
p
xF
qt
xF

xF

1
1
J 2 L1
pwf s Ap

qt
xF A
xF

1
1
A
J 2 L1
pwf s p

qt
xF
A
xF

1
qt

1
qt

J 2 L1
pwf s
xF

J 2 L1
pwf s

xF

2xe Lh 2y e
p
A
xF

p xe

xe xF

119

(A.91)

where xe xF 1 if the drainage area is assumed to be the stimulated reservoir

volume and zero otherwise. From Eq. 3.2.23, we obtain

141.2qB
1
pwf s
xF
kf hft s
xF CFD F tanh

141.2qB
1

kf hft s
xF CFD F tanh

CFD F tanh F

xF

(A.92)

kf hft s

2
pwf
s CFD F tanh
141.2qB
xF

consider

CFD F tanh
xF

CFD
F F tanh F

xF

CFD tanh

F
xF

CFD F 1 tanh2

F
F
xF

1
F CFD tanh
xF

CFD tanh

where

120

(A.93)

F 1 tanh2

F
F
xF

F
1
2F sxF2

xF
2 F xF 2 F xF CFD F
F

CFD
sxF
1

2 F

CFD F xF CFD F xF F F
F
xF

1
CFD

(A.94)

F F
sxF

F xF xF F F

In Eq. A.94

xF xF

xF O tanh O y e
2

w
w O

O tanh O y e xF tanh O y e
2
2 xF

xF O

tanh O y e
xF
2

w
w O

O tanh O y e xF tanh O y e
2
2 xF

w
w O

xF O 1 tanh2 O y e y e
2
2 xF

O tanh O y e
2

2
F2
F

xF O
xF O
O

w O
y e

2 xF

where

121

(A.95)

 O
xF

O
1
O

2 O xF 2f O xF CRD y e

O CRD
1 O u

O CRD y e xF CRD y e xF xF

O
1 O u

O CRD xF y e CRD y e xF xF

2f
2f

(A.96)

In Eq. A.96,
O
f

xF xF xF
f
f

1
xF

O
s

tanh
xe xF
O

O
s

1
tanh
xe xF

xF xF
O

1
s

xe xF
2 tanh
xF
O

s
2 s
1 tanh
xe xF
O
O

x
x
O f
F O2 1 e
xF
xF xF O f

xe
1
xF

(A.97)

and
f s
u
s
0
xF
xF

(A.98)

thus

O
xF

s x2

x
2O f F O2 1 e
O f

xF
2f CRD O xF y e

(A.99)

122


s x2

x
2O f F O2 1 e

xF

O f

F
F2
w
ye
xF O
2
O
xF O

F F
1

xF xF 2f CRD O xF y e

(A.101)

F
2F
sxF

xF
xF CFD F F F

s x2

x
2O f F O2 1 e
O f

xF
2f CRDCFD F O xF y e

F
F2
w
ye
xF O
2
O
xF O

(A.102)

and

CFD F tanh
xF

1
F
F CFD tanh

xF

CFD tanh

F 1 tanh2

2F
sxF

xF CFD F F F
1

2f CRDCFD F O xF y e

(A.103)

s x2 2
x
2O f F O 1 e
O f
xF

F2
F

xF O
xF O
O
therefore

123

w
ye

kf hft s
2

pwf s
pwf s
xF
141.2qB
1

xF

F CFD tanh

CFD tanh

F 1 tanh2

2 F
sxF

xF CFD F F F
1

2f CRDCFD F O xF y e

s x2 2
x
2O f F O 1 e
O f

xF

F2
F xF O
xF O
O

w
ye

(A.104)

and thus

1
nF kf hft
2
J

J 2L1 s pwf s
2
xF 141.2qt B
xF

CFD tanh

F F 1 tanh2

F CFD tanh

2F
sxF

xF CFD F F F
1

2f CRDCFD F O xF y e

s x2 2
x

2O f F O 1 e
O f

xF

F2
F

xF O
xF O
O
1
p xe
J2
qt
xe xF

ye

(A.105)

124