Professional Documents
Culture Documents
(Sustainable Energy Engineering) Fred Aminzadeh - Hydraulic Fracturing and Well Stimulation (Sustainable Energy Engineering) - Wiley-Scrivener (2019)
(Sustainable Energy Engineering) Fred Aminzadeh - Hydraulic Fracturing and Well Stimulation (Sustainable Energy Engineering) - Wiley-Scrivener (2019)
and
Well Stimulation
Scrivener Publishing
100 Cummings Center, Suite 541J
Beverly, MA 01915-6106
Publishers at Scrivener
Martin Scrivener (martin@scrivenerpublishing.com)
Phillip Carmical (pcarmical@scrivenerpublishing.com)
Book Series
Sustainable Energy Engineering
Hydraulic Fracturing and
Well Stimulation
Edited by
Fred Aminzadeh
This edition first published 2019 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA
© 2019 Scrivener Publishing LLC
For more information about Scrivener publications please visit www.scrivenerpublishing.com.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or other-
wise, except as permitted by law. Advice on how to obtain permission to reuse material from this title
is available at http://www.wiley.com/go/permissions.
For details of our global editorial offices, customer services, and more information about Wiley prod-
ucts visit us at www.wiley.com.
ISBN 978-1-119-55569-8
Set in size of 11pt and Minion Pro by Manila Typesetting Company, Makati, Philippines
10 9 8 7 6 5 4 3 2 1
Contents
Foreword xiii
Part 1: Introduction 1
1 Hydraulic Fracturing, An Overview 3
Fred Aminzadeh
1.1 What is Hydraulic Fracturing? 4
1.2 Why Hydraulic Fracturing is Important 5
1.3 Fracture Characterization 8
1.4 Geomechanics of Hydraulic Fracturing 11
1.5 Environmental Aspects of Hydraulic Fracturing 14
1.6 Induced Seismicity 18
1.7 Case Study: Fracturing Induced Seismicity in California 23
1.8 Assessment of Global Oil and Gas Resources Amenable
for Extraction via Hydraulic Fracturing 27
1.9 Economics of HF 27
1.10 Conclusions 28
Acknowledgement 30
References 30
v
vi Contents
Since its inception in 1947, hydraulic fracturing has had a greater influ-
ence on hydrocarbon production than any other technology. Drilling and
fracturing of horizontal wells starting in the late 1980’s enabled further
increases in production. The oil industry took the new technology to a
new level, creating the shale revolution, making the USA the world’s largest
producer of crude oil.
The advancement of technology also brought new challenges in tech-
nological, environmental, human, and geopolitical dimensions. Each chal-
lenge will have to be understood and addressed. This book collects papers
that address the environmental and human dimensions in simple terms in
order to clarify issues and dispel misconceptions.
Exploring the technological dimension, the book includes chapters that
address many of the issues that are very unique to unconventional reser-
voirs, including chapters on Reservoir-Fluid interaction, optimization of
fracture spacing, and optimization of fracture placement.
xiii
Part 1
INTRODUCTION
Abstract
This article provides an overview of the state of the art in hydraulic fracturing, a
controversial topic of the last decade. To Frack or not to Frack, That is the Question;
was posed at a meeting of the Western Regional Society of Petroleum Engineers.
The fact is, we have witnessed an intense debate over hydraulic fracturing’s eco-
nomic benefits and the role it has played in securing energy independence, and
its ill effects (perceived or real) during the past decade. Many, in particular those
in the fossil energy industry, consider shale oil/gas, with the associated horizontal
drilling and hydraulic fracturing, as one of the major developments in the oil and
gas industry of the past two decades. Others, especially many environmentalists,
consider fracking proponents as public enemy number one.
Different sections of this entry attempt to highlight different scientific facts about
hydraulic fracturing, the common-sense environmental concerns, and the respective
economic ramifications. After a brief overview of the principles of hydraulic fractur-
ing in section “What Is Hydraulic Fracturing?”, we discuss the importance of hydrau-
lic fracturing in section “Why Hydraulic Fracturing Is Important.” This is followed
by fracture characterization (section “Fracture Characterization”) and geomechanics
(section “Geomechanics of Hydraulic Fracturing”). They examine the natural frac-
tures in the subsurface and how one can characterize them, how hydraulic fracturing
helps to expand the natural fractures and/or create new (stimulated) fractures as well
as the underlining rock mechanics properties and the related stress regime.
Section “Environmental Aspects of Hydraulic Fracturing” addresses different
environmental concerns about hydraulic fracturing. This includes the potential
ground water contamination, amount of water used for hydraulic fracturing the
water disposal process, and methane emission concerns. Another environmental
concern is the risk of induced seismicity or man-made earthquakes. Given the
Email: faminzad@usc.edu
*Adopted from Aminzadeh, F., 2018, Hydraulic Fracturing, An Overview, Journal of Sustainable
Energy Engineering, Vol. 6, Issue 4, pp. 204–228.
3
4 Hydraulic Fracturing and Well Stimulation
Private well
USDW
Municipal water well:
<1000 ft.
Shale fractures
Additonal steel
casings and cement
to protect
groundwater
Protective steel casing
at peak in March 2015 there were 1900 active oil rigs operating in the var-
ious US basins [7]. Following the price collapse, the rig count dropped
to about 400 in June 2016 and then began climbing again in response to
higher oil prices. As of early 2018 the number of oil rigs stood at about
1000. The relative speed with which these operations can be turned on
and off has a profound influence on the ability of oil producers to raise oil
prices by limiting their own production.
The increased production of shale oil in the USA has made it easier
for the world oil market to withstand several incidents when production
and export from certain countries was halted from war or other political
upheavals [8].
Producing shale oil can cost upwards of $60/ barrel. Prior to 2006 oil pric-
ing hovered around $40/barrel, making HF cost prohibitive. Subsequently,
largely in response to increasing global demand, the oil price rose to above
$100/barrel (with some spikes as high as $140/barrel). At these higher
prices, shale oil production by HF was profitable and it spurred drilling
activities in many US basins. The profitability can vary significantly from
one play to the next. Figure 1.2 shows an earlier peak for shale production
for different US plays that arrived in 2015, after which it dipped through
2017, with the rate of decline being an indication of the price sensitivities
in different plays. Since 2017, production increased in all plays.
Modern HF combined with horizontal drilling allows multiple wells to
be drilled from one surface location, reducing the size of the drilling area
to burning natural gas. In 2005, electricity generation from coal and gas
in the US amounted to 2,000 TWh and 760 TWh, respectively; by 2017
the two sources were each generating about 1,200 TWh, while the total
electricity generation has held steady at 4,000 TWh. Thus, over 700 TWh
of coal generation has switched to gas with the attendant environmental
benefit of reduced carbon dioxide (CO2) emissions.
The positive economic impact of shale-related natural gas and oil
resources is undeniable. By helping to lower power and materials costs,
as well as stimulating economic activity and positive impact on the labor
market for a variety of businesses like service and supply companies, HF
has supported growth across an economy that otherwise has struggled in
recent years. According to an IMF Report, the shale gas revolution has had
a macroeconomic impact between 0.3% and 1% of US GDP [10]. Many
performance factors are used to assess the effectiveness and safety of HF.
Among them are better understanding of fracture azimuth/geometry/
length/distribution, Stimulated Reservoir Volume, and potential hazards
(induced seismicity and HF fluid leakage). We will revisit the environmen-
tal implications of HF in a different section.
To summarize, one of the key factors impacting shale production is the
economics and oil and gas prices, as discussed earlier. However, there are
other factors that have impacted the economics of shale resources, includ-
ing the evolution of the technologies, including subsurface characterization
and advances in HF, ability to monitor creation of new fractures, and asso-
ciated microseismic events. The next two sections provide some insight on
the science behind HF and how the advances in the related science and
technology can further improve the economics of production from shale
and other tight formations.
σv
σh σh
σv
Normal fault
Figure 1.3 Generation of fractures due to application of horizontal and vertical stress and
fluid injection during HF.
C B'
A
D'
400 C'
B A'
800
Depth (m)
1200
1600
2000
100
50 D 80
60
FZIn = F{фu,Zu,Vpu, Vsu, Pu, VEu,} 100 40
20
150
Figure 1.4 Fracture mapping using MEQ, seismic, & petrophysical data using ANN-
based hybrid FZI attribute mapping, from Maity and Aminzadeh [15].
Hydraulic Fracturing, an Overview 11
Figure 1.5 The process of generating a 3D mesh of the domain for two-way coupled flow
geomechanics [20].
Hydraulic Fracturing, an Overview 13
1
0.5 1
Aquifer 0.4
0.9
Impervious Layer 0.3 0.5 0.8 99th %ile= 9.57
0.7 95th %ile= 6.03
0.2 90th %ile= 5.12
0.1 0 2 0 0.6 50th %ile= 0.82
10 10
Time |y| 0.5
F
0.5 1 HI
0.4
Shale Reservoir 0.3 0.5
0.2
0.1 0 2 0
10 10
Time |y|
Figure 1.6 Numerical modeling to quantify risk factors for water contamination
associated with HF, Jabbari et al. [26].
Hydraulic Fracturing, an Overview 17
36.5°N
1cm 1cm
36.3°N
(a)
36.1°N (c)
1cm
Parkfield
35.9°N
35.7°N
(b) (d) (e)
Figure 1.7 A comparison of induced seismicity in a field and laboratory setting with the
associated faulting.
and ISC submitted its recent study on “HF and Induced Seismicity in
California-A case study” [34].
From these, an event may be scored with more “yes” answers indicating
an increased likelihood of an event having been induced.
California. This study evaluated 30 years of seismic data in areas of oil pro-
duction where HF has been used in California and found little or no cor-
relation between oil field activities in general and seismic activity.
The ISC study compared seismic activity from 1980 to 2013 with oil
field activity, including HF, at locations throughout California. Oil field
activities and well locations were obtained from the California Division
of Oil, Gas, and Geothermal Resources (DOGGR) and FracFocus [41].
Seismic activity data was compiled from the Northern and Southern
California Earthquake Centers and included epicenter locations of seismic
activity and the magnitudes and depths of seismic events. Database maps
were prepared to display oil and gas well locations, well type (e.g. oil or
gas producer, water flood injector, steam flood injector, water disposal),
type of activity (drilling, producing, injecting, completing, stimulating, or
abandoning), and the depth of the activity. For more details, see Section 8
where a couple of case histories are highlighted.
Composite maps were prepared, comparing locations of seismic events
and known geologic faults with oil and gas well locations and oil field
activities. Statistical analyses were conducted to determine whether there
was any correlation between seismic events and oil field activities. Little or
no correlation was found. For example, in northern California, a total of
303,609 seismic events were recorded from 1980–2013. The main area of
northern California where oil and gas activities occurred is the Sacramento
area, far from known fault zones, where oil is produced from shallow depths
(less than 6 kilometers). Of the total 303,609 seismic events in northern
California, only 210 events were in the Sacramento area and only 3 seismic
events had magnitudes greater than 3; none were greater than 4.
f. Does induced seismicity from wastewater injection by the oil industry differ
from California in other states?
The practice of deep wastewater injection is commonly used in states
throughout the U.S. and is strictly regulated by state and federal laws.
California has strict regulations governing subsurface wastewater injec-
tion. Injection wells used by the oil industry in California are different
from the injection disposal wells linked to earthquakes in other states like
Ohio, Oklahoma, Texas, and Arkansas. In California oil fields, wastewa-
ter is reinjected back into the formation after the oil is removed. In other
areas of the U.S., wastewater is often injected into rocks with little poros-
ity and storage capacity. Pore pressures in the rock increases dramatically
as the limited available pore space is filled and additional injected water
causes the rock to break, causing an induced seismic event. As a result,
a University of California Santa Barbara geophysicist, Craig Nicholson,
22 Hydraulic Fracturing and Well Stimulation
concluded that “very little of the state’s [California] earthquake activity can
be tied in any way to reinjection…there’s not a connection like there is in
the central and eastern United States.”
the US Seismic Hazard Map (Patterson et al., 2014) and Induced Seismicity
Map (ISM), [43] derived from the publicly available data in California
[44], understanding the potential seismic sensitivity of the region to
hydraulic stimulation takes on greater importance.
The prolific nature of seismic activity in California, owing to an extensive
network of faulting, hinders the task of differentiating potential induced
seismic events from natural events. The task of locating potential induced
events benefits from extensive data, however these data are not organized
in a fashion conducive manner.
The case history we are highlighting here is from the San Joaquin Valley,
California. The seismic catalogue used in this study was provided by the
Southern California Seismic Network (SCSN). The catalog gives the loca-
tion, date, depth, magnitude, and parameter uncertainties for over 470,000
events since 1932. The magnitude used in this study was local magnitude
(ML), with a magnitude of completeness (MC) demonstrated by Hutton and
Jones [45] to have significantly improved with installation of more seis-
mic stations. In Aminzadeh [1], there is a review of the induced seismicity
study conducted in San Joaquin Valley, California. A detailed statistical
assessment of the induced seismicity potential within this region can be
found in Goebel et al. [46]. Figure 1.8 is an example showing the distribu-
tion of different types of seismicity in the San Joaquin Valley. A first spatio-
temporal test was conducted using records of hydraulic fracture operations
conducted in the Bakersfield area from 2000 to 2013. The maximum true
vertical depth for the fractured well dataset is just over 12,900 feet (3.9 km)
and, thus, hypocenters for potential induced seismic events are expected to
fall within a reasonable vicinity of the True Vertical Depth (TVD). Based on
a simulation of pressure transfer using COMSOL and taking into account
error in depth calculation of hypocenter, we considered events less than
10 km depth for our induced seismicity study. This is a more lenient depth
constrain compared to many previous studies of induced seismicity, for
which the limit was chosen around a five-kilometer depth. The left side of
Figure 1.8 displays locations of HF operations and earthquake epicenters
from 2000 to 2013. The right map in Figure 1.8 displays HF operations and
only those earthquakes with a hypocentral depth less than ten kilometers.
Following the filter treatment, no correlation remains.
The number of HF and other SFIP operations were examined from 2000
to 2013. HF operations took place in five oilfields in the region. However,
no spatial-temporal relationship was found between seismic events and
fracturing activities. Some seismic events have been observed in some
places between fracturing and induced seismicity, but it would be unrealis-
tic to assume such correlations as a universal phenomenon.
Hydraulic Fracturing, an Overview 25
Belridge South
Elk Hills
Figure 1.8 All seismic events in study area, San Joaquin Valley, California (left), filtering of
seismic events based on an event depth less than 10 km in San Joaquin Valley (right), Jabbari
et al. [27].
North
Rose Shafter
Round
Mountain
Lost Hills
Belridge
Elk Hills
Buena Vista
0 5 10 15
No Frac Wells
Figure 1.9 Spatial variations in B-values in the study region. (Left), color coded number of
HF in a few sizable oil and gas fields in San Joaquin Valley, (Right) [29].
with a number of active oil and gas fields. Figure 1.9 (bottom) shows a few
large oil fields with the color coded number of HF jobs.
The probability of activating faults due to anthropogenic influences is
likely highest in regions where active faults, fluid injection wells, and low
b-values have been encountered. This is the case, e.g., within the southern
part of the study region. In the proximity of active faults, both different
types of seismicity may occur and more detailed studies are required to
differentiate induced from natural (tectonic) seismicity. Areas of relatively
high b-values are likely connected to smaller ambient stresses and seismic
events are less likely to grow to large sizes.
See additional references on: What can microseismic tell us about
hydraulic fracturing [47], Geomechanical approach for microseismic frac-
ture mapping [48], Finite element method based modeling of hydraulic
fracturing [49], Flowback of Fracturing Fluids with upgraded visualization
of hydraulic Fracturing and its Implication of on Overall Well Performance
[50], Simulation of Hydraulic Fracturing-Induced Permeability Stimulation
Hydraulic Fracturing, an Overview 27
using Coupled Flow and Continuum Damage Mechanics [51], and Oil,
the Next Revolution [52].
1.9 Economics of HF
The total investment for a HF well must cover not only the cost of drill-
ing and completing the well, but also of obtaining land rights and per-
mits. These costs vary with location, and there is a broad range of costs.
According to the Energy Information Agency, the costs can range between
three and ten million dollars a well. Depending on the geology, the wells
produce different amounts of oil and gas, with oil being the more valuable
28 Hydraulic Fracturing and Well Stimulation
product, and the one that determines the economics. There are wells that
only produce gas, but, because there is already an oversupply of gas such
that no new dry wells are being developed, only those with takeoff con-
tracts are in production. It is a shame that the associated gas from oil-
producing wells in Bakken and other basins are being flared for lack of
pipeline infrastructure.
For new HF wells, the average breakeven cost of crude oil from differ-
ent basins currently ranges between $30 and $55 per bbl. The total cost
depends on the geographic location, depth of the target, horizontal length
of the well (offset), number of HF stages, and many other factors. For
example, in the Wolfcamp Delaware basin, the average breakeven cost is
$42/bbl, while in the Bakken basin it is $52. These costs are considerably
lower than they were in 2008 (~$70/bbl), when fracking first became com-
mercially viable, thanks to the spike in crude oil prices. Innovations such
as methods for better locating resources, more efficient drilling, and mul-
tipad drilling have markedly reduced costs. In face of losing market share,
in 2012, the Organization of the Petroleum Exporting Countries (OPEC)
and Russia decided to increase their production and drive down the price
of oil from $100/bbl to $50/bbl. At $50/bbl, producing conventional oil is
still profitable. The expectation was that the competition would be driven
out of business at this reduced price.
However, because of the improved technologies, producing oil from
many of tight formations was still profitable. Operations at those that were
not profitable were simply turned off temporarily and could be brought
on-line quickly when conditions became favorable. The rapidity with which
HF wells can be turned off or brought back on stream is a game changer in
the geopolitics of oil. Furthermore, the new focus on “refracking” to extend
the life of the shale oil and gas fields may become a new normal.
1.10 Conclusions
The key factors impacting shale production are environmental concerns
and economics. In California, environmental impacts of HF practices are
limited when compared to other energy resource extraction, such as urban
oil drilling. However, this is only the case because of extensive State regu-
lations on oil drilling and waste disposal in California and chance geologic
conditions and reservoir parameters that allow for safe disposal, and con-
clusions about California may not be applied elsewhere.
More generally, States engaging in HF, including California, will
aggravate water shortages where local groundwater supplies are limited.
Hydraulic Fracturing, an Overview 29
Acknowledgement
The author acknowledges the financial support of the USC Induced
Seismicity Consortium (isc.usc.edu) members, valuable input from
Ripudaman Malhotra, the co-author of A Cubic Mile of Oil, and Kelly Rose
of NETL.
References
1. Department of Energy. Hydraulic Fracturing Technology, https://energy.
gov/fe/hydraulic-fracturing-technology
2. G.E. King, HF 101: What every representative, environmentalist, regulator,
reporter, investor, university researcher, neighbor and engineer should know
about estimating frac risk and improving frac performance in unconven-
tional gas and oil. In: SPE 152596, SPE HF Technology Conference, pp.
3. California Council on Science and Technology, Advanced Well Stimulation
Technologies in California: An Independent Review of Scientific and
Technical Information, July (2016). Accessible by: https://ccst.us/wp-content/
uploads/160708-blm-report.pdf
4. Colorado Oil & Gas Conservation Commission, Frequently Asked Questions
About HF (2011). Accessible by: https://cogcc.state.co.us/Announcements/
Hot_Topics/Hydraulic_Fracturing/Frequent_Questions_about_Hydraulic%20
Fracturing.pdf
5. U.S. Energy Information Administration (2018) Petroleum and other liquids
data. https://www.eia.gov/petro leum/data.php#crude
6. BP Statistical Review of Global Energy (2017). Accessible by: https://
www.bp.com/en/global/corporate/energy-economics/statistical-review-
of-world-energy.html, https://www.forbes.com/sites/arielcohen/2018/12/21/
americas-oil-and-gas-reserves-double-with-massive-new-permian-
discovery/#328c4cd92c91
7. Baker Hughes, Baker Hughes Rig Count. Accessible by: http://phx.corporate-
ir.net/ phoenix.zhtml?c=79687&p=irol-rigCountOverview
8. FT (2018). Accessible by: https://www.ft.com/content/a8314d06-0741-11e8-
9650- 9c0ad2d7c5b5
9. API (2018). Accessible by: https://www.ft.com/content/a8314d06-0741-
11e8-9650- 9c0ad2d7c5b5or http://www.api.org/oil-and-natural-gas/wellsto-
consumer/exploration-and-production/hydr aulic-fracturing
10. L. Minh-Thong Le (2018) An assessment of the potential for the development
of the shale gas industry in countries outside of North America. Heliyon
4:e00516. https://doi.org/10.1016/j.heliyon.2018.e00516
11. Warpinski NR, Mayerhofer MJ, Vincent MC, CipollaCL, Lolon EP (2009)
Stimulating unconventional reservoirs: maximizing network growth while
Hydraulic Fracturing, an Overview 31
Abstract
Design and control of hydraulic stimulation jobs in naturally fractured rock is
challenging due to interaction between injection-induced fractures and natural fra
ctures. Dynamic stress transfer between the two fracture systems and the result-
ing slip response of the natural fractures are key to understanding the temporal
evolution of hydraulically- and mechanically-stimulated reservoir volumes. Here,
we present a computational model of interaction between induced and natural
fractures based on numerical simulation of coupled flow, deformation, and fail-
ure processes. By capturing the time dependence of stress and energy transfer
mechanisms between hydraulic and natural fracture systems, our model allows
us to better understand the mechanical response of natural fractures to hydraulic
fracturing for the purpose of upscaling such responses in field-scale simulation
models.
2.1 Introduction
Hydraulic stimulation of naturally fractured rock is a common operation
for fluid extraction from low permeability hydrocarbon-bearing rock and
geothermal reservoirs [1–3]. Yet, our understanding of the interaction
between hydraulically induced fractures and natural fractures in these
Email: bjha@usc.edu
37
38 Hydraulic Fracturing and Well Stimulation
NF1 NF3
HF
Ly 1
NF2 NF4
y
x
Lx
Figure 2.1 Schematic of the physical domain used to study interaction between a hydraulic
fracture (HF) and four natural fractures (NF1, NF2, NF3, and NF4). The midpoint of the
hydraulic fracture is shown as a blue dot. Midpoints of the natural fractures are shown as
red dots. The maximum principal stress is oriented along the x-direction and the minimum
principal stress is oriented along the y-direction. The other two boundaries are fixed in the
respective normal directions to prevent translation or rotation of the domain.
40 Hydraulic Fracturing and Well Stimulation
·σ=0 (2.1)
where the total stress tensor σ (x,y,t) is related to the pore pressure p(x,y,t)
and the effective stress tensor σ (x,y,t) through the equation, σ = σ – bp1,
where 1 is the identity tensor and b is the Biot coeffcient. Note that tension
is positive in our sign convention. In 2D, the system of partial differential
equations become
xx xy
0,
x y
(2.2)
xy yy
0.
x y
The total stress can be decomposed as σ = σ0 + δσ, where the initial stress
σ0 results from the boundary tractions and the stress change δσ results from
hydraulic fracturing. The mechanical equilibrium equations become
0 b ( p) 0 b p, (2.3)
where p = p0 + δp with p0 being the initial pressure and δp being the change
in pressure. We assume a slightly compressible single-phase flow for the
flow problem. The pressure can be obtained by solving the fluid mass bal-
ance equation given as [19]
1 p
b V v f (2.4)
M t t
Evolution of Stress Transfer Mechanisms 41
Ly Lx
s
: bp1 d l pf n d l pf n d
0 0 f f
Lx Ly
3 dx 1 dy ,
0 0
(2.5)
Ly Lx
s
: d ld ld
0 0 f f
Lx Ly Ly Lx Ly Lx
s
3 dx 1 dy : 0 d pd
0 0 0 0 0 0
l0 d p f 0nd p f nd
f f f
l0 d p f 0nd p f nd
f f f
(2.6)
u d u_d dd (2.7)
f f f
50
t 120 sec
40
pf, Mpa
30 1 sec
20
10
0
50 100 150
x, m
Figure 2.2 Time evolution of overpressure along HF calculated using the analytical solution
of the fluid mass balance equation for a constant pressure injection in a porous medium.
44 Hydraulic Fracturing and Well Stimulation
ux
100
0
meter
–0.07
y, m
–0.035
–0.035
0
0 x, m 200
(a) 0.0
uy
0.035
0.07
(b)
Figure 2.3 Displacement field at t = 121 sec resulting from a balance of boundary
conditions and hydraulic fracturing induced stress changes. The horizontal displacement,
ux, changes sign as we move from HF to the top and bottom boundaries of the domain.
Its magnitude is asymmetric along x due to asymmetry in the right and left boundary
conditions. Magnitude of the vertical displacement, uy, is asymmetric along y because of
the asymmetry in the top and bottom boundary conditions. For the ease of visualization,
the mesh is distorted with displacements magnified by a factor of 50, and the color scale
is chosen to show the variation in ux and uy and not the minimum and maximum values
of these quantities, which are –0.022 m and 0.026 m for and –0.018 m and 0.092 m for uy.
The dash lines in the upper figure indicate the location of the profile plots in Figure 2.4.
Evolution of Stress Transfer Mechanisms 45
100 100
80 80
60 60
y, m
40 40
xx
20 ux 20 yy
uy
xy
0 0
–0.05 0 0.05 –20 0 20
(a)) Displacement, m ((b) MPa
20
0.04
Displacement, m
MPa
0 0
–0.04
–20
Figure 2.4 Profiles of displacements and stress changes orthogonal (upper row) and parallel
(lower row) to HF at t = 121 sec. Orthogonal profiles are at x = 70 m and parallel profiles are
at y = 25 m along dash lines shown in Figure 2.3a. (a) Displacement discontinuity across HF
is visible. ux changes sign above HF (y > 50 m) and similar to xx . uy is negative below HF
and positive above HF, as expected. (b) The vertical stress change, yy , is compressive due
to opening of HF. The horizontal stress change, xx , is also compressive in the immediate
vicinity of HF but becomes tensile as we approach the traction boundary on the top because
of upward displacement. The shear stress change, xy, due to HF opening is larger above
HF because of a larger upward displacement of the upper surface of HF. (c) Across the center
of HF, ux is asymmetric due to the asymmetry in the horizontal boundary conditions, and
uy is symmetric. Both displacements have discontinuities at x = 25 m and x = 175 m where
we cross NF2 and NF4, respectively. Stresses are piecewise continuous because we use linear
displacement elements.
1 1
N xx yy xx yy cos 2 pf
2 2 (2.8)
1
xx yy sin2
2
0 σxy
NF1
–10 NF2
MPa
σ΄yy NF3
NF4
–20
σ΄xx
–30
0 50 100
Time, sec
Figure 2.5 Time evolution of the effective stresses with opening of the hydraulic fracture.
The evolution is plotted for elements near the midpoints of the four natural fractures. All
natural fractures experience a decrease in compression and an increase in shear magnitude
due to opening of HF. NF4 experiences a relatively smaller decrease in compression and
largest increase in shear, which results in a compressive change in its normal traction
compared to tensile changes observed on the other three natural fractures (Figure 2.6).
0.01 0.01
NF dnormal, m
NF dshear, m
NF1
NF2
0 NF3 0
NF4
–0.01 –0.01
0 50 100 0 50 100
(a) Time , sec (b)
5 –20
–22
NF σ΄n, MPa
NF τ, MPa
–24
0
–26
–28
–5 50 100 –300 50 100
0
(c) (d)
Figure 2.6 Time evolution of slip (top row) and traction (bottom row) at the midpoints
of the four natural fractures during hydraulic fracturing. (a) Positive shear values on NF1
and NF4 indicate left-lateral shear and negative shear values on NF2 and NF3 indicate
right-lateral shear. (b) Normal slip values are zero indicating no opening or closing of the
natural fractures. (c) Shear traction magnitude decreases for all natural fractures as they
slip. This decrease continues longer for NF1, NF2, and NF3 than for NF4. (d) Normal
compression decreases on NF1, NF2, and NF3 thereby destabilizing them. It increases
on NF4, which has a stabilizing effect. Positive values indicate opening or tension and
negative values indicate closing or compression of the fractures.
48 Hydraulic Fracturing and Well Stimulation
3
0.01
2 Time Time
1
NF dshear, m
NF1
NFτ, MPa
0 NF2
NF3 0
–1 NF4
e
–2 Tim
–0.01
–3
–30 –25 –20 0 0.05 0.1
(a) NF σ΄n, MPa (b) HF center dnormal, m
3 –20
2 –22
NF σ΄n, MPa
1
NFτ, MPa
–24
0
–26
–1
–2 –28
–3 –30
0 20 40 0 20 40
(c) HF Ptip, MPa (d) HF Ptip, MPa
Figure 2.7 Evolution of traction and slip on natural fractures due to opening and
pressurization of the hydraulic fracture. (a) Stress paths of tractions at the midpoints of
the four natural fractures showing that NF1 and NF4 start sliding under positive (left-
lateral) shear tractions whereas NF2 and NF3 start sliding under negative (right-lateral)
shear tractions. The dash lines are the failure lines at μf = 0.1. The initial stress state
( N , )t 0 ( 25, 5) MPa falls outside the chosen scales of the plot. (b) Shear slips of
natural fractures increase monotonically with the opening at the center of HF. (c) Shear
tractions on the natural fractures decrease monotonically with the HF tip pressure.
(d) Effective normal compressions decrease on NF1, NF2, and NF3 and increase on NF4
with increasing HF tip pressure.
0.03
NF strain energy/area, MPa
0.5
Moment magnitude , Mω
0.4
0.02
0.3
0.2
0.01 NF1
NF2 0.1
NF3
NF4
0 0
0 2 4 0 0.05 0.1
(a) HF strain energy/area, MPa (b) HF center dnormal, m
Figure 2.8 (a) Energy transfer from the hydraulic fracture to the natural fractures is
monotonic. Growth in NF2 and NF4 strain energy are larger due to larger shear traction
and slip near the fixed bottom boundary. (b) The seismic moment magnitudes of natural
fracture slip events evolve with the opening of hydraulic fracture. Natural fractures are
grouped based on their position with respect to the traction boundary in the minimum
principal stress direction.
is the seismic moment, G is the shear modulus, and |d| is the slip magni-
tude [22]. We find that natural fractures are grouped by their proximity to
the traction boundary in the minimum principal stress direction i.e. NF2
and NF4, which are farther from the top boundary, are in a group with
a faster increase in the moment magnitude compared to NF1 and NF3,
which are closer to the top boundary.
2.7 Conclusion
We investigated the effect of hydraulic fracturing on pre-existing natural
fractures by analyzing the time-evolution of tractions and slips of natural
fractures with opening of the hydraulic fracture. We observe that the pro-
files of normal and shear slip along the natural fractures depend on the rel-
ative position of the natural fracture with respect to the hydraulic fracture,
the fixed displacement boundaries, and the minimum and maximum prin-
cipal stress orientations. We explain the heterogeneity in slip response of
the natural fractures in terms of the displacement and stress profiles paral-
lel and orthogonal to the hydraulic fracture. The energy transfer approach
in our hydraulic fracture–natural fracture interaction model can be used
to capture slip-induced permeability enhancement and microseismic
response of natural fractures during field scale simulations of hydraulic
fracturing. We are enhancing our framework to allow for two-way coupling
50 Hydraulic Fracturing and Well Stimulation
among the processes of fluid flow, deformation, and seismicity. We are also
implementing the capability to allow fracture propagation. The proposed
model allows for improved prediction of stimulated reservoir volume and
induced seismicity during hydraulic stimulation.
References
1. Hubbert, M.K. and Willis, D.G., Mechanics of hydraulic fracturing. Trans.
Am. I. Min. Met. Eng., 210, 6, 153–163, 1957.
2. Hossain, M.M., Rahman, M.K., Rahman, S.S., A shear dilation stimulation
models for production enhancement from naturally fractured reservoirs.
SPE J., 7, 2, 183–195, 2002, doi: 10.2118/78355-PA. SPE-78355-PA.
3. Curtis, J.B., Fractured shale-gas systems. Am. Assoc. Petrol. Geologists Bull.,
86, 1921–1938, 2002.
4. Lamont, G.C. and Jassen, F., The effects of existing fractures in rocks on the
extension of hydraulic fractures. J. Petrol. Technol., 15, 203–209, 1963.
5. Blanton, T.L., An experimental study of interaction between hydraulically
induced and pre-existing fractures. Paper SPE 10847 presented at the SPE/
DOE unconventional gas recovery symposium, Pittsburg, Pennsylvania, May,
1982.
6. Huang, J., Safari, R., Mutlu, U., Burns, K., Geldmacher, I., McClure, M.,
Jackson, S., Natural-hydraulic fracture interaction: Microseismic obser-
vations and geomechanical predictions. unconventional resources technol-
ogy conference (URTeC), Denver, Colorado, USA, 25–27 August, 2014, doi:
10.15530/urtec-2014-1921503.
7. Nagel, N.B., Sanchez-Nagel, M.A., Zhang, F., Garcia, X., Lee, B., Coupled
numerical evaluations of the geomechanical interactions between a hydrau-
lic fracture stimulation and a natural fracture system in Shale Formations.
Rock Mech. Rock Eng., 46, 581–609, 2013, doi: 10.1007/s00603-013-0391-x.
8. Jaeger, J.C., Cook, N.G.W., Zimmerman, R.W., Fundamentals of Rock
Mechanics, fourth edition, Blackwell Publishing, Malden, MA, 2007.
9. Olson, J.E., Bahorich, B., Holder, J., Examining hydraulic fracture-natural
fracture interaction in hydrostone block experiments. SPE 15261, Society of
Petroleum Engineers, 2012, doi: 10.2118/152618-MS.
10. Kresse, O., Weng, X., Gu, H., Wu, R., Numerical modeling of hydraulic frac-
tures interaction in complex naturally fractured formations. Rock Mech. Rock
Eng., 46, 555–568, 2013, doi: 10.1007/s00603-012-0359-2.
11. Warpinski, N.R. and Teufel, L.W., Influence of geologic discontinuities on
hydraulic fracture propagation. J. Petrol. Tech., 209–220, 1987.
12. Zangeneh, N., Eberhardt, E., Bustin, R.M., Investigation of the influence of
natural fractures and in situ stress on hydraulic fracture propagation using a
distinct—element approach. Can. Geotech. J., 52, 926–946, 2015.
Evolution of Stress Transfer Mechanisms 51
13. Aimene, Y.E. and Ouenes, A., Geomechanical modeling of hydraulic frac-
tures interacting with natural fractures-Validation with microseismic and
tracer data from the Marcellus and Eagle Ford. Interpretation-J. Sub., 3,
71–88, 2015, doi: 10.1190/INT-2014-0274.1.
14. Potluri, N., Zhu, D., Hill, A.D., Effect of natural fractures on hydraulic
fracture propagation. SPE European Formation Damage Conference, The
Netherlands, May 2005, 2005, SPE 94568.
15. Gu, H., Weng, X., Lund, J., Mack, M., Ganguly, U., Suarez-Rivera, R.,
Hydraulic fracture crossing natural fracture at nonorthogonal angles: A crite-
rion and its validation. SPE Hydrualic Fracturing Technology Conference and
Exhibition, The Woodlands, Texas, 24–26 January 2011, 2011, SPE 139984.
16. Nassir, M., Settari, A., Wan, R., Prediction of stimulated reservoir volume
and optimization of fracturing in tight gas and Shale with a fully elasto-plas-
tic coupled geomechanical model. SPE J., 19, 5, 771–785, 2014.
17. Cheng, W., Jin, Y., Chen, M., Reactivation mechanism of natural fractures by
hydraulic fracturing in naturally fractured shale reservoirs. J. Nat. Gas Sci.
Eng., 23, 431–439, 2015.
18. Ouenes, A., Aissa, B., Boukhelf, D., Fackler, M., Estimation of stimulated
reservoir volume using the concept of Shale Capacity and its validation
with microseismic and well performance: Application to the Marcellus and
Haynesville. SPE Western North American and Rocky Mountain Joint Regional
Meeting held in Denver, Colorado, USA, 16–18 April 2014, 2014, SPE 169564.
19. Jha, B. and Juanes, R., Coupled multiphase flow and poromechanics: A com-
putational model of pore pressure effects on fault slip and earthquake trigger-
ing. Water Resour. Res., 50, 3776–3808, 2014, doi: 10.1002/2013WR015175.
20. Aagaard, B.T., Knepley, M.G., Williams, C.A., A domain decomposition
approach to implementing fault slip in finite-element models of quasi-static
and dynamic crustal deformation. J. Geophys. Res. Solid Earth, 118, 3059–
3079, 2013, doi: 10.1002/jgrb.50217.
21. Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, second edition,
Clarendon Press, Oxford, 1959.
22. Hanks, T.C. and Kanamori, H., A moment magnitude scale. J. Geophys. Res.,
84, 2348–2350, 1979.
3
Primer on Hydraulic Fracturing
Concerning Initiatives on
Energy Sustainability
Michael Holloway1* and Oliver Rudd2
1
NCH Corporation, Irving, TX, USA
2
Independent Consultant, TX, USA
Abstract
Hydraulic fracturing (also known as fracking, fracing, and other variations) of
rock deep beneath the Earth’s surface to release petroleum product has become
a contentious subject across the globe. This practice is not to be confused with
drilling or extraction. Fracing is the process of using fluid power to fracture rock
to release gas and sometimes crude oil. It is not drilling, although drilling must be
done to establish a well in order to pump fluid thus fracturing rock to release prod-
uct. Certain countries have actually outlawed the practice of hydraulic fracturing
claiming that ground water and air pollution increase due to the practice. There is
also the claim that the comfort of life is adversely affected. Legitimate concerns are
always available and examples to purport a concerned view are magnified, in the
authors’ opinion. The intent of this paper is to provide a correct and balanced view
of fracturing underground rock with fluids in order to release a product to pro-
duce energy. The concept of using water to do work is nothing new. Pumping fluid
below ground in order to fracture rock to release gaseous petroleum is, however, a
relatively new practice. It is done with surprising precision as well as environmen-
tal concern yet it is interesting how the public reacts to the practice in relation to
other techniques used throughout the world. This paper will explore the materials
used as well as the concerns most common to the practice.
53
54 Hydraulic Fracturing and Well Stimulation
it seems human nature dictates that the first opinion heard or the opinion
heard the loudest and with the most hyperbole will be what the public
comes to believe. In time, once something is believed by enough people
and stated as “fact” long enough, the general public will no longer even
bother looking into facts and it will become part of the fabric of beliefs in
our society – for instance, a few examples of this phenomenon are: 1) you,
in fact, cannot see The Great Wall of China from the moon (not even
close); 2) the Sherlock Holmes character never once said “Elementary, my
dear Watson”; and 3) Neil Armstrong actually said, “One small step for
[a] man, one giant leap for mankind.”
As far as hydraulic fracturing is concerned, the aspect given the most
attention by press and most all concerned organizations is the impact it
may have on the environment. The question of environmental impact
through fracking is, to say the least, a very emotional topic and by far the
most polarizing issue; however, a great deal of analysis indicates that the
most significant environmental risks attributed to fracking are similar to
risks long associated with all drilling operations – including groundwa-
ter contamination due to inadequate cementing and/or well construction,
risks associated with trucking, leaks from tanks and piping and spills from
waste handling. This all-encompassing blame has given industry all the
ammunition needed to claim effects attributed to hydraulic fracturing are
overstated, not based on good science, or related to processes other than
hydraulic fracturing.
Due to the great ongoing controversy over alleged impacts from frack-
ing, many public groups have become deeply suspicious of the trustwor-
thiness and overall motives of the oil and gas industry. These suspicions are
continuously intensified by two things:
right and both sides being wrong while never taking steps to come together
on a common goal.
This has become a major argument point for the concerned public
because, basically, “if you have nothing to hide why would you not want
to disclose it?” Take, for example, a small child walks into a room with
his hands behind his back and will not show you what is there for sev-
eral minutes…and then only does when forced. Well, even if it turns out
he was holding something as harmless as a feather behind his back it will
make you suspicious all the same wondering “what was he doing with that
feather!”
To make this contentious subject a little clearer, this section will pro-
vide descriptions of why frac fluids are needed, what general chemicals are
needed and used, relative amounts of chemicals in frac fluid composition,
proppants – the different types and uses, a discussion on slickwater, and a
discussion on present regulations and standards for industry disclosure of
frac fluid compositions.
Now, I am sure many of you that have seen this type of information before
are now expecting to see one of those “other uses” tables telling you that frack-
ing fluid must be safe due to the “ingredients” of fracking fluids having such
everyday uses such as: scale inhibitors having the same chemicals as wind-
shield washer fluid, friction reducers having the same chemicals as many
makeup products, surfactants being basically the same as shampoo products,
proppants being play sand and hydrochloric acid also being swimming pool
cleaner; these may be true in the strictest sense of the word, but this type of
listing can also be very misleading and insincere, in that most all chemicals can
be used for many different things, but are still not something with which you
necessarily want to come in contact. For example, ammonium nitrate is com-
monly used in agriculture as a high-nitrogen fertilizer, nitromethane is a com-
monly used industrial solvent and Ryder trucks are commonly used to move
families and their belongings to their dream homes – while these are also three
of the common “ingredients” used in the tragic April 1995 Oklahoma City
bombing of the Alfred P. Murrah Federal Building, which killed 168 people.
This is, of course, a comparison made for shock value, but it is meant as such to
stick in your memory as how these sorts of comparisons can be manipulated
and to drive home the fact that the best policy is to study upon facts when you
see a comparison like this, and make an informed decision for yourself.
Gelling
agent Scale pH adjusting
KCI 0.056% inhibitor agent
0.06% 0.043% 0.011%
Surfactant Breaker
0.085% 0.01%
Crosslinker
0.007%
Water Other
and sand Iron control
0.49% 0.004%
99.51%
Corrosion
inhibitor
0.002%
Biocide
Friction 0.001%
reducer Acid
0.088% 0.123%
Graph 1 Volumetric percentages of additives in fracturing fluids from Modern Shale Gas
Development in the United States.
3.1.6.1 Proppants
Proppants are pretty hard to make into anything fun, exciting or entertain-
ing…as they are, for the most part, made up of sand or a manufactured
facsimile of sand. Sure, if you want to be poetic you can refer to proppants
as the only materials the operators want to remain downhole in the frac-
tures. If you really want to think poetically, feel free to consider a prop-
pant’s life as one of making its way from origins mined within the Earth
only to return to its final resting place deeper within the Earth’s fractures.
As discussed earlier, proppants are simply materials (typically silica sand,
resin coated silica sand, or manufactured ceramics) used to prop open the
open fractures to promote flow and eventual extraction of hydrocarbons.
As simple as prop-pants may seem, the estimated amount of proppant used
in industry has grown tenfold since 2000. In some regions it is not uncom-
mon to see upwards of four million pounds of proppant used per well and
represent up to 5% of well costs. The growth in proppant usage is generally
attributed to operators realizing better well completion techniques with
more proppant per stage and better well pad techniques with more laterals
and fracturing stages per pad.
62 Hydraulic Fracturing and Well Stimulation
Even considering the accelerated growth in the last decade, the evolu-
tion of proppant usage has been slow to develop over the industry lifetime
as a whole.
Consider that the first frac job was conducted in 1947, utilizing a
reported approximate 20,000 pounds of uncoated frac sand, and man-
ufactured ceramic proppant was not first used until 1983...or 36 years
later. Then, approximately one year later, resin coated proppant was first
introduced. As with most all technologies, as new techniques continue to
develop, proppants will surely evolve further to increase effectiveness and
efficiency in hydraulic fracturing.
No matter the type of proppant used, the most important characteristics
for a proppant are particle size distribution, crush resistance, shape and
sphericity (or roundness). Proppant materials are carefully sorted for size
and sphericity to provide an efficient conduit for production of fluid from
the reservoir to the well-bore. Grain size is critical because a proppant
must reliably fall within certain size ranges to coordinate with downhole
conditions and completion design (Figure 3.1).
Proppant shape and hardness qualities are also very important to the
efficiency and effectiveness of a fracturing operation. A coarser proppant
allows for higher flow capacity due to the larger pore spaces between
grains, but it may break down or crush more readily under high closure
stress, and rounder, smoother proppant shapes allow for better permeabil-
ity (Figure 3.2).
Another important quality that must be taken into consideration is its
hardness with respect to the formation. If the proppant is unable to embed
in the formation, something referred to as point load occurs, which leads
to higher flow capacity, but the proppant will break easier. However, if the
proppant is able to embed in the formation, it is referred to as embedment,
Sufficiently placed
& sized proppant No proppant
- Effective return - - No return -
Individual
fracture
Return flow
Insufficiently placed
& sized proppant
- Ineffective return -
which results in the load pressure spreading out over the proppant area,
increasing the breaking point but also lowering flow capacity. Embedment
is also a function of particle size (Figure 3.3).
Even though most all proppant materials are naturally occurring, includ-
ing manufactured ceramic proppants, with relatively low amounts of addi-
tional engineering necessary, the logistics in procuring and transporting
proppants can be daunting. Logistical considerations include coordination
of manufacturing material resources, transportation costs, and possibly a
substantial monetary investment in equipment necessary for processing
and material handling facilities.
Currently there are two types of resin coated proppants, Pre-cured and
Curable. Pre-cured is the “original” technology in which the resin coating on
the silica sand grains is fully cured prior to injection into the fractures. The
newer, curable technology has often been described, and I believe very well
described, as having a coating that is not completely “baked” or hardened.
Curable resin coated prop-pants are used at a little more than half cure so
that when the proppant is pumped downhole it can finish curing in the frac-
tures with down hole pressure and temperature. The advantage to curable
proppant technology is that it allows the individual proppant grains to bond
together in the fracture – resulting in the grains bonding together uniformly
in strength when temperature and pressure reach a appropriate levels.
A reason for this is because a lot of data is collected for the other aspects
related to fracking operations to try and prove/disprove their existence,
when the nuisances discussed in this section are easily seen as in existence
(just spend a few minutes along any road used for drilling operations, and
this will become abundantly clear). Please keep in mind that there are
many additional nuisances absorbed by those living near drilling loca-
tions or related roadways, so this listing is far from comprehensive. The
following are merely what most see as the most common and are not pre-
sented in any order of magnitude. That decision has to be made by each
individual – one person may be more affected by noise, while another is
much more concerned with dust.
3.3.1 Noise
Noise conditions are usually one of the first things to change and be
noticed by local landowners. An increase in noise is also one of the most
continuous nuisances related to operations. Drilling and completing a
well – from the pad construction to the final completion of the well –
takes several weeks and utilizes many different types equipment. This
additional equipment can include additional trucking, construction and
drilling equipment. The noise concerns usually begin with the additional
traffic brought to an area during pad construction, then continue with
the noises associated with equipment and trucking required to construct
a pad, only to be followed by the large amount of noise related to rig
construction and operation throughout the well drilling process. Then,
once the well site is completed, there may come the additional sounds of
compressors used during ongoing production activities.
When you think of noise concerns related to the oil and gas industry,
the first thing that commonly comes to mind is the big noisy rig or maybe
the noisy traffic coming back and forth. These are, of course, very real and
valid concerns; however, the thing that is quite possibly the most notable
noise nuisance related to the oil and gas industry, due to length of time, is
the compressor. For the most part, the heavy rig work and heavy truck traf-
fic lasts approximately one to two months – while the compressor, while
not as loud, can continue for a much longer amount of time (months to
even years).
Gas compressors are normally the largest equipment remaining after the
well development process is complete and are utilized for something called
gas lift. Gas lift is used in wells that have insufficient reservoir pressure
to produce efficiently on their own. The gas lift process involves inject-
ing gas through the tubing-casing annulus to aerate the fluid to reduce
68 Hydraulic Fracturing and Well Stimulation
its density. Following aeration of the fluid the formation pressure is then
able to better lift the oil column up the wellbore. For pad sites where long
term compressor use is anticipated, especially in rural communities where
serenity is the norm and even the slightest ongoing noise can be heard
clearly for long distances, operators have addressed compressor noise con-
cerns with remote siting (trying to locate the compressors on the part of
the pad farthest from homes), noise tampering sound walls, and directing
compressors with fans away from homes. However, even with the measures
presently taken to mitigate ongoing sound issues, additional work must be
done and technology developed to work toward a solution.
necessary lease roads and drill pad locations, fewer necessary pipelines and
fewer tank batteries.
The second, “do something to improve the roads,” would depend on the
type of road to be improved. Improving and widening gravel type lease and
rural roads is a less daunting task than improving paved city/county roads
due to ease of obtaining the proper materials and the fewer restrictions put
on maintenance. However, making improvement to city and county roads
70 Hydraulic Fracturing and Well Stimulation
would include needing to clear a wider right of way all along the road to be
widened and the city/county would need to have the funds set aside for this
task…which is a time consuming process.
could possibly escape the tanks plus sufficient extra space for “worst case
scenario” rainfall. This amount is calculated for each region of the country
based on historic rainfall data.
Even with attempts to minimize the amount of on-site storage, some
chemical and product storage is unavoidable, and there are very valid
concerns, including potential spills, leaks, tank or container overfill, and
even the chance of traffic accidents on location or roadways leading to
releases of chemicals and/or products. Release events could range from
relatively small amounts from equipment leaks to possibly hundreds of
barrels from tank release. Two regulatory measures in place to manage and
oversee on-site chemical storage conditions are requiring Spill Prevention
Countermeasure and Control (SPCC) plans and SARA reporting.
SPCC plans are documents required by all facilities having the potential
to discharge oil to navigable waters of the U.S. and meeting one or both
of the following: greater than 1,320 gallons (31.4 bbls) aggregate above-
ground storage in equipment, drums, tanks, totes, tanks greater than
55 gallons in size; or, greater than 42,000 gallons total underground storage
capacity. Just to clarify, aggregate refers to adding up separate amounts of
all storage vessels…you can have one 1,320 gallon tank or ten 132 gallon
tanks and they would be equal under the SPCC requirements. Also, the
“having potential to discharge oil to navigable waters of the U.S.” is left up
to regulatory discretion to calculate, and has come to include pretty much
anywhere in the U.S. you could imagine. SPCC plans are, to keep it simple,
engineer-stamped documents that must be created for all facilities meet-
ing the above conditions that include a list of spill response procedures,
an emergency notification phone list, inspection procedures and schedule,
training requirements, site figures, site chemical and product storage vessel
types and sizes and containment calculations to prove sufficient contain-
ment is given to contain the largest possible spill amount.
SARA reporting, or possibly better known as the federal “right to
know,” requires quarterly and annual reporting of chemical storage
details (types of chemicals, amounts and dates of storage) for all facilities
which used more than 10,000 pounds per year of the chemical exceeding
the threshold quantity. This requirement means a facility storing more
than 10,000 pounds of a given chemical in a year must report that chem-
ical and amount. This program is intended as the “right to know” for
emergency responders and emergency services that may respond to an
emergency situation on the location so they will be able to adequately
prepare for what may be stored on site. The drawback of this program
as related to the oil and gas industry is that, with quarterly reporting,
72 Hydraulic Fracturing and Well Stimulation
by the time a chemical has been reported the oilfield function requiring
the chemical has normally been long complete and the chemicals are no
longer on site. This basically means that once the chemical is reported as
being on a location it is no longer there; however, as previously stated,
oilfield operations have become such a streamlined process that if you
know what has been reported for a previous location by a specified oper-
ator you can, for the most part, expect much the same chemicals and
products stored at following locations. If you are really curious about all
the chemicals used at a site, ask to receive a copy of the Material Safety
Data Sheet of the chemicals used.
3.7 Conclusion
One hundred and fifty years ago when crude oil was first being extracted,
the damage done to the environment was nothing short of a nightmare.
In some areas little has changed but in many instances companies take
extraordinary precautions. Many years ago, we did not understand the
ramifications of pollution. Today, much work is underway to address what
is being understood as environmental concerns. The work that goes into
preparing a fracing well site today in assuring that the chemicals used are
innocuous while maintaining the physical integrity of the surrounding
land is taken into consideration primarily for legal reasons as well as busi-
ness concerns. If environmental laws are in place, then work shall be struc-
tured accordingly. The challenge is to enact law that makes sense according
to empirical evidence. It is also vital that those using the process and chem-
icals as well as the public understand the technology. Where industry in
general fails is when it rushes forward and only applies the letter of the law
and does not push for higher requirements. Legislation is often left to the
most verbal and most passionate. It is fair to say that when emotion runs
high, logic wanes. If we are to utilize the gifts that the Earth has bestowed
upon us then it is more than fair to assume a protective role going forward
and exploring proper protocols to ensure that the environment and com-
fort of life remain as balanced as it was found.
78 Hydraulic Fracturing and Well Stimulation
Bibliography
Fjaer, E. (2008). “Mechanics of hydraulic fracturing”. Petroleum related rock
mechanics. Developments in petroleum science (2 ed.). Elsevier. p. 369. ISBN
978-0-444-50260-5. http://books.google.co.uk/books?id=l6CfasFxhzYC&pg=
PA369&lpg=PA369.
Laubach, S.E.; Reed R.M., Olson J.E., Lander R.H. & Bonnell L.M. (2004).
Coevolution of crack-seal texture and fracture porosity in sedimentary
rocks: cathodoluminescence observations of regional fractures“. Journal of
Structural Geology (Elsevier) 26 (5): 967–982. doi:10.1016/j.jsg.2003.08.019.
http://www.sciencedirect.com/science/article/pii/S0191814103001858.
“Definition of frac job”. Oilfield Glossary. Schlumberger. http://www.glossary.oil-
field.slb. com/Display.cfm?Term=frac%20job.
Office of Air and Radiation U.S. Environmental Protection Agency (November
2010). “General Technical Support Document for Injection and Geologic
Sequestration of Carbon Dioxide: Subparts RR and UU Greenhouse Gas
Reporting Program”. Office of Air and Radiation U.S. Environmental
Protection Agency. http://www.epa.gov/cli-matechange/emissions/down-
loads10/Subpart-RR-UU_TSD.pdf.
Jack E. Whitten, Steven R. Courtemanche, Andrea R. Jones, Richard E. Penrod,
and David B. Fogl (Division of Industrial and Medical Nuclear Safety,
Office of Nuclear Material Safety and Safeguards (June 2000). “Consolidated
Guidance About Materials Licenses: Program-Specific Guidance About Well
Logging, Tracer, and Field Flood Study Licenses (NUREG-1556, Volume 14)”.
US Nuclear Regulatory Commission. http://www.nrc. gov/reading-rm/
doc-collections/nuregs/staff/sr1556/v14/#_1_26.
Anthony Andrews et al. (30 October 2009) (PDF). Unconventional Gas Shales:
Development, Technology, and Policy Issues. Congressional Research Service.
p. ?. http://www.fas.org/sgp/crs/misc/R40894.pdf.
Bill McKibben (8 March 2012). “Why Not Frack?”. The New York Review of Books 59
(4). http://www.nybooks.com/articles/archives/2012/mar/08/why-not-frack/.
David Biello (30 March 2010). “What the Frack? Natural Gas from Subterranean
Shale Promises U.S. Energy Independence--With Environmental
Costs”. Scientific American. http://www.scientificamerican.com/article.
cfm?id=shale-gas-and-hydraulic-fracturing.
(PDF) Chemicals Used in Hydraulic Fracturing (Report). Committee on Energy
and Commerce U.S. House of Representatives. April 18, 2011. http://
democrats.energy-commerce.house.gov/sites/default/files/documents/
Hydraulic%20Fracturing%20 Report%204.18.11.pdf
4
A Graph Theoretic Approach for Spatial
Analysis of Induced Fracture Networks
Deborah Glosser1,2* and Jennifer R. Bauer3,4
1
United States Department of Energy, National Energy Technology Laboratory,
Pittsburgh, PA, USA
2
Oak Ridge Institute for Science Education, Oak Ridge, TN
3
United States Department of Energy, National Energy Technology Laboratory
Albany, Oregon, USA
4
United States Department of Energy, National Energy Technology Laboratory, AECOM,
Albany, Oregon, USA
Abstract
Drilling induced fractures are generated when excessive stresses around a bore-
hole cause tensile failure of the wellbore wall. If stress concentrations are great
enough, compressive failures can form in the region surrounding the wellbore,
leading to wellbore breakout, and the potential compromise of wellbore integrity.
Another category of induced fracture networks are hydraulically induced frac-
tures, which are generated by the injection of pressurized fluids into the subsur-
face. Overlapping induced fracture networks between collocated wellbores may
increase pathways in the subsurface, and create the potential for unwanted fluid
leakage. The generation of induced fractures is greatly dependent upon the struc-
tural and geological characteristics. Probabilistic-based simulations are often used
to model fracture systems. Several methods for modeling local fracture networks
have been proposed in the literature. These models often involve the generation of
randomly located fractures, and may have limited capabilities for honoring engi-
neered fractures such as induced fracture networks. We present a graph theoretic
approach for identifying geospatial regions and wellbores at increased risk for sub-
surface connectivity based on wellbore proximity and local lithologic character-
istics. The algorithm is coded in Matlab, and transforms 3 dimensional geospatial
data to graph form for rapid computation of pairwise and topological relationships
between wellbores (nodes), and the spatial radius of induced fractures (edges).
79
80 Hydraulic Fracturing and Well Stimulation
integrity [8]. HIFs are pressure induced fractures that are generated when
fluid is injected at high pressure into subsurface formations. If induced
fracture networks around collocated wellbores intersect, there may be an
increased likelihood of communication between wellbores, and the poten-
tial for unwanted fluid leakage between such networks. This is particularly
of concern for older, structurally unstable wellbores, which may lack ade-
quate casing or cementing necessary for zonal isolation [9]. Furthermore,
wellbore spatial densities are likely to be relatively high in regions with
historical drilling activity: Before regulations requiring minimum well-
bore spacing were implemented, it was a common practice to drill multiple
wellbores in close proximity (Figure 4.1). Furthermore, regions with exten-
sive drilling histories likely have higher wellbore densities as a result of the
multiple centuries of exploration of resources in those regions [1].
The generation of induced fractures is greatly dependent upon the
structural and lithological characteristics of local geology, which is often
difficult to accurately characterize in the absence of expensive geophysi-
cal surveys. Consequently, probabilistic-based simulations are often used
to model such fracture systems. Several methods for modeling local frac-
ture networks have been proposed in the literature [10–12]. These models
often involve the generation of randomly located fractures, with varying
degrees of user defined connectivity controls. Because of the importance
of wellbore locations; spatial densities; and the potential for overlapping
induced fracture networks to create fluid flow pathways, it is important to
Figure 4.2 A drawing of an example graph structure showing vertices (blue circles) and
edges (orange lines).
Acquire data
Define input
parameters
Figure 4.3 Workflow for preforming a graph analysis to spatially assess induced fractures.
Table 4.1 Literature derived values for the average fracture radii for a bounding
lithology type [17, 31–34].
Shale Sand
Induced fracture radius (m) 67 113
and associated values reported in Table 4.1 (or other user defined values),
gives the model flexibility for a range of geologic conditions.
Pennsylvania wellbores
Status of wellbores in bradford eld
Active
Plugged &/or abandoned
Other
N 0 1 .5 3 4. 5 6
Kilometers
Figure 4.4 Distribution and known status of wellbores within the Bradford Oil & Gas
field, Pennsylvania.
88 Hydraulic Fracturing and Well Stimulation
4.3.1.2 Results
The two algorithms – the knn algorithm and the topological algorithm,
were executed on these data. 50 wellbore locations from the Bradford field
were subsampled, and associated bounding lithology values for the well-
bores were chosen. The knn algorithm was run for three nearest neighbor
scenarios: k = 1 (Figure 4.5f); k = 2 (Figure 4.5e); and k = 3 (Figure 4.5d).
For the topological algorithm, the results are presented in 3 dimensions
with the induced fracture radius of influence for each wellbore (Figure
4.5a); in 2D form with the radius of influence (Figure 4.5b); and in simple
graph form showing only the edge connections (Figure 4.5c). Unlike the
knn approach – where edges are drawn to each wellbore (node’s) k nearest
neighbors, in the topological approach, the algorithm draws an edge if and
only if a wellbore (node) is within the induced fracture radius of influ-
ence of another wellbore (node). That is – if the induced fracture radius of
influence of wellbore nodes overlap, then an edge is drawn. It is apparent
from the results that the knn algorithm is, by definition, sensitive to the
number of neighbors, with the edge connections and overall graph con-
nectivity varying greatly based on this value. It is further apparent that
even when only one nearest neighbor is selected, the geometry of the edge
connections is considerably different than the geometry of the edges in the
topologic approach.
0
0.2
0.4
Depth
0.6
0.8 1.2
1 1 1.2
1.2 0.8 1
1 0.6 0.8
0.8 1 0.4 0.6
0.5 0.6 0.4
Long
0 0 0.2 0.4 Lat Long 0.2 0.2 Lat
(a) (b) 0 0
1.2
1 1
0.8 1.2 0.8
1
0.6 1 0.6 0.8
0.8
0.4 0.6 0.4 0.6
Long 0.4 0.4
0.2 0.2
0.2 0.2
(c) 0 0 Lat (d) 0 0
1.2 1.2
1 1
0.8 1.2 0.8
1.2
0.6 1 0.6 1
0.8 0.8
0.4 0.6 0.4 0.6
0.2 0.4 0.2 0.4
0.2 0.2
(e) 0 0 (f) 0 0
Figure 4.5 Representative graph output: Graph representation of wellbore points (n =50).
Blue dots represent wellbores (vertices), red lines represent edges, green ellipsoids
represent radii of influence (a) topological approach in 3D; (b) topological approach in 2D
form; (c) topological approach in 2D form (d) knn approach (k = 3) in 2D form (e) knn
approach (k = 2); (f) knn approach (k = 1). Graph axes are represented non-dimensionally.
fracture network. Like the Bradford field, Armstrong County has an exten-
sive regional history of oil and gas exploration. To test the performance
of the graph-based spatial analysis on a real world leakage scenario, both
the knn algorithm (k = 1) and the topological algorithm were applied to a
subset of wellbores near the pressurized well that caused the leakage event.
90 Hydraulic Fracturing and Well Stimulation
4.3.2.2 Results
Both the knn (k = 1) algorithm and the topological algorithm were applied
to the subset of Armstrong County data. Results are shown in 2D form in
Figure 4.6. Both graph-based analyses identified a cluster of wellbores as
associated with the leakage event: However, the knn algorithm erroneously
identified subsurface connectivity between several more wellbores than
were reported as being associated with the stray gas leakage. Overall, the
Pennsylvania wellbores
Status of wellbores
Active
Dayton incident well
Plugged &/or abandoned
N Other
0 500 1,000 1,500 2,000
Meters Armstrong country, PA
4.4 Discussion
The spatial locations of wellbores, as well as the geologic characteristics
of the reservoir, place physical controls on the formation of induced frac-
tures. Many fracture and fracture flow models do not honor wellbore loca-
tions and related engineered fractures; or, they consider only the spatial or
geologic attributes of the wellbores or reservoir in stochastically generating
random fractures. The importance of considering both spatial and geo-
logic attributes in identifying areas at greater risk for overlapping induced
fracture networks is highlighted by comparing the topological to a simple
distance-based nearest neighbor algorithm. Once identified, these regions
can be targeted for finer scale modelling, and used as part of a cumulative
modeling strategy to constrain uncertainty in the subsurface. The results
of the model are also of value as a standalone piece of data for the develop-
ment of science-based wellbore drilling, injection, and risk management
plans.
As shown in Figure 4.5, a nearest neighbor approach is by nature sensi-
tive to the selected k number of neighbors. The subgraphs produced by the
knn algorithm contrast considerably from the subgraphs produced by the
topological approach. In particular, the knn approach suggests far greater
connectivity between nodes than does the topologic approach, particularly
when higher values of “k” are chosen. Conversely, when low values of “k”
are chosen, several potential connections between nodes may be missed,
since the algorithm chooses the nearest neighbor of a node, and not its
range of influence on other nodes.
Since the purpose of this model is to identify probable regions of sub-
surface connectivity – and wellbores and wellbore clusters at greater risk
for unwanted fluid migration – the knn approach gives a substantially less
accurate representation of the spatial extents that are likely to contain over-
lapping induced fracture networks. First – the k is user defined, and the
92 Hydraulic Fracturing and Well Stimulation
(a) (b)
(c)
Figure 4.7 (a) Results of knn algorithm on Dayton PA area wellbore (k = 1) (n = 77);
(b) Results of topological algorithm on Dayton PA wellbores (n = 77); (c) Results of
topological algorithm on Dayton PA wellbores, zoomed in on wellbores associated with
leakage. Black arrows point to overpressurized wellbore. All results are shown in 2D form
for clearer visual representation of results.
Spatial Analysis of Induced Fracture Networks 93
4.5 Conclusions
Overlapping induced fracture networks between collocated wellbores may
increase communication in the subsurface, and create the potential for
unwanted fluid flow. The generation of induced fractures is greatly depen-
dent upon the structural and lithological characteristics of local geology,
which is often difficult to accurately characterize in the absence of costly
94 Hydraulic Fracturing and Well Stimulation
Acknowledgements
This work was completed as part of National Energy Technology Laboratory
(NETL) research for the Department of Energy’s Complementary Research
Program under Section 999 of the Energy Policy Act of 2005. This
research was supported in part by an appointment to the National Energy
Technology Laboratory Research Participation Program, sponsored by the
U.S. Department of Energy and administered by the Oak Ridge Institute
for Science and Education. The authors wish to acknowledge the techni-
cal feedback from Kelly Rose, Russell Schwartz, Circe Verba, and Grant
Bromhal. Deborah also thanks her family for their support, with special
recognition to Miriam and Michael Miller.
References
1. D. Glosser, K. Rose, and J. Bauer, Spatio-temporal analysis to constrain
uncertainty in wellbore datasets: an adaptable analytical approach
Spatial Analysis of Induced Fracture Networks 95
Abstract
Over the years, productivity studies conducted for horizontal multistage comple-
tions have shown significant stage-wise variability. Optimizing such completions
could hold the key to unlocking true value from shale reservoirs and improving
well economics. Traditional hydraulic fracturing programs use the same fracture
design along laterals without any consideration for changes in the reservoir and
wellbore conditions. Methods using mechanical rock properties require expen-
sive petrophysical logging data, while those involving use of drill cuttings can be
highly resource intensive and time consuming. In this paper, we introduce a novel
approach, which utilizes routinely available data such as measurements made
while drilling and petrophysical data as available within a fracture spacing design
framework. We validate our approach through application on multiple wells by
comparing results from our workflow with post completion production logging.
Finally, we highlight the potential advantages and pitfalls in our approach and
present a roadmap for future implementation in different plays.
5.1 Introduction
While the advent of multistage hydraulic fracturing in long lateral com-
pletions has revolutionized shale oil and gas production, the process still
101
102 Hydraulic Fracturing and Well Stimulation
lacks the robust understanding of what happens downhole within the res-
ervoir and what leads to the significant variability in productivity from
completed stages. A significant amount of work has gone into improving
our understanding of these completions, and some of these observations
have provided enough information to try and improve fracture spacing
design. Traditional approach to completion design involves use of tran-
sient rate-time analysis to identify key design parameters (permeability
and fracture half lengths) and using them to predict well performance [1].
Stage spacing is a critical design parameter, which is impacted by consid-
erations of reservoir permeability [2], stress shadowing effects [3, 4], SRV
considerations [5], and economic considerations such as net present value
[6]. Other more elaborate techniques at optimizing stage spacing include
use of microseismic data [7], and fracture network modeling [8] to name
a few. In practical applications, what is desirable is to take a holistic view
of the completion process and utilize as much data and analysis as pos-
sible for design [9]. While holistic design techniques are in place, most
methods do not account for variability in reservoir and completion effects
along the long laterals. From production logs and distributed acoustic and
temperature sensing data, we know that many clusters show insignificant
to no production, creating zones with very low productivity. This clearly
indicates that the “one size fits all” approach creates sub-optimal fracture
design, and this has been abundantly recognized by the industry [10]. We
believe that while design issues (such as fracturing efficiency) are import-
ant; formation quality is critical, as sections with lower quality should have
modified fracture density to provide for adequate drainage.
Many novel approaches have been suggested in the past few years,
which involve a thorough investigation of the reservoir properties and
the geomechanical aspects of completion in shale formations. One recent
example highlighting an engineered fracture spacing design approach uses
characterization of reservoir and completion quality, which are used to
predict proper stage placement [11]. Microseismic monitoring and other
geophysical tools can also allow for improvements in design based on
observations [12]. Generating pseudo logs for lateral sections of wellbore
based on observations from the vertical pilots is an established technique
[13] for understanding laterals and improving associated completions.
Methods looking at fracability alone and utilizing stochastic optimization
techniques have been evaluated [14]. The need for running wireline petro-
physical logs for at least the vertical pilots, and the need for core analysis
and correlation are well understood. These are not routinely available, and
therefore, we felt there was a need for a technique that can be applied on
any well to provide a quick optimal design suggestion based on the historic
Multistage Hydraulic Fracturing Completions 103
field data available for the play in question and the mud log data from the
well under consideration.
For this study, we wanted to devise a technique that can systematically
distribute fracture stages for more effective drainage of the reservoir with-
out the use of expensive wireline or logging while drilling data. What is
unique about our completion design approach is the use of mud log data
for completion design in the absence of any wireline petrophysical or
geomechanical data from that area. This is expected to work reasonably
well, provided a predictive model for rock properties can be developed. In
our approach, we use a hybrid AI (Artificial Intelligence) based modeling
workflow to predict geomechanical properties where stage spacing design
utilizes mud log gas shows, as well as predicted geomechanical rock prop-
erties, within a predefined design framework.
5.2 Method
For most new field development programs in unconventional plays, verti-
cal pilot wells are drilled, cored, and logged in order to gain a robust under-
standing of the formation before completions can be designed. These pilots
provide valuable insights into the rock, including the mineralogy, in situ
stress state, organic content, lithology, porosity, etc. to name a few. This
wealth of data can be used to predict the behavior of the well laterals drilled
from the pilot. Using data from such pilots, we propose to design predic-
tive models for reservoir properties, which can be obtained from mud log
data alone. This allows for wider applicability of the design methodology
without compromising on upfront well completion costs. We propose a
hybrid neural network (Neuro-Fuzzy) workflow, which uses mud log data
and geomechanical predictive models to design a fracture density model,
which can then be used to place fracture clusters.
Mud logs typically provide estimates for observed gas shows, Gamma
ray for geo-steering purposes, and rate of penetration data. We understand
that Gamma ray logs can provide indications of shale layers, which have
higher natural radioactivity. Gas shows could indicate possible productive or
non-productive zones and also potential naturally fractured zones. However,
the observed gas shows are influenced by rate of penetration which in turn
can be impacted by a multitude of factors, not all of which are due to reser-
voir conditions. The gamma ray tool is also influenced by the erratic drilling
speeds and varying wellbore conditions encountered during drilling, in gen-
eral. In order to develop the suggested design framework, we have to answer
some important questions posed at the outset. Most importantly we want to
104 Hydraulic Fracturing and Well Stimulation
see how these parameters relate to zonal productivity potential, whether the
impact is verifiable, and which parameters are needed for reasonable design
solutions. To find these answers, we use available data from multiple wells
from the Marcellus shale play, and verify through observations the necessary
framework for the proposed design approach.
1.5
Normalized measure
0.5
0
0 1000 2000 3000 4000 5000 6000
(a)
1.5
Normalized measure
0.5
0
0 1000 2000 3000 4000 5000 6000
(b)
1.5
1 R² : 0.81132
Norm. L/W ratio
0.5
−0.5
−0.5 0 0.5 1 1.5
(c) Norm. b-value
Figure 5.1 Stage-wise distribution of microseismic derived (a) ‘b value’ and (b) length-to-
width ratio for a study well. Subplot (c) shows cross plot of the two showing a strong
positive correlation between the two parameters.
106 Hydraulic Fracturing and Well Stimulation
1.5
measure
1
Norm.
0.5
0
(a) 0 1000 2000 3000 4000 5000 6000
1.5
measure
1
Norm.
0.5
0
(b) 0 1000 2000 3000 4000 5000 6000
1.5
measure
1
Norm.
0.5
(c) 0 1000 2000 3000 4000 5000 6000
20
% G as
10
0
(d) 0 1000 2000 3000 4000 5000 6000
Lateral length (ft.)
Figure 5.2 (a) Microseismic derived ‘b value’, (b) image log derived fracture density,
(c) mud log gas shows (red: 1 and blue: 0), and (d) production log for a study well. The
properties (except production log data) are normalized between 0 and 1. The black insert
in subplot (d) indicates the section of the wellbore where production log could not be run.
Multistage Hydraulic Fracturing Completions 107
5.2.2 Workflow
Apart from the pre-completion drilling data, the proposed workflow
requires some rock properties derived from specialty logging such as dipole
sonic or spectral azimuthal gamma for representative lithologic layers. This
data is typically available for a given field, especially new development felds
where a single or multiple pilot wells are drilled before a full field develop-
ment drilling program is implemented. This is necessary to model for the
same properties based on routine mud log data by deriving the necessary
models. The workflow involves the following steps:
• Rule #1: Low modeled brittleness and low gas shows imply
very high density.
• Rule #2: Medium modeled brittleness and low gas shows
imply very high density.
• Rule #3: High modeled brittleness and low gas shows imply
high density.
• Rule #4: Low modeled brittleness and medium gas shows
imply high density.
• Rule #5: Medium modeled brittleness and medium gas
shows imply medium density.
• Rule #6: High modeled brittleness and medium gas shows
imply medium density.
• Rule #7: Low modeled brittleness and high gas shows imply
low density.
• Rule #8: Medium modeled brittleness and high gas shows
imply low density.
• Rule #9: High modeled brittleness and high gas shows imply
very low density.
Here the ‘density’ values indicate final fracture density (or perforation
cluster density) recommendation to be made by the designed fracture den-
sity model. We do note that these rules suggest relatively lower fracture
cluster count for the so called sweet spots. Since the decision on how much
to frac and where is a highly complex one with well economics playing a
major role, the workflow is adaptable enough so that the rules can be flipped
with the high density recommendations changed to low density recommen-
dations and vice versa. This approach is useful in cases where specific well
intervals have a predictable behavior and sensitivity to stimulation.
5.3 Data
We apply this approach to three wells from two separate well pads (sep-
arated by 10’s of miles from one another). The data from Well #1 for Pad
#1 is used as the design data as the well had open-hole logging carried out
for the vertical pilot as well as the horizontal section of the well. Two other
wells were used as application wells for validating the models as each had
a production log available for independent validation. These include Well
#2 associated with the same well pad as Well #1 and Well #3 associated
with Pad #2. Figure 5.3 shows the various lithological layers of relevance
across Well #1. We can clearly observe that the well lateral intersects two
layers (Zone C or the target zone and zone B which is the overburden lower
Multistage Hydraulic Fracturing Completions 111
4900
Marcellus
Onondaga
5000 Top cherry valley
Top zone A
Top zone B
5100 Top zone C
Top zone D
Well #1
5200
TVD (ft)
5300
5400
5500
5600
4000 5000 6000 7000 8000 9000 10000 11000 12000
MD (ft.)
Figure 5.3 Well #1 cross-section showing well track in reference to the varying lithologic
formations in the Marcellus shale play. The lateral is restricted to Zones C & D.
Marcellus layer) of interest, and we use the available data to model the geo-
mechanical properties for these two layers.
Since we have extensive wireline logging done for this long lateral (Well
#1), we can use the data from the logs to estimate geomechanical and other
properties, which in turn form the basis for our model design framework
using artificial neural nets. Apart from the standard mud log gas shows avail-
able from drilling records, other wireline tools run for this well (both for the
horizontal lateral and the vertical sections) include standard measurements
such as density, porosity, and resistivity, as well as lithological tools to iden-
tify mineralogy and organic content. A thorough petrophysical analysis was
carried out for the entire logged wellbore, and the geomechanical proper-
ties were ascertained using the lithology data. Gamma from both the actual
wireline logging run and the mud log data was correlated to validate applica-
bility for other “application” wells which lack similar wireline logs.
Using the geomechanical properties and the mud log data available for
well #1, the entire dataset was pruned such that two separate datasets were
generated. The inputs were expanded using multiple filtering bandwidths
to extract features at different frequency spectrums, which might hold
physical meaning and therefore are valuable in the modeling process. The
inputs were in turn used to develop two separate models for two separate
reservoir (shale) units, namely Zone C and Zone B as discussed earlier and
observed from Figure 5.3. However, based on the well trajectory along the
112 Hydraulic Fracturing and Well Stimulation
lateral, the two sections have varied data density in terms of available sam-
pling points. From the mud logs, while Zone C has approximately 600 data
points, zone B only has approximately 70 data points, which makes results
from the model defined for this zone susceptible to more errors. Figure 5.4
shows sample rock property (Poisson’s Ratio) modeled for these two layers,
and we can observe relatively higher errors for the Lower Marcellus Layer
compared to the target layer due to said mismatch.
For network training, care is necessary to prevent either over-fitting
or non-representative dataset generation. Care is taken in the neural net-
work design with the ratio of number of network nodes to the number of
data samples kept at less than 0.10. Also, segments from each Zone are
combined making sure that they are representative of varying behavioral
aspects of the property being modeled (such as sudden rise or drop in
value). The final network models are chosen based on the minimized error
observed within the network validation process (Figure 5.5).
With the individual geomechanical models as well as the composite
brittleness model ready for use, the next step is to identify the best fuzzy set
definitions corresponding to the rules defined earlier. Figure 5.6 shows the
final identified rules set which corresponds with the best match between
the predicted fracture density models and the observed production log
behavior for Well #1.
Once the model is ready for application, all of the model design param-
eters are stored for later use with application test scenarios. These include
Original
Modeled
(a) 5760 5780 5800 5820 5840 5860 5880 5900 5920
(b) 5760 5780 5800 5820 5840 5860 5880 5900 5920
Original
Modeled
Figure 5.4 (a) Modeled Poisson’s Ratio compared with actual log derived values and
(b) corresponding error mismatch for Zone B and (c) Modeled Poisson’s Ratio compared
with actual log derived values and (d) corresponding error mismatch for Zone C.
Multistage Hydraulic Fracturing Completions 113
1.0 Data
Fit
0.9 Y= T
0.8
Modeled property
0.7
0.6
0.5
0.4
R = 0.9074
0.3
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Target property
Figure 5.5 Validation results from a sample neural network training run showing
minimized error in modeled property (Poisson’s Ratio) compared to target property
(actual log derived data). Note: The target data was normalized before modeling run.
(in)
(a) 0 50 100
in
(in)
(b) 0 50 100
in
Very Low Medium High Very
low high
(in)
(c) 0 50 100
in
Figure 5.6 Optimal fuzzy set definitions based on training well data for (a) modeled
brittleness, (b) mud log gas shows, and (c) output fracture density guidance.
114 Hydraulic Fracturing and Well Stimulation
the number of layers in the network, number of nodes in the hidden, input
and output layers, trained weights associated with each node in the net-
work, activation function s associated with each node, biases within the
network, etc. These saved models are in turn applied to any dataset from
“application” wells to generate hydraulic fracture/cluster density maps. The
final fracture density maps need to be res-called in order to make sure
that sufficient “maximum” and “minimum” fracture spacing is maintained
before final fracture or perforation cluster placement. These maxima and
minima limits can be independently evaluated using other stage design
approaches discussed in the introduction to this study.
5.4 Results
Before the models can be applied to other wells, they need to be catego-
rized based on lithological layers associated with each portion of the well-
bore to be analyzed. This is done by using the same approach as used for
the training well (Well #1) as discussed earlier. Once the wellbore has been
characterized, these segmented subsections are evaluated for rock proper-
ties by using the corresponding rock property predictive artificial neural
net derived models. We will share the design results obtained using the
models for the two application wells (Well #2 and Well #3) and compare
the observations with production log data. This will allow validation of the
observed results using independent production log results, which is critical
as initial production is the key for rapid return on investment in shale gas
wells. We again note that while Well # 2 belongs to the same pad as the
training well (Well #1), Well #3 is located 10’s of miles from the first pad
and incorporates a different completion design.
Before we look into the application wells (Well #2 & Well #3), we apply
the derived models to data from the training well itself (Well #1) using the
segmented modeling approach. Based on the wellbore location in reference to
lithological units (Figure 5.3), separate models are applied to the dataset along
the wellbore. Figure 5.7 shows the results for this particular test case and as
expected, we get a good match between the suggested fracture density derived
from the proposed workflow and the production log data for Well #1.
We note that the original completion for Well #1 involved 18 stages
with 4 perforation clusters per stage and an inter-stage separation of ~280
feet. The modified spacing design as suggested by the workflow is shown
in Figure 5.7(b). As observed from these results, significant mismatch is
observed close to 6900 ft. (measured depth), which could be due to the
wellbore lying either very close to or at the interface of zone B & zone C
Multistage Hydraulic Fracturing Completions 115
Production log
inverse of designed density
%age match: 82.1302 Poor match
Good match
6000 6500 7000 7500 8000 8500 9000 9500 10000 10500
(a) MD(ft.)
5400
TVD(ft.)
5450
5550
6000 6500 7000 7500 8000 8500 9000 9500 10000 10500
(b) MD(ft.)
Designed clusters
# stages: 18 # Clusters/Stage: 4 Confidence - low
Smoothening Window: 40 ° of randomness: 0.01 Confidence - high
Equalspaced separation: 70 feet Rescaling: 0.75
Confidence: 50 Cluster density - rescaled
Cluster density - original
(c) 6000 7000 7500 8000 8500 9000 9500 10000 10500
MD(ft.)
Figure 5.7 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on the
model results for Well #1 incorporating both zone B & zone C models using segmented
modeling approach. Red arrow indicates section with significant mismatch between
production and designed fracture density.
(Figure 5.3). This would make it difficult to interpret which model is the
right model and applicable for corresponding sections of the wellbore.
For Well #2, we ran three tests with the first test incorporating the model
associated with zone C for the entire lateral (Case 2A), the second test incor-
porating the model associated with zone B for the entire lateral (Case 2B),
and the third test incorporating segment wise modeling using both models
based on location of the lateral in relation to the lithological units (Case 2C).
Figure 5.8 shows the modeling and design results for Case 2A, Figure 5.9
shows the results for Case 2B, and Figure 5.10 shows the results for Case 2C.
We note that the original completion profile for Well #2 involved 14
stages with 4 perforation clusters per stage with an inter-stage separa-
tion of ~300 feet. We can clearly see sections along the wellbore where the
design recommendation suggests sparser clusters and other sections which
suggest denser cluster spacing. This correlates well with the production log
results with sections suggesting denser clusters showing lower productivity
and vice versa. This is desirable considering the defined modeling framework
discussed earlier. However, for Case 2A & Case 2B, the predicted fracture
spacing design does not match well with the observed stage-wise productivity
behavior at some locations (identified by red arrow). This can be attributed to
the model applicability issue in certain sections of the wellbore, depending on
116 Hydraulic Fracturing and Well Stimulation
Production log
%age match: 87.2792 inverse of designed density
Poor match
Good match
5450
5500
5550
6000 6500 7000 7500 8000 8500 9000 9500 10000
(b) MD(ft.)
Designed clusters
# stages: 14 # Clusters/Stage: 4 Confidence - low
Smoothening Window: 50 ° of randomness: 0.01 Confidence - high
Equalspaced separation: 78 feet Confidence: 50 Cluster density - rescaled
Rescaling: 0.75 Cluster density - original
(c) 6000 6500 7000 7500 8000 8500 9000 9500 10000
MD(ft.)
Figure 5.8 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on
the model results for Well #2 incorporating model from zone C. Green arrow indicates
section with significant mismatch between production and designed fracture density,
which corresponds with wellbore section falling outside zone C.
Production log
inverse of designed density
%age match: 87.2792 Poor match
Good match
(a) 6000 6500 7000 7500 8000 8500 9000 9500 10000
MD(ft.)
5400
TVD(ft.)
5450
5500
5550
(b) 6000 6500 7000 7500 8000 8500 9000 9500 10000
MD(ft.)
Designed clusters
# stages: 14 # Clusters/Stage: 4 Confidence - low
Smoothening Window: 50 ° of randomness: 0.01 Confidence - high
Equalspaced separation: 78 feet Rescaling: 0.75
Confidence: 50 Cluster density - rescaled
Cluster density - original
(c) 6000 6500 7000 7500 8000 8500 9000 9500 10000
MD(ft.)
Figure 5.9 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on
the model results for Well #2 incorporating model from zone B. Green arrows indicate
sections with significant mismatch between production and designed fracture density,
some of which corresponds with wellbore section falling outside zone B.
Multistage Hydraulic Fracturing Completions 117
Production log
inverse of designed density
%age match: 84.1999 Poor match
Good match
(a) 6000 6500 7000 7500 8000 8500 9000 9500 10000
MD(ft.)
5400
5450
TVD(ft.)
5500
5550
6000 6500 7000 7500 8000 8500 9000 9500 10000
(b) MD(ft.)
Designed clusters
# stages: 14 # Clusters/Stage: 4 Confidence - low
Smoothening Window: 50 ° of randomness: 0.01 Confidence - high
Equalspaced separation: 78 feet Rescaling: 0.75
Confidence: 50 Cluster density - rescaled
Cluster density - original
(c) 6000 6500 7000 7500 8000 8500 9000 9500 10000
MD(ft.)
Figure 5.10 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on the
model results for Well #2 incorporating both zone B & zone C models using segmented
modeling approach. We do not see any significant mismatch between the predicted
fracture density behavior and the productivity of completed perforation clusters.
whether the well track is within the zone defining the applied geomechanical
model or not.
For Case 2C, we observe a much better match along the entire completed
lateral, and this is due to segmented modeling approach where the correct
model (based on the location of the wellbore in reference to the lithological
units) is used (Figure 5.10). This validates the applicability of the proposed
approach for wells within proximity of the well used in training our models.
For Well #3, the completion design was significantly different with 27
stages and 4 perforation clusters per stage with an inter-stage separation of
200 feet. Once again we generate results incorporating the model associ-
ated with zone C for the entire lateral (Case 3A), the second test incorpo-
rating model associated with zone B for the entire lateral (Case 3B) and the
third test incorporating segment wise modeling using both models based
on location of lateral (Case 3C). Since Well #3 is at an offset of 10’s of miles
from Well #1, we expect the results to be not as robust as was the case with
Well #2. Figure 5.11 shows the modeling and design results for Case 3A,
Figure 5.12 shows the results for Case 3B, and Figure 5.13 shows the results
for Case 3C.
For Case 3A & Case 3B, the predicted fracture spacing design does not
match well with the observed stage-wise productivity behavior at sections
118 Hydraulic Fracturing and Well Stimulation
Production log
inverse of designed density
%age match: 84.4068 Poor match
Good match
8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
(a)
MD(ft.)
8200
TVD(ft.)
8300
8400
8500
8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
(b) MD(ft.)
(c)
8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
MD(ft.)
Figure 5.11 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on
the model results for Well #3 incorporating model from zone C. Green arrows indicate
sections with significant mismatch between production and designed fracture density,
some of which corresponds with wellbore section falling outside zone C.
Production log
%age match: 84.6568 inverse of designed density
Poor match
Good match
8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
(a)
MD(ft.)
8200
TVD(ft.)
8300
8400
8500 8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
(b) MD(ft.)
(c) 8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
MD(ft.)
Figure 5.12 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on the
model results for Well #3 incorporating the model from zone B. Green arrows indicate
sections with significant mismatch between production and designed fracture density,
some of which corresponds with wellbore section falling outside zone B.
Multistage Hydraulic Fracturing Completions 119
Production log
inverse of designed density
%age match: 86.7141 Poor match
Good match
(a) 8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
MD(ft.)
8200
TVD(ft.)
8300
8400
8500
8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
(b) MD(ft.)
# stages: 27 # Clusters/Stage: 4 Designed clusters
° of randomness: 0.01 Confidence- low
Smoothening Window: 50 Confidence- high
Equalspaced separation: 51 feet Confidence: 50
Rescaling: 0.75 Cluster density - rescaled
Cluster density - original
(c) 8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500
Figure 5.13 Figure showing (a) the comparison between modeled fracture density and
available production log, (b) TVD behavior along lateral with traditional equally spaced
perforation locations, and (c) the actual fracture spacing recommendation based on the
model results for Well #2 incorporating both zone B & zone C models using segmented
modeling approach. Green arrows indicate sections with significant mismatch between
production and designed fracture density.
highlighted using red arrows. This can be attributed to the model applica-
bility issue in certain sections of the wellbore, depending on whether the
track is within the zone defining the applied geomechanical model or not,
as observed with the earlier test case. Other issues include the robustness
of the model at separation of 10’s of miles as well as issues with inadequate
data for zone B model as highlighted earlier.
For Case 3C (Figure 5.13) using segmented modeling approach, we
observe a much better match along the completed lateral. However once
again, there are small sections of the lateral where the fracture placement
recommendation based on density model does not match well with the
production log observations.
Next we highlight the issue of model robustness due to data inadequacy.
As observed from Figure 5.12, a certain section of the modeled fracture
density along the wellbore shows significantly poor results suggested by the
consistent high values from ~9300 feet to ~11700 feet (measured depth).
This is because the model from zone B is poorly defined due to lack of
adequate data as discussed earlier. While the laterals for both Well #1 and
Well #2 fall mostly within the target zone C, significant sections of the well-
bore corresponding to the identified depth interval for Well #3 fall within
overburden zone B (as observed in Figure 5.13 from ~10300 feet measured
depth to ~11000 feet measured depth). Therefore, these erroneous artifacts
120 Hydraulic Fracturing and Well Stimulation
are observed for both Case 3B and 3C, which makes use of geomechanical
models from zone B.
(a)
(b)
(c)
Figure 5.14 (a) Well schematic showing localized sweet spot due to intersecting natural
fracture swarm (green box) and (b) hydraulic fracture density framework suggested in
this study with (c) potential alternate framework along the wellbore lateral.
Multistage Hydraulic Fracturing Completions 121
work in most situations provided proper care is taken before applying this
approach. However, the proposed approach needs to be validated as it may
not hold under many situations depending on economic considerations.
For future work, we propose to carry out extensive modeling studies and
generate guidelines for applicability under varying scenarios as suggested
in this work.
In the future, we plan on using a fuzzy or probabilistic classifier to decide
which model to be used depending on the closeness of the well track to a
particular lithologic boundary. This is due to significant uncertainty rang-
ing from 10’s to 100’s of feet when it comes to layer boundaries and exact
well location, which can make the decision making on models to be used
for design very non-representative. We hope to test this approach on mul-
tiple wells in other shale plays (Permian Basin). We also expect to conduct
similar design work for multiple wells which are geographically spread out
and validate these observations.
Acknowledgement
This work was supported by Research Partnership to Secure Energy for
America (RPSEA) project number 11122-20. We also acknowledge WPX
Energy for providing access to active hydraulically fractured wells of
opportunity in Marcellus shale play and Schlumberger for acquiring the
wireline and production logs used in this study.
References
1. J.W. Crafton and D. Anderson, Use of extremely high time-resolution pro-
duction data to characterize hydraulic fracture properties, Presented at
the SPE Annual Technical Conference and Exhibition, San Antonio, Texas,
24–27 September, SPE-103591-MS, http://dx.doi.org/10.2118/103591-MS
(2006).
2. P. Ye, L. Chu, I. Harmawan, et al., Beyond linear flow analysis in an
unconventional reservoir. Presented at the SPE Unconventional Resources
Conference, The Woodlands, Texas, 10–12 April, SPE-164543-MS, http://dx.
doi.org/10.2118/164543-MS (2013).
3. K. Wu and J.E. Olson, Investigation of the impact of fracture spacing and
fluid properties for interfering simultaneously or sequentially generated
hydraulic fractures. SPE Production & Operations 28(4), 427–436 (2013).
4. M. Jonathan and J.L. Miskimins, Optimization of hydraulic fracture spacing in
unconventional shales. Presented at the SPE Hydraulic Fracturing Technology
Multistage Hydraulic Fracturing Completions 123
Abstract
In an effort to better understand the well performance in one of the Chevron’s
assets in San Joaquin Valley, a study was conducted to evaluate the perforation
strategies and capture best practices. Well completion through perforation is typ-
ically performed using bare essential technology such as wireline logs and per-
foration guns. For basic reservoir formations, simple rules-of-thumb are used
for perforation spacing and interval lengths. These are rarely validated by other
methods, such as production logging and micro-seismic monitoring. For more
challenging lithology, a more appropriate approach would be to place perforation
clusters in target formations with similar properties. The research presents an effi-
cient use of fuzzy clustering technology for identification of the optimum perfo-
ration strategy in a challenging waterflood diatomite reservoir. The methodology
was applied on all newly drilled wells in the reservoir (within the last two years),
and we found that this new approach improved our understanding over previ-
ous practices, not only by designing optimum perforations, but also an increased
production was observed. Cluster analysis is the task of grouping a set of objects
in such a way that objects in the same group (called a cluster) are more similar to
each other (in some sense or another) than to those in other clusters. There are two
commonly used types of clustering methods: hard and fuzzy clustering. In hard
clustering, data is divided into distinct clusters, where each data element belongs
to exactly one cluster. In fuzzy clustering, data elements can belong to more than
one cluster, and associated with each element is a set of membership levels. The
fuzzy clustering algorithm, also known as Fuzzy C-Mean (FCM) algorithm, was
applied to log data of wells from different areas of a reservoir. Based on the cluster-
ing results, the workflow then identified whether the perforation was performed
125
126 Hydraulic Fracturing and Well Stimulation
6.1 Introduction
The Lost Hills oil field is one of several significant oil fields located in the
San Joaquin basin of central California [1]. The field is located about 45
miles northwest of Bakersfield in Kern County, Figure 6.1. The producing
structure is a northwest-southeast trending double-plunging anticline [2].
The anticline is about 12 miles long and about 1 mile wide with high dip-
ping flanks reaching up to approximatively ~20 degrees.
The main reservoir is relatively shallow, and the producing interval var-
ies from 600 feet to a maximum of 1200 feet along the crest of the anti-
cline. The most productive unit is the Belridge diatomite of the Miocene
6.3 Technology
The well perforation strategy for this field was strongly linked to the number
of completion frac stages planned for the well, and vice-versa. Therefore,
a large number of frac stages, which would provide better frac coverage
across the productive formation, would require an increased number of
perforation sets. However, it was shown that one likely non-uniformity
in well production is that fluid that is injected with intention to gener-
ate fractures from all perforation clusters is diverted to only a portion of
the clusters leaving many clusters unstimulated [4]. Industry experience
shows that there is an opportunity for improvement of recovery efficiency
by stimulating effectively all perforation clusters along the well. Several
approaches to improve stimulation uniformity have been proposed. In
the “engineered completion” approach [4], well log data was used so that
perforation clusters can be placed in regions of similar stress or brittle-
ness index. Cipolla details new algorithms and an integrated workflow that
could improve fracture treatment staging in both vertical and horizontal
wells [5]. In a newer approach, Pierce and Bunger, promoted uniform stim-
ulation of shale gas reservoirs by strategically locating entry points so that
Clustering-Based Optimal Perforation Design 129
n k 2
arg min s wi , j x1 j
i 1 j 1
(6.1)
where:
1
wi , j 2
m 1 (6.2)
k x1 j
l 1 x1 j
Clustering-Based Optimal Perforation Design 131
Area 1
Area 2
Area 3
Figure 6.3 Field map showing the division of the three zones.
Figure 6.4 A snapshot of the sample data showing collected log tracks and perforation
completion.
1. For every half foot, the DRES, GR, and NPHI log data was
input into the FCM algorithm. Also, the measurements that
were less than 1000 feet were disregarded as not being part
of the reservoir.
2. Run the FCM algorithm on xi = [DRESi, GRi, NPHIi], where
i represents each half-foot measurement, on all the available
log data for the well. The number of clusters, k, was set to be
5 in all the experiments.
3. The mean of each of the five clusters was returned from
the FCM algorithm. The clusters were ranked by the DRES
Clustering-Based Optimal Perforation Design 135
log, where the one with the least DRES log was labeled as
the “bad” (shale) region and the one with the largest DRES
was labeled as the “good” (sand) region. An example of the
cluster means returned by the FCM algorithm is shown in
Figure 6.5.
4. The results of the clustering allow for associating each well
with a cluster by depth. The clustering results for two sam-
pled wells are depicted in Figure 6.5.
5. The final step was to calculate the perforation footage in the
cluster label “good.”
–100 –50 0 50 100 150 200 –100 –50 0 50 100 150 200
1000 1000
1200 1200
1400 1400
Depth, ft
Depth, ft
1600 1600
1800 1800
2000 2000
2200 2200
2400 2400
DRES C1 DRES C1
GR C2 GR C2
NPHI C3 NPHI C3
Perforated C4 Perforated C4
C5 C5
Figure 6.5 Clustering results by depth for two sample wells. In this figure, C1 represents
the shale region, and C5 represents the sand region. The actual perforation regions were
also labeled by the dark purple regions.
136 Hydraulic Fracturing and Well Stimulation
column to the far right represents the “cluster” track where every half foot
is assigned to one of the five clusters.
This study was carried out using only fuzzy clustering. Other method-
ologies, such as hard clustering of ants colony optimization, can be applied
to cluster the data.
The procedure presented is explained through the two plots presented
in Figure 6.5 above. The continuous lines represent the DRES, GR, and
NPHI log data every half foot, while the blue “dots” at the right of the
graphs represent the footage perforation existing in the wells. Lastly, the
column to the far right represents the “cluster” track where every half foot
is assigned to one of the five clusters.
It should be noted that current field practices were rather inconsis-
tent and selection of the perforations was accomplished through a visual
inspection of each log separately. Certain logs may show particular signa-
tures in the “good regions”, however, this was not very intuitive. The clus-
tering allowed not only for a better definition of the regions but also for
corroboration with the perforation intervals and production response.
1800
Figure 6.6 3-month peak production vs. total perforation footages for all sampled wells.
1800
Peak 3-month production (bopd)
1600
1400
1200
1000
Zone 1
800
Zone 2
600
Zone 3
400
200
0
0 50 50 100 150
Footage perforated in "good" zones (C5) (ft)
Figure 6.7 3-month peak production vs. total perforation footages on good regions
identified by the FCM algorithm.
Strat well
31.33 min
0.000
0.550
1.100
Figure 6.8 Fracture modeling. Proppant coverage at the end of pumping, perforations at
2680 ft.
Strat well
368.63 min
0.000
0.450
0.900
Figure 6.9 Fracture modeling. Proppant coverage at the end of pumping, perforations at
2750 ft.
6.8 Conclusions
The paper presents a simple, yet efficient methodology to define the “good”
footage regions for potential perforation in complex and heterogeneous
reservoirs using FCM clustering. The study summarizes a step-by-step pro-
cedure that can assist the geologist and completion engineers in designing
the perforation plan. The methodology demonstrates that the perforation
design can be improved if one uses all the log data available for the well
(such as DRES, GR, and NPHI) as well as the additional information defin-
ing the behavior of the logs. Using Artificial Intelligence (AI) techniques
such as fuzzy clustering analysis methods, the well log data is effective in
identifying rock intervals with similar properties.
The clustering techniques demonstrate an unmatched value by exploit-
ing hidden information from multiple attributes and drawing knowledge
to assist in practical engineering applications. The methodology is flexible
and can be applied to any well where complex lithology creates a challenge
in defining the optimum perforation intervals.
Acknowledgements
The authors would like to thank Chevron North America Exploration and
Production for allowing the publication of this work.
References
1. S.A. Graham and L.A. Williams, Tectonic, depositional, and diagenetic
history of Monterey Formation (Miocene), central Sand Joaquin basic,
California. AAPG Bulletin 69, 385–411 (1985).
2. J. Brink and T. Patzek, Lost hills field trial - incorporating new technology
for reservoir management. SPE 77646-MS, SPE Annual Technical Conference
and Exhibition, 29 September-2 October, San Antonio, Texas (2002).
3. M.S. Bruno and C. Bovberg, Reservoir compaction and subsurface subsid-
ence above the Lost Hills Field, California, Proc. 33rd US Symp. On Rock
Mechanics, AA Balkema, Rotterdam, pp. 263–272 (1992).
140 Hydraulic Fracturing and Well Stimulation
Abstract
Recent drilling and hydraulic fracturing activities in the Utica-Point Pleasant shale
play have recorded true vertical depth of over 13,500 feet. Drilling wells at this
depth is very costly and challenging. With the current commodity pricing, drilling
in such conditions becomes unaffordable.
One immediate solution to the current low energy prices is optimizing well
spacing to enhance hydrocarbon recovery and, thus, the commercial feasibility
of the project. Horizontal well spacing constitutes a fundamental parameter for
the success of a shale-drilling venture. Determining the optimum horizontal well
spacing in shale reservoirs represents a challenging task because of the complexity
of controlling factors. These factors can be categorized into three groups: geologi-
cal, engineering, and economic.
Geological modeling and reservoir simulation are the standard tools utilized in
the industry to integrate these controlling factors. In this study, we employed these
tools to perform sensitivity analysis of reservoir characteristics and future produc-
tion optimization for a deep drilling case study in the Utica-Point Pleasant forma-
tions. We sought to find the optimum horizontal well spacing scenario as well as
hydraulic fracturing design, in order to attain the highest net present value for 50
years of gas production. Our reservoir model represented a portion of Utica-Point
Pleasant formations at the depth of 13,000 feet and the dry gas window. A com-
mercial reservoir simulator was coupled with an optimization algorithm to reach
the best solution with a minimum simulation cost. Although the outcome of our
study is subjective to the chosen asset, the workflow provides a good example of
horizontal well spacing and hydraulic fracturing design optimization.
141
142 Hydraulic Fracturing and Well Stimulation
7.1 Introduction
Inter-lateral spacing (well spacing) is one of the biggest decisions in a field
development in unconventional shale plays. Not having optimized well
spacing can either cost the operators a lot of money or leaves a lot of money
on the table. The well spacing must be close enough to help create as much
of the stimulated reservoir volume (SRV) as possible; however, it must be
wide enough to minimize fracture interference [1] (well to well interfer-
ence and “frac hits”) and over-capitalization in a field development [2].
Various tools such as rate transient analysis (RTA) and numerical simula-
tions can be used to optimize well spacing [3, 4]. At the very beginning of
a field development, little data is available; therefore, it is very important to
run sensitivity analysis and uncertainty investigation to assess the effect of
uncertainties involved in reservoir characteristics of the field performance.
Once pressure and production history data are available, the model can be
calibrated to insure all of the assumptions are practically feasible.
Well spacing is impacted by a combination of three types of factors
defining the geological characteristics, the engineering process, and the
economic constraints. The geological and reservoir characteristics assure
the quality of the reservoir and the hydrocarbon presence. The engineering
design provides the deliverability of the well. Finally, the economic factors
warrant that the engineering project leads to profit [5].
The most important parameters that must be heavily studied are matrix
permeability, fracture half-length (impacted by completions design), reser-
voir properties, capital expenditure, operating costs, and hydrocarbon pric-
ing. The hydraulic fracturing enhances the reservoir permeability and leads to
more recovery. The fracture permeability can be upscaled for fluid flow simu-
lation using Oda’s method [6, 7]. In a high commodity pricing environment,
it would make more sense to place the wells closer apart; however, low com-
modity pricing will dictate farther well spacing. A rock that has high matrix
permeability would be better suited for a larger well spacing, but a rock that
has lower matrix permeability is better optimized in a tighter well spacing.
When designing well spacing, it is crucial to take completions design into
account. For example, if 1,000 feet of well spacing is designed for a field devel-
opment, it is very important to design the completions job in a fashion that
would yield 500 feet fracture half-length by pumping enough sand and water,
along with optimizing stage spacing, cluster spacing, and perforations design.
Hydraulic Fracturing Design Optimization 143
Therefore, a big portion of achieving the desired well spacing is creating com-
pletion jobs that would guarantee the attainment of such goals.
Deep dry Utica located in both Pennsylvania and Ohio has sparked the
interest of a lot of the operating companies in the past two years due to its
attracting geology, massive production volumes, and the advantageous time
value of money created in the production of as much volume as possible in
the first 3 to 5 years of the life of the well. The biggest challenges in developing
deep dry Utica have been high capital expenditure, lack of production his-
tory and reservoir rock characteristics, and optimal management of pressure
drawdown to avoid prop-pant crushing/embedment and geo-mechanical
effects. All of these challenges will be resolved and understood with time as
this has been the case for all of the other shale plays developed to date.
In order to create long-term value for the shareholders, the well spacing
must be selected based on a spacing paradigm that yields the highest NPV.
Production volumes could be important to the strategic development of a
company, but the single economic parameter that creates long-term value
for the shareholder of a company is NPV [5]. In this study, the well spacing
was optimized using NPV. Optimal economical well spacing is achieved
by having a solid understanding of fracture half-length, conductivity, and
matrix permeability, which are typically very hard to obtain in unconven-
tional shale plays and reservoir modeling. Therefore, various sensitivities
must be run to understand the impact of these variables.
In this study, we utilized a numerical reservoir simulation in order to
model a dry gas reservoir 13,000 feet deep into a formation that resembles
the Utica-Point Pleasant play. Initially by drilling one well, we studied the
recovery after 50 years of production over a constant drainage area. Next,
we looked for increasing the recovery factor by adding multiple wells.
Finally, we wanted to find the optimum number of wells that leads to the
highest NPV over 50 years of production.
7.2 Methodology
7.2.1 The Base Reservoir Simulation Model
Our reservoir model covers a 1,003 acres area. For the sake of comparison,
we kept the drainage area constant for the different well spacing scenarios.
The drainage area is divided into 35,000 grid blocks, and the grid block size
is 50 feet by 50 feet. The top of the reservoir is 13,000 feet deep, and the pay
zone thickness is 200 feet, allotting into two layers of 100 feet thickness.
These layers resemble the structure of Utica-Point Pleasant play. The reser-
voir contains dry gas, methane.
144 Hydraulic Fracturing and Well Stimulation
13,100
13,090
13,080
13,070
13,060
13,050
13,040
13,030
13,020
13,010
13,000
Figure 7.1 Reservoir top 3D model built in CMG-BUILDER. The drainage area covers
1,003 acres and the reservoir includes 200 feet pay zone divided into two layers.
Hydraulic Fracturing Design Optimization 145
1.00
Krw vs Sw
0.80
Kr - relative permeability
0.60
0.40
0.20
0.00
0.20 0.36 0.52 0.68 0.84 1.00
Sw
0.80
Krg vs Sl
0.64
Kr - relative permeability
0.48
0.32
0.16
0.00
0.20 0.35 0.50 0.65 0.80 0.95
Sl
Figure 7.2 Relative permeability curves for this study. These curves are modified from the
correlations available in the CMG-BUILDER.
curves utilized in this study. These curves are modified from the correla-
tions available in the CMG-BUILDER. The base reservoir model has eight
percent porosity and one micro-Darcy permeability.
The wells are drilled and perforated in the bottom layer of the Point
Pleasant formation. The length of the horizontal lateral is 7,000 feet. In
146 Hydraulic Fracturing and Well Stimulation
Well-S
Well-3
Well-4
Well-E
Well-S
Table 7.3 Five scenarios of inter-lateral spacing and the corresponding well
spacing over a constant drainage area (1,003 acres) in this study.
Scenario Well spacing (ft) #Wells
1 500 8
2 750 6
3 1,000 5
4 1,250 4
5 1,500 3
Figure 7.4 The workflow for the well spacing and completion parameters optimization study.
range that the user has provided earlier. Then the optimization algorithm
calls the simulator, and, in turn, the simulator runs the models and calculates
the NPV values based on the economic analysis parameters. The output of
this workflow is a collection of reservoir models, which allows us to select
the optimum parameters based on the highest field NPV value. In this study,
the whole process of optimization was carried out using CMG-CMOST [8].
10 years
20 years
50 years
5 years
1 year
Figure 7.5 The cumulative gas production, gas flow rate, and pressure distribution for
different time steps (1, 5, 10, 20, and 50 years).
245
245.09
240
235 235.27 240.02
230
225
223.27
220
215
213.73
210
500 750 1000 1250 1500
Well spacing,ft
8 6 5 4 3
Number of wells
Figure 7.6 Field cumulative gas production after 50 years for five drilling scenarios.
100.0 96.0
90.0
80.0 74.3
70.0
500 750 1000 1250 1500
Well spacing, ft
8 6 5 4 3
Number of wells
Figure 7.7 The results of net present value optimization study for five drilling scenarios.
Figure 7.9 compares the capital and operating costs of these projects
with the NPV values. As it is expected, a higher number of wells will lead to
a higher capital cost. However, completion projects typically require a sig-
nificant share of capital cost. In addition, a higher production will increase
the operational cost.
Hydraulic Fracturing Design Optimization 151
66.3% 65.5%
70.0%
59.0% 56.4%
60.0% 54.3%
50.0% 42.7%
40.0%
30.0%
20.0%
10.0%
0.0%
500 750 1000 1250 1500
Well spacing, ft
8 6 5 4 3
Number of wells
Figure 7.8 Field recovery factor (RF) over different time periods for all the drilling scenarios.
96.0
(88)
90 74.3
(66) (67)
60 (53.6)
(40.2)
30
0
500 750 1000 1250 1500
Well spacing (ft)
8 6 5 4 3
Number of wells
Figure 7.9 Comparison of capital and operating costs with NPVs for different drilling scenarios.
of 500 and 750 feet, fracture half-lengths are 250 and 350 feet, respectively.
Obviously, higher values of fracture half-length will cause interference
between the fractures (“frac hit”) and loss of injected fluids. The number of
fracture stages depends on the space between the stages (fracture spacing).
In addition, the cost of completion has a direct relationship with the num-
ber of stages. Therefore, an economic optimization analysis is required. We
could suggest a higher number of stages for a lower number of wells per a
constant drainage area and vice versa.
8 6 5 4 3
Number of wells
80.0% increase
increase
70.0% 63.6% 66.3% 66.2%
59.0% 56.4% 59.0%
60.0% 54.3% 54.3%
50.0% 42.7% 44.7%
40.0%
30.0%
20.0%
10.0%
0.0%
500 750 1000 1250 1500
Well spacing, ft
8 6 5 4 3
Number of wells
Figure 7.11 Field recovery factor comparison after 5 years between two completion costs
scenarios.
5 years of gas production shale reservoirs could bear more than 70% of
ultimate production. The main reason behind the increase of recovery
factor values goes back to having a higher number of completion stages
due to the reduction of cost (Table 7.6). However, for the well spacing
of 500 feet, the even increase of stages from 28 to 35 will not impact the
production level after five years. This might be because of the tight well
spacing that already covers all of the reservoir volume.
7.5 Conclusion
In addition to the reservoir quality and engineering factors, the commod-
ity costs and capital expenditure guide oil and gas development plans. In
this project, we utilized geological modeling and reservoir simulation to
study the profitability of a shale dry gas asset in the Utica-Point Pleasant
formations. Due to the very tight formations, horizontal drilling and multi-
stage hydraulic fracturing were planned to increase the reservoir deliver-
ability. Our reservoir model simulates a portion of Utica-Point Pleasant
formations at 13,000 feet deep and 200 feet thick. The drainage area covers
1,003 acres. Our simulation results showed that due to the tight shale for-
mation, drilling just one well would leave a significant amount of natural
gas behind. We wanted to find the optimum number of lateral wells over
Table 7.6 Optimized completion parameters for the completion cost scenario of $200K per stage.
Acknowledgments
The authors wish to extend their appreciation to Computer Modeling Group
(CMG) for providing the software applications for reservoir simulation.
References
1. A. Awada, M. Santo, D. Lougheed, D. Xu, and C. Virues, Is that interference?
A work flow for identifying and analyzing communication through hydrau-
lic fractures in a multiwell pad. SPE J. (2016). DOI: 10.2118/178509-PA
2. R. Barree, S. Cox, J. Miskimins, J. Gilbert, and M. Conway, Economic opti-
mization of horizontal well completions in unconventional reservoirs. PE
Hydraulic Fracturing Technology Conference, The Woodlands, Texas, USA,
Society of Petroleum Engineers (2014). DOI: 10.2118/168612-MS
3. A.S. Ziarani, C. Chen, A. Cui, D.J. Quirk, and D. Roney, Fracture and wellbore
spacing optimization in multistage fractured horizontal wellbores: Learnings
from our experience on Canadian unconventional resources. International
Hydraulic Fracturing Design Optimization 157
Introduction
The role of discontinuities and the major impact they can have on the out-
come of hydraulic fracturing has been recognized since the early days of
field experiments carried by the Department of Energy and Gas Research
Institute. Warpinski and Teufel [1] described many of these issues and
attempted to model them with the best tools they had at their disposal at
that time. Unfortunately, that initial effort was not continued, and the shale
revolution was built on that initial knowledge and did not expand it sig-
nificantly. With the oil crisis of 2014–2017, followed by financial pressures
imposed by investors in 2018–2019 who were not informed about the risks
of frac hits and well interreferences, the entire industry has to catch up on
years of “natural fracture denial” where the hydraulic fracturing design of
thousands of wells used software and technologies that does not include
any representation of the natural fractures. New evidence from field exper-
iments such as the one conducted at the Hydraulic Fracturing Test Site I
industry consortium (HFTS-1) is again highlighting the importance of the
natural fractures.
The recent findings of HFTS-1 are well summarized in the September
2018 Journal of Petroleum Technology article titled “Real Fractured Rock is
So Complex it’s Time for New Fracturing Models” [2]. 600ft of core taken in
a hydraulically fractured Wolfcamp reservoir in the Permian Basin, USA
showed a more complex reality than what is accounted for in most hydrau-
lic fracturing design and analysis software. This includes the interaction
between hydraulic and natural fractures, which is largely ignored or poorly
accounted for in most software currently used to model hydraulic fractures
and their resulting geometry. Rassenfos [2] emphasized in his summary of
multiple recent publications describing HFTS 1 findings, that the “fracture
height is overrated. While microseismic testing indicated that fractures grew
up about a 1000ft, the height of the propped fractures- the fractures most
likely to produce oil and gas was about 30ft.” Rassenfos summary article
discusses the important role played by the natural fractures.
The impact of natural fractures and discontinuities in general on hydraulic
fracturing can be seen at different scales and manifest itself in different forms.
In this section, four articles attempt to illustrate this impact while showing
the variety of data that can be used to enhance our understanding and val-
idate our new models. The four articles also illustrate the various numerical
approaches which may be used to incorporate the effects of discontinuities.
In Umholtz and Ouenes, the effects of faults and discontinuities is exam-
ined at a regional scale where local changes in stress created by a complex
network of faults could inhibit or promote earthquakes resulting from
stress perturbations created by deep water injection or hydraulic fracturing.
Most of the past work in this area was limited to the conventional study
by seismologists of fault stability which frequently was restricted to one
or a limited number of these faults thus not capturing the true stress field
created by a complex fault network. The authors used the Material Point
Method (MPM) to incorporate a large number of faults over a regional scale
to numerically compute the various stress properties that could be used to
compute an Induced Seismicity Potential (ISP) that was validated using
limited public data from Oklahoma. The use of numerical methods such
as MPM applied to continuum mechanics augmented with discontinuities
could provide the necessary means to capture the effects of the faults and
fractures if adequate data is provided to model these critical discontinuities.
Maity’s section shows the various data that can be used to capture the
effects of natural fractures during hydraulic fracturing. In this article, some
of challenges related to data collection and availability are highlighted.
Maity studied two Marcellus wells that had not only microseismic data
but also production and image logs which provided a quantification of
the natural fractures at different stages. To quantify the performance of
the hydraulic fracturing at each stage, Maity used a modified Nolte-Smith
diagnostic tool to correlate different diagnostic parameters to production
logs, microseismic and fracture density estimated from image logs. The
microseismic data was quantified using the b-value. When plotting the
various diagnostic plots against the production logs, b-value from micro-
seismicity, and natural fracture from images logs, Maity found some weak
correlations when using the production performance. However, stronger
correlations were found between the b-value from microseismic with the
extensional growth parameters compared to natural fracture interaction
related parameters. The b-value showed no correlation with dilatational
fracture swelling related parameters. Lastly, Maity observed no appar-
ent correlation between the derived fracture density and the modeled
Part 161
diagnostic parameters. This was expected since the image logs provide
indication on the near wellbore region and cannot estimate the presence
or absence of actual fractures in the formation. The author emphasized the
multiple challenges associated with each type of data and the pitfalls that
could result from not addressing data quality issues.
Given all the challenges capturing the effects of natural fractures with
the data that could possibly be available, Alzahabi et al. used a more engi-
neering oriented approach to capture the effects of natural fractures and
the resulting interaction with hydraulic fractures is homogenized into a
Stimulated Reservoir Volume (SRV) with a certain average property. Given
a certain SRV distribution estimated with simple geometrical properties
of a fracture geometry, the goal is to find the best well positions using an
optimization problem based on Integer Programming.
Continuing into more computational approaches, Cai et al. use a semi-
analytical approach to predict the well performance in the case of refractur-
ing. The authors use the concept of fracture-matrix cross-flow for a single
fractured wellbore to derive their equations. The semi-analytical approach
assumes some reservoir properties and focuses on the number of hydraulic
fractures and their impact on the well productivity. Testing the approach
against actual performances shows that the approach overestimates the
well performance by 10%. The authors attribute this overestimation to
the uncertainties in the reservoir parameters which will always remain an
issue when using analytical and semi-analytical approaches. The sensitivity
analysis conducted by the authors shows the increase in productivity as the
number of hydraulic fractures increases but the benefits of such an increase
levels out when the fracture conductivity reaches 2000 md-in.
With these four articles, the reader is exposed to the variety of problems
encountered in representing of the natural and hydraulic fractures and
the consequences they have on the well performance. The important take
away, is that multiple approaches exist and are in continuous development
to create a toolbox that engineers can use to solve their many hydraulic
fracturing challenges.
References
1. Warpinski, N.R. and Teufel, L.W., Influence of geologic discontinuities on
hydraulic fracture propagation. J. Petroleum Technol., 39, 02, pp.209-220,
1987.
2. Rassenfos, S., Real Fractured Rock is so Complex it’s Time for New Fracturing
Models, Journal of Petroleum Technology. 19 September 2018.
8
Geomechanical Modeling of Fault Systems
Using the Material Point Method –
Application to the Estimation of
Induced Seismicity Potential to Bolster
Hydraulic Fracturing Social License
Nicholas M. Umholtz and Ahmed Ouenes*
Abstract
In an effort to promote awareness and understanding of the phenomena of induced
seismicity, geomechanical modeling is applied to large publicly available datasets
to demons trate the potential for bolstering social license. The Material Point
Method (MPM) is used to simulate the interaction of fault systems with regional
stresses. By combining mechanical results of the simulation to create induced seis-
micity potential (ISP) proxies, maps are generated to express areas of high and
low inducement potential of seismic events. The results are compared to recent
earthquake epicenter and injection well data. High coincidence of earthquake epi-
centers with regions of predicted high induced seismicity potential suggests the
workflow presented could be deployed to quantify the risk of induced seismicity
associated with the location of high-volume injection wells. The addition of a tool
to assess the impact of location, not only injection volumes, is another critical
step towards responsible regulation of injection wells, and mitigation of induced
seismic events.
163
164 Hydraulic Fracturing and Well Stimulation
8.1 Introduction
Recent media coverage has highlighted a major shortcoming in the indus-
try of unconventional resource production, inducement of seismic events.
Seismicity in certain areas of the midcontinent North America has increased
drastically in the past few years. This has come to be understood as a result
of pore pressure, and stress field perturbation due to water injection. Water
disposal and hydraulic fracturing practices both introduce quantities of
water to the subsurface which may alter the mechanical quasi-equilibrium
established through geologic time. Such rapid perturbation of established
stress fields can in some cases cause earthquakes. Because many of these
operations are based in historically seismically quiescent regions, even
magnitude 3–5 earthquakes may cause significant damage. The public’s
perception and media’s coverage of this phenomena has produced a signif-
icant discussion topic, which impedes the rise of Shale 2.0, which will seek
to achieve a more efficient development of shale resources. It is clear that
operators do not understand the mechanical impact of such injections, as
unpredicted seismic events have been induced at an increasing rate and of
increasing magnitude. Regulators in Oklahoma have taken steps to reduce
the risk of operators inducing seismic events, however the lack of means to
quantify induced seismicity potential has hindered deployment of effective
policy. The new tougher regulations imposed in Oklahoma are limited to
reducing injection volumes and shutting down a limited number of water
disposal wells around areas affected by recent earthquakes. In other words,
the regulators are mostly reacting to the increasing earthquakes rather than
proactively regulating the disposal process. With the advent of induced
seismicity, institutionalized trust is compromised once again, and this time
during a serious shift in the economies of unconventional production,
underpinning not just the importance of the social license to operate, but
the effect such public support can have when attempting to deploy the vast
infrastructure necessary to realize unconventional petroleum resources at
half of the value which they were previously produced.
In order to address these concerns, a workflow is proposed, to com-
bine the Material Point Method (MPM) and continuous fracture modeling
(CFM) technologies with large geologically and geophysically constrained
datasets. This workflow is applied to investigate the relationship between
complex regional fault systems, and their collective impact on stress fields
in a region. The mechanical outputs of this modeling workflow can be
combined into proxies, and used to investigate areas of the model with
stress fields more likely to be perturbed through high-volume injections
Geomechanical Modeling of Fault Systems 165
Trust
Psychological Institutionalized
boundary
identification trust
Credibility
Approval boundary
Socio-political Interactional
Acceptance legitimacy trust
Withheld/ Legitimacy
withdrawn boundary Economic
legitimacy
Figure 8.1 Diagrams showing the importance of credibility and trust in promoting
social license. Critical to the highest level of public support, institutionalized trust, is
sociopolitical legitimacy, and a transition from credible operation to trusted operations.
Modified from Thomson and Boutilier [1].
166 Hydraulic Fracturing and Well Stimulation
Figure 8.2 Fault map of Oklahoma [12]. These fault data are used as input to the
geomechanical models simulated in this study.
Figure 8.3 Available earthquake and Underground Injection Control (UIC) well data
from 2006-Present [13, 14]. Note the lack of earthquakes in multiple areas where large
volumes were injected.
168 Hydraulic Fracturing and Well Stimulation
(c)
(a)
(b)
(d)
Figure 8.4 (a) Extent of the Montney/Duvernay shale play and considered area of
interest, (b) Regional faults mapped by the Alberta Geologic Survey, (c) Faults in the
study area between the cities of Fox Creek and Saint John where earthquakes of 4.2 and
4.5 magnitude, respectively attributed by Alberta Geologic Survey to hydraulic fracturing,
were recorded in 2015, (d) major faults extracted for the geomechanical simulation and
subjected to the regional stress.
Cf = τβ + μ(σβ + ρ)
where τβ is the shear stress on the fracture plane, σβ is the normal stress on
the fracture plane, p is the pore pressure, and μ is the coefficient of friction
along the fracture plane [7].
In these models, the Coulomb stress change, which is the difference
between the Coulomb stress distribution before and after a major fault slip,
highlights regions of stress perturbation, in many cases associated with
subsequent seismicity or aftershocks. To better understand these complex
stress fields and their relation to earthquakes on a field or regional scale,
a new geomechanical workflow will be used and compared to previous
results.
170 Hydraulic Fracturing and Well Stimulation
useful for solving solid mechanics problems. Nairn [10] extended MPM
to handle explicit cracks, resulting in the CRAMP method. By imposing
discrete discontinuities in the model, CRAMP is able to simulate fractured
materials using elastic fracture mechanics.
The major inputs for MPM simulations applied to a study area are four-
fold: 1) the distribution of rock geomechanical properties, such as Young’s
Modulus, in the study area, 2) the distribution and properties of the discon-
tinuities such as natural fractures or faults, 3) the distribution of the pore
pressure, and 4) the boundary conditions representing the far field stresses
acting on the study area. The MPM formulation is used to solve the clas-
sical dynamic continuum mechanics equations that include the momen-
tum equation. The results of the MPM simulation is the usual stress and
strain at the end of the dynamic simulation when the study area reached a
quasi-equilibrium. More details on MPM and its use in the geomechani-
cal modeling of multiple fractures and faults can be found in Aimene and
Ouenes [3].
The MPM geomechanical computation was used to investigate a variety
of geomechanical and engineering issues related to unconventional petro-
leum production at the well scale [3, 4]. Through several field validations,
at the well and pad scales, this workflow has been demonstrated to success-
fully capture local stress variations, and predict complicated microseismic
distribution in a number of stimulated wells. In this study, the workflow
is now deployed on a regional scale inquiry to capture the interaction
between large, complex, regional fault systems, and regional stresses. At
the regional scale, the distribution of elastic properties and pore pressure
used in the MPM simulation are considered constant. The discontinuities
are represented by the interpreted regional faults. To test the validity of
these assumptions at the regional scale, the results from MPM simulations
are compared to previous studies conducted by seismologists studying the
distribution of earthquakes near a fault.
A simple case study using a boundary element method, from King
and Cocco [7], is reproduced using MPM. The model presented by King
and Cocco [7] plots a Coulomb stress change (Figure 8.5a), as a result of
movement of the fault in the model. An approximation of Coulomb stress,
captured by outputs of the MPM model, is compared to the distribution
of stress in the original model, and is strikingly similar (Figure 8.5a,b).
A proxy for Coulombs Stress (PCS) (Figure 8.5b) is defined as:
Figure 8.5 (a) Coulombs stress change as computed by King & Cocco [7], (b) An
approximation of Coulomb stress computed in this work using MPM, (c) An Induced
Seismicity Potential, ISP1, computed in this work using MPM.
where τ13 is shear stress, σ11 is the maximum principle stress at the par-
ticle in the simulation, σ33 is the minimum principle stress at a particle,
σH is the input maximum horizontal stress, and σh is the minimum hor-
izontal stress input to the model. This quantity captures a particle’s sub-
jection to shear stress and stress field perturbation due to the presence of
a fracture. Plotting the proxy highlights areas near the fault that display
an asymmetric distribution of aftershocks proximal to and oriented along
the fault (approximately N-S), with quiescent regions surrounding the ele-
vated occurrence of aftershocks, and with a secondary concentration of
aftershocks distributed E-W further from the simulated fault (Figure 8.5c).
Advances in processing power, and using MPM to simulate fracture
mechanics, is desirable when investigating earthquakes as it allows for
more realistic and comprehensive inputs to be considered by the model.
Geomechanical Modeling of Fault Systems 173
Simulation 2
Study area
Simulation 1
EFM
EFM
Figure 8.6 Equivalent Fracture Models (EFM) derived from an area of interest in
Oklahoma for input into geomechanical models.
dv du
ISP ( H max h max )
dx dy
ISP
High
Low
Figure 8.7 Result of the geomechanical proxy for ISP. Compared to historical earthquakes
and underground injection control well data.
ISP
High
Low ISP regions Prague fault: high ISP
Large injection volumes Many earthquakes
Limited earthquakes Limited injec
Low
Figure 8.8 Zoom of Figure 8.7 that includes to the east the Prague fault, Oklahoma City
and very large injection volumes in areas with no earthquakes. Notice that most of the
earthquakes occur in high ISP areas and large volumes are injected north and west of the
Prague fault in low ISP areas without causing any earthquake in their vicinity. Size of blue
circles represent injected volumes. Size of red stars represent magnitude of earthquakes.
0 60
km
Figure 8.9 Zoom of Figure 8.8.3 to show seismicity NE of interpreted fault in Major
County.
Geomechanical Modeling of Fault Systems 177
High
Low
(a) (b)
Figure 8.10 (a) Equivalent Fracture Model derived from the major fault map of Figure
8.2d (b) Resulting geomechanical output showing the complex distribution of the areas of
high potential induced seismicity.
ISP
High
Low
Figure 8.11 Derived ISP maps plotted along with injection volumes and earthquake data
over a large area of the state of Oklahoma.
178 Hydraulic Fracturing and Well Stimulation
High
Low
Figure 8.12 Derived ISP maps plotted in an area of Alberta where multiple large
magnitude earthquakes attributed to hydraulic fracturing were recorded in Fox Creek
and Fort St. John in 2015.
Acknowledgements
The authors thank Arman Khodabakhshnejad for his help with the MPM
simulation of the one fault case shown in Figure 8.6.
References
1. I. Thompson and R.G. Boutilier, Modeling and measuring the social license
to operate: Fruits of a dialogue between theory and practice, http://social-
icense.com/publications/Modelling%20and%20Measuring%20the%20
SLO.pdf (2011).
2. T.H. Darrah, A. Vengosh, R.B. Jackson, N.R. Warner, and R.J. Poreda,
Noble Gases Identify the Mechanisms of Fugitive Gas Contamination
180 Hydraulic Fracturing and Well Stimulation
Abstract
Hydraulic fracturing of shales and other tight formations has gained tremendous
prominence over the past decade or so, and is set to grow in importance with
increasing global energy demands. In recent years, new methods, which develop
upon older techniques, have been proposed for the interpretation of multi-stage
hydraulic fracturing data. A major issue has been validating these methods for
wider applicability within the industry, considering some limitations that these
techniques pose as a result of certain inherent assumptions.
In this paper we take up a recently proposed diagnostic technique and analyze
completion data from multiple shale gas wells. We correlate the results with avail-
able microseismic monitoring data and validate the interpretations made using
this approach. Using actual stage-wise production contributions information from
production logs, we also highlight the limitations of the current approach when it
comes to predicting overall stage-wise completion quality.
9.1 Introduction
Analyzing pressure data as a means to understand treatment behavior has
been a part and parcel of the oil and gas industry for many decades now.
Pressure data can help identify fracturing behavior, proppant transport
issues, screen out situations, limited entry calculations, etc., to name just a
Email: Debotyam.Maity@gastechnology.org
181
182 Hydraulic Fracturing and Well Stimulation
few of its uses. However, with the increasing use of multi-stage hydraulic
fracturing treatments, many of these traditional techniques no longer hold
validity, and newer techniques are considered desirable for improved treat-
ment diagnostics. In this paper we consider a recently proposed diagnostic
technique, which builds upon a very popular real time fracturing analysis
technique, and demonstrate the utility as well as pitfalls associated with
using such techniques.
9.2 Method
9.2.1 Pressure Data Analysis
Simple fracture models were proposed by Perkins and Kern [1] who
suggested that the fracturing pressure at the wellbore should be a power
function of treatment time (Eq. 9.1) with a large value of the exponent
indicating better fluid containment within the developing fracture.
p(t) te (9.1)
This and other work by Nordgren [2] formed the basis for the real time
completion analysis technique as proposed by Nolte and Smith [3]. They
concluded that fracture propagation may follow one of four predefined
modes based on the slope of net pressure plotted against time as shown
in Figure 9.1. The basic power law equation guiding the propagation of
hydraulic fractures is given as follows:
p e
t ti eC t ti (9.4)
t
Correlating Pressure with Microseismic 183
p
t ti e p pi (9.5)
t
dp / dt
et (9.6)
pn pi / tn ti
pn pi
ct et (9.7)
tn ti
tn
1
E et dt (9.8)
tn ti
ti
tn
1
C ct dt (9.9)
tn ti
ti
184 Hydraulic Fracturing and Well Stimulation
III
Net pressure
II
I
IV
Time
Figure 9.1 Nolte-Smith Interpretation Guide (log Pnet vs. log T). Mode I indicates
constrained height with unrestricted extensional growth; Mode II indicates Stable growth;
Mode III indicates restricted extension or screenout conditions; Mode IV indicates
unstable fracture height growth.
Next we can estimate BHP by using the E and C values and identify
the error from the actual BHP value available either through downhole
pressure instrumentation or surface pressure data corrected for downhole
conditions.
x 10 4
1.202 Calc. BHP
Modeled BHP 15
1.2
Δ Pressure (psi)
Pressure (psi)
1.198 10
1.196
5
1.194
0
64 66 68 62 64 66 68
(a) Time (minutes) (b) Time (minutes)
Figure 9.2 (a) Comparison of calculated BHP and estimated BHP (Eq. 9.10) and (b)
observed ΔP mismatch for a short window extracted from completion data for a sample
hydraulic fracturing stage.
0.5
Exponent, e
-0.5
-1
62 63 64 65 66 67 68 69
Time (minutes)
Figure 9.3 Modeled modified Nolte-Smith exponent ‘e’. The region highlighted from e =
0.75 to 1 indicates restricted flow and fracture dilation, the region highlighted from e = 0.1
to 0.3 indicates unrestricted fracture extension and from e = -1 to -0.1 indicates natural
fracture interactions or rapid height growth.
by defining certain diagnostic parameters computed for each stage. For the
three modes (I, III and IV) identified, we calculate two separate diagnostic
parameters as defined below:
tend
tend
Here δ represents the Kronecker delta, tend represents the end of evalu-
ation period, and tS is the set with the time stamps that correspond with
the value of exponent ‘e’ falling within the three modes in question. Finally,
Δt is the time step in seconds. The reason for using the above definitions
for the diagnostic parameters DP1-S and DP2-S was that the extent that the
exponent ‘e’ stays within each modal region should provide an indication
of how much extensional growth, dilatational growth as well as interaction
with natural fractures occurs during completion.
N = No. earthquakes M
102 b=1
100
6 7 8 9
Magnitude (M)
9.3 Data
We apply this technique on two separate wells from the Appalachian Basin
(Marcellus shale gas play). Both the wells were completed as essentially dry
gas wells with reasonable water cuts. The choice was based on the fact that
both these wells involved microseismic monitoring during completion,
188 Hydraulic Fracturing and Well Stimulation
9.4 Results
As stated earlier, we apply both these techniques, namely modified Nolte-
Smith approach, to fracture diagnostics as well as stage wise b-value
mapping on the data from the two wells under study. For both wells, we
generate the diagnostic parameters (DP1-S and DP2-S) for each of the frac-
turing modes under study defined by subscript S (Mode I, III and IV).
Based on the identified parameter values, we cross-correlate the same with
available production log data (fractional gas flow) from post completion
production logging runs. Figure 9.5 shows how the modeled diagnostic
parameter results for Well # 1 correlate with observed stage-wise produc-
tivity from production logs. We notice weak positive correlation for both
mode I and mode IV results but no correlation with mode III results.
Figure 9.6 shows the same parameters evaluated for Well # 2 under
study and their correlation with stage-wise productivity.
Once again, we observe weak positive correlation for both the mode
I and mode IV parameters and no correlation whatsoever with mode III
results. Since observations from both sets of analysis seem to validate each
other, we can argue that this observation is non-unique and should hold
for more completions. We do note that the correlations are weakly posi-
tive and may not signify much. However, in essence, these observations
indicate that the productivity from any hydraulically fractured stage shows
slight correlation with the extent of extensional fracture growth taking
place during the treatment as well as any interaction with natural frac-
tures observed during treatment. Moreover, we can clearly state that the
degree of dilatational fracture growth has no impact on the productivity
of the completed zone. These observations are also intuitive as we would
Correlating Pressure with Microseismic 189
140 600
120 500
100 400
80
DP 2-l
300
DP 1-l
60
200
40
100
20 y = 353.25x + 73.603 y = 1502.6x + 286.31
R2 = 0.3066 R2 = 0.3206
0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
(a) Fractional gas (b) Fractional gas
1600 1800
1400 1600
1200 1400
1200
1000
DP 1-lII
DP 2-lII
1000
800
800
600
600
400 400
200 y = 397.23x + 905.92 200 y = 478.99x + 989.48
R2= 0.0032 R2 = 0.0041
0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
(c) Fractional gas (d) Fractional gas
0 1400
0 0.05 0.1 0.15
y = −1178.8x − 318.41 1200
−100
R2 = 0.2997
1000
−200
DP 2-lV
800
DP1-lV
−300 600
−400 400
Figure 9.5 Production data (fractional gas flow) compared with extensional diagnostic
parameters (a) (DP1-I), (b) DP2-I, dilatational diagnostic parameters (c) DP1-III, (d) DP2-III,
natural fracture interaction diagnostic parameters (e) DP1-IV and (f) DP2-IV for Well # 1.
190 Hydraulic Fracturing and Well Stimulation
180 700
160 600
140
500
120
400
DP1-l
DP2-l
100
80 300
60
200
40
y = 2.9381x + 81.959 100 y = 11.798x + 324.91
20
R2 = 0.4891 R2 = 0.4554
0 0
0 10 20 30 0 10 20 30
(a) Fractional gas (b) Fractional gas
1800 2000
1600 1800
1400 1600
1200 1400
1200
DP2-lII
DP2-lII
1000
1000
800
800
600
600
400 400
200 y = 5.4387x + 1058.1 y = 3.4788x + 1089.9
200
R2 = 0.0122 R2 = 0.0041
0 0
0 10 20 30 0 10 20 30
(c) Fractional gas Fractional gas
(d)
0 1000
0 10 20 30 900
−50
y = −7.1587x −225.18 800
−100 R2 = 0.3822
700
−150
600
DP2-lV
−200 500
DP1-lV
−250 400
300
−300
200
−350 y = 13.746x + 572.19
100
R2 = 0.4271
−400 0
0 10 20 30
−450
Fractional gas Fractional gas
(e) (f)
Figure 9.6 Production data (fractional gas flow) compared with extensional diagnostic
parameters (a) (DP1-I), (b) DP2-I, dilatational diagnostic parameters (c) DP1-III, (d) DP2-III,
natural fracture interaction diagnostic parameters (e) DP1-IV and (f) DP2-IV for Well # 2.
Correlating Pressure with Microseismic 191
not expect fractures ballooning due to fluid fill-up to have any impact on
productivity as any far field fracturing due to stress perturbation may not
be contributing due to them being spatially isolated. Extensional growth
and growth into natural fractures, on the other hand, should lead to more
productivity due to higher fractured area through the connected hydrauli-
cally created as well as natural fractures.
Next we look at the b-value analysis results and how they correlate with
the modeled parameters shared above. Careful selection was made from
the microseismic event catalogs to make sure that the analysis was valid,
including removing outliers. Those stages with very low event count were
not considered in this analysis (stages 1 through 5 for well # 1). Finally,
fitting to identify slope was done using maximum likelihood technique as
mentioned earlier. We understand that higher b values (and consequently
higher fractal dimensions) are indicative of more complex fractured net-
work or values higher than 1, indicating extensional fracture growth [18].
In our analysis, for all stages studied, we found b-values close to or higher
than 1. Any b-value close to or higher than 2 could be a result of microseis-
mic data quality or could indicate fluid-rich completions. Also, as observed
from Figures 9.7 and 9.8, b-value shows a much stronger correlation with
extensional growth parameters compared to natural fracture interaction
related parameters. Finally, b-value shows no correlation with dilatational
fracture swelling related parameters.
This is again expected as extensional growth through fracture propa-
gation should create extensive three dimensional microseismicity and so
should interaction with natural fractures. However, we note that with fluid
flling up dilating fractures, seismicity will be limited in size (small shear
tip or far field failures). Next, we look at correlation with OBMI log in
Figure 9.9.
As observed from Figure 9.9, we can clearly see no apparent cor-
relation between the derived fracture density from OBMI logs and the
modeled diagnostic parameters. This is expected since the OBMI logs
provide a snapshot of wellbore or near wellbore fracturing and cannot
estimate the presence or absence of actual fractures in the formation.
Eyeballing the modeling results using the pressure data can provide
indicators to make judgement calls by identifying stages with significant
interactions with natural fractures during treatment. Figure 9.10 shows
sample stages from Well # 1 showing the modeled exponent. We validate
the observations by comparing the corresponding b-values as shown in
Figure 9.11.
As observed from results for the two stages shared in Figure 9.10, the
stage corresponding to Figure 9.10a (stage ‘A’) shows a relatively lower
192 Hydraulic Fracturing and Well Stimulation
3.5 3.5
3 3
2.5 2.5
b-value
b-value
2 2
1.5 1.5
1 1
3.5 3.5
3 3
2.5 2.5
b-value
b-value
2 2
1.5 1.5
1 1
3.5 3.5
3 3
2.5 2.5
b-value
b-value
2 2
1.5 1.5
1 1
3 3
2.5 2.5
2 2
b-value
b-value
1.5 1.5
1 1
2.5 2.5
2 2
b-value
b-value
1.5 1.5
1 1
1.5 500
400
1 300
200
y = −0.0022x + 1.421 0.5 y = 364.95x −68.817
100
R2 = 0.4581 R2 = 0.4384
0 0
−600 −400 −200 0 0 1 2 3
DP1-lV DP2-lV
(e) (f)
100 100
80 80
OBMI
OBMI
60 60
40 40
20 20
0 0
0 50 100 150 0 200 400 600
(a) DP1-l (b) DP2-l
140 y = 0.0271x + 14.913 140 y = 0.0249x + 14.756
R2 = 0.0268 R2 = 0.0262
120 120
100 100
80 80
OBMI
OBMI
60 60
40 40
20 20
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
DP1-lII (d) DP2-lII
(c)
y = 0.0865x + 70.236 140 140 y = −0.0443x + 66.422
R2 = 0.025 R2 = 0.0199
120 120
100 100
80 80
OBMI
OBMI
60 60
40 40
20 20
0 0
−600 −400 −200 0 0 500 1000
(e) DP1-lV (f) DP2-lV
Figure 9.9 OBMI log derived fracture density compared with extensional diagnostic
parameters (a) (DP1-I), (b) DP2-I, dilatational diagnostic parameters (c) DP1-III, (d) DP2-III,
natural fracture interaction diagnostic parameters (e) DP1-IV and (f) DP2-IV for Well # 1.
Correlating Pressure with Microseismic 195
0
e
-1
20 25 30 35 40 45 50 55 60 65
(a) Time (minutes)
1
0
e
-1
25 30 35 40 45 50 55 60 65
(b) Time (minutes)
Figure 9.10 Modeled exponent ‘e’ for two stage treatments during proppant injection
phase. Subplot (a) shows lower productivity for stage ‘A’ compared to subplot (b) showing
productivity for stage ‘B’ from production log data.
103
102
Cumulative number
Cumulative number
102
101
101
Figure 9.11 Results from b-value analysis for (a) stage ‘A’ and (b) stage ‘B’ using
Maximum Likelihood approach.
9.5 Conclusions
Novel completion diagnostic techniques such as those applied in this study
provide valuable tools, which can be very useful in helping people under-
stand the behavior of long lateral multi-stage hydraulically fractured wells.
Judicious selection of data, analysis methodology and a careful consider-
ation of potential pitfalls are also necessary in order to add value to any
such diagnostic workfow. In this study, we have demonstrated two ways
of identifying potential fracture growth mechanisms available today and
have tried to correlate the two to highlight a reasonable match between
the results as per our observations. However, a more careful analysis of
Correlating Pressure with Microseismic 197
methods as well as better models are deemed necessary before any such
technique can find widespread use within the fracking industry. This is
because of a high degree of uncertainty and the non-uniqueness possible
in the interpretations.
9.6 Acknowledgements
This work was supported by RPSEA project number 11122-20. I also
acknowledge WPX Energy for providing access to active hydraulically frac-
tured wells of opportunity in the Marcellus shale play, and Schlumberger
for acquiring the production log data used in this study. I would also like to
acknowledge valuable contributions and constructive suggestions by Iraj
Salehi and Jordan Ciezobka from the Gas Technology Institute.
References
1. T.K. Perkins and L.R. Kern, Widths of hydraulic fractures. J. Petr. Tech. 13,
937–949 (1961).
2. R.P. Nordgren, Propagation of a vertical hydraulic fracture. Soc. Petrol. Eng.
J. 253, 306– 314 (1972).
3. K.G. Nolte and M. Smith, Interpretation of fracturing pressures. J. Petr. Tech.
33, 1767– 1775 (1981).
4. E. Pirayesh, M.Y. Soliman, and M. Rafiee. Make decision on the fly: A new
method to interpret pressure-time data during fracturing – application to
frac pack. SPE Annual Technical Conference and Exhibition (2013).
5. M.Y. Soliman, M. Wigwe, A. Alzahabi, and E. Pirayesh, Analysis of fractur-
ing Pressure data in heterogeneous shale formations. Hydraulic Fracturing J.
1(2), 8–13 (2014).
6. K.G. Nolte, Determination of fracture parameters from fracturing pressure
decline, SPE Annual Technical Conference and Exhibition (1979).
7. N.R. Warpinski, S.L. Wolhart and C.A. Wright, Analysis and prediction of
microseismicity induced by hydraulic fracturing. SPE J. 9(1), 24–33 (2004).
8. M.F. Kanninen and C.H. Popelar: Advanced Fracture Mechanics, pp. 563,
Oxford University Press, New York (1985).
9. B. Gutenberg and C.F. Richter: Seismicity of the Earth and Associated
Phenomena, pp. 16–25, Princeton University Press, Princeton, New Jersey
(1949).
10. A.M. Dziewonski, T.A. Chou, and J.H. Woodhouse, Determination of earth-
quake source parameters from waveform data for studies of global and
regional seismicity. J. Geophys. Res. 86, 2825–2852 (1981).
198 Hydraulic Fracturing and Well Stimulation
Abstract
This paper introduces a grid-based fracture-stimulated reservoir volume (SRV)
concept. SRV is defined as a volume occupied by fluid in a fracture, whether cre-
ated or caused by intersection with natural fractures.
Fracturing of the optimum zones is believed to contribute to higher hydrocar-
bon production from shale and tight formations. This requires choosing placement
of fractures along the designed path of horizontal wells to maximize expected SRV.
Additionally, a new linear programming-based approach to mathematically
optimize the placement of SRV in shale reservoirs is presented. The approach may
be useful in pad drilling and fracturing as well as development of applications for
use with shale formations.
This work aims to globally optimize the placement of surface well pads, the
location and number of wells attached to the pads, and the location of the frac-
tures throughout the wells. This optimum placement will also take into account
numerous practical constraints, including the length of wells, the number of wells
associated with a pad, numerous overlap constraints inherent in unconventional
gas and oil well development, the spacing between wells and fractures, etc.
199
200 Hydraulic Fracturing and Well Stimulation
10.1 Introduction
Simulated reservoir volume (SRV) has a long history of use in defining
the effect of fracturing in shales. Substantial evidence from sonic logs and
production data from shale wells shows that certain segments of the wells
make up 70% of the total production of wells. This paper presents a con-
cept for identifying SRV in shale rock. Creating hydraulic fractures leads to
fracture network growth. Fracture growth interaction with existing natural
fractures causes complexity. Complexity is a resultant network of induced
and existing fractures. SRV is used to account for this resultant complexity.
These complex networks have a substantial impact on well performance in
shale and tight rocks. The shape of SRV can be predicted from stimulated
and shear propped fractures, while the volume can be correlated with frac-
ture network length. Britt et al. [1] & Cipollaet et al. [2] discussed geome-
chanics of a shale prospective and fracture complexity.
The size of the SRV is correlated to treatment volume based on micro-
seismic measurement. Figure 10.1 shows the relationship between treat-
ment volume and fracture network length for five vertical Barnett shale
wells, modified after Fisher et al. [3, 4].
Mayerhofer et al. (2008) introduced the SRV concept as a 3D size of
created fracture network, and defined SRV as a correlation parameter for
well performance. Mayerhofer et al. [5] linked SRV with well performance
of shale reservoirs. A direct relationship is demonstrable between fracture
network length and SRV, as shown in Figure 10.1 for Barnett shale wells
(modified after Fisher et al. [3]).
Anderson et al. [6] defined SRV linked to the horizontal well by stimu-
lated reservoir width, areal extent and fracture half-length. Zhou et al. [7]
introduced a method for identifying anisotropic regions in unconventional
hydrocarbon reservoirs. Anisotropy can be indicative of sweet spot zones
for fracturing and for drilling a productive well. Seismic amplitude data
from receivers along two orthogonal lines radiating from a seismic source
Fracture Stimulated Reservoir Volume Mapping 201
30000
Fracture network length = –4E–05 Vol.2 + 2.108 Vol. + + 1920.2
R2 = 0.9839
25000
Fracture network length, ft.
20000
15000
10000
5000
0
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
Fluid volume bbl
Figure 10.1 Fracture network length as a function of fluid volume injected for five
vertical wells of Barnett (modified after Fisher et al. [3]).
where Hf is the fracture height; SRA is the stimulated reservoir area; Hp and
is the reservoir thickness. One limitation is implicit: it seems to be only
valid when hp = hf.
Cheng et al. [10] presented an SRV formula as a function of average
fracture length, average height of fractures, maximum number of fractures,
maximum horizontal stress and minimum horizontal stress as follows:
H
nc
2 h
SRV 4 x hi i 2
(10.2)
1 H
1
h
202 Hydraulic Fracturing and Well Stimulation
where
xi = Average length of the fractures in the ith cluster
hi = Average height of the fractures in the ith cluster, m
nc = Number of clusters
σH = Maximum horizontal in situ stress, N/m2
σh = Minimum horizontal in situ stress, N/m2
Store model
End
Collect user-designed
optimization parameters
End
(a) (b)
Figure 10.5 Four different SRV’s are located, whether staggered, or staggered and
overlapping designs. (a) Staggered design. (b) Staggered and overlapping design.
10.3.1 Methodology
The following describes the numerical formulation of the newly developed
technology.
10.3.3.1 Sets
1. Amn, total FI values unlocked by the stage from node m to
node n
2. Fk, total FI values unlocked by fracturing at node k
3. Lmn, length of stage from node m to node n
4. EI(n), edges inbound to node n
5. EO(n), edges outbound from node n
6. S(n), set of stages that connect to or pass through node n,
represented as edges in the network
7. Ψ(n), set of starting nodes mutually exclusive with w
8. E, set of valid edges for the network
9. Λ(m, n), set of fractures accessible from the stage between
node m and node n
10. Φ(n), set of nodes whose fracture SRV would intersect the
fracture SRV of node n
11. Ω(n), set of stages intersected or interfered with by fractur-
ing node n
10.3.3.2 Variables
1. Pi,j,k, total FI values unlocked by fracturing or placing well at
node i, j, and k
10.3.3.6 Constraints
The problem is subject to the following constraints:
SRV 2 SRV 1
SRV 3
Figure 10.6 Drainage volume for different SRV for different fracture stages branched out
from a single horizontal well.
Figure 10.7 One Horizontal well passes through four different SRV’s. Model domain
shows the molded portion of our reservoir along one horizontal well of 10,000 ft. The four
different separate regions represent four different volume SRV’s.
Figure 10.8 Combination of 4 wells and SRV for the model before optimization, where
the blue cells represent the FI below the cut-off.
The sequential steps in this process, is shown in the flow charts provided
in Figure 10.3 and Figure 10.4.
The automated process is a direct tool utilizing two software programs:
the global optimizer and the reservoir simulator. The first step begins
with a spreadsheet input file that is built to facilitate the entry data of the
Fracturability Indices data, or any other input maps. The currently used set
of input data maps is generated using the correlation published in Alzahabi
et al. [13].
The number of optimum wells, number of optimum fractures within the
wells, and the spacing between wells and fractures will be suggested using
the approach. It is believed that fracturing the optimum zones will contrib-
ute to higher hydrocarbon production from shale and tight formations.
construct a detailed geological model that is used here for testing. Many
relationships of parameters (e.g. porosity, permeability, quartz, clay con-
tent, E) were tested in this work to understand the Wolfcamp shale reser-
voir. The reservoir model has many 2D layers. The 2D nodes have FI values
assigned to it. The commercial reservoir Simulator Eclipse and Petrel were
used to populate the properties, then the correlation of FI was programmed
to generate the quality maps of Fracturability Index values, denoted by
(FI), as shown in Figure 10.9. The terminology quality map was introduced
by Da Cruz et al. (2004) and is commonly used in conventional reservoirs
in identifying producing regions. The heterogeneous properties of the res-
ervoir model are represented in the FI values. These input maps serve as
input for the optimization developed model.
These quality map generations were applied for each layer in the reser-
voir model.
The importance of the new model proposed here lies in its simplicity
and relative accuracy for the theory used in this work. More importantly,
it is based on easily obtained maps that are becoming more available in
today’s applications of shale characterization. Unlike many of the avail-
able SRV prediction tools, the new model does not require information
obtained from real microseismic data. Therefore, the SRV can be estimated
before drilling many wells in the reservoir.
Figure 10.10 compares a generated SRV before and after applying the
filter of FI = 0.5 on a chosen fracture stage of the shale model. Figure 10.10
shows the difference in grid-based modeling of one SRV before and after
removing the cells which have values of FI < 0.5.
0.4 25
20
0.3
15
0.2
10
0.1 5
0 10 20 30 40 50
Figure 10.9 Fracturability index distribution for the middle layer in the 80 × 80 × 5 model.
214 Hydraulic Fracturing and Well Stimulation
(a) (b)
Figure 10.10 A comparison between SRV, before and after applying the filter of FI. Note that
some grids with assigned values of FI < 0.5 were removed in Figure 10.10b. (a) One chosen
SRV before applying the filter of FI > 0.5. (b) One SRV after applying the filter of FI > 0.5.
Optimized
Cum oil prod., BBls thousands
120
Uniform distribution
100
80
60
40
20
0
0 2 4 6 8 10 12 14
Number of fracks
Figure 10.11 Optimized placement of the same size SRV versus uniform distribution of
SRV along one horizontal well.
Fracture Stimulated Reservoir Volume Mapping 215
Table 10.1 Parameters of the reservoir model used in validating the developed
model.
Well type Horizontal wells
Reservoir dimension, ft. 10,000 × 10,000
Fracture half length, ft. 500, 600, 650, 700, 750
Number of hydraulic fractures/well 5–50
Number of wells/pad 2–40
Length of horizontal well (6,000–10,000 ft.)
Half length of fracture, SRV half (200–600 ft.)
length
Spacing (wells, fractures) (500–1,600 ft.)
Stage width 200 ft.
1. Input Parameters
i. 3D FI excel sheet heat map
ii. Min Fracture stage (SRV) spacing
iii. Reservoir dimensions X, Y, Z
iv. Number of wells: 2 wells (budget constraints), assum-
ing we have two pads at the surface, each pad has 2
wells
v. Horizontal well length 6,000 ft
vi. Overlap constraints
vii. Spacing between wells and neighboring fractures
2. Output Parameters
viii. Number of fractures stages (SRV) per well and per the
pad
ix. The location of surface well pads per the model.
x. The locations Fracture stages (SRV) along the hori-
zontal wells
xi. Optimum scheduling of fracture stages (numbered
rank based on sum of FI values per the stage).
xii. Well spacing
xiii. Fracture dimensions (length & width assuming frac-
ture propagates as a network through the whole frac-
ture stage)
216 Hydraulic Fracturing and Well Stimulation
Assumption
Table 10.2 Properties of the reservoir model used in validating the developed
model.
Mineralogical properties
Parameters Minimum value Maximum value
Quartz, wt. % 6.00 75.0
Calcite wt. % 0.00 84.0
Clay wt. % 3.00 49.0
Pyrite wt. % 0.00 8.00
Petrophysical properties
Photoelectric Index (Pe), 2.61 5.71
barns/electron
Density ρz, g/cc 2.41 2.71
Geomechanical properties
E, psi 0.38 E6 9.75 E6
ν, ratio 0.02 0.38
Reservoir properties
Initial reservoir pressure, psia 5,330
Thickness, ft. 900
Model dimensions 80 × 80 × 5
Porosity, % Avg. : 9
Permeability, Nano-darcy Avg. : 208
achieving maximum contact area, while respecting the physical and eco-
nomic constraints.
References
1. Britt, L.K. and Schoeffer, J., The geomechanics of a shale play: What makes a
shale prospective? 2008. SPE-125525.
2. Cipolla, C.L., Warpinski, N.R., Mayerhofer, M.J., Lolon, E.P., Vincent, M.C.,
The relationship between fracture complexity, reservoir properties, and frac-
ture treatment design. Presented at the 2008 SPE Annual Technical Conference
held in Denver, Colorado, USA, 21–24 September, 2008, SPE-115769, doi:
10.2118/115769-MS.
3. Fisher, M.K., Wright, C.A., Davidson, B.M., Goodwin, A.K., Fielder, E.O.,
Buckler, W.S., Steinsberger, N.P., Integration fracture mapping technologies
to optimize stimulations in the Barnett Shale. Presented at the SPE Annual
Technical Conference and Exhibition held in San Antonio, Texas, 2002,
SPE-77441.
4. Fisher, M.K., Heinze, J.P., Harris, C.D., Davidson, B.M., Wright, C.A., Dunn,
K.P., Optimizing horizontal completions techniques in the Barnett Shale
using microseismic fracture mapping. Paper 90051 presented at the SPE
Annual Technical Conference and Exhibition, Houston, 26–29 September,
2004, doi: 10.2118/90051-MS.
5. Mayerhofer, M.J., Lolon, E.P., Warpinski, N.R. et al., What is stimulated
reservoir volume? Presented at SPE Shale Gas Production Conference, Fort
Fracture Stimulated Reservoir Volume Mapping 219
Appendix A
Abbreviations
E Young’s modulus
v Poisons’ ratio
FCD Dimensionless conductivity
K Permeability
Δσh Difference between minimum and maximum hori-
zontal stress
FI Fracturability Index
E Young’s Modulus, psi
E Plane Strain Modulus, psi
E Normalized Plane Strain Modulus
v Poisson’s ratio
ρ Density, lb./ft3
x(i,j) X_Y location of a fracture in the reservoir represents
the location (I, j) in the shale formation grid
X Coordinate axis along well path, ft.
Y Coordinate axis along fracture path, ft.
BHP Bottomhole pressure, psi
Qg Gas fow rate, Mscf/d
cf Formation compressibility, psi-1
Lf Fracture half length, ft.
Pi Initial reservoir pressure, psi
L Well Lateral length, ft.
ΔX,ΔY Model grid dimensions in x and y direction, ft.
Pnet Net pressure, psi
Xe, Ye Rectangular Reservoir shape dimensions in x and y
direction, ft.
W Horizontal well
F Fracture stage
Dmin Min well spacing, ft.
Appendix B
Definition of the Fracturability Index Used in the Well
Placement Process
In this model every cell is assigned a number based on the calculated FI
using geomechanical properties assigned to each cell. The FI consists of a
Fracture Stimulated Reservoir Volume Mapping 221
Appendix C
Pad
D1
D2
Figure A10.13 Stimulated reservoir volume (SRV) for one fracture stage.
Well origin
Wn Xmn
(n)
Yk = 0
Yk = 1
Node m Node n
Po tential fracture
locations (^(m, n))
Min frac
distance
Invalid frac
(too close to
other frac)
Frac
height
Frac
width
Invalid frac*
Invalid frac* (intersects SRV)
(intersects SRV)
*Handled by
Φ(n)
Frac
width
Invalid frac
Invalid well path* (intersects SRV)
(intersects SRV)
*Handled by
Ω(n)
Fracture height
Frac center,
well node
σmin
σmax Discretized SRV (sampled as approximated ellipse
centered at frac with height and width determined
by frac type)
Frac center
Abstract
Refracturing is a cost effective method enhancing the sustainability of both
depleted and good quality reservoirs. It is generally believed that refracturing
technology does not always work in all reservoir conditions. It is highly desirable
to have a quantitative method for predicting well productivity to select refractur-
ing well candidates. Assuming radial fractures around wellbore are uniformly dis-
tributed, a semi-analytical model was derived in this study to predict productivity
of refractured oil wells under pseudo steady state flow conditions. Result of a field
case study indicates that the mathematical model over-estimates well productivity
by about 10%. Sensitivity analysis with the mathematical model shows that the
productivity of refractured wells increases non-linearly with the number of radial
fractures around the wellbore. Sensitivity analysis with the mathematical model
also shows that the productivity of refractured wells increases non-linearly with
fracture conductivity, but the benefit of increasing fracture conductivity levels out
beyond 2000 md-inch. Therefore, there is no need to create very high-conductivity
fractures in refracturing operations. This paper provides petroleum engineers a
handy tool to assess well candidates for refracturing.
227
228 Hydraulic Fracturing and Well Stimulation
11.1 Introduction
Seeking sustainable methods of fossil fuels and addressing energy sustain-
ability are major contents of sustainable energy engineering. Refracturing
is considered one of the important technologies used to strengthen the
sustainability of the oil and gas reservoirs. It is a hydraulic fracturing oper-
ation conducted on a well which has already been hydraulically fractured.
Refracturing can sustainably increase the production rate and enhance
ultimate oil/gas recovery from not only the depleted reservoir, but also the
good quality reservoir. In the down turn of the oil and gas industry, com-
panies are realizing that refracturing old wells is a more budget-friendly
way to achieve cash flow than drilling and completing new wells [1, 2],
so it is also a sustainable method used to increase the cost effciency of oil
companies.
While some people are certain that it works quite well, others are still
unsure of the success rate of refracturing due to variation of reservoir con-
ditions [3]. It is highly desirable to have a quantitative method for assessing
reservoir conditions to select wells for refracturing.
For previously fractured wells, it is possible to create a secondary frac-
ture that is perpendicular to the first. The secondary orthogonal fracture
can be created within a certain time window that, in turn, depends on res-
ervoir properties [4]. The optimum time window for refracturing is in the
order of months to years. Roussel and Sharma [5] presented a systematic
methodology for selecting candidate wells for refracturing using produc-
tion data. It considers the stress reorientation occurring around a fractured
well, causing the refracture to propagate orthogonally to the initial fracture
in underdepleted sections of the reservoir.
Ren et al. [6] presented a novel method to determine the refracturing
strategy and priorities of the candidate wells by comprehensive analysis,
which considers reservoir properties, remaining oil distributions, and
induced stress fields. According to Miller et al. [7], reservoir depletion
from parent wells can create localized “pressure sinks.” When a nearby
infill well is fractured, the hydraulic fracture grows towards areas of lowest
stress (the depleted area) which can damage the parent well. Subsequently,
the production potential of the new well is decreased because it is now
draining from the same reservoir area as the parent well. With the advent
of modern diversion-based refracturing techniques, it is possible to create
a “pressure barrier” around the parent well that helps redirect the new frac-
turing treatment into other, virgin areas of the reservoir. By refracturing
the parent well first, this may not only preserve the original production of
Productivity of Refractured Oil Wells 229
the parent well, but it may also increase the production of the parent well,
as well as the new infill wells. Although refracturing adds to the capital
expenditure of the project, project economics may be more favorable with
an investment in parent well refracturing.
A gap exists between reservoir data and well refracturing design. Particularly,
a simple and accurate mathematical model is needed for quantitatively pre-
dicting the productivity of refractured oil wells. This study fills the gap.
No literature has been found to disclose any method for predicting pro-
ductivity of refractured wells where multiple radial fractures are created.
Methods for predicting productivity of a well with a single planar fracture
of infinite and finite conductivity are well documented in the literature
[8–11]. The concept of matrix-fracture cross flow presented by Guo and
Schechter [10] for single-fractured wellbore is employed in this study to
develop a semi-analytical model for quantitatively predicting productivity
of refractured wells.
Radial Fractures
k= kXky
hf
Lf
Lf
4nkh f Pd
Q dx (11.1)
B0 tan( ) x
rw
4k
c (11.3)
wk f (tan )
pdT12
C1 (11.4)
T02T11 T12T01
pdT11
C2 (11.5)
T02T11 T12T01
pd pe pw (11.10)
Table 11.1 Estimated reservoir and fracture properties in the Naricual-5 formation.
Horizontal permeability kx 0.08 md
Horizontal permeability ky 0.08 md
Fluid viscosity 8 cp
Oil formation volume factor 1.1 rb/stb
Wellbore radius 0.328 ft
Number of fractures 6
Fracture length 20 ft
Fracture height 4.5 ft
Fracture width 0.1 inch
Fracture permeability 1000 md
Reservoir pressure 12472 psia
Wellbore pressure 5339 psia
250
Well production rate (stb/day)
200
150
100
50
0
0 2 4 6 8 10
Number of fractures
350
Well production Rate (stb/day)
300
250
200
150
100
50
0
0 2000 4000 6000 8000 10000
Fracture conductivity (md-inch)
11.5 Conclusions
Assuming uniformly distributed radial fractures around wellbore, a
semi-analytical model was derived in this study to predict productivity of
refractured oil wells under a pseudo steady state flow condition. Field case
study and sensitivity analyses were performed with the semi-analytical
model. The following conclusions are drawn:
Acknowledgements
The authors are grateful to Chevron USA and the UNOCAL Corporation
for providing the LA Board of Regents Chevron I and Chevron II
Professorships in Petroleum Engineering and UNOCAL Professorship in
Engineering throughout this study.
References
1. R. Khusainov, B. Ganiev, A. Karimova, and O. Karpova, Refracturing is the
best way to develop hard-to-recover reserves in Romashkino oilfield con-
ditions. Presented at SPE Russian Oil and Gas Exploration and Production
Technical Conference and Exhibition, Moscow, Russia, 14–16 October
(2014).
2. E. Urban, D. Orozco, A. Fragoso, K. Selvan, and R. Aguilera, Refracturing
Vs. infill drilling - A cost effective approach to enhancing recovery in shale
reservoirs. Presented at Unconventional Resources Technology Conference,
San Antonio, Texas, USA, 1–3 August (2016).
3. M. C. Vincent, Refracs: Why do they work, and why do they fail in 100 pub-
lished field studies? Presented at SPE Annual Technical Conference and
Exhibition, Florence, Italy, 19–22 September (2010).
Productivity of Refractured Oil Wells 235
p
ry
da
un
Bo
ow
-fl
No
r a A Lf
o x
x
Δx
Fractures
1 V
ct (A11.1)
V p
p V
ctV q( x ) (A11.2)
t t
Productivity of Refractured Oil Wells 237
x tan( ) ( x Δx )tan( )
V Δh f h f tan( )xΔx (A11.3)
2
p q( x ) q( x )
(A11.4)
t ctV ct h f tan( )xΔx
2
p ct p
2 (A11.5)
y k t
2
p q (x )
2 (A11.6)
y kh f tan( )xΔx
p q( x )
y c1 (A11.7)
y kh f tan( )xΔx
p
0. (A11.8)
y y x tan( )
238 Hydraulic Fracturing and Well Stimulation
q( x )
c1 (A11.9)
kh f Δx
p q( x ) y
1 (A11.10)
y kh f Δx x tan( )
q( x ) y
dp 1 dy (A11.11)
kh f Δx x tan( )
q( x ) y2
p y c2 . (A11.12)
kh f Δx 2 x tan( )
where the integration constant, c2, can be determined using the boundary
condition at the fracture face
p y 0 p f ( x ). (A11.13)
where pf(x) is the pressure in the fracture at point x. Applying Eq. (A11.13)
to Eq. (A11.12) gives
c2 = pf(x) (A11.14)
q( x ) y2
p p f (x ) y (A11.15)
kh f Δx 2 x tan( )
Productivity of Refractured Oil Wells 239
Along the no-flow boundary, y = x tan(a), the pressure is pe. Eq. (A11.15)
demands
2kh f Δx
q( x ) [ pe p f ( x )] (A11.16)
x tan( )
The fluid velocity in the matrix in the y-direction at the fracture face at
point x can be expressed as
q( x ) 2k
v( x ) [ pe p f ( x )] (A11.17)
h f Δx x tan( )
Lf x
2kh f
Q( x ) 2 v( x )h f dx 2 [ pe p f ( x )]dx (A11.18)
x tan( )
x Lf
Q( x )
v f (x ) (A11.19)
wh f
k f dp f ( x )
v f (x ) (A11.20)
dx
Q( x ) k f dp f ( x )
wh f dx (A11.21)
240 Hydraulic Fracturing and Well Stimulation
Substituting Eq. (A11.18) into Eq. (A11.21) and rearranging the latter
gives
x
dp f ( x ) 2k
2 [ pe p f ( x )]dx (A11.22)
dx wk f x tan( )
Lf
d 2 p f (x ) 4k
2
[ pe p f ( x )] (A11.23)
dx wkx x tan( )
pd = pe – pf (x). (A11.24)
and
4k
c (A11.25)
wk f tan( )
d 2 pd c
pd (A11.26)
dx 2 x
pd x rw pd pe pw (A11.27)
pd|x Lf 0. (A11.28)
Productivity of Refractured Oil Wells 241
pdT12
C1 (A11.30)
T01T11 T12T01
and
pdT11
C2 (A11.31)
T02T11 T12T01
where
Hydraulic fracturing is an old technique which has been around for sev-
eral decades. Extensive use of fracturing in recent years is attributed to
advances in horizontal drilling and the capability of reaching deep target
formations with lower costs. Concerns have been raised on wide-spread
application of hydraulic fracturing in oil and gas and other operations
despite of the benefits of this technology [3, 4]. Unique safety and environ-
mental considerations must be taken to mitigate concerns and ensure safe
operations [5]. In particular, little is known about long-term environmen-
tal consequences given the relatively short history of extensive horizontal
drilling.
Potential environmental impacts can be classified into three major cat-
egories: air, induced seismicity, and water and wastewater. Air pollution
in general is linked to natural gas development activities, including frac-
turing. For example, a human health risk assessment study by McKenzie
et al. [6] on a gas field in Garfield County, Colorado suggested that people
living at a distance greater than half a mile from the site were in less dan-
ger than the ones living at a lesser distance. According to a more recent
study by Moore et al. [7], particulate matters smaller than 2.5 μm in size
(PM2.5) and Nox are found in diesel emissions during preproduction (i.e.
drilling and hydraulic fracturing activities) of unconventional natural gas
production.
Subsurface injection operations normally generate low magnitude
(smaller than 2 in Richter scale) seismic events which are termed as
micro-earthquakes or microseismic [8]. In a few cases of hydrofracking
jobs, seismic events have been felt by nearby residents and the operator has
been forced to stop the injection. In 2011, two seismic events (2.3 and 1.5
magnitude in Richter scale) were recorded close to Blackpool, Lancashire
in the U.K. -- the operator had to halt the fracturing operation in nearby
Bowland Shale formation [9]. Generally, the risk of generating serious
earthquakes as a result of hydraulic fracturing is low when compared to the
deep-well injection process, which shows higher probabilities of observed
larger seismic events [8, 10, 11]. Historically, the oldest injection-induced
seismic events are those of the Rocky Mountain Arsenal waste injection
site in Denver, Colorado [12]. The most recent events were generated in
Youngstown, Ohio [13] and central Oklahoma [14], both in 2011. The
former is also known to be the largest injection-induced event with a 5.6
magnitude [8].
Potential surface water [15, 16] and groundwater contaminations are
caused by either fracturing fluid chemicals or volatile compounds from
deep formations [17–20]. The returned fluid handling and treatment is
another source of possible environmental complications [16, 21–24].
Part 245
References
1. U.S. Energy Information Administration, 2018.
2. Spellman, F., Environmental Impacts of Hydraulic Fracturing, CRC Press,
2012.
3. Kargbo, D.M., Wilhelm, R.G., Campbell, D.J., Natural gas plays in the
Marcellus Shale: challenges and potential opportunities. Environ. Sci.
Technol., 44, 15, 5679–84, 2010.
4. USEPA, Assessment of the Potential Impacts of Hydraulic Fracturing for Oil
and Gas on Drinking Water Resources, 2015.
5. Jabbari, N., Ashayeri, C., Meshkati, N., Leading Safety, Health, and
Environmental Indicators in Hydraulic Fracturing. SPE Western Regional
Meeting, Garden Grove, CA, 27-30 April 2015, 2015.
6. McKenzie, L.M. et al., Human health risk assessment of air emissions from
development of unconventional natural gas resources. Sci. Total Environ.,
424, 79–87, 2012.
7. Moore, C.W. et al., Air Impacts of Increased Natural Gas Acquisition,
Processing, and Use: A Critical Review. Environ. Sci. Technol., 2014.
8. Ellsworth, W.L., Science (New York, N.Y.), 341, 6142, 1225942, 2013.
9. Pater, C. De and Baisch, S., Geomechanical study of Bowland Shale seismicity,
2011.
10. Goebel, T. et al., A probabilistic assessment of waste water injection induced
seismicity in central California. American Geophysical Union Fall Meeting,
San Francisco, 2014.
11. Aminzadeh, F., Tiwari, A., Walker, R., Correlation between Induced
Seismic Events and Hydraulic Fracturing activities in California. American
Geophysical Union Fall Meeting, San Francisco, 2014.
12. Healy, J. and Rubey, W., The Denver earthquakes. Science (New York, N.Y.),
161, 3848, 1301–1310, 1968.
246 Part
13. Ohio Department of Natural Resources, Preliminary Report on the North Star
1 Class II Injection Well and the Seismic Events in the Youngstown, Ohio, Area,
2012.
14. Holland, A., 2011.
15. Hammer, R. and VanBriesen, J., In fracking’s wake: new rules are needed to
protect our health and environment from contaminated wastewater, Natural
Resources Defense Council, 2012.
16. VanBriesen, J., Wilson, J., Wang, Y., 2014.
17. Osborn, S.G. et al., Methane contamination of drinking water accompanying
gas-well drilling and hydraulic fracturing. Proc. Natl. Acad. Sci. U.S.A., 108,
20, pp.8172–6, 2011.
18. Vengosh, A. et al., The Effects of Shale Gas Exploration and Hydraulic
Fracturing on the Quality of Water Resources in the United States. Procedia
Earth Planet. Sci., 7, pp.863–866, 2013.
19. Llewellyn, G.T. et al., Evaluating a groundwater supply contamination inci-
dent attributed to Marcellus Shale gas development. Proc. Natl. Acad. Sci.,
112, 20, pp.6325–6330, 2015.
20. Jabbari, N., Aminzadeh, F., de Barros, F.P.J., Hydraulic fracturing and the
environment: risk assessment for groundwater contamination from well cas-
ing failure. Stochastic Environ. Res. Risk Assess., 31, 1527–1542, 2017, doi:
10.1007/s00477-016-1280-0.
21. Gregory, K.B., Vidic, R.D., Dzombak, D.A., Water Management Challenges
Associated with the Production of Shale Gas by Hydraulic Fracturing.
Elements, 7, 3, pp.181–186, 2011.
22. USEPA, Study of the Potential Impacts of Hydraulic Fracturing on Drinking
Water Resources, 2012.
23. Gordalla, B.C., Ewers, U., Frimmel, F.H., Hydraulic fracturing: a toxicolog-
ical threat for groundwater and drinking-water? Environ. Earth Sci., 70, 8,
3875–3893, 2013.
24. Glazer, Y.R. et al., Potential for Using Energy from Flared Gas for On-Site
Hydraulic Fracturing Wastewater Treatment in Texas. Environ. Sci. Technol.
Lett., 1, pp.300–304, 2014.
25. Jabbari, N., University of Southern California, Los Angeles, California,
2016, Retrieved from http://digitallibrary.usc.edu/cdm/ref/collection/
p15799coll40/id/231045.
12
The Role of Human Factors Considerations
and Safety Culture in the Safety of
Hydraulic Fracturing (Fracking)
Jamie Heinecke1, Nima Jabbari2* and Najmedin Meshkati3
1
Senior Student, Daniel J. Epstein, Department of Industrial & Systems
Engineering, Viterbi School of Engineering, University of Southern California
2
Ph.D. Candidate, University of Southern California, Sonny Astani Department of Civil
and Environmental Engineering, Viterbi School of Engineering, University of Southern
California, Los Angeles, CA
3
Professor, Sonny Astani Department of Civil/Environmental Engineering,
Daniel J. Epstein, Department of Industrial & Systems Engineering,
Viterbi School of Engineering, University of Southern California
Abstract
Hydraulic fracturing is a well stimulation method frequently used in oil and gas
operations in order to facilitate the flow movement and increase production in
tight and low permeability formations. Along with horizontal drilling, hydraulic
fracturing has helped the United States in meeting its enormous need for energy.
Despite of the advantages such as boosting economy and energy-independency,
there are potential environmental and safety issues associated which rise many
questions on the future of this technology. More specifically, recent accidents and
incidents have heightened public and industry concerns. This paper addresses
major contributing factors to the safety of fracturing by analyzing different steps
of the operation and providing associated human factors and safety culture con-
siderations. Primary human factor root-causes of three recent fracking accidents
have been analyzed, and recommendations as how to proactively address these
considerations in the entire hydraulic fracturing life-cycle process are proposed.
247
248 Hydraulic Fracturing and Well Stimulation
12.1 Introduction
The United States currently has access to some of the largest shale oil and gas
reserves in the entire world. “Since 2007, discoveries of unconventional gas
including shale gas have more than doubled the estimate of North American
reserves to 3,000 trillion cubic feet, enough to meet 100 years of demand”
[1]. Shale rock is a unique type of rock formation that lies over a mile below
the surface and contains a significant amount of gas trapped inside tiny
pockets within the rock [2]. Until recently, companies found it difficult to
extract these precious resources using conventional drilling methods.
However, the improvement of hydraulic fracturing (also known as fracking)
in recent years, has allowed companies to take advantage of these reserves.
Fracking allows gas to flow to the surface by pumping water under extreme
pressure down into the well in order to fracture the rock formation. The pro-
cess behind hydraulic fracturing is not entirely new, but became especially
advantageous when combined with another technique called horizontal
drilling. This combination is so effective that according to the U.S. Energy
Information Administration (EIA), production of natural gas will meet con-
sumption needs by 2019 and 12% of all production will be exported by 2040
[3], as shown in Figure 12.1. In fact, it is primarily due to fracking in the U.S.
that already, according to a recent article in the Wall Street Journal (August
12, 2014), “since March 2008, oil production has increased 58% and
natural-gas output has risen 21%, making the U.S. the world’s largest producer
of both fuels, according to federal and international agency statistics”
10
Net imports
0
-10
1990 2000 2010 2020 2030 2040
Figure 12.1 Total U.S. natural gas production, consumption, and net imports in the
reference case, 1990–2040 (trillion cubic feet) [3].
The Role of Human Factors Considerations 249
(emphasis added); and “jobs directly related to oil and gas production have
nearly doubled in the past 10 years to 697,000, government data shows” [4].
Fracking will become an increasingly important topic of discussion as it
starts to play a larger role in America’s energy future. Human factors con-
sideration should be the foundation of this rapidly growing technology to
ensure that it is implemented correctly in order to minimize the drawbacks
of the fracking process.
The potential benefits of fracking could be enormous for the United
States, but only if executed properly. Due to the increasing pressure to find
alternative energy sources, hydraulic fracturing is being implemented in a
greater capacity than ever before. “In a decade, shale gas has risen from 2
percent of US natural gas production to 37 percent. The US has overtaken
Russia as the world’s largest natural gas producer” [5]. However, such rapid
and enormous growth poses significant safety risks. As companies rush
to capitalize on this opportunity, they tend to over-look many safety and
environmental concerns such as increased seismicity as a result of waste
injection [6] and groundwater contamination [6]. These challenges have
been crucial enough in some cases to make the cities take serious actions to
regulate, monitor, and even halt the fracking operations just like the cases
of Pavillion, Wyoming in 2009 [7] and New York in 2010 [8].
Furthermore, according to the National Institute of Occupational Safety
and Health (NIOSH) [9], “the oil and gas extraction industry has an annual
occupational fatality rate of 27.5 per 100,000 workers (2003–2009) - more
than seven times higher than the rate for all U.S. workers. The oil and gas
extraction industry employed approximately 435,000 workers in 2010. The
annual occupational fatality rate in this industry is highly variable; this
variation is correlated with the level of drilling activity in the industry.
Fatality rates are higher when there is an increased number of active drill-
ing and workover rigs.” An accident in drilling-related operations, like in
other contexts, can be characterized as “an error with sad consequences”
[10]. Nevertheless, a correct understanding of the root-causes of the afore-
mentioned so-called “error” in terms of instances of human-machine,
human-task, and/or human-organization mismatches can be greatly con-
tributed to its prevention [11–14].
To minimize accidents which could result in fatalities and nonfatal
injuries and the possibility of other negative effects associated with frack-
ing, one must consider the human factors issues involved in each stage of
the process. By examining each stage of hydraulic fracturing, the specific
human factors issues involved in the overall process can be addressed and
ultimately both the safety and the efficiency of the entire system can be
improved.
250 Hydraulic Fracturing and Well Stimulation
on the people and environment around it, as well as some grey areas
in industry regulation [16].
the San Joaquin Basin of California, it has been shown that the seismic
events related to the oil and gas activities can be distinguished from the
natural events using semiology measures such as fractal dimension and
b-value [23].
The final major area of concern for critics is the composition of frack-
ing fluids used to break open the shale formation. Although many critics
argue these fluids include harmful chemicals, “evaluating the relative vol-
umes of the components of a fracturing fluid reveals the relatively small
volume of additives that are present. Overall the concentration of addi-
tives in most slickwater fracturing fluids is a relatively consistent 0.5% to
2% with water making up 98% to 99.2%” [24]. However, it is important
to remember when analyzing these percentages that millions of gallons
of water are used in each drilling operation and the long term effects of
what these chemicals will do in the ground is unknown. Furthermore,
the exact composition of the fracking fluid used by each company is
not fully revealed because it is considered proprietary, which causes
additional unease. These risks and benefits of fracking are affected by
operations in each stage and can be improved through human factors
management.
In the end, if these allegations are true, experts say that it is not due to
the scientific fracking technique but instead is a result of human error. This
illustrates the key obstacle fracking must overcome: the science behind
fracking is valid, but in accounting for human factors, it becomes a much
less reliable process that may be prone to significant and costly problems.
Also, it is worthwhile to note that the concerns and worriedness about the
fracturing are important issues to be addressed through communications
with the public media. It, thus, necessitates proactively conducting sessions
and talks on the scientific facts regarding the hydraulic fracturing and try
to stay away from the myths [25–27].
Vertical Well is spudded vertically up to a specified Using proper lifting and material handling
Drilling depth. techniques.
Human-machine interface
Communications
Monitoring/Supervisory Control
Job hazards (identification & conrol);
PPEs; (”Electricity hit”) [29]
Vertical Vertical casing is inserted in place to Standard operating procedures
Casing function as a barrier to the surrounding
environment.
Horizontal
Horizontal casing is placed after the Following cementing standards and
Casing horizontal drilling is finished. Horizontal protocols
section in entirely located in the target Information monitoring (”Supervisory
formation. Control”)
Monitoring/information processing/signal
Perforation Horizontal casing is perforated using detection.
electrical charges.
Situational awareness
1Will be further discussed in the section of hydraulic fracturing process as well as the section of human factors
Figure 12.2 Hydraulic fracturing steps (inspired from [28]) and related major human factors.
hazards. The first is the possible release of excess drilling mud which can get
on the workers skin or create a slippery floor surface. In order to counter
this, it is important to follow the proper protocol of shutting down the mud
pumps and using a kelly that has a mud saver valve. The second hazard is
the possibility of being struck by slip handles when the drill string is spun.
There are alternative technologies such as a pipe spinner that companies
should consider investing in to increase worker safety. The last hazard is
the possibility of getting struck by the kelly when it is being placed over
the hole. Again, proper training can prevent any incidents of this hazard
occurring [31].
The third stage requires workers to latch elevators onto the pipe to pre-
pare for lifting or lowering the pipe. Potential risks include getting hands
pinched, caught, or being struck by elevators. Supervisors need to make
sure that workers are correctly following latching procedures and elevators
should be inspected and maintained after each use.
Tripping the pipe also requires the worker to climb the oil derrick and
work from an area called the monkeyboard (component 4 of Figure 12.4).
There are numerous hazards to address in this stage. Workers can fall when
they are climbing the ladder or while working from the monkeyboard. It
is very important for the worker to use a climb assist device and to wear
a body harness to protect against falls. They should be wearing the proper
protective clothing including a hard hat, gloves, and safety-toed footwear.
Companies need to ensure that slip-resistant coatings are put on all work-
ing surfaces. Additional hazards include getting caught between the pipe
and various objects and being struck by dropped objects. Workers need to
exercise additional caution when work is being done over head and to use
proper hand placement [31].
The next stage involves using a pair of tongs to break out and disconnect
the pipe. Tongs are tools that consist of two long arms that aid in seizing or
holding an object. If these tongs slip or backlash during operations, seri-
ous injury could occur to the worker. It is important to implement proper
break out procedure and workers should be outside of the 4 foot tong rota-
tion area (as seen in Figure 12.5). Communication is essential between the
driller and the floor hands that are operating the tongs [31].
The final stage is to rack the pipes. Crew members need to be wary of
getting their hands pinched between pipes or getting feet crushed under a
stand of pipe. It is important for workers to use proper hand positioning
and to properly position their feet away from pipe stands.
The risks in each process described above may be mitigated through
human factors consideration. In the drilling stage, proper training and exe-
cuting proper lifting and handling techniques will eliminate many risks.
256 Hydraulic Fracturing and Well Stimulation
Hazardous
Area
Mouse
Hole
Hole
Center RADIUS
4 FEET
Caution Area
Figure 12.5 Drilling rig floor hazardous area layout with tong swing radius [31].
The Role of Human Factors Considerations 257
Due to the fracking boom, more and more inexperienced workers are being
hired and put to work without sufficient training. This puts both the workers
and the environment at risk. Companies must invest in thorough training to
keep their workers safe. Such training is also in the company’s best interest
because the reduction of personal hazards improves job performance.
Perforating Cement
Gun
Detonation
Cord
Jet
Charge Casing Formation
creating pathways for the gas to continually flow into the wellbore and up
to the surface to be harvested. Each stage of this fracking process presents
a few key human factor issues that must be addressed.
12.8 Wastewater
The second human factors topic associated with the fracking process is
the flow back wastewater. Flow back fluids contain not only the harsh
The Role of Human Factors Considerations 259
chemicals found in fracking fluids but also other contaminants and natu-
rally occurring radioactive materials from exposure to the shale formation.
The fluid is usually stored in temporary pits or steel tanks until it can be
properly treated or disposed of. Correctly containing these fluids within
a lined pit is especially important for reducing the risk of contaminating
shallow ground water, a common concern among critics:
The failure of a tank, pit liner, or the line carrying fluid (“flowline”)
can result in a release of contaminated materials directly into surface
water and shallow ground water. Environmental clean-up of these
accidentally released materials can be a costly and time consuming
process. Therefore, prevention of releases is vitally important [24].
12.9.1.1 Microergonomics
Microergonomics, also called human engineering, addresses the relationship
between human, equipment and physical environment. It is focused on the
human-machine system level and is, for example, concerned with the design
of individual workstations, work methods, tools, control panels and displays.
Microergonomics includes studies of the human body sizes, known as anthro-
pometrics, physical and psychological abilities and limitations, information
processing, and human decision-making and error. It is noteworthy that, in
the context of the control room environment:
12.9.1.2 Macroergonomics
Ergonomics at the macro level, macroergonomics, is focused on the overall
people-technology system level and is concerned with the impact of techno-
logical systems on organizational, managerial, and personnel (sub-) systems.
Macroergonomics includes areas such as training, management, the plan-
ning process, information systems, internal review/inspection programs,
The Role of Human Factors Considerations 261
Interactive
Effect
Human
Organization
Technology
Volume of Output
(2013), that could (and should) be adopted and utilized for this purpose by
the fracking industry [47].
Output
Productivity Ratio =
Input
Final Outcomes
Safety, Health, Environmental, Economical, and Social Impacts
Figure 12.8 Risks/loses and impacts of fracking mishaps and accidents (adapted from [48]).
264 Hydraulic Fracturing and Well Stimulation
~50 meters NE
GL
Deep Monitoring
Well
Accidental
perforations at
136 mMD
177 mm Surface
Casing set to
500 mMD 606 mMD
Base of Ground
Water Protection
600 m 114 mm Production
KOP 697 mMD Previous to incident perforated and
Casing Set to
hydraulically fractured intervals
2503 mMD
Bridge Plug
1557 mMD
1000 mMD
Planned Perforation
Interval
1486–1486.6 mMD
its training program. The company should also evaluate the monitoring
equipment to ensure that it is ergonomically correct and easy to use. If
equipment is not adjusted to ergonomic standards then the chance for
operator error is much higher.
Acknowledgment
Authors would like to express their sincere gratitude to Dr. Iraj Ershaghi,
Dr. Donald Paul of University of Southern California (USC), and Dr. Wayne
D. Pennington of Michigan Technological University for their insights and
analyses concerning technical issues affecting the hydraulic fracturing pro-
cess. This work, however, should not necessarily be construed as their rep-
resentative position(s).
References
1. S.L. Sakmar, Shale gas developments in North America: An overview of the
regulatory and environmental challenges facing the industry, SPE North
American unconventional gas conference and exhibition, The Woodlands,
Texas. pp. 1, 3, 7 (2011).
2. K. Barillaro, Process of Fracking, http://shalestuff.com/education/fracking/
fracking (2012).
3. J.J. Conti, U.S. Energy Information Administration, Office of Integrated and
International Energy Analysis, Annual energy outlook 2013, http://www.eia.
gov/forecasts/aeo/ (2013).
The Role of Human Factors Considerations 267
19. T. Henry, How oil and gas disposal wells can cause earthquakes, https://
stateimpact. npr.org/texas/tag/earthquake/.p. 1, (2012).
20. S. Horton, Disposal of hydrofracking waste fluid by injection into subsurface
aquifers triggers earthquake swarm in central Arkansas with potential for
damaging earthquake. Seismol. Res. Lett. 83(2), 250–260 (2012).
21. N.R. Warpinski, Understanding Hydraulic Fracture Growth, Effectiveness,
and Safety through Microseismic Monitoring, ISRM International
Conference for Effective and Sustainable Hydraulic Fracturing. International
Society for Rock Mechanics, (2013).
22. W.D. Pennington, Reservoir geophysics. Geophysics 66(1), 25–30 (2001).
23. F. Aminzadeh, T.H.W. Goebel, Identifying induced seismicity in active tec-
tonic regions: a case study of the San Joaquin Basin, California. AGU Fall
Meeting, Abstract: S31F-06, San Francisco, California. (2013).
24. Frac Focus, Chemical Disclosure Registry, http://fracfocus.org/ (2013).
25. F. Aminzadeh, To Frack or not to Frack. That is the question. Los Angeles
Science Center Series on “Science Matters”, June 8 (2013).
26. N. Jabbari, F. Aminzadeh, F. de Barros, Hydraulic Fracturing, Points and
Counter Points. Urban Water Institute 20th Annual Water Policy Conference
San Diego, CA, August 16 (2013).
27. F. Aminzadeh and D. Paul, Induced Seismicity Intersection of Science,
Public and Regulation, IOGCC Annual Meeting, November 5. Long Beach
California. (2013).
28. I. Ershaghi and Q. Qianru, Aspects of Oilfield Related Accidents. SPE Annual
Technical Conference and Exhibition. Society of Petroleum Engineers.
(2013).
29. M. Economides and T. Martin, Modern Fracturing: Enhancing Natural Gas
Production, Energy Tribune Publishing Inc., Houston. (2007).
30. American Petroleum Institute, Hydraulic fracturing operations- well con-
struction and integrity guidelines, API Guidance Document HF1, doi:
Product No. GHF101, (2009).
31. OSHA, Oil and gas well drilling and service etool, http://www.osha.gov/
SLTC/etools/oilandgas/index.html (2014).
32. Clean Water Action, Fracking: The process, http://www.cleanwateraction.
org/page/fracking-process (2013).
33. J.A. Merchant, U.S. Department of Health and Human Services, National
Institute for Occupational Safety and Health, Occupational respiratory dis-
eases, Publication No. 86-102, p. 266, (1986).
34. F.L. Rice, U.S. Department of Health and Human Services, National Institute
for Occupational Safety and Health, Health effects of occupational exposure
to respirable crystalline silica, Publication No. 2002-129.p. 5, (2002).
35. National Academy of Engineering and National Research Council. Interim
Report on Causes of the Deepwater Horizon Oil Rig Blowout and Ways to
Prevent Such Events. The National Academies Press, Washington, DC. (2010).
The Role of Human Factors Considerations 269
36. I. Ershaghi and D. Luna, Validation process for human factor in com-
plex upstream oil and gas operations. SPE 147654, SPE Annual Technical
Conference and Exhibition, Denver, CO, (2011).
37. M. Tabibzadeh and N. Meshkati, Learning from the BP Deepwater Horizon
accident: Risk analysis of human and organizational factors in negative pres-
sure test. Environ. Syst. Decis. 34(2), 194–207 (2014).
38. United States Chemical Safety and Hazard Investigation Board. Investigation
Report, Refinery Explosion and Fire, BP Texas City, Texas. Report No. 2005-
04-I-TX. (2007).
39. Eastman Kodak Company. Ergonomic Design for People at Work (Vol. 1),
Lifetime Learning Publications, Belmont, California. (1983).
40. N. Meshkati, Critical Human and Organizational Factors Considerations
in Design and Operation of Petrochemical Plants. Proceedings of the First
International Conference on Health, Safety & Environment in Oil and Gas
Exploration and Production, Society of Petroleum Engineers (SPE), 11-14
November, The Hague, The Netherlands, Volume I, 627–634 (Paper # SPE
23275). (1991).
41. A.D. Swain and H.E. Guttmann, Handbook of Human Reliability Analysis
with Emphasis on Nuclear Power Plant Applications. Final Report (NUREG/
CR-1278). Washington, D.C. U.S. Nuclear Regulatory Commission. pp. 11–15,
(1983).
42. H.W. Hendrick, Macroergonomics: A concept whose time has come. Human
Factors Soc. Bull. 30(2), 1–3, (1987).
43. J.T. Reason, A Life in Error: From Little Slips to Big Disasters. Ashgate,
Burlington, VT, (2013).
44. N. Meshkati, Human factors in process plants and facility design, in Cost-
Effective Risk Assessment for Process Design, R. Deshotels and R. Zimmerman
(Eds.), pp.113, 115, McGraw-Hill, Inc., New York, NY. (1995).
45. N. Meshkati, Cultural context of nuclear safety culture: A conceptual
model and field study, in Nuclear Safety: A Human Factors Perspective,
J. Misumi, B. Wilpert, R. Miller (Eds.), pp. 61–75, Taylor and Francis,
London. (1999).
46. M.J. Gelfand, M. Frese, E. Salmon, Cultural influence on errors: Preventions,
Detections, and Management, in Errors in Organizations, D.A. Hofmann and
M. Frese (Eds.), pp. 273–315, Taylor & Francis, Routledge, New York, NY.
(2011).
47. Institute of Nuclear Power Operations. Traits of a Healthy Nuclear Safety
Culture. Institute of Nuclear Power Operations. INPO 12-012. http://
pbadupws.nrc.gov/docs/ ML1303/ML13031A707.pdf. (2013).
48. N. Meshkati, Preventing Accidents and Explosions in the Petroleum
Industry: The Critical Role of Human Factors. Invited plenary presentation
at the First International Congress on Medicine in the Petroleum Industry,
organized by Pemex, the State-owned oil company of Mexico. (1992).
270 Hydraulic Fracturing and Well Stimulation
49. J. Osborne, Exxon mobil subsidiary faces criminal charges over 2010 fracking
spill, Dallasnews, http://www.dallasnews.com/business/energy/20130911-exx-
on-mobil-subsidiary-faces-criminal-charges-over-2010-fracking-spill.ece. p. 1,
(2013).
50. C. Linnitt, Alberta finds mismanagement of errors causes fracking water
contamination. http://www.desmogblog.com/2012/12/22/alberta-finds-
mismanagement-errors-causes-fracking-water-contamination-alberta.
p.1.(2012).
51. R. Swift, Spill in state forest moves gas drilling moratorium debate, Times-
Tribune, http://thetimes-tribune.com/news/spill-in-state-forest-moves-gas-
drilling-morato-rium-debate-1.705590, p. 1, (2010).
52. M. Christou and M. Konstantinidou, Safety of offshore oil and gas operations:
Lessons from past accident analysis, JRC Scientific and Policy Reports, http://
publications.jrc.ec.europa.eu/repository/bitstream/111111111/27463/1/off-
shore-accident-analysis-draft-final-report-dec-2012-rev6-online.pdf (2012).
53. R. Rapier, Both sides mislead when it comes to fracking, http://blogs.wsj.
com/experts/2013/11/14/both-sides-mislead-when-it-comes-to-fracking
p. 1, (2013).d
13
Flowback of Fracturing Fluids with
Upgraded Visualization of Hydraulic
Fractures and Its Implications on
Overall Well Performance
Khush Desai* and Fred Aminzadeh†
Abstract
The increasing popularity of hydraulic fracturing follows some achievements in the
USA, where it has become a proven technology in the stimulation of tight reser-
voirs. Nevertheless, the physics behind the process are not completely understood,
particularly in the domain of post fracturing fluid recovery. In many instances, the
recovery of large portions of injected fracturing fluid has not been successful. In
this research, the goal was to identify and evaluate the responsible factors.
The scope of our study includes determining the fate of the fracturing fluid
within the reservoir, and calculating the loss in incremental production as a conse-
quence of that outcome. The information can be used to more effectively predict the
performance of stimulated wells with hydraulic fractures over time. Also, an estima-
tion of incremental oil recovery post treatment can be calculated more accurately.
This new knowledge is beneficial to many of the participants - service provider
companies will enjoy a clear understanding of the treatment; operating companies
should perform more reliable economic analyses; the individual states could real-
ize increased accuracy in the quantification of reserves; and regulatory agencies
can better determine the probability of these fluids in groundwater contamination.
271
272 Hydraulic Fracturing and Well Stimulation
13.1 Introduction
A streamtube model has been developed as a tool to predict well perfor-
mance post hydraulic fracturing treatment. It is also useful in ascertaining
various possibilities of representative streamtube models that would be
characteristic of the reservoir.
As stated earlier, major portions of injected fracturing fluid cannot be
recovered. Various physical phenomena impact that reality. Several sen-
sitivity analyses have been conducted in order to correctly prioritize the
contributing issues.
The loss of potential production has been calculated using different
input parameters resulting from damage caused by not recovering injected
fracturing fluids. Exponential decline has been applied for the calculation
of well production rates over time. Results were obtained for administering
the simulation for a finite time period.
An ongoing integrated work considering the causes of partial flowback
is the in situ stress analysis in a reservoir. The purpose is to incorporate the
stress distribution along the horizontal section. More importantly, the frac-
ture closing might be a function of proppant travelling and stress regimes
surrounding the wellbore. Each fracture stage will be classifed based upon
stress regime at a distance from the wellbore using seismic data, and with
image and sonic logs as secondary data.
13.2 Assumptions
Our assumptions from this research are as follows:
Low velocity
streamtubes
7 mm
High velocity
streamtubes
7 mm
13.3.2 Results
All the points in Plot 13.1 are the possibilities of the initial production
rate of the well post hydraulic fracturing treatment. The magnitudes of the
increase in the well production rate comply with the fact that the produc-
tion rate of the well will be high in case of higher cumulative volume of
connected tubes (higher reservoir permeability) and vice versa for a fixed
number of set (The term ‘set’ or ‘sets’ here and onwards in this paper means
the number of possibilities of range of radius of the streamtube in the dis-
tribution). Also, for a fixed volume of connected tubes the production rate
150
100
6 8 10 12 14 16
Sets
of the well would decrease with increasing number of sets. These observa-
tions validate the model. Below are observations from Plot 13.1:
1/5
Q 3t 4 E
L 51.46
(1 v 2 ) h 4
1/ 4
(1 v 2 )Q L
Wmax 0.389
E
Wavg 0.628 Wmax
Where,
L is fracture half-length in ft.
Q is injection rate in BPM
t is time in minutes
h is formation thickness in ft.
E is young’s modulus in PSI
v is poisson’s ratio which is unitless
μ is viscosity in cp
10
9
Depth of penetration (ft)
8
7
6
5
4
3
2
1
0
0 100 200 300 400 500 600 700 800
Fracture half length (ft)
transience gets deeper into the reservoir. The pressure differential should be
enough for the fracturing fluid to flow within the reservoir. However, frac-
turing fluids react with reservoir rock and fluids over time, which decrease
their mobility, and thus require higher differential pressure to overcome the
capillary pressure and flow within the reservoir. In reality, applying such
high differential pressures in the reservoir may not be practical, leading to
unproduced injected fracturing fluids within the reservoir. This phenome-
non is a major factor in high volumes of fracturing fluid remaining in the
reservoir.
Calculated Values:
5000
Additional irrecoverable volume (bbls)
4500
4000
3500
3000
2500
2000
1500
1000
500
0
0 5 10 15 20 25 30
(cp)
10000
Additional irrecoverable volume (bbls)
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
0 200 400 600 800 1000 1200 1400
t (hrs)
900
800
700
600
qo (bbls/day)
500
400
300
200
100
0
0 100 200 300 400 500 600 700 800 900 1000
Time, post comingled flow (days)
Without skin due to fracturing fluid With skin due to fracturing fluid
13.7 Conclusions
We conclude that by combining the streamtube model approach with
the knowledge of geomechanics of the field, the well performance after
hydraulic treatment can be predicted with reasonable accuracy.
Partial flowback of fracturing fluids is primarily a result of:
13.8 Limitations
All the estimates of the magnitude of loss of fracturing fluid are low when
considering only one fracture. However, in reality, the estimates would
increase substantially due to the complex geometry of fracture where a
single fracture does not exist.
In actuality, the distribution of the proppants within the fracture is not
uniform. For that reason, the fracture may close off in the early or middle
region, making the rest of the length of the fracture highly ineffective. In
that case the loss of fracturing fluid due to closing of fractures may be much
more than what has been previously calculated. Usually, the concentration
of proppants is greater in the lower parts of a fracture due to the effect of
gravity.
One would encounter losses attributable to heterogeneity, natural frac-
tures and faults in real life which have not been taken into account in this
model. That could be one of the contributing factors for further losses of
injected fracturing fluid impacting the observation of 2–26% flowback [6].
References
1. T.K. Perkins and L.R. Kern, SPE-89-PA widths of hydraulic fractures. J Pet
Technol 13(9), 937–949 (1961).
2. T. Martin and P. Valko, Hydraulic Fracture Design for Pressure Enhancement,
Modern Fracturing Enhancing Natural Gas Production, Houston, Texas.
3. R.A. Woodroof, M. Asadi, and M.N. Warren, SPE-82221-MS Monitoring
fracturing fluid flowback and optimizing fracturing fluid cleanup using
chemical Frac Tracer, SPE European Formation Damage Conference, 13–14
May, The Hague, Netherlands (2003).
4. http://www.corelab.com/ps/hydraulic-fracture-design
5. http://www.fekete.com/SAN/WebHelp/FeketeHarmony/Harmony_
WebHelp/Content/HTML_Files/Reference_Material/Analysis_Method_
Theory/Blasingame_Theory.htm
6. Q. Zhou, R. Dilmore, A.N. Kleit, and J. Yilin, SPE-173364-MS Evaluating
Fracturing Fluid Flowback in Marcellus Using Data Mining Technologies.
282 Hydraulic Fracturing and Well Stimulation
7. http://biomedical.materialise.com/cases/cfd-simulation-water-flow-
sandstone-using-micro-ct
8. http://www.slb.com/services/completions/stimulation/sandstone/hiway_
channel_fracturing.aspx?t=2
Appendix - A
Calculations for Upgraded Visualization:
R2 p
Vavg
8 x
Where,
Vavg is average velocity in m/s
R is radius in m
μ is viscosity in kg/m/s
P is change of pressure in kg/m2
x is change in distance in m
k r
S frac S 1 ln s
ks rw
k0
( Pwf Pi ) h
0
q0
1688 Ct rw2
162.6 BO log (0.868 S )
t
Where,
q is flowrate in bbl/day
Pwf is wellbore flowing pressure in psi
Pi is the initial reservoir pressure in psi
k0 is relative oil permeability and its unitless
μ0 is oil viscosity in cp
h is thickness of formation in ft
Bo is the formation volume factor in rb/stb
ϕ is porosity and it’s unitless
Ct is the total compressibility of the reservoir in psi–1
rw is wellbore radius in ft
t is time in hrs
S is skin and its unitless
qt = qi * exp(–ta)
Where,
qi is initial production rate in stb/day
qt is production rate at time t in stb/day
t is time in days
a is decline factor and its unitless
14
Assessing the Groundwater
Contamination Potential from a Well
in a Hydraulic Fracturing Operation
Nima Jabbari1*, Fred Aminzadeh2 and Felipe P. J. de Barros1
1
University of Southern California, Sonny Astani Department of Civil and
Environmental Engineering, Los Angeles, CA, USA
2
University of Southern California, Mork Family Depart of Chemical Engineering
and Material Science, Los Angeles, CA, USA
Abstract
The introduction of hydraulic fracturing as a new technology for shale gas produc-
tion from low-permeability geologic formations has revitalized natural gas as an
abundant, economic, and cleaner source of energy. However, hydraulic fracturing
has raised several environmental concerns, from air pollution and drinking water
contamination to on-site accidents and risks associated with potential seismic
activities induced by subsurface fluid injection. Some of the potential incidents
and accidents could be mitigated by adding more precautions to the different
stages of an operation. Integrity of the injection well is one of the most important
factors to be considered when dealing with water resources contamination. This
work focuses on groundwater contamination potential in a leakage accident from
the casing during hydraulic fracturing. A numerical flow and transport modeling
approach is adopted to explore and quantify the associated risk of groundwater
contamination under various hypothetical scenarios. The goal is to investigate a
few cases with different geological and operational parameters and, also, to eval-
uate the importance of the well integrity, especially along the interval in which
well casing is inside an aquifer. The results of the work show that if the well integ-
rity is compromised, the groundwater will be contaminated by the chemicals in a
relatively short duration. It is also shown that the result of the stochastic work is
dependent on how well one knows the failure scenario. The appropriate choice of
285
286 Hydraulic Fracturing and Well Stimulation
the probability distribution function for the leakage point is of high importance in
simulations such as these.
14.1 Introduction
Hydraulic Fracturing is an old reservoir stimulation method that has been
acknowledged in petroleum industry since fifty years ago. The technology
aims at increasing the oil and gas production through a sequence of pro-
cesses including drilling vertical or/and horizontal wells, injecting highly
pressurized fluid, cracking the matrix formation, and creating fissures,
which, in turn, increases permeability values of the reservoir [1]. As stated
by Daneshy [2], hydraulic fracturing boosts the production through chang-
ing the flow pattern regime toward the well from radial to linear. At the
same time, horizontal wellbores help in obtaining a larger reservoir-well
contact which, in turn, results in higher production numbers. A combina-
tion of advanced horizontal drilling technologies and hydraulic fracturing
has become very popular lately in producing oil and natural gas from shale
and other tight formations [3, 4]. Despite the popularity and the common
use, this technique still has imperfections that raise questions on the safety
and environmental aspects of the process [5, 6]. In addition to the safety,
environmental problems are also important since the long-term effects of
the hydraulic fracturing activities are not very well understood yet [7]. This
fact creates lots of questions for the environmentalist and the public. The
environmental impacts can be viewed in three categories: air, water, and
induced seismicity.
Air pollution, in general, is linked to several different activities car-
ried out during natural gas development including drilling and fracturing
[8–10].
Subsurface injection operations normally generate low magnitude
(smaller than 2 in Richter scale) seismic events which are termed as
micro-earthquakes or microseismic [11]. In a few cases of hydrofracking
jobs, seismic events have been felt by the nearby residents and make the
operator stop the injection. As an example, in 2011 two seismic events
(2.3 and 1.5 magnitude in Richter scale) were recorded near Blackpool,
Lancashire in the United Kingdom which made the operator halt its frack-
ing operation in the nearby Bowland Shale formation [12]. Generally,
the risk of generation of serious earthquakes as a result of hydraulic frac-
turing is low when compared to the deep-well injection process, which
Groundwater Contamination Potential 287
shows higher probabilities of observing larger seismic events [11, 13, 14].
Historically, the oldest injection-induced seismic events are those of the
Rocky Mountain Arsenal waste injection site in Denver, Colorado [15].
The most recent events were generated in Youngstown, Ohio [16] and cen-
tral Oklahoma [11], both in 2011. The former is also known to be the larg-
est injection-induced event with a 5.6 magnitude [11].
Chemicals used in a typical fracturing job vary and range from benign
(e.g. guar gum), to toxic (e.g. Tetramethylammonium chloride), to very
toxic materials (e.g. Kathon which is a biocide) [17]. Injecting these
chemicals under the ground, if not practiced under safe and sound condi-
tions, can an increase in the risk of having ground or surface water bodies
contaminated. The importance of a solid well system with full integrity is
pronounced here. Handling and management of returned fluid is another
aspect of the fracturing operation which can potentially be a threat to the
health of drinking water resources [18–21].
The well integrity issue is a quite well-known phenomenon in injec-
tion operations such as carbon sequestration and deep waste injection
[22–27]. If the well integrity is not maintained, groundwater can be a tar-
get for the contaminants originating from the initial injectant or found in
the returned fluid [28, 29]. From the human health point of view, ground-
water pollution is critical, more specifically for regions with water short-
age and high demand for groundwater tables [3, 9]. Underground failure
and upward fluid migration has been listed as one of the important risk
pathways to the shallow groundwater [30]. Damages in well integrity
and leakage of fracking fluid into adjacent formation, poor design of the
fracturing job and fracture growth in overburdened formations, reactiva-
tion of faults as permeable pathways for the fluid, and fluid flow through
abandoned wells are along risk pathways discussed by the United States
Environmental Protection Agency (USEPA) [3, 18]. However, among the
discussed risk pathways, the chance of upward fluid migration from deep
formations to the shallow aquifers is reduced as a result of a number of
factors. Absence of a hydraulic connectivity between the aquifer and shale
formation because of geometrical features and physical barriers, makes it
unlikely for reservoir fluid to migrate upward, mostly, as stated by several
authors. The separation between a shallow aquifer and the shale forma-
tion is too large to make the upward fluid migration into the aquifer hap-
pen in a short time-scale [4, 31, 32]. Furthermore, shale layers are usually
capped with tight formations acting as natural barriers [33]. Artificially
created fractures may penetrate into the overburden formation, however,
the distance to which they can propagate upward is limited and the frac-
ture is very unlikely to pass through the overburden, reach the aquifer, and
288 Hydraulic Fracturing and Well Stimulation
connect it to the shale layer [31, 34, 35]. In addition, when the production
starts fluid flows toward the horizontal section of the well, upward migra-
tion would be less probable to take place [32, 36]. Meanwhile, hydraulic
fracturing operations near conductive faults might change the story and,
as declared by Gassiat, Gleeson, Lefebvre, and McKenzie [37], under some
circumstances, reservoir fluids could reach the aquifer through the per-
meable fault, although, due to the slow rate of movement, the mechanism
takes place in a long time (thousands of years).
Well integrity issues described in scenario 1 are further divided into two
categories [28]: Behind the casing movement of the fluid toward the well
head in which fluid finds the space between cementing and casing as a
Groundwater Contamination Potential 289
possible pathway (annular flow) and; leakage from the well in a radial pat-
tern so that the fluid can reach the nearby formation and the chemicals can
undergo different transport mechanisms (leak flow).
King [1] has also named surface casing rupture as one of the poten-
tial leakage scenarios in a fracturing operation. Real-time pressure mon-
itoring helps increase the chance of leakage discovery and, in a way, the
pressure drop followed from a major leakage in pipe enables the opera-
tors to discover the leakage in five minutes [17]. However, when there is
a smaller leakage event happening, the failure remains undetected during
the hydraulic fracturing operation [17]. It is therefore, important to closely
investigate the aftermath of a leakage event, particularly when the rupture
is within the aquifer or at a point close to the aquifer.
On the same subject, a probability bound analysis study has been done
on risk of water pollution in Marcellus shale [30]. The study concluded
that, among other risk pathways such as spills from on-site activities and
wastewater disposal, well casing failure is an important contributor to the
contamination risk where it shows lower than 0.01 m3 and 9 m3 contami-
nation volumes for best-case 50th percentile and worst-case 50th percen-
tile respectively [30]. Furthermore, a study by Browning and Smith [24] on
deep well injection in Louisiana, Nebraska, Michigan, and Pennsylvania
suggests malfunctioning casings and tubings as two reasons for gaining
failure in mechanical integrity tests. According to the same study, out of
10,000 wells that had undergone the integrity test, 50% of them fail to pass
with casing failure as the main contributor to the failure rate (responsible
for 45% to 85% of failure). Among the failed tests due to casing problems,
22% showed failure in the only protective layer so that the injection waste
was able to leak into the near formation. Although the data from this study
might not be directly applicable to fracking wells because of the difference
between waste injection and fracking fluid injection, it still gives ideas on
chance of possible failures of casing in a fracking well [24].
state that the failure scenario of concern in this study falls into the type
of incidents classified as low probability. The small probability is mainly
attributed to the pressure monitoring during the operations and the fact
that the leakage incident normally causes an observable pressure drop [17].
However, the leakage phenomenon is still worth studying, as the risk asso-
ciated could be significant. One should note that risk, here, is defined as
the multiplication of probability of an event by the consequence, and the
aftermath of such a leakage scenario near or into the aquifer can be quite
remarkable. With the low probability and high consequence, the hypothet-
ical case studied here is considered to be a “rare” event. To investigate such
a scenario a 2D numerical simulation is used and the effects of different
parameters on the results are studied. The parameters of interest in this
study are selected to be the concentration magnitude of the chemical and
the arrival time at the well location.
(c )
(c u D c) q (14.1)
t
1
u k( p g z) (14.2)
where dm is the molecular diffusion coefficient and dl and dt are the longi-
tudinal (parallel to the flow) and transverse (perpendicular to the flow in
vertical direction) mechanical dispersion coefficients. The Euclidian norm
of the specific discharge is given by:
1 u12 u1u2
E(u) u)
and E (u ) I E(u (14.5)
|u|2 u2u1 u22
ρu · υ = g2 on Γ (14.6)
where u is the unit normal outward to Γ and ρu·υ gives the projection of
the flux vector on a unit normal vector of the boundary. In equation 14.6,
g2 stands for the prescribed flux on the boundary. The initial condition is
given as:
with p0(x) being the hydrostatic pressure and Ω denoting the entire domain
of interest.
Injection Monitoring
well well
Vadose zone
Aquifer
Impervious layer
Sand layer
Shale reservoir
Figure 14.1 Schematic representation of failure in vertical pipe and contamination plume
in the aquifer.
layer extending from a depth of 840 to 900 meters. In order to investigate the
effect of groundwater flow movement on the contamination, a monitoring
well is placed 100 meters away from the injection well. The numbers here are
adapted from average properties of shallow shale plays reported by the U.S.
Energy Information Administration [39].
In the numerical analysis, the aquifer and the impervious layer are
modeled using a two dimensional ECLIPSE grid with dimensions of 210
m × 20 m. ECLIPSE is a simulation software for subsurface multi-phase
and multi-component flow and transport equations (see equations 14.1
through 14.7) [40]. The stochastic simulations are then conducted in order
to find the probability of exceedance from the threshold in the monitor-
ing well. The chemicals used in the fracturing fluid are modeled using a
passive tracer and the results are reflected in measurements of one quan-
tity of interest which is the concentration level at the monitoring well. The
Monte-Carlo simulation is adapted to quantify the probability of crossing
the specified concentration threshold.
The fracturing operation is assumed to be carried out in 6 stages, each
2.5 hours long with 10 hours of relaxation time, with the injection rate of
Groundwater Contamination Potential 293
11,500 m3/d (50 bbl/ min). It is also presumed that the chemical concen-
tration is equal to two weight percent of the fracturing slurry and the sim-
ulation time is 500 years. The parameters of concern for this study include
porosity and anisotropic permeability of the aquifer, as well as the location
of the leakage point with respect to the aquifer top and the leakage rate.
Each parameter is randomly drawn from different probability distribution
functions. Table 14.1 shows the parameters which are constant over the
course of simulation.
Stochastic simulations are conducted in order to evaluate the sensitivity
of the model to different parameters and also to determine the probability
of passing the contaminant threshold in the aquifer. The threshold is set to
be 1 mg/l. Figure 14.2 shows the workflow used to incorporate randomness
in the model. In here, the Monte-Carlo simulation is used to calculate the
probabilities. The number of Monte-Carlo simulations and the probability
Simulation run
density functions (PDF) used for selecting the parameters are shown in
Table 14.2. Log-normal and truncated normal distributions are chosen in
a way to pick aquifer permeability and porosity values from ranges 150 to
270 mD and 0.10 to 0.30, respectively. Selected PDFs for permeability and
porosity are in line with the petroleum engineering convention. The ranges
chosen for permeability and porosity represent a permeable aquifer with
an average permeability of 220 mD and an average porosity of 0.2. Also, the
vertical permeability is selected as one-tenth of the horizontal permeability
by convention. For the leakage location, two different PDFs are chosen. The
first one is a uniform one extending from one to eighteen meters inside the
aquifer with the assumption of absence of any information on the statistics
of leakage on the wall of a vertical well. Whereas, the second is set to be a
Poisson’s distribution in which the probabilities increase as the depth takes
larger values in a manner where the maximum probability is gained for the
depth equal to the aquifer bottom. The main reason for working with two
different PDFs assigned for the leakage location is to assess the changes in
results when one takes into account the fact that exertion of more stresses
on the well casing in higher depths increases the probability of failure.
0.27 yr 2.74 yr
1 1
CDF
CDF
0.5 0.5
00 1 2 3 4 5 0
1 2 3
C/C 0 x 10–5 C/C0 x 10–4
27.4 yr 100 yr
1 1
Concentration
Threshold
CDF
CDF
0.5 0.5
0 0
0 2 4 6 8 0 1 2 3 4 5
C/C0 x 10–3 C/C0 x 10–5
0.27 yr 2.74 yr
1 1
CDF
CDF
0.5 0.5
0 0
0 1 2 3 4 5 6 8
–5
C/C0 x 10 C/C0 x 10–4
27.4 yr 100 yr
1 1
Concentration
Threshold
CDF
CDF
0.5 0.5
00 2 4 0 5
0
6 8 0 1 2 3 4
C/C x 10–3 C/C0 x 10–5
the concentration values show higher numbers as opposed to the case with
uniform PDF. The probability of exceedance is also larger when using the
Poisson’s PDF. The same holds for the next time-step shown (27.4 years),
with a difference that the concentrations are roughly ten folds larger. In
year 100 and beyond, the concentration values in the monitoring well
show small numbers, mainly because of the mass transport from the open
296 Hydraulic Fracturing and Well Stimulation
1
0.9
0.8
0.7
0.6
F(time)
0.5
0.4
0.3
0.2
0.1
0
2 4 6 8 10 12 14 16
(a) Time (y)
1
0.9
0.8
0.7
0.6
F(time)
0.5
0.4
0.3
0.2
0.1
0
0 5 10 15
(b) Time (y)
Figure 14.5 Arrival time CDF – Uniform (a) and Poisson’s (b) PDFs for the leakage depth.
Groundwater Contamination Potential 297
14.7 Conclusion
Hydraulic fracturing operations are not yet without deficiencies and can
potentially be harmful to the environment if not practiced in a sound and
safe manner. This work focuses on possible groundwater contamination
issues stemmed from hypothetical failure and leakage in the well casing
when the injection takes place. Permeability and porosity of the aquifer,
leakage depth, and leakage rate are the parameters considered in this study.
Higher importance is given to the leakage depth by assigning two different
PDFs for this parameter and comparing the results. The results are reported
as the concentration value of a tracer as well as the arrival-time of a specific
threshold in the monitoring well. It can be inferred from the results that,
under the special scenario and specific geological settings introduced in
this study, the concentration in the monitoring well can exceed the thresh-
old limit. It is also shown that using Poisson’s PDF results in more critical
results as the concentrations reported increase and the average arrival-time
decreases. Keeping the importance of leakage location in mind, one can
298 Hydraulic Fracturing and Well Stimulation
References
1. G.E. King, Hydraulic Fracturing 101: What Every Representative,
Environmentalist, Regulator, Reporter, Investor, University Researcher, Neighbor
and Engineer Should Know about Estimating Frac Risk and Improving Frac
Performance in Unconventional Gas and Oil Wells, SPE Hydraulic Fracturing
Technology Conference, The Woodlands, TX, pp. 1–80 (2012).
2. A. Daneshy, Hydraulic fracturing to improve production. SPE Tech. 101,
14–17 (1990).
3. USEPA (US Environmental Protection Agency), Study of the Potential
Impacts of Hydraulic Fracturing on Drinking Water Resources, (2012).
4. J.D. Arthur, B.K. Bohm, B.J. Coughlin, M.A. Layne, and A L L Consulting,
Evaluating Implications of Hydraulic Fracturing in Shale Gas Reservoirs,
SPE Americas E&P Environmental and Safety Conference, pp. 23–25 (March
2009).
5. J. Heinecke, N. Jabbari, and N. Meshkati, The role of human factors consider-
ations and safety culture in the safety of hydraulic fracturing (fracking). JSEE
2 (2), 130–151 (September 1, 2014). doi:10.7569/JSEE.2014.629509.
6. N. Jabbari, C. Ashayeri, and N. Meshkati, Leading Safety, Health, and
Environmental Indicators in Hydraulic Fracturing, SPE Western Regional
Meeting, Garden Grove, CA (2015).
7. T. Engelder, R.W. Howarth, and A. Ingraffea, Should fracking stop? Nature
477 (7364), 271–275 (2011).
8. L.M. McKenzie, R.Z. Witter, L.S. Newman, and J.L. Adgate, Human health
risk assessment of air emissions from development of unconventional natu-
ral gas resources. Sci. Total Environ. 424, 79–87 (2012).
9. D.M. Kargbo, R.G. Wilhelm, and D.J. Campbell, Natural gas plays in the
Marcellus shale: challenges and potential opportunities. Environ. Sci. Technol.
44 (15), 5679–5684 (August 1, 2010). doi:10.1021/es903811p.
Groundwater Contamination Potential 299
10. C.W. Moore, B. Zielinska, G. Pétron, and R.B. Jackson, Air impacts of
increased natural gas acquisition, processing, and use: a critical review.
Environ. Sci. Technol. 48 (15), 8349–8359 (March 18, 2014). doi:10.1021/
es4053472.
11. W.L. Ellsworth, Injection-induced earthquakes. Science 341 (6142), 1225942
(2013).
12. C.J. De Pater, and S. Baisch, Geomechanical Study of Bowland Shale
Seismicity, Synthesis Report, 57 (2011).
13. R. Walker, F. Aminzadeh, and A. Tiwari, Correlation between Induced
Seismic Events and Hydraulic Fracturing Activities in California, AGU Fall
Meeting Abstracts, Volume 1, p. 05 (December 2014).
14. T. Göbel, A comparison of seismicity rates and fluid-injection operations
in Oklahoma and California: implications for crustal stresses. The Leading
Edge 34 (6), 640–648 (2015).
15. J. Healy, and W. Rubey, The denver earthquakes. Science 161 (3848), 1301–
1310 (1968).
16. ODNR (Ohio Department of Natural resources), Preliminary Report on the
Northstar 1 Class II Injection Well and the Seismic Events in the Youngstown,
Ohio, Area, (2012).
17. B.C. Gordalla, U. Ewers, and F.H. Frimmel, Hydraulic fracturing: a toxico-
logical threat for groundwater and drinking-water? Environ. Earth Sci. 70
(8), 3875–3893 (August 13, 2013). doi:10.1007/s12665-013-2672-9.
18. USEPA (US Environmental Protection Agency), Assessment of the Potential
Impacts of Hydraulic Fracturing for Oil and Gas on Drinking Water
Resources, (2015).
19. R. Hammer, and J. VanBriesen, In fracking’s wake: new rules are needed to
protect our health and environment from contaminated wastewater. Nat.
Resour. Defense Coun. 11 (May 2012).
20. Y.R. Glazer, J.B. Kjellsson, K.T. Sanders, and M.E. Webber, Potential for using
energy from flared gas for on-site hydraulic fracturing wastewater treatment
in Texas. Environ. Sci. Technol. Lett. 1 (7), 300–304 (2014). doi:10.1021/
ez500129a.
21. J.M. VanBriesen, J.M. Wilson, and Y. Wang, Management of Produced Water
in Pennsylvania: 2010-2012, Shale Energy Engineering, 107–113 (2014).
22. J.M. Nordbotten, M.A. Celia, S. Bachu, and H.K. Dahle, Semianalytical solu-
tion for CO2 leakage through an abandoned well. Environ. Sci. Technol. 39
(2), 602–611 (2005). doi:10.1021/es035338i.
23. C.F. Tsang, J. Birkholzer, and J. Rutqvist, A comparative review of hydro-
logic issues involved in geologic storage of CO2 and injection disposal of liq-
uid waste. Environ. Geol. 54 (8), 1723–1737 (August 4, 2007). doi:10.1007/
s00254-007-0949-6.
24. L.A. Browning, and J.D. Smith, Analysis of the Rate of and Reasons for
Injection Well Mechanical Integrity Test Failure, SPE/EPA Exploration and
Production Environmental Conference, San Antonio, TX (1993).
300 Hydraulic Fracturing and Well Stimulation
25. R.J. Davies, S. Almond, R.S. Ward, R.B. Jackson, C. Adams, F. Worrall,
L.G. Herringshaw, J.G. Gluyas, and M.A. Whitehead, Oil and gas wells
and their integrity: implications for shale and unconventional resource
exploitation. Mar. Petrol. Geol. 56, 239–254 (March 2014). doi:10.1016/j.
marpetgeo.2014.03.001.
26. G.E. King, and D.E. King, Environmental risk arising from well-construction
failure-differences between barrier and well failure and estimates of failure
frequency across common well types locations. SPE Prod. Oper. 28 (4), 323–
344 (October 2013).
27. T. Watson, and S. Bachu, Evaluation of the potential for gas and CO2 leakage
along wellbores. SPE Drill. Completion. 24 (1), 115–126 (March 2009).
28. M.D. Holloway, and O. Rudd, Fracking: The Operations and Environmental
Consequences of Hydraulic Fracturing, John Wiley & Sons, Beverly, MA
(2013).
29. N. Jabbari, F. Aminzadeh, F. de Barros, and B. Jafarpour, Hydraulic Fracturing
and Potential Groundwater Contamination Risk. Presented at the 248th
American Chemical Society National Meeting & Exposition, Paper Number:
315, San Francisco, CA, 10–14 (2014).
30. D.J. Rozell, and S.J. Reaven, Water pollution risk associated with natural
gas extraction from the Marcellus shale. Risk Anal. 32 (8), 1382–93 (August
2012). doi:10.1111/j.1539-6924.2011.01757.x.
31. M. Zoback, S. Kitasei, and B. Copithorne, Addressing the Environmental
Risks from Shale Gas Development, Worldwatch Institute, Washington, DC
(2010).
32. R.W. Howarth, A. Ingraffea, and T. Engelder, Natural gas: should fracking
stop? Nature 477 (7364), 271–275 (September 15, 2011). doi:10.1038/477271a.
33. J.D. Arthur, B.K. Bohm, and M.A. Layne, Hydraulic Fracturing Considerations
for Natural Gas Wells of the Marcellus Shale, The Ground Water Protection
Council Annual Forum, Cincinnati, OH (2008).
34. M. Fisher, and N. Warpinski, Hydraulic-fracture-height growth: real data.
SPE Prod. Oper. 27 (1), 8–19 (November 2012).
35. S.A. Flewelling, M.P. Tymchak, and N. Warpinski, Hydraulic fracture height
limits and fault interactions in tight oil and gas formations. Geophys. Res.
Lett. 40 (14), 3602–3606 (July 28, 2013). doi:10.1002/grl.50707.
36. A. Kissinger, R. Helmig, A. Ebigbo, H. Class, T. Lange, M. Sauter, M. Heitfeld,
J. Klünker, and W. Jahnke, Hydraulic fracturing in unconventional gas reser-
voirs: risks in the geological system, part 2. Environ. Earth Sci. 70 (8), 3855–
3873 (June 22, 2013). doi:10.1007/ s12665-013-2578-6.
37. C. Gassiat, T. Gleeson, R. Lefebvre, and J. McKenzie, Hydraulic fracturing
in faulted sedimentary basins: numerical simulation of potential contami-
nation of shallow aquifers over long time scales. Water Resour. Res. 49 (2)
(December 12, 2013). doi:10.1002/2013WR014287.
Groundwater Contamination Potential 301
38. Z. Chen, G. Huan, and Y. Ma, Computational Methods for Multiphase Flows
in Porous Media (Computational Science and Engineering), Society for
Industrial and Applied Mathematic, Philadelphia, PA (2006).
39. USEIA (U.S. Energy Information Administration), Review of Emerging
Resources: U.S. Shale Gas and Shale Oil Plays, US Energy Information
Administration, (2011).
40. Schlumberger, ECLIPSE: Reservoir Simulation Software, Reference Manual,
(2010).
Index
303
304 Index